Metastatic non-small cell lung cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up

Treatment algorithm for stage IV SCC. aMolecular testing is not recommended in SCC, except in those rare circumstances when SCC is found in a never-, long-time ex- or light-smoker (< 15 pack-years). bIn absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy will be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%. Alternatively, if TMB can accurately be evaluated, and conditioned by the registration and accessibility, nivolumab plus ipilimumab should be preferred to platinum-based standard ChT in patients with NSCLC with a high TMB. cNot EMA-approved. ALK, anaplastic lymphoma kinase; BSC, best supportive care; ChT, chemotherapy; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; Mb, megabase; MCBS, ESMO-Magnitude of Clinical Benefit Scale; nab-PC, albumin-bound paclitaxel and carboplatin; NSCLC, non-small cell lung cancer; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PS, performance status; SCC, squamous cell carcinoma; TMB, tumour mutation burden.
Click here to view original web page at academic.oup.com

Treatment algorithm for stage IV SCC. aMolecular testing is not recommended in SCC, except in those rare circumstances when SCC is found in a never-, long-time ex- or light-smoker (< 15 pack-years). bIn absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy will be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%. Alternatively, if TMB can accurately be evaluated, and conditioned by the registration and accessibility, nivolumab plus ipilimumab should be preferred to platinum-based standard ChT in patients with NSCLC with a high TMB. cNot EMA-approved. ALK, anaplastic lymphoma kinase; BSC, best supportive care; ChT, chemotherapy; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; Mb, megabase; MCBS, ESMO-Magnitude of Clinical Benefit Scale; nab-PC, albumin-bound paclitaxel and carboplatin; NSCLC, non-small cell lung cancer; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PS, performance status; SCC, squamous cell carcinoma; TMB, tumour mutation burden.

Primary lung cancer remains the most common malignancy after non-melanocytic skin cancer, and deaths from lung cancer exceed those from any other malignancy worldwide [1]. In 2012, lung cancer was the most frequently diagnosed cancer in males with an estimated 1.2 million incident cases worldwide. Among females, lung cancer was the leading cause of cancer death in more developed countries and the second leading cause of cancer death in less developed countries [1]. The highest incidence is found in Central/Eastern Europe and Asia with age-standardised incidence rates of 53.5 and 50.4 per 100 000, respectively. European projections for 2017 indicate a 10.7% drop in 5 years with an incidence of 33.3/100 000 in males and a rise of 5.1% and an incidence of 14.6/100 000 in females [2]. Contrary to the United States, the death rate in females is increasing in Europe [3]. The number of lung cancer-related deaths in Europe for 2017 is estimated to represent the leading cause of cancer deaths in both genders, accounting for 24% in males and 15% in females, respectively [2].

Non-small cell lung cancer (NSCLC) accounts for 80%–90% of lung cancers, while small cell lung cancer (SCLC) has been decreasing in frequency in many countries over the past two decades [4]. During the last 25 years, the distribution of histological types of NSCLC has changed: in the United States, squamous cell carcinoma (SCC), formerly the predominant histotype, decreased, while adenocarcinoma has increased in both genders. In Europe, similar trends have occurred in men, while in women, both SCC and adenocarcinoma are still increasing [5].

The World Health Organization (WHO) estimates that lung cancer is the cause of 1.59 million deaths globally per year, with 71% of them caused by smoking. Tobacco smoking remains the main cause of lung cancer and the geographical and temporal patterns of the disease largely reflect tobacco consumption during the previous decades. Both smoking prevention and smoking cessation can lead to a reduction in a large fraction of lung cancers [6]. In countries with active tobacco control measures, the incidence of lung cancer has begun to decline in men and is reaching a plateau for women [1, 7–9]. Several other factors have been described as lung cancer risk factors, including exposure to asbestos, arsenic, radon and non-tobacco-related polycyclic aromatic hydrocarbons. Hypotheses about indoor air pollution (e.g. coal-fuelled stoves and cooking fumes) are made for the relatively high burden of non-smoking-related lung cancer in women in some countries [10]. There is evidence that lung cancer rates are higher in cities than in rural settings but many confounding factors other than outdoor air pollution may be responsible for this pattern.

About 500 000 deaths annually are attributed to lung cancer in lifetime never-smokers [1]. Absence of any history of tobacco smoking characterises 19% of female compared with 9% of male lung carcinoma in the United States [11, 12]. An increase in the proportion of NSCLC in never-smokers has been observed, especially in Asian countries [13]. These new epidemiological data have resulted in ‘non-smoking-associated lung cancer’ being considered a distinct disease entity, where specific molecular and genetic tumour characteristics have been identified [14].

Use of non-cigarette tobacco products such as cigars and pipes has been increasing. A pooled analysis highlighted the increased risk, particularly for lung and head and neck cancers, in smokers (former and current) of cigars and pipes [15].

Familial risk of lung cancer has been reported in several registry-based studies after careful adjustment for smoking [16]. A recent study estimated the heritability of lung cancer at 18% but many of the genetic components remain unidentified. Genome-wide association studies (GWAS) have identified lung cancer susceptibility loci including CHRNA3, CHRNA5, TERT, BRCA2, CHECK2 and the human leukocyte antigen (HLA) region [17–19]. Another trial, including data from 29 266 cases and 56 450 controls from European descent, found 18 susceptibility loci reaching genome-wide significance, among which 10 were previously unknown. Interestingly, while four of the latter were associated with overall lung cancer risk, six were associated with lung adenocarcinoma only [20].

Changes in the therapeutic scenario in the last 15 years have emphasised the need for a multidisciplinary approach in lung cancer. Data show that high-volume centres and multidisciplinary teams are more efficient at managing patients with lung cancer than low-volume or non-multidisciplinary centres, by providing more complete staging, better adherence to guidelines and increased survival [21, 22]. Multidisciplinary tumour boards influence providers’ initial plans in 26%–40% of cases [23]. The absolute need to reach a proper and precise morphological and biological definition often requires challenging tissue sampling, with most treatment decisions depending on the information obtained from the specimen collected at diagnosis.

Bronchoscopy is a technique ideally suited to large, central lesions and offers the advantage of minimal morbidity. Bronchoscopy can be used for bronchial washing, brushing, bronchial and transbronchial biopsy, with a diagnostic yield of 65%–88% [24–26]. By combining direct bronchoscopic airway visualisation with ultrasound-guided biopsy of the lesion, endobronchial ultrasound (EBUS) provides a diagnostic yield of 75%–85% in large, centrally located lesions [27, 28]. Fibre optic bronchoscopy allows for the evaluation of regional lymph nodes by EBUS and/or endoscopic ultrasound (EUS). EBUS-guided transbronchial needle aspiration (TBNA) is less invasive and at least as accurate as mediastinoscopy [29]. Several studies have shown that cytological specimens obtained by EBUS-TBNA are suitable for molecular testing for epidermal growth factor receptor (EGFR), Kirsten rat sarcoma viral oncogene homologue (KRAS) and anaplastic lymphoma kinase (ALK) status [30–33]; however, collection of samples suitable for broader molecular diagnostic testing should be encouraged.

In case of peripheral lesions, transthoracic percutaneous fine needle aspiration and/or core biopsy, under imaging guidance [typically computed tomography (CT)] is proposed [34]. Needle biopsy is associated with a diagnostic accuracy of > 88% yield, a sensitivity of 90% and a false-negative rate of 22% [25, 35–38]. The most significant disadvantage of transthoracic needle biopsy is a procedural risk of pneumothorax, ranging from 17% to 50% [37, 38].

In the presence of a pleural effusion, thoracentesis could represent both a diagnostic tool and a palliative treatment. If fluid cytology examination is negative, image-guided pleural biopsy or surgical thoracoscopy should be carried out. More invasive, surgical approaches [mediastinoscopy, mediastinotomy, thoracoscopy, video-assisted thorascopic surgery (VATS), secondary lesion resection etc.] in the diagnostic workup are considered when the previously described techniques cannot allow for an accurate diagnosis.

Histological diagnosis of NSCLC is crucial to many treatment decisions and should be as exact and detailed as the samples and available technology allow. Diagnosis should be based upon the criteria laid out in the WHO classification [39]. This classification details the complete diagnostic approach for surgically resected tumours but, importantly, also provides guidance for assessing and reporting small biopsy and cytology samples where complete morphological criteria for specific diagnosis may not be met [39–41].

Most patients with NSCLC present with advanced stage unresectable disease, therefore all treatment-determining diagnoses must be made on small biopsy and/or cytology-type samples. Sampling may be carried out of the primary tumour or any accessible metastases, taken under direct vision or more usually with image-guided assistance, which greatly increases the diagnostic yield (hit rate). Sampling metastatic disease may facilitate staging, as well as diagnosis. These diagnostic samples frequently have limited tumour material and must therefore be handled accordingly; ensuring processing is suitable for all likely diagnostic procedures and that material is used sparingly at each step, since many diagnostic tests may be required [42].

Immunohistochemistry (IHC) has become a key technique in primary diagnosis as well as in predictive biomarker assessment. In those cases of NSCLC where specific subtyping is not possible by morphology alone, a limited panel of IHC is recommended to determine the subtype [39, 40]. Thyroid transcription factor 1 (TTF1) positivity is associated with probable diagnosis of adenocarcinoma, p40 positivity with probable diagnosis of SCC; if neither are positive the diagnosis remains NSCLC-not otherwise specified (NOS). IHC staining should be used to reduce the NSCLC-NOS rate to < 10% of cases diagnosed [IV, A]. Pathologists are urged to conserve tissue at every stage of diagnosis, to use only two tissue sections for IHC NSCLC subtyping and to avoid excessive IHC investigation, which may not be clinically relevant.

After morphological diagnosis, the next consideration is therapy-predictive biomarker testing. This practice will be driven by the availability of treatments and will vary widely between different geopolitical health systems [43–45]. Contemporary practice has now evolved into two testing streams, one for the detection of targetable, usually addictive, oncogenic alterations and the other for immuno-oncology therapy biomarker testing. A personalised medicine synopsis table is shown in Table 1.

Biomarker Method Use LoE, GoR
EGFR mutation Any appropriate, validated method, subject to external quality assurance To select those patients with EGFR-sensitising mutations most likely to respond to EGFR TKI therapy I, A
ALK rearrangement Any appropriate, validated method, subject to external quality assurance. FISH is the historical standard but IHC is now becoming the primary therapy-determining test, provided the method is validated against FISH or some other orthogonal test approach. NGS is an emerging technology To select those patients with ALK gene rearrangements most likely to respond to ALK TKI therapy I, A
ROS1 rearrangement FISH is the trial-validated standard. IHC may be used to select patients for confirmatory FISH testing but currently lacks specificity. NGS is an emerging technology. External quality assurance is essential To select those patients with ROS1 gene rearrangements most likely to respond to ROS1 TKI therapy II, A
BRAF mutation Any appropriate, validated method, subject to external quality assurance To select those patients with BRAF V600-sensitising mutations most likely to respond to BRAF inhibitor, with or without MEK inhibitor therapy II, A
PD-L1 expression IHC to identify PD-L1 expression at the appropriate level and on the appropriate cell population(s) as determined by the intended drug and line of therapy. Only specific trial assays are validated. Internal and external quality assurance are essential To enrich for those patients more likely to benefit from anti-PD-1 or anti-PD-L1 therapy. For pembrolizumab, testing is a companion diagnostic for nivolumab and atezolizumab, testing is complementary I, A

Several molecular drivers for oncogene addiction represent strong predictive biomarkers and excellent therapeutic targets. They are generally mutually exclusive of each other [43–45]. These tumours are much more common in never- (never smoked or who smoked < 100 cigarettes in lifetime), long-time ex- (> 10 years) or light-smokers (< 15 pack-years) but they can also be found in patients who smoke. The vast majority of oncogene-addicted lung cancers are adenocarcinomas. Patients, in general, tend to be younger, while female gender and East Asian ethnicity particularly enriches for EGFR-mutant tumours. Nonetheless, guidelines suggest that all patients with advanced, possible, probable or definite, adenocarcinoma should be tested for oncogenic drivers [43–46]. Molecular testing is not recommended in SCC, except in those rare circumstances when SCC is found in a never-, long-time ex- or light-smoker (< 15 pack-years) [IV, A]. Testing for EGFR mutations and rearrangements involving the ALK and ROS1 genes are now considered mandatory in most European countries. BRAF V600E mutations are rapidly approaching this status as first-line BRAF/MEK inhibitors are more widely approved, while HER2 (human epidermal growth factor receptor 2) and MET exon 14 mutations and fusion genes involving RET and NTRK1 (neurotropic tropomyosin receptor kinase 1) are evolving targets/biomarkers [43–46].

EGFR tyrosine kinase inhibitors (TKIs) are established effective therapies in patients who have activating and sensitising mutations in exons 18–21 of EGFR [47]. Prevalence is around 10%–20% of a Caucasian population with adenocarcinoma but much higher in Asian population. Around 90% of the most common mutations comprise deletions in exon 19 and the L858R substitution mutation in exon 21. Any testing approach must cover these mutations [I, A]; however, complete coverage to include exons 18–21 is recommended [III, B]. The T790M exon 20 substitution mutation is only rarely found in EGFR TKI-naive disease using standard techniques but is the most frequent cause of resistance to first- and second-generation EGFR TKIs (50%–60% of cases). Cases of patients carrying germline T790M mutation have also been reported [48]. Further studies to better understand the prevalence, familial penetrance and lifetime lung cancer risk in germline T790M-mutant patients are warranted. Implications of this mutation in TKI-naive disease are unclear, but the availability of TKIs effective against T790M-mutant recurrent disease makes T790M testing on disease relapse mandatory [I, A]. Cell-free DNA (cfDNA) blood testing is an acceptable approach to detect T790M at relapse but lacks sensitivity, so all patients with a negative blood test still require tissue biopsy [II, A] [49]. Tissue biopsy may also be more effective in identifying other resistance mechanisms which may require alternative treatment (SCLC transformation, MET amplification, HER2 alterations etc.).

Fusion genes involving ALK and a number of partners (most commonly EML4) account for around 2%–5% of the same population that is routinely tested for EGFR mutations [50]. ALK-driven adenocarcinoma is very sensitive to several ALK TKIs. Early trials validated break-apart fluorescent in situ hybridisation (FISH) as the test to identify ALK gene rearrangement but the close association between a positive FISH test and modestly elevated ALK protein in tumour cells allows ALK IHC to be used, either to select cases for confirmatory FISH testing or as the primary therapy-determining test [50, 51]. ALK IHC must reliably detect low levels of ALK protein and be validated against alternative tests to detect ALK fusion genes, especially if ALK IHC is used as the therapy-determining assay, without confirmation by FISH [II, A]. Emerging data demonstrate that the presence of the ALK protein (positive IHC staining) is associated with treatment response [I, A] [52, 53]. Recently, IHC has been accepted as an equivalent alternative to FISH for ALK testing [54]. Testing for ALK rearrangement should be systematically carried out in advanced non-squamous NSCLC [I, A]. ALK mutations are emerging as important resistance mechanisms to ALK TKIs and ALK mutation testing may soon become a routine test at relapse as newer-generation ALK TKIs show differential efficacy against different ALK mutations [55].

ROS1 fusion genes are yet another addictive oncogenic driver that occurs in ∼1%–4% of the same testing population. Like ALK, ROS1 has several potential fusion gene partners. Crizotinib, a TKI effective against ALK and MET, is also approved by the European Medicines Agency (EMA) for use in ROS1-rearranged adenocarcinomas. FISH has been the standard approach to detecting ROS1 rearrangements. IHC may be used in a manner similar to ALK testing, to identify candidate tumours for confirmatory FISH testing. The sensitivity of this approach is high, using currently available IHC, but specificity of IHC is low [IV, C]. FISH or other testing is required to confirm the diagnosis; IHC is currently not recommended as the primary treatment determining test [IV, A] [45, 46, 50]. Testing for ROS1 rearrangement should be systematically carried out in advanced non-squamous NSCLC [III, A].

BRAF mutation testing is now required in many countries after the approval of BRAF and MEK inhibitors for BRAF V600-mutant NSCLC. Any method is valid provided that it is adequately sensitive for the samples used and has been appropriately quality-assured, both within the laboratory and through external quality assurance. The V600E mutation is the most common of the BRAF V600 family and, overall, these BRAF mutations are found in ∼2% of cases. BRAF V600 mutations appear mutually exclusive to EGFR and KRAS mutations, ALK and ROS1 rearrangements and are similarly much more common in adenocarcinoma. BRAF V600 mutation status should be systematically analysed in advanced non-squamous NSCLC for the prescription of BRAF/MEK inhibitors [II, A].

For many laboratories, testing for EGFR and BRAF mutations and ALK and ROS1 rearrangements involves individual stand-alone tests. Multiplex, massively parallel, so-called next-generation sequencing (NGS) of various sorts is rapidly being adopted as the standard approach to screening adenocarcinomas for oncogenic targets [III, A] [45, 49, 50, 56]. Platform-specific, commercially available panels can cover genes of interest and provide a comprehensive, multiplex test for mutations and, in some cases, fusion genes. NGS will not address biomarkers that require testing at the protein level (requires IHC) and the question of whether NGS-detected fusion genes require an orthogonal test (IHC, FISH) for confirmation remains open. Whatever testing modality is used, it is mandatory that adequate internal validation and quality control measures are in place and that laboratories participate in, and perform adequately, external quality assurance schemes for each biomarker test [III, A].

The approval of the anti-programmed cell death protein 1 (PD-1) agent pembrolizumab as a standard-of-care first-line treatment in selected patients has made programmed death-ligand 1 (PD-L1) IHC a mandatory test in all patients with advanced NSCLC. Although the PD-L1 IHC 22C3 assay was the only test validated in clinical trials of pembrolizumab, extensive technical comparison studies suggest that trial-validated commercial kit assays based on the 28-8 and SP263 PD-L1 IHC clones may be alternative tests [III, A] [57–61]. If laboratories use, by choice or force of circumstances, a non-trial-validated PD-L1 IHC test, i.e. a laboratory developed test (LDT), there is a high risk that the assay may fail quality assurance and a very careful, extensive validation is essential before clinical use [IV, A] [35, 36]. There is a relationship between the extent of PD-L1 expression on tumour cells, or in some trials in tumour infiltrating immune cells, and the probability of clinical benefit from numerous anti-PD-1 or PD-L1 agents, in first- and second-line therapy [57]. For pembrolizumab, the mandatory treatment threshold is a tumour proportion score (TPS, presence of PD-L1 signal on tumour cell membranes) ≥ 50% in first line and ≥ 1% in second line [62, 63]. PD-L1 expression testing is recommended for all patients with newly diagnosed advanced NSCLC [I, A]. For nivolumab and atezolizumab in second line, PD-L1 testing is not required for drug prescription. PD-L1 IHC is an approved biomarker test for immunotherapeutics in NSCLC but it is not a perfect biomarker; less than half of biomarker-selected patients benefit from treatment and some responses may be encountered in ‘biomarker-negative’ cohorts. Much work is underway to identify alternative, or more likely, additional biomarkers to enrich patient populations for response. Various measures of tumour mutational burden (TMB) are being explored and TMB has been validated prospectively in a unique prospective clinical trial to date [64]. An international effort is ongoing to define a consensus on how TMB should be measured [65–67]. Assessment of tumour inflammation is also of interest, but again, various approaches are being pursued, including histological assessment of immune cell infiltrates and mRNA-based expression signatures of immune-related genes. More data are required before any of these new approaches can be routinely incorporated into NSCLC biomarker testing.

The ability to detect oncogenic driver genomic alterations, or factors associated with disease resistance to treatment in peripheral blood, opens the way to disease monitoring in a way that would not be practically feasible were repeat testing solely based upon tumour biopsy testing. In practice, and with current knowledge, this is more likely to involve the use of cfDNA rather than circulating tumour cells (CTCs); the vast majority of existing data concern EGFR mutation testing in blood [68]. Currently, much EGFR plasma testing is based upon highly sensitive allele-specific polymerase chain reaction (ASPCR). Plasma genotyping may be considered before undergoing a tumour biopsy to detect the T790M mutation. However, if the plasma testing is negative for T790M, the tissue biopsy is strongly recommended to determine T790M status because of the risks of false-negative plasma results [III, A]. NGS techniques can be used; as more biomarkers are identified and validated, more NGS-based gene panels would be available.

Notwithstanding the issues regarding sensitivity of blood testing, potentially clinically valuable information may be derived from serial blood testing during treatment. For example, the disappearance from the blood of the primary sensitising EGFR mutation is associated with clinical and radiological evidence of response to EGFR TKIs and is a good prognostic indicator [IV, C].

After maximum response to EGFR TKI therapy and disappearance of the mutation from the plasma, the reappearance of the primary sensitising mutation, with or without detection of the T790M resistance mutation, may be an indicator of ‘biochemical’ disease relapse. This occurrence may predate radiological relapse, which, in turn, may predate clinical/symptomatic disease relapse. Currently, such findings are essentially exploratory since there is no consensus as to when and how any clinical intervention should be managed. There is no doubt, however, that this kind of molecular monitoring could, in the future, offer benefit to patients in a number of different personalised treatment scenarios.

TMB was evaluated in patient tissue as well as blood samples in different trials. Unique assays and cut-offs are not yet defined but preliminary data from the POPLAR and OAK trials found TMB in blood is associated with improved atezolizumab clinical benefit in patients with NSCLC [69]. Preliminary data suggesting blood TMB as a predictive biomarker for atezolizumab activity have recently been presented [70]. A prospective trial in the first-line setting is exploring the same biomarker [NCT03178552].

A complete medical history with comorbidities, weight loss, performance status (PS) and physical examination must be recorded. An exhaustive smoking habit assessment has to be included, indicating type, quantity and timing.

Standard tests including routine haematology, renal and hepatic function and bone biochemistry tests are required. The routine use of serum markers, such as carcinoembryonic antigen (CEA), is not recommended [IV, B] [71].

The neutrophil to lymphocyte ratio (NLR) is a widely available blood-based data point, validated in numerous oncological settings as a potential prognostic marker. NLR has been considered as a potential dynamic marker but further prospective validations are needed [IV, C] [72, 73].

A contrast-enhanced CT scan of the chest and upper abdomen including complete assessment of liver, kidneys and adrenal glands should be carried out. Imaging of the central nervous system (CNS) is most relevant in those patients with neurological symptoms or signs [IV, A]; however, if available, imaging of the CNS with magnetic resonance imaging (MRI, preferably with gadolinium enhancement) or CT of the brain with iodinated contrast should be carried out at diagnosis [IV, B]. MRI is more sensitive than CT scan [III, B] [74].

Leptomeningeal disease (LMD) is a deadly complication of solid tumours and has a poor prognosis. Adenocarcinomas are the most common tumours to metastasise to the leptomeninges. In case of clinical suspicion, LMD diagnostic imaging should include the brain and the spinal cord, as LMD can impact the entire neuraxis. If metastatic disease has been determined by CT scan of the chest and upper abdomen or by brain imaging, other imaging is only necessary if it has an impact on treatment strategy. If bone metastases are clinically suspected, bone imaging is required [IV, B]. Bone scan or positron emission tomography (PET), ideally coupled with CT, can be used for detection of bone metastasis [IV, B]. PET-CT is the most sensitive modality in detecting bone metastasis [II, B] [75]. Fluorodeoxyglucose (FDG)-PET or PET-CT has higher sensitivity and specificity than bone scintigraphy [76]. FDG-PET-CT scan also has high sensitivity for the evaluation of solitary pulmonary nodules, intra-thoracic pathological lymph nodes and distant metastatic disease [77]. However, the low sensitivity of this exam in small lesions, in lesions close to FDG-avid structures (overprojection) or in lesions that move extensively, such as those just above the diaphragm, should be considered. MRI may complement or improve the diagnostic staging accuracy of FDG-PET-CT imaging, particularly in assessing local chest wall, vascular or vertebra invasion and is also effective for identification of nodal and distant metastatic disease. NSCLC is staged according to the American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) system (8th edition) and is grouped into the stage categories shown in Tables 2 and 3 [78, 79].

Primary tumour (T)
TX Primary tumour cannot be assessed, or tumour proven by the presence of malignant cells in sputum or bronchial washings but not visualised by imaging or bronchoscopy
T0 No evidence of primary tumour
Tis Carcinoma in situa
T1 Tumour 3 cm or less in greatest dimension, surrounded by lung or visceral pleura, without bronchoscopic evidence of invasion more proximal than the lobar bronchus (i.e. not in the main bronchus)b
T1mi Minimally invasive adenocarcinomac
T1a Tumour 1 cm or less in greatest dimensionb
T1b Tumour more than 1 cm but not more than 2 cm in greatest dimensionb
T1c Tumour more than 2 cm but not more than 3 cm in greatest dimensionb
T2 Tumour more than 3 cm but not more than 5 cm; or tumour with any of the following featuresd
  • -Involves main bronchus regardless of distance to the carina, but without involvement of the carina

  • -Invades visceral pleura

  • -Associated with atelectasis or obstructive pneumonitis that extends to the hilar region either involving part of or the entire lung

T2a Tumour more than 3 cm but not more than 4 cm in greatest dimension
T2b Tumour more than 4 cm but not more than 5 cm in greatest dimension
T3 Tumour more than 5 cm but not more than 7 cm in greatest dimension or one that directly invades any of the following: parietal pleura, chest wall (including superior sulcus tumours) phrenic nerve, parietal pericardium; or separate tumour nodule(s) in the same lobe as the primary
T4 Tumour more than 7 cm or of any size that invades any of the following: diaphragm, mediastinum, heart, great vessels, trachea, recurrent laryngeal nerve, oesophagus, vertebral body, carina; separate tumour nodule(s) in a different ipsilateral lobe to that of the primary
Regional lymph nodes (N)
NX Regional lymph nodes cannot be assessed
N0 No regional lymph node metastases
N1 Metastasis in ipsilateral peribronchial and/or ipsilateral hilar lymph nodes and intrapulmonary nodes, including involvement by direct extension
N2 Metastasis in ipsilateral mediastinal and/or subcarinal lymph node(s)
N3 Metastasis in contralateral mediastinal, contralateral hilar, ipsilateral or contralateral scalene, or supraclavicular lymph node(s)
Distant metastasis (M)
M0 No distant metastasis
M1 Distant metastasis
M1a Separate tumour nodule(s) in a contralateral lobe; tumour with pleural or pericardial nodules or malignant pleural or pericardial effusione
M1b Single extrathoracic metastasis in a single organf
M1c Multiple extrathoracic metastasis in a single or multiple organs
Occult carcinoma TX N0 M0
Stage 0 Tis N0 M0
Stage IA T1 N0 M0
Stage IA1 T1mi N0 M0
T1a N0 M0
Stage IA2 T1b N0 M0
Stage IA3 T1c N0 M0
Stage IB T2a N0 M0
Stage IIA T2b N0 M0
Stage IIB T1a-c T2a,b N1 M0
T3 N0 M0
Stage IIIA T1a-c T2a,b N2 M0
T3 N1 M0
T4 N0, N1 M0
Stage IIIB T1a-c T2a,b N3 M0
T3, T4 N2 M0
Stage IIIC T3, T4 N3 M0
Stage IV Any T Any N M1
Stage IVA Any T Any N M1a, M1b
Stage IVB Any T Any N M1c

In the presence of a solitary metastatic lesion on imaging studies, including pleural and pericardial effusion, efforts should be made to obtain a cytological or histological confirmation of stage IV disease [IV, A].

Response evaluation is recommended after two to three cycles of chemotherapy (ChT) or immunotherapy, using the same initial radiographic investigation that demonstrated tumour lesions [IV, B]. The same procedure and timing (every 6–9 weeks) should be applied for the response evaluation in patients treated with targeted therapies and/or immunotherapy [IV, B]. Follow-up with PET is not routinely recommended, due to its high sensitivity and relatively low specificity [IV, C].

Measurement of lesions should follow Response Evaluation Criteria in Solid Tumours (RECIST) v1.1 [IV, A] [80]. The adequacy of RECIST in evaluating response to EGFR or ALK TKI in respective genetically driven NSCLC is still debatable even if this remains the standard method of evaluation for these patients [IV, B]. In these two subgroups of patients (and in other actionable oncogene alterations), treatment beyond RECIST progression is a common approach, pursuing clinical benefit more than morphological response. This approach differs from what was carried out historically with cytotoxic agents. The conventional radiological response criteria are unable to describe pseudoprogression (PsPD) and can result in underestimation of the therapeutic benefit of immune checkpoint blockade. Several radiological criteria have been developed specifically for immunotherapy, to better define the tumour response in this context. Two-dimensional immune-related response criteria (irRC) were proposed in 2009 and modified in 2013 with the immune-related RECIST (irRECIST) [81, 82]. More recently, the RECIST working group published a proposition of new criteria called immune-RECIST (iRECIST), to standardise response assessment among immunotherapy clinical trials [83]. A subsequent adaption of RECIST designed to better capture cancer immunotherapy responses has been published: immune-modified RECIST (imRECIST) [84]. More data are needed to compare the RECIST, iRECIST, imRECIST and irRECIST to quantify the differences in outcome estimation before using of them in clinical practice. Non-conventional responses and PsPD are very rarely observed in NSCLC, ranging generally under 5% of all cases, and RECIST v1.1 should still be used in routine practice [IV, B] [85–88].

Management of advanced/metastatic NSCLC

The treatment strategy (Figures 1–7) should take into account factors such as histology, molecular pathology, age, PS, comorbidities and the patient’s preferences. Treatment decisions should ideally be discussed within a multidisciplinary tumour board who can evaluate and change management plans, including recommending additional investigations and changes in treatment modality [89]. Systemic therapy should be offered to all stage IV patients with PS 0–2 [I, A].

Treatment algorithm for stage IV SCC. aMolecular testing is not recommended in SCC, except in those rare circumstances when SCC is found in a never-, long-time ex- or light-smoker (< 15 pack-years). bIn absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy will be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%. Alternatively, if TMB can accurately be evaluated, and conditioned by the registration and accessibility, nivolumab plus ipilimumab should be preferred to platinum-based standard ChT in patients with NSCLC with a high TMB. cNot EMA-approved. ALK, anaplastic lymphoma kinase; BSC, best supportive care; ChT, chemotherapy; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; Mb, megabase; MCBS, ESMO-Magnitude of Clinical Benefit Scale; nab-PC, albumin-bound paclitaxel and carboplatin; NSCLC, non-small cell lung cancer; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PS, performance status; SCC, squamous cell carcinoma; TMB, tumour mutation burden.

Treatment algorithm for stage IV SCC. aMolecular testing is not recommended in SCC, except in those rare circumstances when SCC is found in a never-, long-time ex- or light-smoker (< 15 pack-years). bIn absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy will be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%. Alternatively, if TMB can accurately be evaluated, and conditioned by the registration and accessibility, nivolumab plus ipilimumab should be preferred to platinum-based standard ChT in patients with NSCLC with a high TMB. cNot EMA-approved. ALK, anaplastic lymphoma kinase; BSC, best supportive care; ChT, chemotherapy; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; Mb, megabase; MCBS, ESMO-Magnitude of Clinical Benefit Scale; nab-PC, albumin-bound paclitaxel and carboplatin; NSCLC, non-small cell lung cancer; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PS, performance status; SCC, squamous cell carcinoma; TMB, tumour mutation burden.

Treatment algorithm for stage IV NSCC, molecular tests negative (ALK/BRAF/EGFR/ROS1). aIn absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy will be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%. Alternatively, if TMB can accurately be evaluated, and conditioned by the registration and accessibility, nivolumab plus ipilimumab should be preferred to platinum-based standard ChT in patients with NSCLC with a high TMB. bNot EMA-approved. ALK, anaplastic lymphoma kinase; BSC, best supportive care; ChT, chemotherapy; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; IO, immuno-oncology; Mb, megabase; MCBS, ESMO-Magnitude of Clinical Benefit Scale; nab-PC, albumin-bound paclitaxel and carboplatin; NSCC, non-squamous cell carcinoma; NSCLC, non-small cell lung cancer; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PS, performance status; TMB, tumour mutation burden.

Treatment algorithm for stage IV NSCC, molecular tests negative (ALK/BRAF/EGFR/ROS1). aIn absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy will be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%. Alternatively, if TMB can accurately be evaluated, and conditioned by the registration and accessibility, nivolumab plus ipilimumab should be preferred to platinum-based standard ChT in patients with NSCLC with a high TMB. bNot EMA-approved. ALK, anaplastic lymphoma kinase; BSC, best supportive care; ChT, chemotherapy; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; IO, immuno-oncology; Mb, megabase; MCBS, ESMO-Magnitude of Clinical Benefit Scale; nab-PC, albumin-bound paclitaxel and carboplatin; NSCC, non-squamous cell carcinoma; NSCLC, non-small cell lung cancer; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PS, performance status; TMB, tumour mutation burden.

Treatment algorithm for stage IV NSCC, molecular tests positive (ALK/BRAF/EGFR/ROS1). aMCBS score for the combination of bevacizumab with gefitinib or erlotinib. bNot EMA-approved. ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; MCBS, ESMO-Magnitude of Clinical Benefit Scale; NSCC, non-squamous cell carcinoma.
View large Download slide

Treatment algorithm for stage IV NSCC, molecular tests positive (ALK/BRAF/EGFR/ROS1). aMCBS score for the combination of bevacizumab with gefitinib or erlotinib. bNot EMA-approved. ALK, anaplastic lymphoma kinase; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; MCBS, ESMO-Magnitude of Clinical Benefit Scale; NSCC, non-squamous cell carcinoma.

Treatment algorithm for stage IV lung carcinoma with EGFR-activating mutation. aMCBS score for the combination of bevacizumab with gefitinib or erlotinib. bNot EMA-approved. cfDNA, cell-free DNA; ChT, chemotherapy; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; MCBS, ESMO-Magnitude of Clinical Benefit Scale; PS, performance status; RT, radiotherapy.

Treatment algorithm for stage IV lung carcinoma with EGFR-activating mutation. aMCBS score for the combination of bevacizumab with gefitinib or erlotinib. bNot EMA-approved. cfDNA, cell-free DNA; ChT, chemotherapy; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; MCBS, ESMO-Magnitude of Clinical Benefit Scale; PS, performance status; RT, radiotherapy.

Treatment algorithm for stage IV lung carcinoma with ALK translocation. aNot EMA-approved. ALK, anaplastic lymphoma kinase; ChT, chemotherapy; EMA, European Medicines Agency; MCBS, ESMO-Magnitude of Clinical Benefit Scale; RT, radiotherapy; TKI, tyrosine kinase inhibitor.
Treatment algorithm for stage IV lung carcinoma with ROS1 translocation. ChT, chemotherapy; MCBS, ESMO-Magnitude of Clinical Benefit Scale; TKI, tyrosine kinase inhibitor.
Treatment algorithm for stage IV lung carcinoma with BRAF V600 mutation. ChT, chemotherapy; MCBS, ESMO-Magnitude of Clinical Benefit Scale; MEK, mitogen-activated protein kinase.

In any stage of NSCLC, smoking cessation should be highly encouraged: it can improve outcome and smoking may interact with systemic therapy [II, A]. For example, smoking reduces erlotinib bioavailability [90, 91]. Given the established relationship between smoking and lung cancer, patients who have smoked may feel stigmatised or guilty after diagnosis and more pessimistic about their illness and likely outcomes, all of which may have adverse implications for health-related quality of life (QoL) [92].

For these reasons, healthcare professionals should give clear advice about the adverse implications of continued smoking and include smoking cessation programmes in the therapeutic algorithm.

Lung cancers were previously considered poorly immunogenic, with minimal benefit seen in historical studies of cytokine modulation or vaccines. However, the recent development of immune checkpoint inhibitors has upended this belief and provided proof of principle that immunotherapy can play an important role in the treatment of patients with lung cancers.

The phase III KEYNOTE-024 study has established the role for pembrolizumab as first-line treatment in patients with untreated, advanced NSCLC and tumour characterised by PD-L1 expression ≥ 50% [62], in absence of EGFR mutation or ALK translocations. In KEYNOTE-024, 1934 patients were screened to identify 500 patients (30%) with tumour PD-L1 expression ≥ 50%. Of these patients, 305 patients were randomised to receive 200 mg pembrolizumab every 3 weeks (up to 2 years) or four to six cycles of standard platinum-doublet ChT. All efficacy measures favoured pembrolizumab, including objective response rate (ORR 45% versus 28%), progression-free survival [PFS, hazard ratio (HR) 0.5, 95% confidence interval (CI) 0.37–0.68, P < 0.001] and overall survival (OS, HR 0.6, 95% CI 0.41–0.89, P = 0.005). Safety and QoL also favoured pembrolizumab [93]. Continued follow-up has further emphasised the effectiveness of pembrolizumab, with median OS (mOS) doubled in those who received pembrolizumab compared with ChT (30 versus 14 months) [94].

Pembrolizumab is considered a standard first-line option for patients with advanced NSCLC and PD-L1 expression ≥ 50% who do not otherwise have contraindications to use of immunotherapy (such as severe autoimmune disease or organ transplantation) [I, A; European Society for Medical Oncology-Magnitude of Clinical Benefit Scale (ESMO-MCBS) v1.1 score: 5].

Other studies, KEYNOTE-042 and CheckMate 026, examined a lower threshold for PD-L1 [66, 95]. Recent results from KEYNOTE-042, a phase III study of patients with PD-L1 ≥ 1% who were randomised to either pembrolizumab or ChT, demonstrated improved OS in patients treated with pembrolizumab at three thresholds of PD-L1: ≥ 50%, ≥ 20% and ≥ 1%. The HR for OS was 0.69, 0.77 and 0.81, respectively. Overall, the preponderance of the OS benefit was driven by patients with ≥ 50%, while no significant increase was seen in those patients with 1%–49% PD-L1 expression (HR 0.92, 95% CI 0.77–1.11).

In CheckMate 026, patients with untreated, advanced NSCLC and PD-L1 ≥ 1% (analysis based on 5% threshold) were randomised to nivolumab or platinum-doublet ChT [66]. There were no improvements in any efficacy metrics. However, an exploratory retrospective and unplanned analysis examined the impact of TMB on benefit of nivolumab. A total of 312 patients (58% of randomised patients) had sufficient tissue for whole exome sequencing. In those patients with the highest tertile of TMB (> 243 missense non-synonymous somatic mutations per sample), ORR (47% versus 28% with ChT) and PFS (HR 0.62, 95% CI 0.38–1.0) favoured those who received nivolumab. Meanwhile, among patients with low or medium TMB, ORR was numerically better in those who received ChT (33% versus 23% with nivolumab).

Overall, these results confirm the benefit of pembrolizumab in the first-line setting seen in KEYNOTE-024, restricted to patients with high PD-L1 expression (≥ 50%).

Recently, results of the phase III trials KEYNOTE-189, IMpower150 and IMpower132 have brought new options for the therapeutic choices in first line of non-squamous NSCLC and trials KEYNOTE-407 and IMpower131 for patients with squamous NSCLC.

In KEYNOTE-189, patients with metastatic non-squamous NSCLC, PS 0–1, without sensitising EGFR or ALK mutations, were randomised to receive pemetrexed and a platinum-based ChT plus either 200 mg of pembrolizumab or placebo every 3 weeks for 4 cycles, followed by pembrolizumab or placebo for up to a total of 35 cycles plus pemetrexed maintenance therapy [96]. The mOS was not reached in the pembrolizumab/ChT arm versus 11.3 months (95% CI 8.7–15.1) in the ChT arm (HR 0.49; 95% CI 0.38–0.64; P < 0.001). The PFS also favoured the pembrolizumab/ChT combination (HR 0.52; 95% CI 0.43–0.64; P < 0.001). The OS benefit of pembrolizumab/ChT was observed in all PD-L1 tumour subgroups. Notably, among the PD-L1 TPS < 1%, there was not a clear PFS benefit with pembrolizumab/ChT (HR 0.75, 95% CI 0.53–1.05), such that the degree of durable benefit in this group remains limited. Still, based on the results from KEYNOTE-189, pembrolizumab in combination with pemetrexed and a platinum-based ChT should be considered a standard option in metastatic non-squamous NSCLC [I, A; ESMO-MCBS v1.1 score: 4].

In IMpower150, the addition of atezolizumab to bevacizumab plus ChT significantly improved PFS and OS among patients with metastatic non-squamous NSCLC, regardless of PD-L1 expression [97]. The PFS was longer in the atezolizumab/bevacizumab/ChT arm compared with bevacizumab/ChT [median PFS (mPFS) 8.3 versus 6.8 months; HR 0.59; 95% CI 0.50–0.70; P < 0.001]. Survival was longer in the atezolizumab/bevacizumab/ChT arm compared with bevacizumab/ChT (mOS 19.2 versus 14.7 months; HR 0.78; 95% CI 0.64–0.96; P = 0.02; 12-month OS 67% versus 61%). Results from IMpower150 place the combination of atezolizumab and bevacizumab with carboplatin and paclitaxel as a therapeutic option in patients with PS 0–1 with metastatic non-squamous NSCLC, in absence of contraindications to use of immunotherapy [I, A]. Of note, this is the only trial to date also including patients with EGFR or ALK genetic alterations and demonstrating a stringent OS benefit (PFS HR 0.59, 95% CI 0.37–0.94; OS HR 0.54, 95% CI 0.29–1.03). This association in EGFR- or ALK-positive NSCLC patients defines a treatment opportunity for this subgroup after targeted therapies have been exploited [I, A in unselected non-squamous NSCLC including EGFR- and ALK- driven NSCLC, specifically [III, A] for EGFR and [III, B] for ALK subgroups; not EMA-approved].

Recently, the combination of carboplatin or cisplatin with pemetrexed and atezolizumab has been shown, in the context of the IMpower132 trial, to be superior to the ChT doublet. An improvement in mPFS from 5.2 to 7.6 months was observed (HR 0.6; 95% CI 0.49−0.72; P < 0.0001) while OS was not statistically significantly increased at the time of analysis with mOS of 18.1 versus 13.6 months (HR 0.81; 95% CI 0.64−1.03; P = 0.0797), suggesting another potential treatment opportunity [I, B, not EMA approved] [98].

KEYNOTE-407 is a randomised, placebo-controlled study of patients with metastatic squamous NSCLC [99]. Patients were randomised 1:1 to receive carboplatin and paclitaxel every 3 weeks, or albumin-bound paclitaxel (nab-P) weekly plus pembrolizumab or placebo for 4 cycles, followed by pembrolizumab or placebo for a total of 35 treatments. The combination of ChT plus pembrolizumab was associated with improved ORR (58.4% versus 35.0%, P = 0.0004) and improved OS (HR 0.64, mOS 15.9 versus 11.3 months, P = 0.0008). The benefit in OS was seen across PD-L1 expression strata (TPS < 1% HR 0.61, TPS 1%–49% HR 0.57, TPS ≥ 50% HR 0.64). No new safety concerns were observed. Results from KEYNOTE-407 place the combination of pembrolizumab plus carboplatin and paclitaxel or nab-P as a standard choice in patients with metastatic squamous NSCLC [I, A; not EMA-approved].

Atezolizumab was studied in patients with metastatic squamous NSCLC in the IMpower131 study. Patients were randomised to atezolizumab/carboplatin/paclitaxel, atezolizumab/carboplatin/nab-P or carboplatin/nab-P (nab-PC) [100]. Atezolizumab/carboplatin/nab-P had improved PFS compared with nab-PC (HR 0.715, P = 0.0001), but no improvement in OS was seen at the first interim analysis (mOS 14 versus 13.9 months). More mature data are needed to evaluate long-term benefit of the strategy; with the use of atezolizumab with nab-PC today representing an option in patients with metastatic squamous NSCLC [I, B; not EMA-approved].

One key area of uncertainty is among PD-L1 TPS ≥ 50%, as none of these trials provide a direct comparison between ChT plus checkpoint inhibitors versus pembrolizumab monotherapy. However, cross-trial comparison between trials suggest similar OS outcomes among PD-L1 ≥ 50%, with very different toxicity profiles, suggesting that pembrolizumab monotherapy may remain a reasonable choice for patients with PD-L1 ≥ 50% [101].

TMB has shown encouraging results as a predictive biomarker in retrospective studies in NSCLC and SCLC. The first pre-specified analysis of TMB as a biomarker was reported in the phase III trial CheckMate 227, evaluating nivolumab plus ipilimumab versus ChT in first-line NSCLC [64]. The TMB cut-off of 10 mutations per megabase (Mb) was determined based on data from CheckMate 568 based on receiver operating characteristic (ROC) curves and clinical impact analysis [102]. Patients with newly diagnosed advanced NSCLC were randomised based on PD-L1 expression. Those who had PD-L1 TPS ≥ 1% received nivolumab/ipilimumab, nivolumab monotherapy or ChT; and those with a PD-L1 TPS < 1% received nivolumab/ipilimumab, nivolumab/ChT or ChT. In patients with high TMB (≥ 10 mutations/Mb, 44% of assessable patients), nivolumab/ipilimumab was associated with longer PFS than ChT (HR 0.58; 97.5% CI: 0.41–0.81; P < 0.001), and more than tripling of 1-year PFS (42.6% versus 13.2%). The PFS benefit with nivolumab/ipilimumab was seen irrespective of PD-L1, wherein the HR similarly favoured nivolumab plus ipilimumab in patients with a PD-L1 TPS ≥ 1% and those < 1% (HR 0.62 and HR 0.48, respectively). A similar benefit was seen in both squamous and non-squamous histologies (squamous HR 0.63, non-squamous HR 0.55). Of importance, there was no difference in PFS among patients with < 10 mutations/Mb (HR 1.07; 95% CI 0.84–1.35).

Grade 3–4 treatment-related adverse events (AEs) leading to discontinuation were more common with ChT than nivolumab plus ipilimumab (36% versus 31%), with more subsequent discontinuations with immunotherapy (12% versus 5%).

CheckMate 227 continues for the coprimary endpoint of OS in PD-L1 selected patients. For now, nivolumab plus ipilimumab represents an optional treatment regimen for patients with NSCLC with a high TMB [I, A; not EMA-approved]. Important questions remain regarding the role of immunotherapy combinations versus PD-1 monotherapy in PD-L1 TPS ≥ 50% and how TMB may inform the optimal use of PD-(L)1 plus ChT versus immunotherapy alone combinations in NSCLC. Additional clinical data and evaluation of long-term benefit of these new strategies are needed. Physicians and patients will need to conduct individualised discussions regarding benefit and risks of available therapies over time.

Overall, the results from the KEYNOTE-024, IMpower150, KEYNOTE-189, IMpower132, CheckMate 227, KEYNOTE-407 and IMpower131 trials suggest that introducing immunotherapy will be a standard new approach for most patients with newly diagnosed NSCLC.

ChT with platinum doublets should be considered in all stage IV NSCLC patients without an actionable oncogenic driver, without major comorbidities and PS 0–2 [I, A]. Benefits of ChT versus best supportive care (BSC), namely a 23% reduction of risk of death, a 1-year survival gain of 9% and a 1.5-month absolute increase in median survival and improved QoL, were observed irrespective of age, sex, histology and PS in two meta-analyses [103–105]. The survival benefit of two-agent over one-agent ChT regimens was reported in a meta-analysis in 2004; no survival benefit was observed for three-agent over two-agent regimens [106]. Based on a 2006 meta-analysis, revealing a statistically significant reduction (equal to 22%) in the risk of death at 1 year for platinum over non-platinum combinations, without induction of unacceptable increase in toxicity, platinum-based doublets are recommended in all patients with no contraindications to platinum compounds [I, A] [107]. Neither a large individual trial nor a meta-analysis found an OS benefit of six versus fewer cycles of first-line platinum-based doublets, although a longer PFS coupled with significantly higher toxicity was reported in patients receiving six cycles [108, 109]. Therefore, four cycles of platinum-based doublets followed by less toxic maintenance monotherapy [I, A], or four cycles in patients not suitable for maintenance monotherapy [I, A], up to a maximum of six [IV, B], are currently recommended.

Several platinum-based regimens with third-generation cytotoxics (paclitaxel, gemcitabine, docetaxel, vinorelbine) have shown comparable efficacy [110, 111]. The expected toxicity profile should contribute to the selection of the ChT regimen, taking into account that:

  • A recent Cochrane review including 10 trials with 3973 patients available for meta-analysis could not demonstrate any difference between carboplatin-based and cisplatin-based ChT in OS. Cisplatin had higher ORRs in an overall analysis but trials using paclitaxel or gemcitabine plus a platinum agent in both arms had equivalent response. Cisplatin caused more nausea or vomiting and carboplatin caused more thrombocytopaenia and neurotoxicity, while no difference in the incidence of grade 3-4 anaemia, neutropaenia, alopaecia or renal toxicity was observed [112].

  • The nab-PC regimen has been shown in a large phase III trial to have a significantly higher ORR compared with solvent-based paclitaxel/carboplatin (sb-PC), and less neurotoxicity [I, B] [113]. The benefits were observed in both SCC and non-SCC (NSCC), with a larger impact on response in SCC. For this reason, the nab-PC regimen could be considered a chemotherapeutic option in advanced NSCLC patients, particularly in patients with greater risk of neurotoxicity, pre-existing hypersensitivity to paclitaxel or contraindications for standard paclitaxel premedication [I, B].

Most individual trials and meta-analyses evaluating ChT options in the first-line treatment of advanced NSCLC did not report any differential efficacy in patients with SCC [104]. Therefore, platinum-based doublets with the addition of a third-generation cytotoxic agent (gemcitabine, vinorelbine, taxanes) are recommended in advanced SCC patients without major comorbidities and PS 0–2 [I, A] (Figure 1).

Necitumumab, an immunoglobulin G1 (IgG1) monoclonal antibody against EGFR, did not demonstrate a significant impact in first-line treatment of metastatic NSCC when added to cisplatin/pemetrexed [114]. However, outcomes were different when necitumumab was combined with different ChT regimens in SCC. In the SQUIRE trial, the addition of necitumumab to cisplatin/gemcitabine produced a significant OS improvement (11.5 versus 9.9 months, HR 0.84, 95% CI 0.74–0.96; P = 0.01) and PFS improvement, with a 1-year survival equal to 48% in the experimental arm versus 43% in the control arm [115]. In a retrospective analysis, the group of patients expressing EGFR (assessed by IHC) showed an improvement in OS and PFS [mOS 11.7 months versus 10.0 months; HR 0.79, 95% CI 0.69–0.92; P = 0.002; mPFS 5.7 versus 5.5 months, HR 0.84, 95% CI 0.72–0.92; P = 0.018] [116]. Based on these results, due to the limited clinical improvement, the addition of necitumumab to cisplatin and gemcitabine has not been adopted as a standard in Europe for advanced SCC and its use should be carefully evaluated [I, C; ESMO-MCBS v1.1 score: 1].

Any platinum-based doublets with a third-generation agent including gemcitabine, vinorelbine or taxanes can be used in NSCC (Figure 2). The incorporation of pemetrexed and bevacizumab into individual treatment schedules should be considered, based on the following:

  • Pemetrexed-based combination ChT represents a therapeutic option, based on the results of a recent meta-analysis that showed a slight but significant survival benefit compared with gemcitabine- or docetaxel-based combinations and of a pre-planned subgroup analysis of a large randomised phase III trial [II, A] [117, 118]. Pemetrexed use should be restricted to NSCC in any line of treatment in advanced disease [II, A] [119, 120].

  • The survival benefit of carboplatin in combination with pemetrexed has been investigated in a meta-analysis (exploratory subgroup analysis); survival benefit for pemetrexed plus platinum held true for cisplatin-containing regimens but not for carboplatin-based regimens; however, results from prospective randomised studies investigating this question are not yet available [117]. The combination of carboplatin with pemetrexed can be an option in patients with a contraindication to cisplatin [II, B].

  • Findings of two randomised clinical trials revealed that bevacizumab improves OS when combined with paclitaxel/carboplatin regimens in patients with NSCC and PS 0–1 and, therefore, may be offered in the absence of contraindications in eligible patients with advanced NSCC (bevacizumab should be given until progression) [I, A] [121, 122]. A randomised phase III trial evaluating gemcitabine/cisplatin combination with or without bevacizumab demonstrated an ORR and modest PFS advantage, but no OS benefit [123].

Two meta-analyses showed a consistent significant improvement in ORR, PFS and OS for the combination of bevacizumab and platinum-based ChT, compared with platinum-based ChT alone in eligible patients with NSCC [124, 125]. Bevacizumab might therefore be considered with platinum-based regimens beyond paclitaxel/carboplatin in the absence of contraindications [II, B]. Treatment with bevacizumab has also shown encouraging efficacy and acceptable safety in patients with NSCC and asymptomatic, untreated brain metastases [126].

Decision-making about maintenance therapy must take into account histology, residual toxicity after first-line ChT, response to platinum doublet, PS and patient preference. Several trials have investigated the role of maintenance treatment in patients with good PS (0–1) either as ‘continuation maintenance’ or as ‘switch maintenance’. ‘Continuation maintenance’ and ‘switch maintenance’ therapies refer to the maintained use of an agent included in first-line treatment or the introduction of a new agent after four cycles of platinum-based ChT, respectively. One randomised phase III switch maintenance trial has reported improvements in PFS and OS with pemetrexed [120] and erlotinib [127] versus placebo, following four cycles of platinum-based ChT. In the case of pemetrexed, this benefit was seen only in patients with NSCC [I, B]. Furthermore, the phase III IUNO study (maintenance erlotinib) failed to meet its primary endpoint of OS (HR 1.02; 95% CI 0.85–1.22; P = 0.85) [128]. Maintenance treatment with erlotinib is only recommended for NSCC patients with an EGFR-sensitising mutation [III, B]. Randomised trials investigating continuation maintenance have shown an improvement in PFS and OS. A large phase III randomised trial of continuation maintenance with pemetrexed versus placebo after four induction cycles of cisplatin plus pemetrexed ChT demonstrated a PFS and OS improvement in patients with a PS 0–1, confirmed at long-term follow-up [129, 130]. mOS was 13.9 months (95% CI 12.8–16.0) with pemetrexed and 11.0 months (95% CI 10.0–12.5) with placebo, with 1- and 2-year survival rates significantly longer for patients given pemetrexed (58% and 32%, respectively) than for those given placebo (45% and 21%). Another phase III study comparing maintenance bevacizumab, with or without pemetrexed, after first-line induction with bevacizumab, cisplatin and pemetrexed showed a benefit in PFS for the pemetrexed/bevacizumab combination but no improvement in OS [131], although a trend towards improved OS was seen when analysing 58% of events of 253 patients randomised for this study [132]. In the PointBreak trial, which compared carboplatin/paclitaxel/bevacizumab followed by bevacizumab with carboplatin/pemetrexed/bevacizumab followed by pemetrexed/bevacizumab, OS was comparable in both arms (12.6 versus 13.4 months; HR 1.00; 95% CI 0.86–1.16; P = 0.949) [133]. In a phase III trial, it was also shown that continuation maintenance with gemcitabine significantly reduces disease progression (mPFS, 3.8 versus 1.9 months; HR 0.56; 95% CI 0.44–0.72) with a non-significant OS improvement in patients with advanced NSCLC treated with four cycles of cisplatin/gemcitabine as first-line ChT [I, C] [134]. Continuing pemetrexed following completion of four cycles of first-line cisplatin/pemetrexed ChT is, therefore, recommended in patients with NSCC, in the absence of progression after first-line ChT and upon recovery from toxicities from the previous treatment [I, A]. Of note, three studies, one employing bevacizumab and the other two using monoclonal antibodies against EGFR (cetuximab or necitumumab) administered concomitantly to ChT and further continued as monotherapy until disease progression, have demonstrated survival benefits; however, the specific role of the maintenance phase cannot be appreciated in this context [115, 121, 135].

ChT prolongs survival and improves QoL in NSCLC patients with PS 2 when compared with BSC [I, A] [136, 137].

A recently published meta-analysis of randomised trials comparing the efficacy and safety of platinum-based doublets versus single-agent regimens in the first-line therapy of PS 2 patients revealed platinum-based regimens to be superior in terms of ORR and survival despite an increase in toxicities (mainly haematological) [138]. The superiority of carboplatin-based combinations over monotherapy in PS 2 patients has been identified within two large phase III trials [137, 139], with an acceptable toxicity profile. Therefore, platinum-based (preferably carboplatin) doublets should be considered in eligible PS 2 patients [I, A]. Single-agent ChT with gemcitabine, vinorelbine, docetaxel [I, B] or pemetrexed (restricted to NSCC) [II, B] is an alternative treatment option [139, 140].

All phase III studies with immunotherapies reported until today excluded patients with PS ≥ 2. Reported in abstract form only, CheckMate 153 included 108 patients with advanced NSCLC and PS 2 treated with single-agent nivolumab [141]. mOS was 3.9 months and 1-year survival 23%, being lower than observed in patients with PS 0–1. Toxicities associated with treatment were comparable between PS 0–1 and PS 2 patients. Interestingly, an improvement in patient-reported outcomes was observed for non-squamous NSCLC patients in the context of this trial. In a European-based safety phase II trial (CheckMate 171), among 809 patients enrolled, 98 PS 2 patients were treated with nivolumab; the safety was comparable to the overall population with an mOS of 5.4 months [142]. In conclusion, insufficient data are available to date on the use of checkpoint inhibitors for these patients, but this treatment option can be considered [III, B].

Poor PS (3–4) patients should be offered BSC in the absence of documented sensitising alterations such as EGFR mutations, ALK or ROS1 rearrangements or BRAF V600 mutation [III, B].

In the early 2000s, based on several phase III trials, single-agent ChT over BSC was established as the standard of care for first-line therapy of advanced NSCLC patients aged > 70 years [140, 143]. A recent systematic review identified platinum-based combination ChT as the preferred option for patients > 70 years of age with PS 0–2 and adequate organ function [144]. Here, data from 13 randomised controlled trials (RCTs) with 1705 patients > 70 years of age showed that the addition of platinum agents resulted in improvement in OS (HR 0.76; 95% CI 0.69–0.85), PFS (HR 0.76; 95% CI 0.61–0.93) and ORR (RR 1.57; 95% CI 1.32–1.85) compared with non-platinum containing therapy. Carboplatin was associated with an OS benefit (HR 0.67; 95% CI 0.59–0.78) whereas cisplatin was not (HR 0.91; 95% CI 0.77–1.08). Treatment with platinum-based combinations comes at the expense of more treatment-related morbidity, mainly anaemia, thrombocytopaenia, emesis, diarrhoea and peripheral neuropathy; this should be weighed against the expected survival benefit. It is noteworthy that those RCTs that included formal QoL analysis found no difference in QoL between treatment with platinum-based combinations or single agents in this population [137, 145]. Nevertheless, concerns about treatment-related toxicity in the elderly population has led to the study of comprehensive geriatric assessment (CGA) as a selection tool for treatment with either platinum-based regimens, single-agent therapy or BSC based on patient’s fitness or frailty. The sole prospective randomised trial reported failed to demonstrate an improvement in time to treatment failure and OS for advanced NSCLC patients > 70 years when treatment (carboplatin doublet, single-agent ChT or BSC) was allocated based on CGA alone or a combination of PS and age. Also, the incidence of grade 3–4 toxicities was not different between the two arms in this study [146]. Carboplatin-based doublet ChT is recommended in eligible elderly patients with PS 0–2 and with adequate organ function [I, A]. For those patients not eligible for doublet ChT, single-agent ChT remains the standard of care [I, B].

Evidence is accumulating for immune checkpoint inhibitors in elderly patients with advanced NSCLC. Although no studies dedicated to elderly patients were reported yet, it can be inferred that ORRs and survival are not different between patients ≤ 65 years and those > 65, based on subgroup analysis of the randomised second-line trials [63, 147–150]. Of note, no differences in toxicities were observed [149]. In KEYNOTE-024, comparing first-line pembrolizumab with combination ChT in advanced NSCLC patients whose tumours expressed PD-L1 > 50%, half the randomised patients were > 65 years of age. In the subgroup analysis, the beneficial effect of pembrolizumab was not different between patients aged ≤ 65 years and > 65 years of age (HR 0.61 versus 0.45) [62]. Likewise, in CheckMate 026, comparing nivolumab with combination ChT in unselected first-line advanced NSCLC patients, there was no difference in survival outcomes between patients treated with nivolumab aged ≤ 65 years and those > 65 years [66]. Immunotherapy should therefore be considered according to standard recommendations in elderly patients [III, A].

In the few years since benefit was shown with PD-1 blockade in lung cancers, three PD-1 or PD-L1 therapies have been approved by the United States Food and Drug Administration (FDA) and the EMA in the second-line setting.

The three approved therapies in the immunotherapy-naive, second-line setting include nivolumab, pembrolizumab and atezolizumab. Each has been approved on the basis of phase III studies demonstrating improved OS in comparison with docetaxel. Results are summarised below. Overall, there are no major differences in terms of efficacy or safety among these three therapies to inform a single optimal choice, and no comparative studies have been conducted. There are two key distinctions between the three approved therapies, which can affect choice and use:

  1. PD-L1 expression: nivolumab and atezolizumab are approved in patients with previously treated, advanced NSCLC irrespective of PD-L1 expression, while pembrolizumab is approved only in patients with PD-L1 ≥ 1%.

  2. Schedule of administration: atezolizumab and pembrolizumab are approved to be given once every three weeks, while nivolumab is given once every two weeks based on current EMA approval. Of note, the FDA has recently approved a 4-weekly schedule for nivolumab.

Overall, any of these three therapies represents reasonable standard therapy for most patients with advanced, previously treated, PD-L1-naive NSCLC. Treatment of patients with a history of autoimmune disease should be considered only with caution and after discussion of risks/benefits. Because of the risk of graft rejection, anti-PD-1/PD-L1 agents should be avoided in patients with solid organ transplantation. For reference, we summarise the key data from the relevant phase III studies here:

  • Nivolumab: two phase III studies, CheckMate 017 and CheckMate 057, have established the effectiveness of nivolumab in the second-line setting [147, 148]. In CheckMate 017, 272 patients with squamous NSCLC were randomised to nivolumab or docetaxel. OS was significantly improved in those who received nivolumab (HR 0.59, 95% CI 0.44–0.79, P < 0.001). In CheckMate 057, 582 patients with non-squamous NSCLC were randomised to nivolumab or docetaxel. OS was significantly improved with nivolumab (HR 0.73, 95% CI 0.59–0.89, P = 0.002). In a recent update of these studies, 2-year OS favoured nivolumab in both squamous (29% versus 16% with docetaxel) [I, A; ESMO-MCBS v1.1 score: 5] and non-squamous NSCLC (23% versus 8%) [I, A; ESMO-MCBS v1.1 score: 5]. Tolerability also favoured nivolumab, with 10% of patients experiencing grade 3–4 treatment-related AEs compared with 55% with docetaxel.

  • Pembrolizumab: The KEYNOTE-010 trial randomised 1034 patients with previously treated NSCLC with PD-L1 expression on at least 1% of tumour cells to receive pembrolizumab (tested at two doses, 2 mg/kg or 10 mg/kg, each given every three weeks) or docetaxel 75 mg/m2 every 3 weeks [63, 151]. OS was significantly longer for pembrolizumab versus docetaxel (2 mg/kg, HR 0.71, 95% CI 0.58−0.88; P < 0.001; 10 mg/kg, HR 0.61, 95% CI 0.49−0.75; P < 0.001), with a recently reported 2-year OS rate of 14.5% versus 30.1% (2 mg/kg group) [I, A; ESMO-MCBS v1.1 score: 5]. Grade 3–5 treatment-related AEs were less common with pembrolizumab than with docetaxel (13%–16% versus 35%). There was no significant difference in the efficacy or safety of pembrolizumab at 2 or 10 mg/kg.

  • Atezolizumab: The OAK trial [149] evaluated 850 patients with advanced NSCLC previously treated with one or two prior lines of ChT, who were randomised to atezolizumab or docetaxel. OS was significantly improved with atezolizumab (HR 0.73, 95% CI 0.62–0.87, P<0.001). Tolerability was also better with atezolizumab, with 15% of patients experiencing a grade 3–4 treatment-related toxicity compared with 43% of those treated with docetaxel [I, A; ESMO-MCBS v1.1 score: 5].

There is a general trend across each of the phase III studies in second-line (nivolumab, pembrolizumab and atezolizumab versus docetaxel) for enriched efficacy of anti-PD-1/PD-L1 agents in patients with higher PD-L1 expression compared with those with no/less PD-L1 expression. However, unselected patients may still have improved survival and tolerability with anti-PD-1/PD-L1 agents compared with docetaxel [I, A].

Therefore, anti-PD-1/PD-L1 agents are the treatment of choice for most patients with advanced, previously treated, PD-L1-naive NSCLC, irrespective of PD-L1 expression [I, A].

Combination ChT regimens failed to show any OS benefit over single-agent treatments in second line. Single agents improve disease-related symptoms and OS. Docetaxel has shown improved efficacy compared with BSC in randomised trials with a significant improvement in OS in the TAX 320 trial for those patients who received docetaxel at a dose of 75 mg/m2 every 3 weeks [152, 153]. Similar efficacy, but more favourable tolerability for the weekly schedule, could be confirmed in randomised trials comparing 3-weekly to weekly schedules of docetaxel [I, B] [154, 155].

Pemetrexed demonstrated comparable OS to docetaxel in a randomised phase III trial but had a more favourable toxicity profile, with lower rates of neutropaenia, alopaecia and gastrointestinal events [156]. A retrospective analysis confirmed a predictive impact of histology with an improved efficacy of pemetrexed compared with docetaxel in patients with non-squamous NSCLC (mOS 9.0 versus 8.3 months; HR 0.78; 95% CI 0.61–1.0, P = 0.004) [119].

While registration trials of pemetrexed and docetaxel did not limit therapy to a set number of treatment cycles, second-line treatment duration should be individualised. Treatment may be prolonged if disease is controlled and toxicity acceptable [II, B].

Docetaxel and pemetrexed (for NSCC only) are confirmed treatment options in second-line ChT, with comparable efficacy [I, B], taking into account that immunotherapy is now the current standard second-line systemic therapy and that these agents have not been formally assessed after checkpoint inhibitors.

In several trials, the combination of antiangiogenic agents with ChT has been investigated in patients with pretreated advanced NSCLC. In the REVEL trial, ramucirumab, a vascular endothelial growth factor receptor 2 (VEGFR2) antibody, in combination with docetaxel, showed a superior OS (mOS 10.5 versus 9.1 months, HR 0.86; 95% CI 0.75–0.98, P = 0.032) and PFS (mPFS 4.5 versus 3 months, P < 0.0001) compared with docetaxel and placebo regardless of histology [ESMO-MCBS v1.1 score: 1] [157]. The main AEs associated with ramucirumab consisted of myelotoxicity, oedema and mucositis. The efficacy of this combination was also preserved in the poor prognosis group of patients who did not show any response to first-line ChT [157, 158]. Nintedanib, an oral angiokinase inhibitor, improved PFS in combination with docetaxel compared with ChT alone in the LUME-1 trial (mPFS 3.4 versus 2.7 months, HR 0.79; 95% CI 0.68–0.92; P = 0.0019) [159]. A significant prolongation of OS was observed in the group of patients with adenocarcinoma histology (mOS 12.6 versus 10.3 months; HR 0.82, 95% CI 0.7–0.99, P = 0.0359). Gastrointestinal events and transient elevation of liver enzymes were the most frequent AEs associated with nintedanib. However, the QoL analyses did not show any impact on QoL measurements for this combination. Again, improved efficacy was seen in the poor prognostic group of patients with non-responding or fast progressing tumours [159, 160]. The efficacy of the combination of antiangiogenic agents and ChT was confirmed in the ULTIMATE trial, which showed prolongation of PFS for the combination of weekly paclitaxel and bi-weekly bevacizumab compared with docetaxel (mPFS 5.4 versus 3.9 months, HR 0.62; 95% CI 0.44–0.86; P = 0.005) with no difference in OS [161]. The combination of ramucirumab and docetaxel represents a treatment option for patients with NSCLC progressing after previous ChT or immunotherapy, with PS 0–2 [I, B]. The combination of nintedanib and docetaxel represents a treatment option for patients with adenocarcinoma progressing after previous ChT or immunotherapy [II, B]. Combination of paclitaxel and bevacizumab is another treatment option [I, C; not EMA-approved].

Erlotinib represents a potential second-/third-line treatment option, in particular for patients not suitable for immunotherapy or second-line ChT in unknown EGFR status or EGFR wild-type (WT) tumours [II, C]. Erlotinib has shown superiority in OS compared with BSC in pretreated patients not eligible for further ChT (mOS 6.7 versus 4.7 months, HR 0.7; 95% CI 0.58–0.85, P < 0.001) [162]. In two additional trials, comparable efficacy of erlotinib and ChT has been reported for patients with refractory NSCLC (progression during first-line platinum-based ChT) or in second-/third-line therapy [163, 164].

In the recent years, a growing number of reports revealed an inferior efficacy of EGFR TKIs in pretreated patients with EGFR WT tumours compared with ChT [165]. In a meta-analysis summarising the results of 6 randomised trials with 900 patients, PFS for EGFR TKI was significantly inferior to ChT in the group of patients with EGFR WT tumours (HR 1.37, 95% CI 1.20–1.56, P < 0.00001). However, these results did not translate into an OS difference (HR 1.02, 95% CI 0.87–12, P = 0.81) [166]. An additional analysis of the Biomarkers France study reported a significant improvement in PFS or OS for second-line ChT compared with second-line EGFR TKI in 1278 patients with pretreated NSCLC (PFS 4.3 versus 2.83 months, HR 0.66, 95% CI 0.57–0.77, OS 8.39 versus 4.99 months, HR 0.7, 95% CI 0.59–0.83, P < 0.0001) [167].

In patients with advanced SCC, afatinib was investigated versus erlotinib in the LUX-Lung 8 trial. PFS and OS were improved in favour of afatinib (PFS 2.4 versus 1.9 months, HR 0.82, 95% CI 0.68–1.00, P = 0.041; OS 7.9 versus 6.8 months, HR 0.81, 95% CI 0.69–0.95, P = 0.0077) [168]. Afatinib was associated with improved prespecified disease-related symptoms and health-related QoL [169].

Afatinib could be a therapeutic option in patients with advanced SCC with PS 0–2 unfit for ChT or immunotherapy, progressing on or after ChT with unknown EGFR status or EGFR WT [I, C; ESMO-MCBS v1.1 score: 2].

In conclusion, patients clinically or radiologically progressing after first-line therapy with PS 0–2 should be offered second-line therapy, irrespective of administration of maintenance treatment [I, A]. So far, no prospective trials have determined the best second-line therapy following failure of first-line treatment with pembrolizumab; however, according to the first-line trial results, the preferred recommendation would be a platinum-based ChT, as discussed above [V, B] [62].

EGFR mutation is the best established oncogenic target for management of advanced stage NSCLC [170, 171]. The predictive power of EGFR mutation is confirmed in multiple randomised phase III studies comparing first- (erlotinib or gefitinib) or second-generation (afatinib) EGFR TKIs with standard platinum-based ChT [I, A] [172–177]. The benefit of improvement in ORR and PFS is consistent across all age groups, genders, smoking status and PS. Notably, none of the above studies have shown any benefit in OS for an EGFR TKI over platinum-based ChT, likely due to the high level of crossover. EGFR TKIs represent the standard of care as first-line treatment for advanced EGFR-mutated NSCLC [I, A] (Figures 3 and 4). Patients with PS 3–4 may also be offered an EGFR TKI as they are likely to receive a similar clinical benefit as patients with good PS [III, A] [178]. Patients who have benefited from EGFR TKI treatment may continue to receive the same therapy beyond initial radiological progression as long as they are clinically stable [II, A] [179]. Patients with localised distant progression and ongoing systemic control, continuation of treatment with EGFR TKI in combination with local treatment of progressing metastatic sites may be considered [III, B]. Continuous use of EGFR TKI in combination with ChT is not recommended as it was not associated with PFS improvement [I, A] and showed a detrimental effect on OS [II, B] [180].

The choice between first- and second-generation EGFR TKIs was investigated in two randomised studies. LUX-Lung 7 is a randomised phase IIB study that compares afatinib with gefitinib [181]. The study reported similar tumour ORR and a modest difference in PFS (mPFS 11.0 versus 10.9 months; HR 0.73, 95% CI 0.57–0.95, P = 0.0165). The other co-primary endpoint for this study was OS and was not statistically different (mOS 27.9 versus 24.5 months; HR 0.86, 95% CI 0.66‒1.12, P = 0.258) [182]. More specifically, there was no difference in OS in patients with EGFR exon 19 mutation, which is contrary to the earlier claim of benefit in this subgroup from the pooled analysis of LUX-Lung 3 and LUX-Lung 6 studies [183].

ARCHER 1050 is a randomised phase III study that compares dacomitinib with gefitinib in stage IV EGFR-mutated lung cancer patients without CNS metastasis [184, 185]. The study reported significant improvement in PFS (mPFS 14.7 versus 9.2 months; HR 0.59, 95% CI 0.47–0.74, P < 0.0001). The mOS was 34.1 months with dacomitinib versus 26.8 months with gefitinib (HR 0.76, 95% CI 0.58–0.993, P = 0.04). The OS probabilities at 30 months were 56.2% and 46.3% with dacomitinib and gefitinib, respectively. Both afatinib and dacomitinib are associated with higher incidence of grade 3 skin and gastrointestinal toxicity and a significant proportion of patients require dose reduction. Erlotinib, gefitinib and afatinib are recommended as first-line therapy in patients with advanced NSCLC who have active sensitising EGFR mutations, regardless of their PS [I, A]. Dacomitinib will be added to the list when the drug is approved by regulatory agencies, the United States FDA and the EMA [I, A; not EMA-approved]. There is no consensus preferring any of the three currently available first-line EGFR TKIs over others [IV, C].

Osimertinib is a third-generation EGFR TKI that targets both sensitising EGFR mutation and the resistant exon 20 T790M mutation [186]. The drug was compared with a standard first-generation EGFR TKI (gefitinib or erlotinib) in the FLAURA phase III study [187]. Significant improvement in PFS was observed (mPFS 18.9 versus 10.2 months; HR 0.46, 95% CI 0.37–0.57, P < 0.0001). More importantly, a similar degree of improvement was observed in the subgroup of patients with CNS metastasis (mPFS 15.2 versus 9.6 months; HR 0.47, 95% CI 0.30–0.74, P = 0.0009). OS data were immature, while authors reported an HR of 0.63, which was not statistically significant. First-line osimertinib is now considered one of the options for NSCLC patients with sensitising EGFR mutations [I, A; MBCS score v1.1 score: 4].

The combination of ChT with gefitinib, at progression with gefitinib, has not shown any clinical benefit (IMPRESS Trial) [188]. The NEJ009 trial is the first phase III study that evaluated the efficacy of a combination of EGFR TKI (gefitinib) and platinum doublet ChT (carboplatin/pemetrexed) in untreated advanced NSCLC patients with EGFR mutations [189]. Carboplatin/pemetrexed/gefitinib demonstrated significantly better PFS (mPFS: 20.9 versus 11.2 months, HR 0.49, 95% CI 0.39–0.62) and OS (mOS: 52.2 versus 38.8 months, HR 0.69, 95% CI 0.52–0.92) compared with gefitinib, in advanced EGFR-mutated NSCLC, representing a first-line therapy option [I, B; not EMA-approved].

The combination of EGFR TKI and antiangiogenesis was first investigated in Japan. A randomised phase II study compared the combination of erlotinib and bevacizumab with erlotinib alone as first-line therapy for patients with EGFR-mutant NSCLC. Seto et al. reported mPFS of 16.4 and 9.8 months (HR 0.52, 95% CI 0.35–0.76), respectively [II, A] [190, 191]. However, the significant difference of PFS did not translate into a difference of OS between these treatments (mOS: 47 versus 47.4 months). A similar PFS was described in a European phase II trial that also evaluated the combination of erlotinib and bevacizumab, which was determined to be suitable as a front-line treatment option in EGFR-mutated NSCLC [III, B] [192]. A phase III trial (NEJ026) comparing bevacizumab/erlotinib to erlotinib in this patient population reported encouraging interim analysis results with significant benefit on PFS (mPFS 16.9 versus 13.3 months, HR 0.60, 95% CI 0.41–0.87); survival results are pending [II, A] [193]. While active research is ongoing, the EMA has approved the use of the combination of erlotinib and bevacizumab [ESMO-MCBS v1.1 score: 3]. Erlotinib/bevacizumab represents a front-line treatment option in EGFR-mutated tumours [II, B].

Almost all patients who benefit from EGFR TKIs will eventually develop clinical resistance. About half of the resistance is explained by the acquired EGFR exon 20 T790M mutations [194]. Osimertinib and several other third-generation EGFR TKIs were developed targeting the T790M mutation. To date, the only approved medication for patients with T790M mutation is osimertinib. AURA3 is a randomised phase III study that compared osimertinib with pemetrexed/platinum in patients with proven T790M mutation at time of progression on first-/second-generation EGFR TKI [195]. Tumour ORR was 71% and 31%, respectively (HR 5.39, 95% CI 3.46–8.48, P < 0.001). The primary endpoint of PFS was also significantly different (mPFS 10.2 versus 4.4 months; HR 0.30, 95% CI 0.23–0.41, P < 0.0001). Osimertinib also showed a significantly longer CNS PFS (11.7 months) and higher CNS ORR (70%, 95% CI 51–85) compared with ChT (CNS PFS 5.6 months, CNS ORR 31%, 95% CI 11–59) in patients with CNS metastases at baseline [196]. The probability of experiencing a CNS progression event was lower for osimertinib than for ChT at both 3 months (2.7% versus 8.2%, respectively) and 6 months (11.5% versus 28.2%, respectively). This study has established a new paradigm: all patients with clinical resistance to first-/second-generation EGFR TKIs should be tested for the presence of T790M mutation and osimertinib should be offered as standard treatment for patients who test positive [I, A; ESMO-MCBS v1.1 score: 4].

Molecular mechanisms of resistance to EGFR TKIs were complex and heterogenous in patients without T790M mutation. These include MET amplification, HER2 amplification, PIK3CA alternations, BRAF mutation, KRAS mutation and small cell transformation. The current standard in this scenario is platinum-based doublet ChT [I, A] and the expected ORR and PFS are 31% and 5.4 months, respectively [188], and should be considered as a therapeutic option in patients with EGFR-mutated tumour, PS 0–1, in absence of contraindications to use of immunotherapy after targeted therapies have been exploited [III, A; not EMA-approved] [97].

The anti-tumour activity of crizotinib was initially demonstrated in two multicentre single-arm studies, with significant ORR and PFS advantages, as well as a survival advantage, compared with other treatment options [197, 198]. The phase III study, PROFILE 1014, compared crizotinib with platinum–pemetrexed (without maintenance pemetrexed) as first-line treatment in ALK-rearranged advanced NSCLC. It demonstrated a significantly longer PFS (mPFS 10.9 versus 7.0 months; HR 0.45; 95% CI 0.35–0.60; P < 0.001) and higher ORR with crizotinib compared with ChT [199]. First-line treatment with crizotinib is a treatment option for patients with ALK-rearranged NSCLC [I, A; ESMO-MBCS v1.1 score: 4] (Figures 3 and 5).

Ceritinib and alectinib are second-generation ALK inhibitors that have shown robust antitumour efficacy, along with intracranial activity, in patients with ALK-rearranged NSCLC. The ASCEND-4 trial compared ceritinib (750 mg/day) with platinum-based ChT (cisplatin or carboplatin plus pemetrexed followed by maintenance pemetrexed) in untreated advanced ALK-rearranged non-squamous NSCLC [200]. Overall, ceritinib improved ORR over ChT: 72.5% (95% CI 65.5–78.7) compared with 26.7% (95% CI 20.5–33.7). mPFS was 16.6 months (95% CI 12.6–27.2) with ceritinib versus 8.1 months (95% CI 5.8–11.1) with ChT (HR 0.55, 95% CI 0.42–0.73, P < 0.01). At baseline, 59 patients in the ceritinib arm and 62 patients in the ChT arm had CNS metastasis. Among them, the intracranial ORR by RECIST was 72.7% (95% CI 49.8–89.3) with ceritinib versus 27.3% (95% CI 10.7–50.2) with ChT. In patients without baseline brain CNS metastasis, the mPFS with ceritinib was 26.3 months (95% CI 15.4–27.7), versus 8.3 months (95% CI 6.0–13.7) in the ChT arm. The most common AEs (all grades) in the ceritinib group were diarrhoea (85%), nausea (69%), vomiting (66%) and an increase in alanine aminotransferase (ALT, 60%) [ESMO-MCBS v1.1 score: 4]. Considering the safety profile of ceritinib, the influence of food on its oral bioavailability and the fact that food may improve gastrointestinal tolerability, a trial was conducted with a lower dose of ceritinib taken with a low-fat meal (ASCEND-8) [201]. A 450 mg dose of ceritinib taken once daily with food provides similar systemic exposure as the currently approved daily dose of 750 mg in a fasted state, and preliminary safety results demonstrated a reduction of the gastrointestinal toxicities when compared with the 750 mg fasted dose. These results suggest this dosing regimen as an alternative to the ceritinib 750 mg fasted dose [III, B].

The efficacy of alectinib was tested in a phase III head-to-head trial comparing this molecule [300 mg twice daily (b.i.d.)] with crizotinib (250 mg b.i.d.) in ALK TKI-naive ALK-rearranged advanced NSCLC Japanese patients (J-ALEX trial), demonstrating the superiority of alectinib as an initial targeted treatment [202]. The PFS HR of the alectinib arm compared with the crizotinib arm was 0.34 (95% CI 0.17–0.70, P < 0.0001). mPFS was not reached [95% CI 20.3–not evaluable (NE)] in the alectinib arm, while it was 10.2 months (95% CI 8.2–12.0) in the crizotinib arm. A similar global trial in ALK-rearranged treatment-naive patients was conducted (ALEX trial). Patients were randomised to receive either alectinib (600 mg b.i.d.) or crizotinib (250 mg b.i.d.) [53]. The investigator-assessed mPFS with alectinib was 34.8 (95% CI 17.7–not reached), compared with 10.9 months (95% CI 9.1–12.9) with crizotinib [203]. PFS assessed by the independent review committee was also significantly longer with alectinib than with crizotinib (mPFS 25.7 months; 95% CI 19.9–NE versus 10.4 months; 95% CI 7.7–14.6, respectively). In patients with baseline CNS metastases, mPFS was 27.7 months for alectinib versus 7.4 months for crizotinib. The time to CNS progression was significantly longer with alectinib than with crizotinib (cause-specific HR 0.16, 95% CI 0.10–0.28, P < 0.001). The mOS was not estimable in either group. Grade 3–5 AEs were less frequent with alectinib (41% versus 50% with crizotinib) [ESMO-MCBS v1.1 score: 4]. In patients with CNS involvement, front-line use of ALK TKIs is effective, and alectinib [III, A] or ceritinib [IV, B] are recommended. While ceritinib represents a better treatment strategy than ChT [I, B] and presumably crizotinib [IV, B], alectinib represents a better treatment option than ChT [III, A] and crizotinib [I, A].

The benefit of crizotinib over second-line ChT in TKI-naive patients with previously treated ALK-rearranged NSCLC was confirmed in the phase III PROFILE 1007, with better ORR and PFS [204]. The mPFS was 7.7 months (95% CI 6.0–8.8) in the crizotinib group, compared with 3.0 months (95% CI 2.6–4.3) in the ChT group. Any patient with NSCLC harbouring an ALK fusion should receive crizotinib as next-line therapy, if not received previously [I, A]. Despite improved outcome in patients with tumours harbouring ALK rearrangements and treated with crizotinib (mainly in first line), all patients will eventually experience disease progression through primary or acquired resistance. Furthermore, crizotinib penetration into the cerebrospinal fluid (CSF) is negligible, and this pharmacological limitation is extremely relevant in treatment decisions, taking into account the high propensity of ALK-rearranged NSCLC to metastasise to the brain [205]. Ceritinib (ASCEND-5) and alectinib (ALUR) were compared with ChT in patients with ALK-positive NSCLC previously treated with crizotinib and ChT [206, 207]. Both trials showed a significant improvement in mPFS compared with ChT (5.4 months, 95% CI 4.1–6.9 for ceritinib versus 1.6 months, 95% CI 1.4–2.8 for ChT; HR 0.49, 95% CI 0.36–0.6, P < 0.0001 and 9.6 months, 95% CI 6.9–12.2 for alectinib versus 1.4 months, 95% CI 1.3–1.6 for ChT; HR 0.15, 95% CI 0.08–0.29; P < 0.001). CNS ORR was 54.2% and 35% with alectinib or ceritinib, respectively, versus 0% or 5% with ChT in the ALUR and ASCEND-5 trials, respectively [206–208]. Based on this data, ceritinib and alectinib are recommended in patients with ALK-positive advanced NSCLC who progress on treatment with or are intolerant to crizotinib [I, A; ESMO-MBCS v1.1 score: 4].

In ALK-rearranged patients progressing on crizotinib with CNS progression, treatment with next-generation ALK TKIs, such as alectinib or ceritinib, is recommended [I, A]. The next-generation ALK inhibitors, such as brigatinib or lorlatinib, have a wider coverage of ALK resistance mutations, and sequential therapy with these ALK inhibitors is the preferred treatment approach in crizotinib-resistant and/or the second generation-resistant populations. The ALTA trial evaluated the brigatinib in crizotinib-resistant ALK-rearranged NSCLC patients. Patients were randomised (1:1) to receive oral brigatinib 90 mg once daily (arm A) or 180 mg once daily with a 7-day run-in at 90 mg (arm B) [209]. The ORR was 46% (arm A) and 55% (arm B) and mPFS was 9.2 months in arm A and 15.6 months in arm B. mOS was not reached in arm A and was 27.6 months in arm B. CNS ORRs were 50% and 67% in arms A and B, respectively. In results from a phase I study, lorlatinib demonstrated significant activity reporting ORRs of 46% and 42% among ALK-rearranged patients pretreated with one or with two or more ALK TKIs, respectively, including patients with CNS metastases at baseline (intracranial ORR 42%) [210]. A phase II study at the recommended dose (100 mg once a day) is demonstrating 69% RR in crizotinib pretreated patients and 39% after two or more previous ALK TKIs [211]. Of interest, in patients previously treated with one or more second-generation ALK TKIs, a higher proportion of patients harbouring an ALK secondary mutation responded to treatment with lorlatinib compared to those without detectable ALK mutations (ORR: 61% versus 26%) [212]. Lorlatinib and brigatinib are in phase III testing to investigate whether upfront treatment with the next generation can further improve clinical outcomes for patients with advanced ALK-rearranged NSCLC compared with crizotinib treatment [213, 214]. At the first interim analysis, brigatinib was shown to improve PFS compared with crizotinib (HR 0.49; 95% CI 0.33−0.74; P < 0.001) [I, B, not EMA approved] [215]. In patients with baseline CNS metastases, intracranial objective response rate was 78% for brigatinib versus 29% for crizotinib. In patients who progress after a second-generation ALK TKI, the next-generation ALK inhibitors such as brigatinib or lorlatinib are recommended if available [III, B]. They are currently not approved by the EMA.

On the basis of the available preclinical data, the phase I PROFILE 1001 study of crizotinib was amended to include patients with ROS1-rearranged NSCLC in the expansion cohort [216]. Among 50 patients with ROS1-rearranged NSCLC in this trial cohort, the ORR to crizotinib was 72%, with a disease control rate equal to 90% and an mPFS of 19.2 months. In a prospective French phase II study and in the retrospective EUROS1 study of crizotinib for ROS1-rearranged NSCLC, mPFS was 10 and 9.1 months and ORR was 72% and 80%, respectively, although both of these studies enrolled only approximately 30 patients [217, 218]. In a larger East Asian phase II study of crizotinib, the mPFS among 127 patients with ROS1-rearranged lung cancer was 13.4 months [219]. Each study included patients who had received varying numbers of prior lines of systemic therapy, although for all of these patients, crizotinib remained the first ROS1-directed TKI. Single-agent crizotinib is recommended in the first-line setting or as second line in patients with stage IV NSCLC with ROS1 rearrangement [III, A; ESMO-MBCS v1.1 score: 3] (Figures 3 and 6). If patients have received crizotinib in the first-line setting, then they may be offered platinum-based ChT therapy in the second-line setting [IV, A].

Ceritinib is a potent and selective ALK inhibitor that also inhibits ROS1. In a Korean phase II study, 32 patients with ROS1-rearranged advanced NSCLC were treated with ceritinib, 750 mg daily [220]. Among crizotinib-naive patients, the ORR was 67%, with a disease control rate of 87%. The mPFS was 9.3 months for the entire cohort and reached 19.3 months for crizotinib-naive patients. Of note, in those two patients who had received prior crizotinib, no clinical response was observed. Ceritinib might be considered in crizotinib-naive patients but is currently not approved by the EMA [III, C].

Brigatinib, lorlatinib and entrectinib also have a potential anti-ROS1 activity on the basis of preclinical studies and limited phase I/II encouraging clinical data [221].

The most common BRAF mutation, V600E (Val600Glu), is observed in 1%–2% of lung adenocarcinomas [222–224], more frequently in patients with smoking history. In a retrospective multicentre cohort study in Europe, patients with advanced BRAF-mutant lung cancer received treatment with either vemurafenib (n = 29), dabrafenib (n = 9) or sorafenib (n = 1) [225]. Of the BRAF mutations, 83% were BRAF V600E. The ORR was 53% and the PFS and OS were 5 and 10.8 months, respectively.

In a vemurafenib basket trial (VE-BASKET), patients with various BRAF V600 mutation-positive non-melanoma tumours were enrolled in six prespecified cancer cohorts, including an NSCLC cohort with 20 patients [226]. A total of 19 NSCLC patients were evaluable for response. Overall, one patient was treatment-naive and 50% and 45% of patients received one or two or more lines of therapy before study inclusion, respectively. The ORR, mPFS and mOS were 42%, 7.3 months and not yet reached, respectively.

A prospective multicentre multicohort phase II study of dabrafenib monotherapy (cohort A), or combination therapy with a MEK inhibitor (trametinib) (cohort B, beyond first-line and cohort C in first-line treatment) in patients with BRAF V600E-mutant metastatic NSCLC (BRF113928) was reported. With dabrafenib monotherapy (cohort A, n = 78), the ORR was 33% and mPFS and median duration of response (mDoR) were 5.5 and 9.6 months, respectively [227]. With combination dabrafenib and trametinib in pretreated patients (cohort B, n = 57), the ORR was 66% and mPFS and mDoR were 10.2 and 9.8 months, respectively [228, 229]. With combination dabrafenib and trametinib therapy in unpretreated patients (cohort C, n = 36), the ORR was 64% and mPFS and DoR were 10.9 and 10.4 months, respectively [230]. The mOS was 24.6 months and half of the patients were still alive at two years from treatment beginning. The EMA and the United States FDA have approved dabrafenib in combination with trametinib for the treatment of patients with BRAF V600 mutation-positive advanced or metastatic NSCLC. BRAF/MEK inhibition using dabrafenib with trametinib is recommended in patients with BRAF inhibitor naive, stage IV NSCLC with BRAF V600E mutation [III, A; ESMO-MBCS v1.1 score: 2] (Figures 3 and 7). If patients have received BRAF/MEK inhibition in the first-line setting, then they may be offered platinum-based ChT in the second-line setting [IV, A].

Several other molecular targets have been identified harbouring somatic variants with therapeutic potential, including RET, MET, HER2 and NTRK.

RET fusions are found in 1%–2% of NSCLC and tend to be mutually exclusive to other lung cancer drivers [231, 232]. Although RET-selective inhibitors have not yet been developed, several multitarget agents with anti-RET activity have been evaluated in preclinical models and clinical trials. The activity of multikinase inhibitors (cabozantinib, vandetanib, sunitinib, sorafenib, alectinib, lenvatinib, nintedanib, ponatinib, regorafenib) in patients with RET-rearranged NSCLC (ORR 16%–47% and mPFS 2.3–7.3 months) is clearly inferior to the responses and survival outcomes seen with selective TKIs in other oncogene-addicted NSCLC models [233–236]. These studies are small and subject to selection bias and results of heterogeneous benefit [III, C]. Further studies are needed to confirm the benefit of current treatments as well as the development of more specific inhibitors (i.e: BLU-667, LOXO-292) [237]. Targeting RET is not currently routinely recommended and recruitment into open trials is encouraged [III, C].

Somatic dysregulation at MET occurs through a number of different non-exclusive mechanisms in NSCLC including overexpression, amplification, mutation and gene-rearrangement. Previous trials aimed at targeting MET overexpression (e.g. onartuzumab or tivantinib) have failed, and as the relationship between expression and genomics is now better understood, focus has shifted to targeting genomic variants [238–240]. Two major MET variants may play a key role as NSCLC oncogenic drivers: MET exon 14 variants (METex14) and MET amplification. MET amplification can occur as either acquired (as a resistance mechanism to EGFR TKI therapy) or de novo. While a promising target, targeting MET dysregulation by MET amplification is not currently routinely recommended and recruitment into open trials is encouraged [III, C]. METex14 mutations are similarly as common as ALK rearrangements, occurring in 3%–4% of NSCLC. They are more frequently but not exclusively identified in adenocarcinoma and sarcomatoid carcinoma histological subtypes (especially those with an adenocarcinoma component), observed in current, ex- and never-smokers, more frequently observed in older than in younger patients. METex14 mutations are extremely diverse and result in aberrant splicing and exon 14 skipping, resulting in loss of the MET Y1003 c-Cbl binding site and reduced MET degradation, detectable as increased expression by IHC. Moreover, METex14 mutations are mutually exclusive to other drivers (EGFR, ALK, BRAF), further reinforcing MET status as an oncogenic driver, more often encountered in smokers. Multiple case series and cohorts have now demonstrated durable ORRs with MET-targeting TKIs including crizotinib, capmatinib and cabozantinib in METex14 patients, with the PROFILE 1001 trial METex14 cohort reporting an ORR of 44% and a global retrospective series demonstrating a PFS of 7 months, both with crizotinib [241, 242]. A variety of MET-directed TKIs are undergoing development against this target. For METex14 variants, while evidence of benefit is stronger, recruitment into open trials is encouraged [III, C]. Crizotinib has demonstrated potential clinical efficacy that needs to be confirmed [III, C].

HER2 dysregulation is another promising target for advanced NSCLC and is abrogated via different mechanisms including exon 20 mutations, transmembrane domain mutations, amplification and protein overexpression. Mutations in exon 20 were the first HER2 mutations described and occur in 1%–5% of patients, over-represented in young patients, never-smokers, females, patients without ethnic clustering and typically in adenocarcinomas [243]. Such mutations are analogous to EGFR exon 20 insertions, being mutually exclusive to other oncogenic drivers, and are usually 3–12 bp in-frame insertions between codons 775–881, the most common being the YVMA insertion at codon 775. HER2 insertions are typically resistant to HER-targeting TKIs afatinib, dacomitinib and neratinib [244, 245], although some specific genotypes, e.g. those resulting in Gly770 insertion, may retain sensitivity [246]. Afatinib and poziotinib have demonstrated some activity in HER2-mutated NSCLC in small series [247, 248]. More recently, targeting HER2 mutation with ado-trastuzumab emtansine (TDM-1) has shown promise with two cohorts demonstrating responses including mutants with no copy-number change [249]. Abnormal gene copy-number is also identified at HER2, although is typically polysomy, with HER2 exon 20 insertions and amplification usually mutually exclusive [243]. Targeting HER2 amplification or protein expression with trastuzumab monotherapy has not consistently demonstrated benefit, but may have a role in HER2-mutant NSCLC, although data are usually based on cases confounded by concurrent ChT and variable HER2 expression. The antibody–drug conjugate TDM-1 has shown very modest activity in HER2-overexpressing NSCLC [250]. Rarer HER2 variants include transmembrane domain mutations (e.g. V659, G660) that have reported sensitivity to afatinib and TDM-1 [251]. Nevertheless, given the paucity of robust data, targeting HER2 dysregulation is not currently recommended and recruitment into open trials is encouraged [III, C].

Somatic fusions involving the neurotropic tropomyosin receptor kinase genes (NTRK1-3) are rare oncogenic drivers occurring at low prevalence (< 1%) in a variety of tumours including NSCLC [252], again typically in adenocarcinomas (although non-adenocarcinoma cases are reported) and never-smokers. The rarity of these fusions across different cancer types has resulted in basket trial design for drug development. NRTK1-3 fusions encode oncogenic TRKA-C fusion proteins, respectively, that can be targeted by therapies in development, including larotrectinib (LOXO-101) and entrectinib (RXDX-101) [253–256]. Both have demonstrated marked durable responses in NTRK fusion-positive NSCLC in early reports from ongoing single arm basket studies but are not currently recommended for routine care and recruitment into open trials is therefore encouraged [III, C].

External beam radiotherapy (EBRT) plays a major role in the symptom control of metastases, such as painful chest wall disease, painful bone metastasis, superior vena cava syndrome, soft tissue or neural invasion. EBRT is indicated in cases of haemoptysis and symptomatic airway obstruction [III, B]. A Cochrane systematic review of palliative EBRT regimens for patients with thoracic symptoms from NSCLC included 14 RCTs (3576 patients) [257]. Doses of radiation ranged from 10 Gy in 1 fraction to 60 Gy in 30 fractions, with a total of 19 different dose/fractionation regimens. There was no strong evidence that any regimen achieved a greater level of palliation [II, B]. Furthermore, higher dose regimens were associated with higher rates of acute toxicity. However, it should be noted that the studies were heterogeneous and most were conducted in the 1980s and 1990s, therefore using dated radiotherapy (RT) techniques. There are few data on the optimal timing of thoracic RT and systemic therapy in the stage IV NSCLC setting. Furthermore, there is no evidence to date that the concurrent administration of ChT, targeted agents or immunotherapy to palliative RT is beneficial in this group of patients.

Another method of palliation of thoracic symptoms is endobronchial brachytherapy (EBB). The effectiveness of EBB compared with EBRT or other alternative endoluminal treatments was assessed in a Cochrane systematic review [258]. The authors concluded that EBRT alone is more effective for palliation than EBB alone [II, B]. However, evidence was limited with regard to the comparison of EBB plus EBRT over EBRT alone for symptom relief. For patients previously treated by EBRT who are symptomatic from recurrent endobronchial central obstruction, EBB may be considered in selected cases [III, C].

CNS metastases are commonly identified with NSCLC, predominantly with adenocarcinoma. LMD is a deadly complication of solid tumours and is associated with a poor prognosis. Adenocarcinomas are the most common tumours to metastasise to the CNS. Of the patients with NSCLC, 30%–64% have CNS metastases, of which 4%–7% present LMD [260]. Incidence and prevalence of LMD are both increasing due to brain metastases screening, better imaging modalities as well as prolongation of patients’ survival.

Presence of malignant cells on CSF cytology provides the gold-standard for diagnosing leptomeningeal (LM) carcinomatosis. Abnormalities on imaging can be found in 70%–80% of patients with LMD and the imaging modality of choice is high quality, T1-weighted MRI with gadolinium contrast, which has been shown to be more sensitive compared with contrast-enhanced CT [261, 262].

The treatment of patients with brain metastases, with/without LM involvement and no driver mutations, is dependent on the prognosis. Prognosis can be estimated based on the Radiation Therapy Oncology Group recursive partitioning analysis (RPA): class I patients are those < 65 years old, with a good PS [Karnofsky Index (KI) ≥ 70%], no other extracranial metastases and a controlled primary tumour; class III patients have a KI < 70%; and class II represents all other patients [263]. In RPA class III patients, RT is not recommended in view of the dismal prognosis [I, A]; only BSC is recommended, as their median survival is generally < 2 months. The role of whole-brain RT (WBRT) in unselected patients has been questioned by the QUARTZ trial data, a phase III non-inferiority study, in which patients were randomised to either BSC including dexamethasone plus WBRT 20 Gy in 5 daily fractions or to the same BSC without WBRT [264]. This trial demonstrated no difference between the treatment arms regarding the relief of symptoms, steroid use, OS, QoL or quality-adjusted life years in the intention-to-treat (ITT) population, confirming no benefit for WBRT in the RPA class III subset [I, A]. However, the median survival in the trial was poor (8.5–9.2 weeks) and the trial recruited over 7 years, a time during which considerable advances in molecular selection, systemic therapy, stereotactic radiosurgery (SRS) patient selection and MRI brain surveillance have been implemented. A signal for WBRT benefit was seen for younger patients with better Karnofsky PS and either controlled primary or no extracranial disease. WBRT can therefore be considered for patients contingent on prognostic factors of better survival such as driver mutations [III, C].

The most frequent WBRT schedules are 20 Gy in 5 fractions or 30 Gy in 10 fractions, with no difference in outcome [I, A] [265]. For most patients with symptomatic brain metastases and/or significant oedema, dexamethasone or equivalent corticosteroid is recommended [III, A] [266]. Tapering of the dose and, if possible, cessation after RT, are recommended. Corticosteroids are not recommended in the case of asymptomatic brain metastases. WBRT may be associated with delayed progressive cognitive impairment in responders, as tumour progression affects this parameter more than radiation toxicity [267]. Neuroprotective agents have not shown a convincing role and are not recommended for routine use [II, C], with a small phase III trial of memantine on 149 assessable patients (RTOG 0614) suggesting benefit [268]. Hippocampus avoidance WBRT has been shown to be probably safe [269], but is still undergoing trial evaluation and is not currently recommended for routine care [III, C].

Recent data showed that SRS can be considered as another standard of care for this patient population as a less toxic alternative to WBRT. SRS of the surgical cavity in patients who have had complete resection of 1–3 brain metastases significantly lowers local recurrence compared with that noted for observation alone [270].

In case of single brain metastases surgical resection can be considered [III, B] [271–273]. Postoperative WBRT or SRS is generally recommended after surgical resection [I, A] [274].

Another treatment strategy, in the case of a limited number of metastases and RPA class I and II, is SRS alone [III, B] [275–278]. The randomised trials evaluating SRS have included patients with 1–4 brain metastases. SRS has increasingly become the favoured modality due to reduced morbidity compared with WBRT, but it should be noted that there is no randomised trial comparing SRS alone to WBRT. A survival advantage in favour of WBRT plus SRS has been demonstrated against WBRT but only for the subgroup of patients with a single brain metastasis [279]. The majority of studies evaluating WBRT in addition of SRS or neurosurgery have shown a decline in cognitive function in the combined arm [278, 279]. SRS alone with close follow-up, without WBRT consolidation, is therefore a recommended strategy [III, B].

Although it is generally accepted that SRS should generally be considered in patients with ≤ 4 brain metastases, a prospective observational study from Japan challenged this prevailing concept [280]. The study enrolled 1194 eligible patients (76% had lung cancer) with 1 to 10 newly diagnosed brain metastases, longest diameter < 3 cm, largest tumour < 10 mL in volume and a total cumulative volume of ≤ 15 mL. OS did not differ between patients treated with SRS with 2-4 metastases and those with 5-10 metastases. This study therefore suggested the use of tumour volume and absolute size, rather than the number of metastases, as treatment criteria. In some territories, the indication for SRS is now based on total tumour volume rather than number of metastases, as the risk of radionecrosis increases with tumour volume [III, C] [278]. In patients undergoing SRS, radionecrosis is a challenging complication to manage.

In patients with asymptomatic brain metastases who have not yet received prior systemic therapy (i.e. ChT, TKIs), treatment with upfront systemic therapy and deferred RT should be considered, with trial data suggesting similar intracranial and extra-cranial ORRs [II, B] [281, 282]. In patients suitable for first-line immune-checkpoint inhibitor therapy, CNS metastases were generally mandated to have been treated before therapy, with evidence of intracranial response. There is currently limited trial data demonstrating safety and efficacy of immunotherapy in patients with small-volume untreated CNS metastases [III, B] [283].

Among those patients with an actionable oncogenic driver (e.g. EGFR, ALK), between 44% and 60% develop brain metastases in the course of their disease [284, 285]. In such patients, the use of CNS-penetrant next-generation TKIs (e.g. osimertinib, alectinib, ceritinib) may restore control of brain disease, thereby potentially delaying cranial RT [II, A] [53, 187, 200]. Moreover, next-generation TKIs may also reduce the incidence of new CNS metastases, thereby significantly postponing the time to need CNS RT [53].

LMD may present with non-specific neurological symptoms (headaches, nausea, vomiting) as well as discrete signs due to the CNS area involved (gait difficulties, cranial nerve palsies), and a high index of suspicion is required, particularly in those with actionable oncogenic drivers due to higher prevalence [V]. Diagnostic modalities include cerebrospinal MRI with contrast enhancement, ideally before CSF intervention. CSF sampling with cytological assessment is diagnostic but limited by low sensitivity but high specificity [IV, A]. The prognosis from LMD due to NSCLC is poor, and treatment aim is to prolong survival with acceptable QoL. Patients with actionable oncogenic drivers may derive benefit from a CNS-penetrant next-generation TKI as per those with brain metastases [III, B]; otherwise, systemic therapy strategies vary widely across Europe. ChT and bevacizumab may have activity both extra-cranially and intra-cranially, and also in the context of LMD [IV, C] [126, 286]. Intra-CSF pharmacotherapy may be considered through either repeated lumbar punctures, a reservoir or ventricular device, although consideration should be given to patient factors, e.g. PS, extra-cranial control and likely benefit [V, C]. No randomised data exist to support the role of RT for LMD. In exceptional cases, focal RT can be considered for circumscribed, notably symptomatic, lesions [V, C].

As prognosis in the majority of patients with stage IV NSCLC is poor, the role of surgery is traditionally limited in this patient group. However, with the widespread introduction of targeted therapies and immunotherapy improving prognosis in specific subcategories, the role of thoracic surgery is currently redefined. At the present time, surgery may be indicated for diagnosis, evaluation of response to systemic therapy and palliation, and highly selected patients may be considered for lung resection with therapeutic intent or even for a salvage procedure. In the last two categories, surgery can be carried out with a mortality < 2%, a low morbidity rate and 5-year survival rates in the range of 11%–30% in retrospective series [IV, C] [287, 288]. Whether there is a significant difference between synchronous and metachronous metastases and between different distant sites has not been clearly established and more prospective data are needed.

When metastatic disease is suspected on PET scanning, invasive surgical procedures such as incisional biopsies, mediastinoscopy, thoracoscopy (VATS) or laparoscopy may be required to obtain relevant biopsy samples. Examples include patients with contralateral lung nodules, distant metastases or suspicion of mediastinal nodal involvement who do not qualify for minimally invasive biopsies or in whom results of the latter are equivocal. Adequate samples should be provided to the pathologist for detailed routine staining, IHC and molecular genetic testing [III, B].

Palliative interventions may be useful in case of local complications related to the primary tumour or metastatic foci which cannot be managed by conservative measures, e.g. lung abscess, empyema, massive haemoptysis, spinal cord compression and pathological bone fractures.

In the 8th edition of the tumour, node, metastasis (TNM) classification a new subcategory was introduced comprising patients with one metastasis in a single distant organ, defined as M1b disease, in contrast to patients with multiple metastases in one or more distant organs, currently M1c disease [289]. There is no general consensus on the precise definition of oligometastatic disease and clear evidence for surgical treatment is limited, as only relatively small prospective series are available [III, B] [290–292]. Prospective series suggest that complete surgical resection is necessary to obtain long-term survival and that mediastinal nodal involvement carries a poor prognosis [293]. This is further discussed in the section ‘Treatment of oligometastatic NSCLC’.

A specific subgroup consists of patients with malignant pleural nodules or malignant pleural effusion [293]. Extensive surgical procedures consisting of extrapleural pneumonectomy sometimes in combination with intraoperative ChT or hyperthermic ChT, have been described when extrathoracic metastases or mediastinal lymph node involvement have been excluded [294, 295]. However, these interventions carry a higher operative risk and prospective studies are currently not yet available [IV, D]. Persisting or recurrent pleural effusions are usually managed by pleurodesis to improve dyspnoea. Talc is the preferred agent and thoracoscopic poudrage may be better than injection of talc slurry in patients with primary lung cancer [II, B] [296, 297]. In case of a trapped lung by a thickened visceral pleural peel, indwelling pleural catheters or pleuroperitoneal shunts provide symptomatic relief [IV, B] [298, 299].

Lastly, salvage surgery may be considered in case of residual or progressive disease in the primary tumour or metastatic site when no other treatment options remain or specific complications occur, such as formation of a lung abscess in a necrotic cavity [300]. Long-term survival may be obtained in selected patients with limited distant involvement, but only case reports have been published so far [V, C] [301].

In a recent retrospective analysis of the National Cancer Database, a cohort of 300 572 patients with stage IIIA, IIIB or IV NSCLC were studied, of whom 4568 had a surgical intervention for stage IV disease [302]. A surgical selection score could be constructed comprising histology, tumour size, TNM status, Charlson comorbidity index, age, race, facility type, insurance and income. In a logistic regression model this score was found to be a good predictor of survival. However, it should be noted that further prospective validation is necessary, and that the relative contribution of surgery versus RT in a multimodality setting for stage IV disease was not studied in this analysis.

The growing interest in oligometastases is based on the concept that long-term disease control, or even cure, may be achieved in some subgroups of these patients with aggressive local treatment of distant metastases (surgery or high-dose RT) [303]. The term ‘oligometastases’ refers to a limited number of distant metastases, although there is no consensus on the appropriate cut-off to define the oligometastatic state. Almost all published clinical trials investigating local treatment of oligometastatic disease have limited inclusion to patients with ≤ 5 metastases. In addition, the vast majority of the trials included patients with ≤ 3 metastases and in an individual patients meta-analysis published in 2014, almost 90% of the patients had a single metastasis [303]. Some studies also limited the number of organs in which these metastases are present [304]. It should be noted that many of these studies did not include PET-CT staging.

Oligometastases can be either synchronous, when a patient presents with a limited number of metastases at initial diagnosis, or metachronous when metastases are identified after treatment of the primary tumour [305]. The biology of synchronous and metachronous oligometastases may differ as illustrated by the fact that patients with metachronous presentation have a better prognosis [303]. In patients receiving systemic therapy (mainly in tumours with driver mutations treated with TKIs), the term oligoprogression can be also applied in the case of progression of a limited number of metastatic lesions, when all the other lesions remain stable. Clinical trials are ongoing in this setting.

In this heterogeneous group of patients with oligometastases, the specific approach to oligometastases in the brain has been discussed above. Another subgroup requiring further discussion is that of patients with a solitary lesion in the contralateral lung. The International Association for the Study of Lung Cancer (IASLC) Staging and Prognostic Factors Committee carried out a systematic literature review, aiming at distinguishing a second primary from a metastasis in patients who have more than one pulmonary nodule [306]. This review concluded that few features are definitive, with many commonly used factors being suggestive, but carry a substantial risk of misclassification as the majority of second primary lung tumours are of the same histology. For these cases, the IASLC recommended a careful review by a multidisciplinary tumour board, and pursuit of radical therapy, such as that for a synchronous secondary primary tumour, when possible. Both surgery [307, 308] and SRS [309, 310] have been shown to result in long-term survivors in this setting [IV, B].

A systematic literature review identified 757 NSCLC patients treated with 1–5 (88% single metastases) synchronous or metachronous metastases [303]. These patients had a median age at diagnosis of 61 years, 98% had a good PS and two-thirds of patients had early-stage intrathoracic disease staged IA–IIB (after excluding metastatic disease). Surgery was the most common treatment modality for both primary (n = 635, 83.9%) and metastases (n = 339, 62.3%). Predictive factors for OS were synchronous versus metachronous metastases (P < 0.001), N-stage (P = 0.002) and adenocarcinoma histology (P = 0.036). RPA for risk groups identified a good prognosis (low-risk) group presenting with metachronous metastases (5-year OS of 48%), an intermediate-risk group presenting with synchronous metastases and N0 disease (5-year OS of 36%) and, finally, a high-risk group presenting with synchronous metastases and intrathoracic N1/N2 disease (5-year OS of 14%). Caution is warranted before concluding that positive outcomes in these patients are due solely to the treatment intervention, rather than patient selection or other biases [305].

Stage IV patients with limited synchronous metastases at diagnosis may experience long-term disease-free survival (DFS) following systemic therapy and local consolidative therapy [LCT: high-dose RT including stereotactic ablative body RT (SABR) or surgery] [III, B]. Five phase II trials evaluating LCT in patients with NSCLC and synchronous oligometastases have been published. Three of these studies are small, single arm studies which generally showed durable PFS in a subgroup of patients [290, 291, 311]. Two out of the five studies are randomised phase II studies that were stopped early after interim analysis. The first study randomised NSCLC patients between maintenance therapy (RT or surgery) in patients with ≤ 3 metastases, without progression after first-line systemic therapy (n = 49). There was a significant difference in PFS time between the two groups (mPFS 11.9 months in the LCT group versus 3.9 months in the maintenance group; HR = 0.35, P = 0.005) [292]. The second study randomised patients with ≤ 5 metastatic sites between maintenance ChT alone versus SABR followed by maintenance ChT (n = 29) [312]. So far, there are no published data on the impact of LCT on OS and long-term toxicity. Several clinical trials are ongoing to evaluate these important endpoints.

Stage IV patients with limited metachronous metastases may be treated with a local treatment as some may experience long-term DFS [IV, B]. However, this is based mainly on retrospective data. Although operative risk is low and long-term survival may be achieved, current evidence for surgery in oligometastatic disease is limited, and the relative contribution of surgery versus RT as local treatment modality has not yet been established. Solitary lesions in the contralateral lung should, in most cases, be considered as synchronous secondary primary tumours and, if possible, treated with curative-intent therapy [IV, B].

Similarly, there are few prospective data to support this treatment approach in patients with driver mutations who present with oligoprogression on molecular-targeted therapies [IV, C]. Furthermore, there is little data on the safety of combining SABR with molecularly targeted agents.

Some recommendations for the implementation of standard of care and advanced imaging modalities for identifying and following up patients with oligometastatic disease have been published by the European Organisation for Research and Treatment of Cancer (EORTC) imaging group [313]. In the synchronous, metachronous and oligoprogessive setting, because of the limited evidence available, inclusion in clinical trials is preferred.

Given the incidence of bone metastases in NSCLC (30%–40% of patients with NSCLC develop bone metastases), it may be reasonable to evaluate for bone disease upon disease diagnosis. In general, the management aim is to palliate symptoms and prevent complications. Palliative RT is highly effective, usually with rapid pain relief. Both standard EBRT and SABR can be used to palliate painful, uncomplicated bone pain. However, the data on efficacy and safety of SABR are mainly from retrospective single institution studies. Systematic reviews of palliative RT trials for bone metastases showed that single and multiple fraction regimens provided equal pain relief; however, retreatment rates were significantly higher in patients receiving single fraction treatment [I, A] [314, 315].

Zoledronic acid reduces skeletal-related events (SREs) (pathological fracture, radiation/surgery to bone or spinal cord compression) [II, B] [316]. Denosumab shows a trend towards superiority to zoledronic acid in lung cancer in terms of SRE prevention [II, B] [317]. In an exploratory analysis of a large phase III trial, denosumab was associated with improved mOS in the subgroup of 702 metastatic NSCLC patients [318]. In the study of denosumab versus zoledronic acid in patients with advanced cancers, the time extent to which pain interfered with daily life (used as surrogate for QoL) was longer in patients treated with denosumab and with no pain or mild pain interference at baseline [319]. Both agents are associated with increased risk of osteonecrosis of the jaw. Zoledronic acid or denosumab are thus recommended in selected patients with advanced lung cancer with bone metastases [I, B]. Patients should be selected if they have a life expectancy of > 3 months and are considered at high risk of SREs.

Endoscopy has a role to play in palliative care, notably in case of symptomatic major airway obstruction or post-obstructive infection, where endoscopic debulking by laser, cryotherapy or stent placement may be helpful [III, C]. Endoscopy is useful in the diagnosis and treatment (endobronchial or by guiding endovascular embolisation) of haemoptysis [III, C]. Vascular stenting might be useful in NSCLC-related superior vena cava compression [III, B].

Early palliative care intervention is recommended, in parallel with standard oncological care [I, A], with evidence demonstrating that palliative care interventions significantly improve QoL. Two randomised trials evaluating the impact of introducing specialised palliative care early after diagnosis of stage IV disease on patient QoL in ambulatory patients were able to show improvements in QoL and mood, and, in one trial, a reduction in aggressive treatment and an improvement in mOS [320, 321].

The optimal approach to post-treatment management of patients with NSCLC, including the role of radiological evaluation, is controversial, with very limited literature available. Due to the aggressive nature of this disease, generally close follow-up, at least every 6–12 weeks after first-line therapy, is advised to allow for early initiation of second-line therapy but should also depend on individual retreatment options [III, B].

These Clinical Practice Guidelines were developed in accordance with the ESMO standard operating procedures for Clinical Practice Guidelines development, http://www.esmo.org/Guidelines/ESMO-Guidelines-Methodology. The relevant literature has been selected by the expert authors. A summary of recommendations is provided in Table 4. A MCBS table with ESMO-MCBS scores is included in Table 5. ESMO-MCBS v1.1 was used to calculate scores for new therapies/indications approved by the EMA since 1 January 2016 [322]. Levels of evidence and grades of recommendation have been applied using the system shown in Table 6; some statements may be accompanied by a grade of recommendation alone. Statements without grading were considered justified standard clinical practice by the experts and the ESMO faculty.

Diagnosis
  • Bronchoscopy is a technique ideally suited to central lesions and can be used with bronchial washing, brushing, bronchial and transbronchial biopsy

  • EBUS and/or EUS allows evaluation of regional lymph nodes

  • Transthoracic fine needle aspiration and/or core biopsy, passing a needle through the parenchyma under imaging guidance (typically CT), is indicated in case of mid to peripheral lesions

  • In presence of a pleural effusion, thoracentesis could represent both a diagnostic tool and a palliative treatment

  • More invasive, surgical approaches (mediastinoscopy, mediastinotomy, thoracoscopy etc.) in the diagnostic workup can be considered when the previously described techniques cannot allow for an accurate diagnosis

  • With systematic collaboration and constant communication between pathologists and procedure performers, diagnostic yields will be significantly greater than with blind biopsies

Pathology/molecular biology
  • Adequate tissue material for histological diagnosis and molecular testing should be obtained to allow for individual treatment decisions

  • Pathological diagnosis should be made according to the 2015 WHO classification of lung tumours

  • Specific subtyping of all NSCLCs is necessary for therapeutic decision making and should be carried out wherever possible. IHC stains should be used to reduce the NSCLC-NOS rate to fewer than 10% of cases diagnosed [IV, A]

  • EGFR mutation status should be systematically analysed in advanced NSCC [I, A]. Test methodology should have adequate coverage of mutations in exons 18–21, including those associated with resistance to some therapies [III, B]. At a minimum, when resources or material are limited, the most common activating mutations (exon 19 deletion, exon 21 L858R point mutation) should be determined [I, A]

  • The availability of TKIs effective against T790M-mutant recurrent disease makes T790M testing on disease relapse mandatory [I, A]

  • All patients with a negative cfDNA blood test still require tissue biopsy [II, A]

  • Testing for ALK rearrangement should be systematically carried out in advanced non-squamous NSCLC [I, A]

  • Detection of the ALK translocation by FISH remains a standard, but IHC with high-performance ALK antibodies and validated assays may be used for screening [III, A] and have recently been accepted as an equivalent alternative to FISH for ALK testing

  • Testing for ROS1 rearrangement should be systematically carried out in advanced NSCLC [III, A]. Detection of the ROS1 translocation by FISH remains a standard; IHC may be used as a screening approach [IV, A]

  • BRAF V600 mutation status should be systematically analysed in advanced NSCLC for the prescription of BRAF/MEK inhibitors [II, A]

  • Molecular EGFR and ALK testing are not recommended in patients with a confident diagnosis of SCC, except in unusual cases, e.g. never/former light smokers or long-time ex-smokers [IV, A]

  • If available, multiplex platforms (NGS) for molecular testing are preferable [III, A]. Whatever testing modality is used, it is mandatory that adequate internal validation and quality control measures are in place and that laboratories participate in, and perform adequately in, external quality assurance schemes for each biomarker test [III, A]

  • PD-L1 IHC should be systematically determined in advanced NSCLC [I, A]

  • Testing is required for pembrolizumab therapy but may also be informative when nivolumab or atezolizumab are used [I, A]

Staging and risk assessment
  • A complete history including a precise smoking history and comorbidities, weight loss, PS and physical examination must be recorded

  • Laboratory: standard tests including routine haematology, renal and hepatic functions and bone biochemistry tests are required

  • Routine use of serum tumour markers, such as CEA, is not recommended [IV, B]

  • Contrast-enhanced CT scan of the chest and upper abdomen including the liver and the adrenal glands should be carried out at diagnosis

  • Imaging of CNS should be considered at diagnosis for all patients with metastatic disease [IV, B] and is required for patients with neurological symptoms or signs [IV, A]. MRI is more sensitive than CT scan [III, B]

  • If bone metastases are clinically suspected, bone imaging is required [IV, B]

  • Bone scan or PET, ideally coupled with CT, can be used for detection of bone metastasis [IV, B]. PET-CT is the most sensitive modality in detecting bone metastasis [II, B]

  • NSCLC is staged according to the UICC system (8th edition) and is grouped into the stage categories shown in Tables 2 and 3

  • In the presence of a solitary metastatic site on imaging studies, efforts should be made to obtain a cytological or histological confirmation of stage IV disease [IV, A]

  • Response evaluation is recommended after two to three cycles of ChT or immunotherapy, using the same initial radiographic investigation that demonstrated tumour lesions [IV, B]. The same procedure and timing (every 6–9 weeks) should be applied for the response evaluation in patients treated with targeted therapies and/or immunotherapy [IV, B]. Follow-up with PET is not routinely recommended, due to its high sensitivity and relatively low specificity [IV, C]

  • Measurements and response assessment should follow RECIST v1.1 [IV, A]. The adequacy of RECIST in evaluating the response to EGFR or ALK TKI in respective genetically driven NSCLC is debatable [IV, B]

  • In the case of immune checkpoint inhibitor therapy, RECIST should be used, although irRECIST, iRECIST, imRECIST may have a role in the overall assessment of therapy [IV, B]

Management of advanced/metastatic disease
  • The treatment strategy should consider the histology, molecular pathology, age, PS, comorbidities and the patient’s preferences

  • Systemic therapy should be offered to all stage IV patients with PS 0–2 [I, A]

  • In any stage of NSCLC, smoking cessation should be highly encouraged, because it improves the outcome [II, A]

First-line treatment of EGFR- and ALK-negative NSCLC, PD-L1 50%
  • Pembrolizumab is considered a standard first-line option for patients with advanced NSCLC and PD-L1 expression ≥ 50% who do not have contraindications to use of immunotherapy [I, A; ESMO-MCBS v1.1 score: 5]

First-line treatment of NSCLC without actionable oncogenic driver regardless of PD-L1 status
  • ChT with platinum doublets should be considered in all stage IV NSCLC patients without an actionable oncogenic driver, without major comorbidities and PS 0–2 [I, A]

  • Platinum-based doublets are the recommended ChT option in all stage IV NSCLC patients with no contraindications to platinum compounds [I, A]

  • Four cycles of platinum-based doublets followed by less toxic maintenance monotherapy [I, A], or four cycles in patients not suitable for maintenance monotherapy [I, A], up to a maximum of six [IV, B], are currently recommended

  • The nab-PC regimen could be considered a chemotherapeutic option in advanced NSCLC patients, particularly in patients with greater risk of neurotoxicity, preexisting hypersensitivity to paclitaxel or contraindications for standard paclitaxel premedication [I, B]

  • Combinations of platinum-based ChT and anti-PD-(L1) inhibitors have reproducibly demonstrated superiority to standard platinum-based ChT. In the absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy will be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%.

  • Nivolumab plus ipilimumab represents an optional treatment regimen for patients with NSCLC with a high TMB [I, A; not EMA-approved]

First-line treatment of SCC
  • Platinum-based doublets with a third-generation cytotoxic agent (gemcitabine, vinorelbine, taxanes) are recommended in advanced SCC patients without major comorbidities and PS 0–2 [I, A]

  • The addition of necitumumab to cisplatin/gemcitabine has not been adopted as a standard in Europe for advanced SCC and its use should be carefully evaluated [I, C; ESMO-MCBS v1.1 score: 1]

  • Combination of pembrolizumab and carboplatin with paclitaxel or nab-P is a standard choice in patients with metastatic squamous NSCLC [I, A; not EMA approved]

  • The use of atezolizumab with nab-PC today represents an option in patients with metastatic squamous NSCLC [I, B; not EMA-approved]

  • Other combinations of platinum-based ChT and anti-PD-(L1) inhibitors will demonstrate superiority to standard platinum-based ChT. In the absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy should be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%.

  • Nivolumab plus ipilimumab represents an optional treatment regimen for patients with SCC with a high TMB [I, A; not EMA-approved]

First-line treatment of NSCC
  • Pemetrexed-based combination ChT is preferred to gemcitabine- or docetaxel-based combinations in patients with non-squamous tumours [II, A]

  • Pemetrexed use is restricted to NSCC in any line of treatment in advanced disease [II, A]

  • The combination of carboplatin with pemetrexed can be an option in patients with a contraindication to cisplatin [II, B]

  • Pembrolizumab in combination with pemetrexed and a platinum-based ChT should be considered a standard option in metastatic non-squamous NSCLC [I, A; ESMO-MCBS v1.1 score: 4]

  • Atezolizumab in combination with pemetrexed and a platinum-based ChT is a therapeutic option in metastatic non-squamous NSCLC [I, B; not EMA-approved]

  • Combination of atezolizumab and bevacizumab with carboplatin and paclitaxel is a therapeutic option in patients with PS 0-1 with metastatic non-squamous NSCLC, in the absence of contraindications to use of immunotherapy [I, A; not EMA-approved]

  • Other combinations of platinum-based ChT and anti-PD-(L1) inhibitors will demonstrate superiority to standard platinum-based ChT. In the absence of contraindications and conditioned by the registration and accessibility of anti-PD-(L)1 combinations with platinum-based ChT, this strategy should be preferred to platinum-based ChT in patients with PS 0-1 and PD-L1 < 50%.

  • Nivolumab plus ipilimumab represents an optional treatment regimen for patients with NSCC with a high TMB [I, A; not EMA-approved]

  • If PD-(L)1 is not available for ChT combinations, bevacizumab combined with paclitaxel/carboplatin may be offered in the absence of contraindications in patients with advanced NSCC and PS 0-1 (bevacizumab should be given until progression) [I, A]

  • Bevacizumab might be considered with platinum-based regimens beyond paclitaxel/carboplatin in absence of contraindications [II, B]

Maintenance
  • Maintenance ChT should be offered only to patients with PS 0–1 after first-line ChT. Decisions about maintenance should consider histology, response to platinum-doublet ChT and remaining toxicity after first-line ChT, PS and patient’s preference

  • In patients with NSCC and PS 0–1, pemetrexed switch maintenance should be considered in patients having disease control following four cycles of non-pemetrexed containing platinum-based ChT [I, B]

  • Pemetrexed continuation maintenance should be considered in patients having disease control following four cycles of cisplatin/pemetrexed [I, A]

  • Continuation maintenance with gemcitabine is an option in NSCLC patients treated with four cycles of cisplatin/gemcitabine [I, C]

  • Maintenance treatment with erlotinib is only recommended for NSCC patients with an EGFR-sensitising mutation [III, B]

PS 2 and beyond
  • ChT prolongs survival and improves QoL in NSCLC patients with PS 2 when compared with BSC [I, A]

  • Platinum-based (preferably carboplatin) combination ChT should be considered in eligible PS 2 patients [I, A]. Single-agent ChT with gemcitabine, vinorelbine, docetaxel [I, B] or pemetrexed (restricted to NSCC) [II, B] is an alternative treatment option

  • The use of checkpoint inhibitors for patients with advanced NSCLC and PS 2 can be considered [III, B]

  • Poor PS (3–4) patients should be treated with BSC only in the absence of molecularly targetable alterations, such as EGFR mutations, ALK or ROS1 rearrangements or BRAF V600 mutation [III, B]

Elderly patients
  • Immunotherapy should be considered according to standard recommendations in elderly patients [III, A]

  • Carboplatin-based doublet ChT is recommended in eligible elderly patients with PS 0–2 and with adequate organ function [I, A]

  • For those patients not eligible for doublet ChT, single-agent ChT remains the standard of care [I, B]

Second-line treatment of NSCLC without actionable oncogenic driver
  • Patients clinically or radiologically progressing after first-line therapy with PS 0–2 should be offered second-line therapy irrespective of administration of maintenance treatment [I, A]

  • In patients with progression after first-line immunotherapy with pembrolizumab, platinum-based ChT is recommended as second-line treatment option [V, B]

  • There is a general trend across each of the phase III studies in second-line (nivolumab, pembrolizumab and atezolizumab versus docetaxel) for enriched efficacy of anti-PD-1/PD-L1 agents in patients with higher PD-L1 expression compared with those with no/less PD-L1 expression. However, unselected patients may still have improved survival and tolerability with anti-PD-1/PD-L1 agents compared with docetaxel [I, A].

  • PD-L1 and PD-1 inhibitors (nivolumab, pembrolizumab and atezolizumab) are the treatment of choice for most patients with advanced, previously treated, PD-L1-naive NSCLC, irrespective of PD-L1 expression [I, A]

  • Nivolumab is recommended in both squamous [I, A; ESMO-MBCS v1.1 score: 5] and non-squamous NSCLC [I, A; ESMO-MBCS v1.1 score: 5]

  • Pembrolizumab is recommended in patients with previously treated NSCLC with PD-L1 expression > 1% [I, A; ESMO-MCBS v1.1 score: 5]

  • Atezolizumab is recommended in patients with advanced NSCLC previously treated with one or two prior lines of ChT [I, A; ESMO-MCBS v1.1 score: 5]

  • In patients not suitable for immunotherapy, second-line ChT is recommended. Comparable options as second-line therapy consist of pemetrexed, for NSCC only, or docetaxel, with a more favourable tolerability profile for pemetrexed [I, B]

  • Treatment may be prolonged if disease is controlled and toxicity acceptable [II, B]

  • Nintedanib/docetaxel is a treatment option in patients with adenocarcinoma progressing after previous ChT or immunotherapy [II, B]

  • Ramucirumab/docetaxel is a treatment option in patients with NSCLC progressing after first-line ChT or immunotherapy with PS 0–2 [I, B]

  • Combination of paclitaxel and bevacizumab is another treatment option [I, C] but it is not EMA-approved

  • Erlotinib represents a potential second/third-line treatment option in particular for patients not suitable for immunotherapy or second-line ChT in unknown EGFR status or EGFR WT tumours [II, C]

  • In patients with advanced SCC with PS 0–2 unfit for ChT or immunotherapy, afatinib is a potential option with unknown EGFR status or EGFR WT patients [I, C; ESMO-MCBS v1.1 score: 2]

First-line treatment of EGFR-mutated NSCLC
  • Patients with a tumour with a sensitising EGFR mutation should receive first-line EGFR TKIs including erlotinib, gefitinib or afatinib [I, A]. None of the three EGFR TKIs is consensually considered as a preferred option [IV, C]

  • Dacomitinib will be added to the list when the drug will be approved by regulatory agencies, the United States FDA and the EMA [I, A]

  • First-line osimertinib is now considered one of the options for patients with a tumour with sensitising EGFR mutations [I, A; MBCS score v1.1 score: 4]

  • All patients should be considered for EGFR TKIs irrespective of clinical parameters, including PS, gender, tobacco exposure, histology and line of therapy [I, A]

  • Erlotinib/bevacizumab represents a front-line treatment option in patients with EGFR-mutated tumours [II, B; ESMO-MCBS v1.1 score: 3]

  • Addition of carboplatin and pemetrexed to gefitinib represents a first-line option in patients with EGFR-mutated tumours [I, B; not EMA-approved]

  • Patients who have radiological progression with ongoing clinical benefit may continue with EGFR TKI [II, A]

  • In EGFR-mutated NSCLC patients with localised distant progression and ongoing systemic control, continuation of treatment with EGFR TKI in combination with local treatment of progressing metastatic sites may be considered [III, B]

Second-line treatment of EGFR-mutated NSCLC
  • EGFR TKI should be stopped at the time when patient starts ChT for treatment of TKI resistance [I, A]

  • All tumours with clinical evidence of EGFR TKI resistance, not previously treated with osimertinib, should be tested for presence of EGFR exon 20 T790M mutation [I, A]

  • Liquid biopsy can be used as the initial test for detection of T790M mutation, and if tested negative, re-biopsy should be attempted if feasible [II, A]

  • Osimertinib is the standard therapy for patients whose tumours are tested positive for T790M either in liquid biopsy or re-biopsy, if not received previously [I, A; ESMO-MCBS v1.1 score: 4]

  • In EGFR-mutated NSCLC with CNS disease, osimertinib is highly active

  • Platinum-based doublet is the standard therapy for patients whose tumour is tested T790M negative in either re-biopsy or in liquid biopsy (only when re-biopsy is not feasible) [I, A]

  • Combination of atezolizumab and bevacizumab with carboplatin and paclitaxel should be considered as a therapeutic option in patients with EGFR-mutated tumour, PS 0–1, in absence of contraindications to use of immunotherapy after targeted therapies have been exploited [III, A; not EMA-approved]

First-line treatment of ALK-rearranged NSCLC
  • Patients with ALK-rearranged NSCLC should receive first-line ALK TKI including crizotinib [I, A; ESMO-MBCS v1.1 score: 4], ceritinib [I, B; ESMO-MCBS v1.1 score: 4], alectinib [I, A] or brigatinib [I, B; not EMA-approved]

  • Alectinib is associated with longer PFS and lower toxicity than crizotinib and showed activity against CNS disease in patients previously untreated with ALK-positive NSCLC [I, A]

  • Brigatinib is associated with longer PFS than crizotinib at the first interim analysis and showed activity against CNS disease in previously untreated patients with ALK-positive NSCLC [I, B, not EMA approved]

  • In patients with CNS involvement, front-line use of ALK TKIs is effective, and alectinib [III, A], brigatinib [III, B] or ceritinib [IV, B] are recommended. Ceritinib represents a better treatment strategy than ChT [I, B] and presumably crizotinib [IV, B]; alectinib represents a better treatment option than crizotinib [I, A]; brigatinib represents a better treatment option than crizotinib [I, B; not EMA-approved]

  • In ALK-rearranged NSCLC patients with localised distant progression and ongoing systemic control, continuation of treatment with ALK TKI in combination with local treatment of progressing metastatic sites may be considered [III, B]

Second and further lines of treatment of ALK-rearranged NSCLC
  • Any patient with NSCLC harbouring an ALK fusion should receive crizotinib as next-line therapy, if not received previously [I, A]

  • Ceritinib and alectinib are recommended in patients with ALK-positive advanced NSCLC who progress on treatment with or are intolerant to crizotinib [I, A; ESMO-MBCS v1.1 score: 4]

  • In patients with ALK-positive NSCLC progressing on crizotinib with CNS progression, treatment should be a next-generation ALK TKIs such as alectinib or ceritinib [I, A]. In patients who progress after a second-generation ALK TKI, the next-generation ALK inhibitors such as brigatinib or lorlatinib are an option if available [III, B]. They are currently not approved by the EMA

Treatment of ROS1-rearranged NSCLC
  • Crizotinib is recommended in the first-line setting in patients with stage IV NSCLC with ROS1 rearrangement [III, A; ESMO-MBCS v1.1 score: 3]

  • In patients with ROS1-positive NSCLC, who have not received crizotinib in the first-line setting, single-agent crizotinib may be offered as second-line therapy [III, A]

  • Ceritinib might be considered in crizotinib-naive patients but is currently not approved by the EMA [III, C]

  • If patients have received crizotinib in the first-line setting, then they may be offered platinum-based ChT therapy in the second-line setting [IV, A]

Treatment of BRAF-mutated NSCLC
  • Patients with stage IV NSCLC with BRAF V600 mutation should be exposed in first or second line to BRAF/MEK inhibition using dabrafenib/trametinib [III, A; ESMO-MBCS v1.1 score: 2]

  • If patients have received BRAF/MEK inhibition in the first-line setting, then they may be offered platinum-based ChT in the second-line setting [IV, A]

Patients with NSCLC with other actionable oncogenic driver
  • Targeting RET is not currently routinely recommended and recruitment into open trials is encouraged [III, C]

  • Targeting MET amplification is not currently routinely recommended and recruitment into open trials is encouraged [III, C]

  • Targeting METex14 variants (while evidence of benefit is stronger) is not currently routinely recommended and recruitment into open trials is encouraged [III, C]

  • Crizotinib has demonstrated potential clinical efficacy for METex14 variant NSCLC that needs to be confirmed [III, C]

  • Given the paucity of robust data, targeting HER2 dysregulation is not currently recommended and recruitment into open trials is encouraged [III, C]

  • Targeting NRTK fusions is not currently recommended and recruitment into open trials is encouraged [III, C]

Role of RT in stage IV
  • EBRT is indicated in cases of haemoptysis and symptomatic airway obstruction [III, B]

  • RT can achieve symptom control for a variety of clinical scenarios including haemoptysis, symptomatic airway obstruction, painful chest wall disease and bone metastasis, superior vena cava syndrome, soft tissue or neural invasion [II, B]

  • Administration of high-dose RT does not result in greater levels of palliation [II, B]

  • EBRT alone is more effective for palliation than EBB alone [II, B]

  • For patients previously treated by EBRT who are symptomatic from recurrent endobronchial central obstruction, EBB may be considered in selected cases [III, C]

  • Neurological symptoms from spinal cord compression can be relieved by early RT [II, B]

Brain metastases
  • WBRT can be considered in selected patients, contingent on prognostic factors of better survival [III, C]. WBRT should not be offered in RPA class III patients in view of the dismal prognosis [I, A]; only BSC is recommended

  • The most frequent WBRT schedules are 20 Gy in 5 fractions or 30 Gy in 10 fractions, with no difference in outcome [I, A]

  • For most patients with symptomatic brain metastases and/or significant oedema, dexamethasone or equivalent corticosteroid is recommended [III, A]

  • Neuroprotective agents are not recommended for routine use [II, C]

  • Hippocampus avoidance WBRT is not currently recommended as a standard treatment [III, C]

  • In case of single brain metastases surgical resection can be considered [III, B]

  • Postoperative WBRT or SRS is recommended after surgical resection [I, A]

  • In the case of a limited number of metastasis, SRS alone is the recommended treatment in patients with RPA class I–II [III, B]

  • SRS alone, without WBRT but with close MRI brain imaging follow-up, is an alternative strategy [III, B]

  • The indication for SRS is based on total tumour volume rather than numbers of metastases, as the risk of radionecrosis increases with tumour volume [III, C]

  • In patients with asymptomatically detected CNS metastases at presentation, systemic therapy with deferred RT should be considered due to similar intracranial and extra cranial response [II, B]

  • In patients with an actionable oncogenic driver (e.g. EGFR, ALK) and clinically asymptomatic brain metastases, next-generation TKIs may restore control of brain disease and delay cranial RT [II, A]

  • There is currently limited trial data demonstrating safety and efficacy of immunotherapy in patients with small-volume untreated CNS metastases [III, B]

LM carcinomatosis
  • A high index of suspicion should be borne for LM involvement especially in patients with actionable oncogenic drivers having TKI treatment [V]. CSF sampling is diagnostic of LMD but limited by low sensitivity, albeit with high specificity [IV, A]

  • Patients with actionable oncogenic drivers and LMD can be treated with CNS-penetrant next-generation TKIs [III, B]

  • ChT and bevacizumab may have activity both extra-cranially and intra-cranially, and also in the context of LMD [IV, C]

  • Intra-CSF pharmacotherapy can be considered contingent on clinical factors [V, C]

  • In exceptional cases, focal RT can be considered for circumscribed, notably symptomatic, lesions [V, C]

Surgery in stage IV
  • Surgery may be indicated for diagnosis, evaluation of response to systemic therapy and palliation

  • Highly selected patients may be considered for lung resection with therapeutic intent or even for a salvage procedure [IV, C]

  • When metastatic disease is suspected on PET scanning, invasive surgical procedures such as incisional biopsies, mediastinoscopy, thoracoscopy (VATS) or laparoscopy may be required to obtain relevant biopsy samples. Adequate samples should be provided to the pathologist for detailed routine staining, IHC and molecular genetic testing [III, B]

  • Persisting or recurrent pleural effusions are usually managed by pleurodesis to improve dyspnoea. Talc is the preferred agent and thoracoscopic poudrage may be better than injection of talc slurry in patients with primary lung cancer [II, B]

  • In case of a trapped lung by a thickened visceral pleural peel, indwelling pleural catheters or pleuroperitoneal shunts provide symptomatic relief [IV, B]

Treatment of oligometastatic disease
  • Stage IV patients with one to three synchronous metastases at diagnosis may experience long-term DFS following systemic therapy and local consolidative therapy (high-dose RT or surgery) [III, B]. Because of the limited evidence, these patients should be discussed within a multidisciplinary tumour board [II, B], and inclusion in clinical trials is preferred

  • Although operative risk is low and long-term survival may be achieved, current evidence for surgery in oligometastatic disease is limited, and the relative contribution of surgery versus RT as local treatment modality has not been established yet

  • Stage IV patients with limited metachronous metastases may be treated with a radical local therapy (high-dose RT or surgery) and may experience long-term DFS [IV, B]. However, this is based mainly on retrospective data and inclusion in clinical trials is preferred

  • Stage IV patients with driver mutations, with oligoprogression while on molecular-targeted therapy, may be treated with a radical local treatment (high-dose RT or surgery) and may experience long-term DFS [IV, C]. However, this is based mainly on retrospective data and inclusion in clinical trials is preferred

  • Solitary lesions in the contralateral lung should, in most cases, be considered as synchronous secondary primary tumours and, if possible, treated with curative-intent therapy [IV, B]

Bone metastases
  • Zoledronic acid reduces SREs (pathological fracture, radiation/surgery to bone or spinal cord compression) [II, B] and is recommended in stage IV bone metastatic disease [I, B]

  • Denosumab shows a trend towards superiority to zoledronic acid in lung cancer in terms of SRE prevention [II, B] and is recommended in selected patients with advanced lung cancer with bone metastases [I, B]

  • In the case of uncomplicated painful bone metastases, single fraction EBRT is the recommended treatment on the basis of non-inferiority to multiple fraction RT [I, A]

Role of minimally invasive procedures in stage IV NSCLC
  • In case of symptomatic major airways obstruction or post-obstructive infection, endoscopy debulking by laser, cryotherapy or stent placement may be helpful [III, C]

  • Endoscopy is useful in the diagnosis and treatment (endobronchial or by guiding endovascular embolisation) of haemoptysis [III, C]

  • Vascular stenting might be useful in NSCLC-related superior vena cava compression [III, B]

Palliative care in stage IV NSCLC
  • Early palliative care intervention is recommended, in parallel with standard oncological care [I, A]

Follow-up
  • Close follow-up, at least every 6–12 weeks after first-line therapy, is advised to allow for early initiation of second-line therapy but should depend on individual retreatment options [III, B]

ALK, anaplastic lymphoma kinase; BSC, best supportive care; CEA, carcinoembryonic antigen; cfDNA, cell-free DNA; ChT, chemotherapy; CNS, central nervous system; CSF, cerebrospinal fluid; CT, computed tomography; DFS, disease-free survival; EBB, endobronchial brachytherapy; EBRT, external beam radiotherapy; EBUS, endobronchial ultrasound; EGFR, epidermal growth factor receptor; EMA, European Medicines Agency; ESMO-MCBS, European Society for Medical Oncology-Magnitude of Clinical Benefit Scale; EUS, endoscopic ultrasound; FDA, Food and Drug Administration; FISH, fluorescent in situ hybridisation; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry; imRECIST, immune-modified RECIST; iRECIST, immune RECIST; irRECIST, immune-related RECIST; LM, leptomeningeal; LMD, leptomeningeal disease; MEK, mitogen-activated protein kinase; MRI, magnetic resonance imaging; nab-P, albumin-bound paclitaxel; nab-PC, albumin-bound paclitaxel/carboplatin; NGS, next-generation sequencing; NSCC, non-squamous cell carcinoma; NSCLC, non-small cell lung cancer; NSCLC-NOS, non-small cell lung cancer-not otherwise specified; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PET, positron emission tomography; PFS, progression-free survival; PS, performance status; QoL, quality of life; RECIST, Response Evaluation Criteria in Solid Tumours; RPA, recursive partitioning analysis; RT, radiotherapy; SCC, squamous cell carcinoma; SRE, skeletal-related event; SRS, stereotactic radiosurgery; TKI, tyrosine kinase inhibitor; TMB, tumour mutational burden; UICC, Union for International Cancer Control; VATS, video-assisted thorascopic surgery; WBRT, whole-brain radiotherapy; WHO, World Health Organization; WT, wild-type.

Therapy Disease setting Trial Control Absolute survival gain HR (95% CI) QoL/toxicity ESMO-MCBS scoreb
Afatinib, an irreversible ErbB family blocker Advanced
  • Afatinib versus erlotinib as second-line treatment of patients with advanced squamous cell carcinoma of the lung (LUX-Lung 8): an open-label randomised controlled phase III trial [168, 169]

  • Phase III

  • NCT01523587

  • Erlotinib, as second-line treatment of patients with advanced SCC of the lung

  • Control median OS: 6.8 months

OS gain: 1.1 months OS: HR for death 0.81 (0.69–0.95)
  • Similar toxicity profile

  • Improved overall health-related QoL

2 (Form 2a)
Alectinib, potent ALK tyrosine kinase inhibitor Advanced
  • Alectinib versus chemotherapy in crizotinib-pretreated ALK-positive non-small-cell lung cancer: results from the phase III ALUR study [207]

  • Phase III

  • NCT02604342

  • ChT (pemetrexed or docetaxel) in previously treated ALK-rearranged patients

  • Control PFS (investigator assessment): 1.4 months

PFS gain: 8.2 months PFS: HR 0.15 (0.08–0.29) Improved toxicity profile 4 (Form 2b)
Alectinib, potent ALK tyrosine kinase inhibitor Advanced
  • Alectinib versus crizotinib in untreated ALK-positive NSCLC [202]

  • Phase III

  • NCT02075840

  • Crizotinib in untreated, ALK-rearranged patients

  • Control PFS, investigator-assessed: not reached

  • Control PFS, independent review committee-assessed: 10.4 months

PFS gain (independent review committee-assessed): 15.3 months PFS (independent review committee-assessed): HR 0.50 (0.36–0.70) Improved toxicity profile 4 (Form 2b)
Atezolizumab, humanised engineered IgG1 monoclonal antibody targeting PD-L1 Advanced
  • Atezolizumab versus docetaxel in patients with previously treated NSCLC (OAK): a phase III, open-label, multicentre randomised controlled trial [149]

  • Phase III

  • NCT02008227

  • Docetaxel in squamous or non-squamous patients stage IIIB or IV who had received one to two previous cytotoxic ChT regimens

  • Control OS: 9.6 months

OS gain: 4.2 months OS: HR 0.73 (0.62–0.87) Improved toxicity profile 5 (Form 2a)
Bevacizumab, a humanised anti-VEGF monoclonal antibody, in combination with erlotinib Advanced
  • Erlotinib alone or with bevacizumab as first-line therapy in patients with advanced non-squamous NSCLC harbouring EGFR mutations (JO25567): an open-label, randomised, multicentre, phase II study [190]

  • Phase II

  • Japan Pharmaceutical Information Center, number Japic CTI-111390

  • Erlotinib alone as a first-line therapy until disease progression or unacceptable toxicity

  • Control median PFS: 9.7 months

PFS gain: 6.3 months PFS: HR 0.54 (0.36–0.79)
  • Deteriorated toxicity profile not reaching the toxicity thresholds for penalty

  • No improvement in QoL

3 (Form 2b)
Ceritinib, potent and selective oral tyrosine kinase inhibitor of ALK Advanced
  • Ceritinib versus chemotherapy in patients with ALK-rearranged NSCLC previously given chemotherapy and crizotinib (ASCEND-5): a randomised, controlled, open-label, phase III trial [206]

  • Phase III

  • NCT01828112

  • ChT, pemetrexed or docetaxel (investigator choice), in patients with ALK-rearranged stage IIIB or IV

  • Control PFS: 1.6 months

PFS gain: 3.8 months PFS: HR 0.49 (0.36–0.67)
  • Similar treatment related serious adverse events

  • Improved overall health-related QoL

4 (Form 2b)
Ceritinib, potent and selective oral tyrosine kinase inhibitor of ALK Advanced
  • First-line ceritinib versus platinum-based chemotherapy in advanced ALK-rearranged NSCLC (ASCEND-4): a randomised, open-label, phase III study [200]

  • Phase III

  • NCT01828099

  • Platinum-based ChT in untreated patients stage IIIB/IV ALK-rearranged non-squamous NSCLC

  • Control PFS: 8.1 months

PFS gain: 8.5 months PFS: HR 0.55 (0.42–0.73) Delayed deterioration in overall health-related QoL 4 (Form 2b)
Crizotinib, a small-molecule tyrosine kinase inhibitor of ALK, ROS1 and MET Advanced First-line crizotinib versus chemotherapy in ALK-positive lung cancer [199] Phase III NCT01154140
  • Pemetrexedplus platinum ChT

  • Control PFS: 7.0 months

PFS gain: 3.9 months PFS: HR 0.45 (0.35–0.60) Improved QoL 4 (Form 2b)
Crizotinib, a small-molecule tyrosine kinase inhibitor of ALK, ROS1 and MET Advanced
  • Crizotinib in ROS1-rearranged NSCLC [216]

  • Phase I (expansion cohort)

  • NCT000585195

Cohort study: 50 patients (86% had received at least one previous line) (no control) 72% achieved an overall response and mPFS was 19.2 months ORR: 72% (58–84) mPFS: 19.2 months (14.4–not reached) 3 (Form 3)
Dabrafenib, a selective inhibitor of mutated forms of BRAF kinase and trametinib, a MEK1/MEK2 inhibitor Advanced
  • Dabrafenib plus trametinib in patients with previously untreated BRAF V600E-mutant metastatic NSCLC: an open-label, phase II trial [230]

  • Phase II

  • NCT01336634

Cohort study: 36 patients (no control) Independent review committee-assessed confirmed overall response: 64% mPFS: 10.9 months ORR: 64%, (46–79) mPFS: 10.9 months (7.0–16.6) Serious adverse events: 57% 2 (Form 3)
Dabrafenib, a selective inhibitor of mutated forms of BRAF kinase and trametinib, a MEK1/MEK2 inhibitor Advanced
  • Dabrafenib plus trametinib in patients with previously treated BRAF V600E-mutant metastatic NSCLC: an open-label, multicentre phase II trial [228]

  • Phase II

  • NCT01336634

Cohort study: 57 patients (no control) Independent review committee–assessed confirmed overall response: 63.2% mPFS: 9.7 months ORR: 63.2% (49.3–75.6) mPFS: 9.7 months (6.9–19.6) Serious adverse events: 56% 2 (Form 3)
Erlotinib, an EGFR TKI Advanced
  • Erlotinib as maintenance treatment in advanced NSCLC: a multicentre, randomised, placebo-controlled phase III study [127]

  • Phase III

  • NCT00556712

  • Placebo, as maintenance treatment in advanced NSCLC

  • Control PFS: 11.1 weeks

PFS gain: 1.2 weeks PFS: HR 0.71 (0.62–0.82) Deteriorated toxicity profile 1 (Form 2b)
Necitumumab, a second-generation, recombinant, human IgG1 EGFR antibody in combination with gemcitabine and cisplatin Advanced
  • Necitumumab plus gemcitabine and cisplatin versus gemcitabine and cisplatin alone as first-line therapy in patients with stage IV squamous NSCLC (SQUIRE): an open-label, randomised, controlled phase III trial [115]

  • Phase III

  • NCT00981058

  • Gemcitabine and cisplatin as first-line therapy in patients with stage IV SCC

  • Control OS: 9.9 months

OS gain: 1.6 months OS: HR for death 0.84 (0.74–0.96) Deteriorated toxicity profile 1 (Form 2a)
Nivolumab, a fully human IgG4 PD-1 immune checkpoint inhibitor antibody Advanced
  • Nivolumab versus docetaxel in advanced non-squamous NSCLC [148, 323]

  • Phase III

  • NCT01673867

  • Docetaxel in patients with NSCC that had progressed during or after platinum-based doublet ChT

  • Control OS: 9.4 months

  • OS gain: 2.8 months

  • 2-year survival gain: 13%

  • 2-year OS: 29% versus 16%

OS: HR for death 0.73 (0.59–0.89) Improved toxicity profile 5 (Form 2a)
Nivolumab, a fully human IgG4 PD-1 immune checkpoint inhibitor antibody Advanced
  • Nivolumab versus docetaxel in advanced squamous-cell NSCLC [147]

  • Phase III

  • NCT01642004

  • Docetaxel in patients with advanced SCC who have disease progression during or after first-line ChT

  • Control OS: 6 months

OS gain:
  • 3.2 months

  • 2-year survival gain: 15%.

  • 2-year OS: 23% versus 8%

OS: HR for death 0.59 (0.44–0.79) Improved toxicity profile 5c (Form 2a)
Osimertinib, oral, irreversible EGFR TKI, selective for both EGFR and T790M resistance mutations Advanced
  • Osimertinib in untreated EGFR-mutated advanced NSCLC [187]

  • Phase III

  • NCT02296125

  • Gefitinib or erlotinib in patients with previously untreated, EGFR mutation (exon 19 deletion or L858R)

  • Control PFS: 10.2 months

PFS gain: 8.7 months PFS: HR 0.46 (0.37–0.57) Improved toxicity profile 4 (Form 2b)
Osimertinib, oral, irreversible EGFR TKI, selective for both EGFR and T790M resistance mutations Advanced
  • Osimertinib or platinum-pemetrexed in EGFR T790M-positive lung cancer [195]

  • Phase III

  • NCT02151981

  • Pemetrexed plus either carboplatin or cisplatin in patients with T790M-positive, who had disease progression after first-line EGFR TKI therapy

  • Control PFS: 4.4 months

PFS gain: 5.7 months PFS: HR 0.30 (0.23–0.41)
  • Improved toxicity profile

  • Improved patient-reported outcomes

4 (Form 2b)
Pembrolizumab, an anti-PD-1 monoclonal antibody Advanced
  • Pembrolizumab versus docetaxel for previously treated, PD-L1-positive, advanced NSCLC (KEYNOTE-010): a randomised controlled trial [63]

  • Phase III

  • NCT01905657

  • Docetaxel in patients with previously treated, PD-L1-positive, advanced NSCLC

  • Control OS: 8.5 months

2-year OS rates of 14.5% for docetaxel versus 30.1% for pembrolizumab (2 mg/kg) HR 0.71 (0.58–0.88) Improved toxicity profile 5 (Form 2a)
Pembrolizumab, humanised, IgG4 monoclonal antibody against PD-1 Advanced
  • Pembrolizumab versus ChT for PD-L1-positive NSCLC [62, 94]

  • Phase III

  • NCT02142738

  • Investigator’s choice of platinum-based ChT in stage IV untreated patients with PD-L1 expression on at least 50% of tumour cells

  • Control OS: 14.2 months

OS gain: 15.8 months OS: HR 0.63 (0.47–0.86) Improved toxicity profile 5d (Form 2a)
Pembrolizumab, an anti-PD-1 monoclonal antibody Advanced
  • Pembrolizumab/pemetrexed with platinum ChT in metastatic non-squamous NSCLC without EGFR or ALK mutations [96] Phase III NCT02578680

  • Pemetrexed/platinum ChT alone in patients with advanced or mNSCLC who have not previously received systemic therapy for advanced disease Control OS: 11.3 months

OS gain above the cut-off of 3 months OS: HR 0.49 (0.38-0.64) Improved QoL with delayed deterioration 4 (Form 2a)e
Ramucirumab, a human IgG1 monoclonal antibody that targets the extracellular domain of VEGFR2, in combination with docetaxel Advanced
  • Ramucirumab plus docetaxel versus placebo plus docetaxel for second-line treatment of stage IV NSCLC after disease progression on platinum-based therapy (REVEL): a multicentre, double-blind, randomised phase III trial [157]

  • Phase III

  • NCT01168973

  • Placebo plus docetaxel in patients with SCC or NSCC who had progressed during or after a first-line platinum-based ChT regimen

  • Control OS: 9.1 months

OS gain: 1.4 months OS: HR for death 0.86 (0.75–0.98) No change 1 (Form 2a)

ALK, anaplastic lymphoma kinase; ChT, chemotherapy; CI, confidence interval; EGFR, endothelial growth factor receptor; EMA, European Medicines Agency; ESMO-MCBS, ESMO-Magnitude of Clinical Benefit Scale; HR, hazard ratio; IgG, immunoglobulin G; mPFS, medical progression-free survival; NSCC, non-squamous cell carcinoma; NSCLC, non-small cell lung cancer; ORR, objective response rate; OS, overall survival; PD-1, programmed cell death protein 1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; QoL, quality of life; SCC, squamous cell carcinoma; TKI, tyrosine kinase inhibitor; VEGF, vascular endothelial growth factor; VEGFR2, vascular endothelial growth factor receptor 2.

Levels of evidence
I Evidence from at least one large randomised, controlled trial of good methodological quality (low potential for bias) or meta-analyses of well-conducted randomised trials without heterogeneity
II Small randomised trials or large randomised trials with a suspicion of bias (lower methodological quality) or meta-analyses of such trials or of trials with demonstrated heterogeneity
III Prospective cohort studies
IV Retrospective cohort studies or case–control studies
V Studies without control group, case reports, expert opinions
Grades of recommendation
A Strong evidence for efficacy with a substantial clinical benefit, strongly recommended
B Strong or moderate evidence for efficacy but with a limited clinical benefit, generally recommended
C Insufficient evidence for efficacy or benefit does not outweigh the risk or the disadvantages (adverse events, costs, ...), optional
D Moderate evidence against efficacy or for adverse outcome, generally not recommended
E Strong evidence against efficacy or for adverse outcome, never recommended

DP has reported honoraria from AstraZeneca, Bristol-Myers Squibb, Boehringer Ingelheim, Celgene, Eli Lilly, Merck, Merck Sharp and Dohme Oncology, Novartis, Pfizer, prIME Oncology, Roche; consulting, advisory role or lectures for AstraZeneca, Bristol-Myers Squibb, Boehringer Ingelheim, Celgene, Eli Lilly, Merck Sharp and Dohme Oncology, Novartis, Pfizer, prIME Oncology, Roche and has received travel grants from AstraZeneca, Roche, Novartis, prIME Oncology and Pfizer; SPo has reported honoraria from Pfizer, Boehringer Ingelheim, AstraZeneca, Roche, Lilly, Novartis, Takeda, Guardant Health, Bristol-Myers Squibb and consulting/advisory role for Boehinger Ingleheim, Roche, Lilly, Novartis, Pfizer and research funding from Boehringer Ingelheim, Epizyme, Bristol-Myers Squibb, Clovis Oncology, Roche, Lilly, Takeda; KMK has reported lecture fees and consultancy for AstraZeneca, AbbVie, Boehringer Ingelheim, Bristol-Myers Squibb, Lilly, Merck Sharpe & Dohme, Merck Serono, Novartis, Pfizer, Roche, Roche Diagnostics; SN has reported membership of the speaker bureau of Eli Lilly, Bristol-Myers Squibb, Merck Sharpe & Dohme, AstraZeneca, Boehringer Ingelheim, Roche, Incyte, Takeda; CFF has reported research/travel funding from AstraZeneca and Merck Sharpe & Dohme; TM has reported holding stock in Sanomics Ltd. and Hutchison Chi-Med, conducting research sponsored by AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Clovis oncology, Merck Sharp and Dohme, Novartis, Pfizer, Roche, SFJ Pharmaceuticals and XCovery; has received speaker's fee from AstraZeneca, Roche/Genentech, Pfizer, Eli Lilly, Boehringer Ingelheim, Merck Sharp and Dohme, Novartis, BMS, Taiho, Takeda Oncology, prIME Oncology and Amoy Diagnostics Co, LTD and honoraria from AstraZeneca, Boehringer Ingelheim, Roche/Genentech, Pfizer, Eli Lilly, Merck Sorono, Merck Sharp and Dohme, Novartis, SFJ Pharmaceuticals, ACEA Biosciences Inc, Vertex Pharmaceuticals, Bristol-Myers Squibb, OncoGenex Technologies Inc, Celgene, Ignyta Inc, Cirina, Fishawack Facilitate Ltd, Janssen, Takeda, Hutchison Chi-Med, OrigiMed, Henfrui Therapeutics Inc, Sanofi-Aventis R&D and Yuhan Corporation for attending advisory boards; MH has reported honoraria for consultancy for Roche/Genentech, AstraZeneca, Merck, Bristol–Myers Squibb, Janssen, Mirati, Syndax, Shattuck Labs and has received research funding from Bristol-Myers Squibb; MR has reported honoraria for lectures and consultancy from AstraZeneca, Bristol-Myers Squibb, Celgene, Boehringer Ingelheim, Novartis, Abbvie, Pfizer, Merck Sharpe & Dohme, Merck, Roche, Lilly; SPe has reported educational grants, consultation, advisory boards and/or lectures for Amgen, AstraZeneca, Boehringer-Ingelheim, Bristol-Myers Squibb, Clovis, Eli Lilly, F. Hoffmann-La Roche, Janssen, Merck Sharp & Dohme, Novartis, Merck Serono, Pfizer, Regeneron and Takeda; PEVS has reported no conflicts of interest. ES has not reported any potential conflicts of interest.

46

Kalemkerian

GP

,

Narula

N

,

Kennedy

EB

et al.

Molecular testing guideline for the selection of patients with lung cancer for treatment with targeted tyrosine kinase inhibitors: American Society of Clinical Oncology Endorsement of the College of American Pathologists/International Association for the Study of Lung Cancer/Association for Molecular Pathology Clinical Practice Guideline Update

.

J Clin Oncol

2018

;

36

:

911

919

.

133

Patel

JD

,

Socinski

MA

,

Garon

EB

et al.

PointBreak: a randomized phase III study of pemetrexed plus carboplatin and bevacizumab followed by maintenance pemetrexed and bevacizumab versus paclitaxel plus carboplatin and bevacizumab followed by maintenance bevacizumab in patients with stage IIIB or IV nonsquamous non-small-cell lung cancer

.

J Clin Oncol

2013

;

31

:

4349

4357

.


The incidence of hepatocellular carcinoma (HCC) has been rising worldwide over the last 20 years and is expected to increase until 2030 in some countries including the United States, while in other countries, such as Japan, the incidence has started to decline [1–3]. In 2012, liver cancer represented the fifth most common cancer in men (554 000 new cases) and the ninth in women (228 000 new cases) and the second most common cause of cancer-related death (746 000 estimated deaths), worldwide [3]. The incidence varies from 3/100 000 in Western countries, to 78.1/100 000 in Mongolia, with the highest incidence in Africa and Asia, mapping the geographical distribution of viral hepatitis B (HBV) and hepatitis C (HCV), the most important causes of chronic liver disease and HCC [4]. In Europe, in 2012 the estimated incidence rate was 10.0 in men and 3.3 in women per 100 000, respectively, while the estimated mortality rate was 9.1 and 3.3 per 100 000 in men and women, respectively [3]. The incidence of HCC shows a strong male preponderance and increases progressively with advancing age in all populations. The association of chronic liver disease and HCC represents the basis for preventive strategies, including universal vaccination at birth against HBV [I, A] [5] and early antiviral treatment of viral HBC and HCV [III, A] [6–8].

The prevalence of obesity and type 2 diabetes has greatly increased in the past decades, leading to a rising incidence of non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH), which can lead to fibrosis and cirrhosis and, eventually, HCC [9]. HCC related to NAFLD/NASH is probably underestimated [10] and is expected to rise in the future, possibly overtaking the other aetiologies in some areas of the world [11]. A significant proportion of patients with NAFLD/NASH-associated HCC do not have histological evidence of cirrhosis [12].

The control of other risk factors for chronic liver disease and cancer is more difficult to implement, such as cutting down on the consumption of alcohol and programmes aiming at a healthier lifestyle in the light of the obesity pandemic [13, 14]. In Africa, reduction of exposure to aflatoxin B1, especially in HBV-infected individuals, may lower the risk of HCC. HCC may evolve from subclasses of adenomas; in < 10% of cases HCC occurs in an otherwise normal liver.

Surveillance of HCC involves the repeated application of screening tools in patients at risk for HCC and aims for the reduction in mortality of this patient population. The success of surveillance is influenced by the incidence of HCC in the target population, the availability and acceptance of efficient diagnostic tests and the availability of effective treatment. Cost-effectiveness studies suggest surveillance of HCC is warranted in all cirrhotic patients irrespective of its aetiology [15], as long as liver function and comorbidities allow curative or palliative treatments [III, A]. Surveillance of non-cirrhotic, hepatitis-infected patients should also be considered in chronic HBV carriers or HCV-infected patients with bridging fibrosis (F3, numerous septa without cirrhosis) [III, A], which are at higher risk than the general population. Specifically in Asian patients, serum HBV-DNA above 10 000 copies/mL was associated with a higher annual risk (above 0.2%/year) compared with patients with a lower viral load [16]. Patients with HCV infection and advanced fibrosis remain at increased risk for HCC even after achieving sustained virological response following antiviral treatment [III, A] [8] and, thus, should remain in a surveillance programme.

Japanese cohort studies have shown that surveillance by abdominal ultrasound (US) resulted in an average size of the detected tumours of 1.6 ± 0.6 cm, with < 2% of the cases exceeding 3 cm [17]. In the Western world and in less experienced centres, the sensitivity of finding early-stage HCC by US is considerably less effective [18]. There are no data to support the use of contrast-enhanced computed tomography (CECT) or contrast-enhanced magnetic resonance imaging (CEMRI) for surveillance. Adding the determination of serum alpha foetoprotein (AFP) to US can lead to a 6% gain in the early HCC detection rate, but at the price of false-positive results and of a worse cost-effectiveness ratio [19]. A randomised controlled trial (RCT) of Chinese patients with chronic HBV infection compared surveillance (US and serum AFP measurements every 6 months) versus no surveillance [20]. Despite low compliance with the surveillance program (55%), HCC-related mortality was reduced by 37% in the surveillance arm. Considering the most appropriate surveillance interval, a randomised study comparing a 3- versus 6-month schedule failed to detect any differences [21].

Surveillance of patients at risk for HCC should be carried out by abdominal US every 6 months with or without AFP [II, A].

The diagnosis of HCC is based on histological analysis and/or contrast-enhanced imaging findings [III, A]. The diagnostic work-up of a patient with an HCC-suspicious nodule is given in Table 1.

History and clinical examination
  • Risk factors for chronic liver disease: i.v. drug abuse, alcohol intake, metabolic syndrome (obesity, diabetes, arterial hypertension)

  • Symptoms and signs of chronic liver disease (jaundice, ascites, encephalopathy, bleeding, splenomegaly)

  • PS (distinguish cancer-related symptoms of recent onset with long-standing symptoms associated with cirrhosis) and nutritional state

Laboratory analysis
  • Aetiology of liver disease: HBV (at least HBsAg and anti-HBc), HCV (at least anti-HCV), iron status, autoimmune disease

  • Liver function: prothrombin, albumin, bilirubin

  • Complete blood cell count including platelets

  • Tumour marker: serum AFP

Assessment of portal hypertension
  • Upper endoscopy: varices and/or hypertensive gastropathy

  • Optional: transjugular measurement of hepatic-venous pressure gradient

Imaging studies
  • Liver dynamic (multiple phase) MRI or CT studies for diagnosis and evaluation of tumour extent inside the liver (number and size of nodules, vascular invasion, extrahepatic spread)

  • CEUS can also be used for the non-invasive diagnosis of HCC if CT scan or MRI are not possible, but is not considered appropriate for tumour staging

  • CT of the chest, abdomen and pelvis to rule out extrahepatic spread

Tumour biopsy
  • Useful for nodules with non-diagnostic at imaging

  • Required to diagnose HCC in non-cirrhotic liver

  • Should be carried out according to national or institutional policy in all clinical trials and may support centre-based innovative treatment approaches

  • Ideally, should evaluate tumour and non-tumour tissue when used for scientific purposes

In patients with liver cirrhosis and specific imaging criteria, a formal pathological proof is not mandatory for diagnosis and the clinician can rely on the contrast-enhanced imaging criteria for lesion characterisation [22–24]. These criteria require a multi-phasic CECT or CEMRI. The diagnosis can be established if the typical vascular hallmarks of HCC (hypervascularity in the arterial phase with washout in the portal venous or delayed phase) are identified in a nodule of > 1 cm diameter using one of these two modalities [III, A]. Compared with multiple detector CT (MDCT), multiphasic MRI offers a moderate increase in sensitivity for diagnosing HCC based on the typical vascular hallmarks [III, B] [24–27]. Serum AFP has no role in the diagnostic algorithm [III, A].

Based on techniques such as diffusion-weighted imaging and the use of hepatobiliary contrast agents, MRI may identify and stratify nodules as high-risk nodules (either HCC not displaying the typical imaging hallmarks features or high-grade dysplastic nodules) [IV, B] [28–31]. However, the impact of identification of additional nodules by diffusion-weighted imaging and hepatobiliary contrast agents on the therapeutic algorithm remains unclear and switching to palliative treatments after identification of potential premalignant nodules by these new techniques should be avoided. New imaging criteria for HCC diagnosis called CT/MRI LI-RADS® v2018 (Liver Imaging Reporting and Data System) include arterial phase enhancement, tumour size, washout, enhancing capsule and threshold growth and have been proposed to improve the diagnosis of HCC, especially for small nodules (Table 2) [32, 33].

Untreated observation without pathological proof in patient at high risk for HCC
  • Definitely benign: LR-1

  • Probably benign: LR-2

  • Not categorisable, due to image degradation or omission: LR-NC

  • Definite tumour in vein (TIV): LR-TIV

  • Probably or definitely malignant but not HCC specific (e.g. if targetoid): LR-M

Otherwise, use CT/MRI diagnostic table below
  • Intermediate probability of malignancy: LR-3

  • Probably HCC: LR-4

  • Definitely HCC: LR-5

Arterial phase hyperenhancement (APHE) No APHE APHE (not rim)
Observation size (mm) < 20 ≥ 20 < 10 10-19 ≥ 20
Count major features:
  • ‘Washout’ (not peripheral)

  • Enhancing ‘capsule’

  • Threshold growtha

None LR-3 LR-3 LR-3 LR-3 LR-4
One LR-3 LR-4 LR-4 LR-4/LR-5b LR-5
≥ Two LR-4 LR-4 LR-4 LR-5 LR-5

For contrast-enhanced US (CEUS), an overlap between the vascular profile of HCC and cholangiocarcinoma (CC) has been described. However, recent data suggest CEUS as a suitable technique to diagnose HCC non-invasively in the setting of liver cirrhosis [IV, B] [34–36]. The typical hallmarks for HCC at CEUS differ slightly to those of CT/MRI; at CEUS, hallmarks are arterial hyper-enhancement followed by late (> 60 s) washout of a mild degree.

Angiography and fluorodeoxyglucose-positron emission tomography (FDG-PET) scan are not recommended for HCC diagnosis. When tumour biopsy fails to demonstrate a correlate for a focal lesion, a second tumour biopsy, a different contrast-enhanced imaging modality or (if amenable) direct resection of the lesion may be considered according to tumour size [IV, B]. If the patient is a candidate for resection that can be carried out with an acceptable morbidity and mortality risk, then either biopsy or direct resection may be an option.

Pathological diagnosis of HCC is based on a biopsy or a surgical specimen of the tumour. Concomitant analysis of the non-tumour liver may be useful in order to define its status and potential causative diseases. Assessment of resection and explant specimen follows the valid TNM (tumour, node, metastasis) classification including resection margin evaluation. Usually tumour grade is provided, but currently no uniform grading scheme is used worldwide and data on the independent prognostic value are inconclusive.

Histopathological diagnosis of tumour biopsies relies on standard [e.g. haemotoxylin and eosin (H&E)] and special stains (e.g. reticulin), and—if required—immunohistochemistry (IHC). It should address different challenges: morphologically, highly differentiated HCC must be distinguished from benign/premalignant lesions (dysplastic nodules, hepatocellular adenoma, focal nodular hyperplasia). In particular, poorly differentiated HCC should be distinguished from intrahepatic CC, combined HCC/CC and some types of metastases (e.g. lung cancer, head and neck squamous cell carcinoma, breast cancer, neuroendocrine tumours). For this reason, histological analyses may be supplemented by IHC for lineage-specific markers. It is important to distinguish combined HCC/CC from HCC due to the different therapeutic modalities; however, the mixed differentiation features might not be visible in the biopsy. In addition, significant expression of cytokeratin 19 (CK19) has been evaluated and considered as a sign of poor prognosis in HCC [IV, B].

In highly differentiated HCC, definitive signs of malignancy (interstitial or vascular invasion) are frequently absent from biopsy. Further consented histological (trabecular alterations—more than two cell broad trabeculae, pseudoglands, reticulin loss, capsule formation) and cytological criteria (increased nuclear/cytoplasmic ratio, i.e. ‘nuclear crowding’, increased cytoplasmic basophilia) support HCC diagnosis [III, B] [37]. IHC should be carried out in unclear cases: capillarisation of sinusoids could be assessed using CD34 IHC [IV, B] [37]. Further immunohistochemical markers have been shown to improve the diagnosis of highly differentiated HCC, including glutamine synthetase, glypican 3, general stress protein (CTC), enhancer of zeste homologue 2 (EZH2) and heat shock protein 70 (HSP70) [IV, B]. A combination of the three markers glutamine synthetase, glypican 3 and HSP70 has been consented as a diagnostic panel (2/3 marker positivity has 70% sensitivity and 100% specificity for HCC) and the use of further markers seems to increase the sensitivity [IV, B] [38]. Moreover, histological subtypes of HCC have been defined (e.g. fibrolamellar, chromophobe, macrotrabecular massive) which specifically correlate with clinical and molecular features [39, 40], which may have future clinical impact.

It is now well accepted that the potential risks of tumour biopsy, bleeding and needle track seeding, are infrequent, manageable and do not affect the course of the disease or overall survival (OS) and, therefore, should not be seen as a reason to abstain from diagnostic liver biopsy. In a comprehensive meta-analysis, the risk of tumour seeding after liver biopsy was reported to be 2.7%, with a median time interval between biopsy and seeding of 17 months [41], but even lower rates are expected in experienced centres. It was reported that needle track seeding can be treated well (e.g. by excision or radiation) and did not affect outcome of oncological treatment [42] and OS [41]. In a meta-analysis of the bleeding risk, mild bleeding complications ranged around 3%–4%, while severe bleeding complications, requiring transfusions, were reported in 0.5% of the cases [43].

Staging of HCC is important to determine outcome and planning of optimal therapy and includes assessment of tumour extent, AFP level, liver function, portal pressure and clinical performance status (PS) (Table 1) [III, A]. Relevant techniques to evaluate tumour extent (number and size of nodules, vascular invasion, extrahepatic spread) include CEMRI or helical CT. CT of the chest, abdomen and pelvis is recommended to rule out extrahepatic spread. There is no justification for routine preoperative bone scintigraphy to detect asymptomatic skeletal metastases in patients with resectable HCC [44] and there are no data in the context of advanced HCC. There is no demonstrated clinical benefit of carrying out FDG-PET scan as a staging modality, despite some evidence that there is a correlation of higher FDG uptake with poor differentiation, tumour size, serum AFP levels and microvascular invasion [IV, D] [45, 46].

Liver function is classically assessed by the Child-Pugh scoring system (serum bilirubin, serum albumin, ascites, prothrombin time and hepatic encephalopathy) [III, A]. Within the Child-Pugh A group, measurement of the albumin-bilirubin (ALBI) score (a model incorporating serum albumin and bilirubin levels alone) is able to split that group into good prognosis (ALBI 1) and poor prognosis (ALBI 2), with median survivals of 26 versus 14 months, respectively [IV, B] [47]. A platelet count > 150 × 109 cells/L and a non-invasive liver stiffness measurement < 20 kPa excludes clinically significant portal hypertension (Baveno VI criteria) [48]. Otherwise, the finding of oesophageal varices and/or splenomegaly with blood platelet counts of 100 × 109 cells/L suggests clinically important portal hypertension, which can also be measured invasively by the transjugular route (hepatic-venous pressure gradient > 10 mmHg) [III, A].

Several staging systems—incorporating some or all of the above-mentioned items—have been developed, including TNM, Okuda, Cancer of the Liver Italian Program (CLIP), Japanese Integrated Staging (JIS) Score and the Barcelona Clinic Liver Cancer (BCLC) system. Every system has advantages and drawbacks. The recently released 8th edition of the TNM system (Table 3) contains changes to the T classifications compared with the previous staging system [49]. The staging system includes microvascular invasion that can only be assessed on pathology and is therefore less useful in clinical practice before treatment decision making. Moreover, a recent validation study pointed to potential problems of heterogeneity in the T2 category and the lack of vascular invasion as a prognostic factor in the T3 group [50]. TNM classification provides a means of standardising histopathological reports in patients treated by resection or transplantation.

T—primary tumour
TX Primary tumour cannot be assessed
T0 No evidence of primary tumour
T1a Solitary tumour 2 cm or less in greatest dimension with or without vascular invasion
T1b Solitary tumour more than 2 cm in greatest dimension without vascular invasion
T2 Solitary tumour with vascular invasion more than 2 cm dimension or multiple tumours, none more than 5 cm in greatest dimension
T3 Multiple tumours any more than 5 cm in greatest dimension
T4 Tumour(s) involving a major branch of the portal or hepatic vein with direct invasion of adjacent organs (including the diaphragm), other than the gallbladder or with perforation of visceral peritoneum
N—regional lymph nodes
NX Regional lymph nodes cannot be assessed
N0 No regional lymph node metastasis
N1 Regional lymph node metastasis
M—distant metastasis
M0 No distant metastasis
M1 Distant metastasis
Stage—liver
Stage IA T1a N0 M0
Stage IB T1b N0 M0
Stage II T2 N0 M0
Stage IIIA T3 N0 M0
Stage IIIB T4 N0 M0
Stage IVA Any T N1 M0
Stage IVB Any T Any N M1

The BCLC staging system was developed on the basis of the results of RCTs and cohort studies and links tumour stage, liver function, cancer-related symptoms and PS to an evidence-based treatment algorithm (Table 4). The system identifies those patients with early HCC who may benefit from ablative treatment (stage 0 and A), those at intermediate (stage B) or advanced stage (stage C) who may benefit from intra-arterial or systemic treatments and those with a very poor life expectancy (stage D). Survival without therapy is > 5 years for stage 0 and A, > 2.5 years for stage B, > 1 year for stage C and ∼3 months for stage D [51]. Treatment assignment of the different stages is discussed below. The aetiology of co-existent liver disease has not been identified as an independent prognostic factor. Nevertheless, finding a treatable underlying co-existent liver disease may be very relevant, e.g. antiviral treatment in case of HBV, corticosteroid treatment in autoimmune hepatitis or stopping alcohol intake may result in a marked improvement in liver function and improving prognosis.

Liver decompensation (including jaundice, variceal haemorrhage, ascites or encephalopathy) should be considered a contraindication for any locoregional therapy that may induce subclinical liver damage such as resection, percutaneous ablation or transarterial therapies. The benefit of systemic therapies has not been established in patients with liver decompensation.

Liver resection (LR), orthotopic liver transplantation (OLT) and local destruction methods [radiofrequency ablation (RFA) or microwave ablation (MWA)] comprise potentially curative treatment modalities for patients with HCC (see Figure 1). Selecting the appropriate treatment for the individual patient remains difficult and there are no randomised phase III trials comparing the efficacy of these three approaches; all evidence is based on cure rates in patient series.

HCC treatment options depending on BCLC stage. aSee Table 4 for indication constraints based on tumour burden and liver function. bNot EMA-approved as of August 2018. BCLC, Barcelona Clinic Liver Cancer; BSC, best supportive care; EMA, European Medicines Agency; HCC, hepatocellular carcinoma; LTX, liver transplantation; SBRT, stereotactic body radiotherapy; SIRT, selective internal radiotherapy; TACE, transarterial chemoembolisation.

The predominant arterial vascularisation of HCC resulted in the application of intra-arterial administration of chemotherapy (e.g. doxorubicin, cisplatin), embolising material (e.g. coils, gelatin sponge particles) or radioactive particles. These therapies are generally regarded as palliative treatment options but may provide complete tumour destruction in well-selected candidates.

Single tumours in patients with well-preserved liver function is the mainstay indication for resection, provided a R0 resection (excision whose margins are clear of tumour cells) can be carried out without causing postoperative liver failure due to insufficient reserve in the liver remnant. LR requires a detailed preoperative work-up with the assessment of liver function and future liver remnant volume. The combination of both variables determines the perioperative risk of liver failure and the associated complications. Child-Pugh A patients without significant portal hypertension are considered good candidates for minor/major LRs [III, B]. Child-Pugh C patients are not suitable for surgical therapy. A recent meta-analysis demonstrates that the presence of portal hypertension or Child-Pugh B status might not be an absolute contraindication and provide acceptable results for these cohorts [52, 53]. Therefore, carefully selected patients with Child-Pugh B and/or portal hypertension may be candidates for minor surgical resection [III, A].

Compared with open LR, laparoscopic LR results in reduced intraoperative blood loss, faster postoperative recovery and does not impair oncological outcome [54]. LR in cirrhosis should preferably be carried out as laparoscopic resection [IV, A]. Currently, there is no high-level evidence to recommend surgical resection in cirrhotic HCC patients with advanced tumour burden and macrovascular invasion.

After LR, tumour recurrence can be observed in 50%–70% of cases within 5 years following surgery, which constitutes either intrahepatic metastases (often within 2 years following surgery) or a new HCC in the remaining cirrhotic liver (occurring more often beyond 2 years). Even though the vast majority of HCC recurrences occur within the liver as a result of subclinical micro-metastases and vascular invasion from the primary tumour, the extent of surgical resection [anatomical resection (AR) versus non-anatomical wedge resection (NAR)] is still a subject of ongoing debate. Theoretically, the systematic removal of the hepatic segment through an AR is considered to be more effective in terms of tumour clearance and eradication of micro-metastases [55]. This, however, is rarely possible in cirrhotic HCC patients for whom tissue-sparing NAR is the procedure of choice to reduce the risk of post-operative liver failure [56]. While some groups report superiority of AR, overall conflicting results are reported, and no clear recommendation may be given due to a lack of currently available high-level clinical evidence [57, 58].

Liver transplantation offers the possibility to cure both the tumour and the underlying liver disease [59]. The Milan criteria (one lesion < 5 cm; alternatively, up to three lesions, each < 3 cm; no extrahepatic manifestations; no evidence of macrovascular invasion) are currently the benchmark for the selection of patients with HCC for OLT. OLT is recommended for patients that fit the Milan criteria, for which < 10% recurrence and 70% 5-year survival are expected [II, A] [60]. Among several more liberal proposals [up-to-seven criteria, extended Toronto criteria, University of California San Francisco (UCSF) criteria], only the UCSF criteria (one tumour ≤ 6.5 cm, three nodules at most with the largest ≤ 4.5 cm and total tumour diameter ≤ 8 cm) were prospectively validated and showed similar outcome and, as such, may also be considered for OLT in patients with HCC beyond Milan criteria [III, B] [60, 61]. The use of marginal grafts or living donor liver transplantation could facilitate the treatment of these patients [62–64].

The low availability of liver allografts, however, is a major limitation for OLT, and liver transplant candidates are often confronted with long waiting times, which may be associated with tumour progression beyond the Milan criteria. When a waiting time (> 3 months) is anticipated, patients may be offered resection, local ablation or transarterial chemoembolisation (TACE) in order to minimise the risk of tumour progression and to offer a ‘bridge’ to transplant [III, B].

Adjuvant therapy is not recommended for HCC patients after OLT, LR or local ablation [I, E]. Mammalian target of rapamycin (mTOR) inhibitors are used as immunosuppressant to prevent graft rejection in liver transplantation (sirolimus) but have failed to improve recurrence-free survival in a recently published phase III study [65]. Similarly, sorafenib did not improve median recurrence-free survival of HCC patients after LR or local ablation [66].

Thermal ablation by RFA or MWA may be recommended as first-line treatment in very early-stage disease (BCLC 0) [II, A]. In very early-stage disease (tumours < 2 cm diameter), RFA has demonstrated similar outcomes to LR and thus may be recommended as first-line treatment, specifically in light of its lesser invasiveness and morbidity compared with surgery [64]. In patients with early-stage HCC (up to three lesions ≤ 3 cm), RFA has been adopted as an alternative first-line option irrespective of liver function after demonstrating survival benefit similar to surgery in RCTs and meta-analyses [64, 67–70]. To date, MWA has not been adequately tested in comparison to RFA and the potential advantage for tumours between 3 and 5 cm or the reduced impact of the cooling effect of adjacent large vessels remains unknown. Both methods have limitations in exophytic tumours as well as those close to the gallbladder, liver hilum or with neighbouring intestine, which may be overcome by administering laparoscopic surgery [71]. Chemical tumour ablation (e.g. by ethanol injection) plays no role, since thermal ablation has proven better disease control and outcomes [72]. In very small lesions, superiority of thermal ablation is minimal [73].

High conformal high dose rate (HDR) radioablation and stereotactic body radiotherapy (SBRT) may be considered as alternatives for the ablation of tumours with a high risk of local failure after thermal ablation due to location [III, C]. High conformal irradiation techniques with hypofractionated (SBRT) or single fraction dose regimens (HDR brachytherapy) have evolved as alternatives to thermal ablation in recent years. In contrast to classic fractionated irradiation schemes, high conformal HDR irradiation techniques such as SBRT or CT-guided HDR brachytherapy have proven efficacy with tumour control rates > 90% after 12 months in ≤ 5 cm (SBRT) or ≤ 12 cm tumour diameter (HDR brachytherapy) in single-centre studies [74–77]. However, a recent comparative trial has demonstrated better survival when applying RFA than SBRT in small tumours ≤ 3 cm [78]. In contrast to thermal ablation, high conformal HDR radioablation is not limited by adjacency to large vessels, exophytic growth or central location. Both SBRT and HDR brachytherapy have demonstrated excellent safety profiles [79, 80]. External beam radiotherapy (EBRT) can be used to control pain in patients with bone metastases [III, B]. Any ablation recommendation should be proposed by the local multidisciplinary meeting (MDM) based on liver function, tumour size, tumour location and the medical expertise provided by the given treatment centre.

The almost exclusive arterial vascularisation of HCC resulted in the application of intra-arterial infusion of chemotherapy alone (doxorubicin, cisplatin, mytomicin C or combinations), mixed with the contrast agent lipiodol (ethiodised oil) that is selectively retained by HCC nodules, embolising material (e.g. coils, gelatin sponge pieces or polyvinyl alcohol-calibrated particles) or tiny radioactive particles containing yttrium-90 (90Y).

Absolute contraindications for transarterial therapies are decompensated cirrhosis, extensive tumour burden, reduced portal vein flow, renal failure or any technical contraindication to transarterial therapy. Important relative contraindications include bile duct occlusion or incompetent papilla, reduced PS, impaired liver function (Child-Pugh B), high-risk oesophageal varices, portal vein thrombosis of any kind for TACE or involving the main trunk for selective internal radiotherapy (SIRT) [81].

Overall, the efficacy of TACE has been explored in seven randomised trials compared with best supportive care (BSC) [82]. Only two studies reported a survival benefit for the treatment arm [83, 84]. The benefit of TACE in prolonging OS was demonstrated in selected asymptomatic patients with maintained liver function that belong to the BCLC A stage to early intermediate BCLC B stage, who had a small tumour burden but were not amenable to surgery or local ablation [I, A]. Median OS (mOS) of 30–45 months can be expected in this population [85–87]. Shorter median survival of < 20 months has been reported in real life cohorts when patients with no proven benefit are treated including those in Child-Pugh B stage, with portal vein invasion, large tumour burden or deteriorating liver function under TACE [88–91]. Several scores have been developed to identify patients that benefit from TACE from retrospective cohort studies. Currently, only the hepatoma arterial-embolisation prognostic (HAP) score has been validated in a prospective trial and in multiple large international datasets [92, 93]. The HAP score is able to define four distinct prognostic groups with respect to OS and could be used as a stratification factor for TACE trials in future [88]. Outside clinical trials, the use of therapeutic algorithms based on prognostic scores of unknown predictive values is currently not recommended for the selection of candidates to initial and repeated TACE [III, A].

Conventional lipiodol-based TACE is the standard of practice, although using doxorubicin-eluting bead (DEB)-TACE is an option to minimise systemic side effects of chemotherapy [I, C]. Compared with conventional TACE, in RCTs DEB-TACE is associated with significantly fewer side effects related to the leakage of doxorubicin into the systemic circulation [94] and provides a more standardised way to perform TACE. No prospective trial has so far demonstrated the superiority of conventional TACE, bland embolisation or DEB-TACE. One randomised phase II trial compared cisplatin-based conventional TACE with bland embolisation using polyvinyl alcohol particles alone, and two trials have compared DEB-TACE with bland embolisation using unloaded beads [95–97]. None of these trials showed an apparent clinical benefit in terms of OS for the addition of chemotherapy; non-inferiority was also formally not proven.

The optimal duration and frequency of TACE treatment is not yet defined. TACE should not be repeated if a substantial necrosis is not achieved after the second session, or when a subsequent session fails to induce remission at sites that have initially responded to TACE. Additionally, the indication of TACE should be critically re-evaluated in patients with reduced PS and impaired liver function following TACE treatment.

The combination of TACE with systemic agents such as sorafenib—either sequential or concomitant—is not recommended in clinical practice [I, E]. Five randomised trials with 2468 patients have not shown a clinical meaningful benefit of systemic therapy (sorafenib, brivanib or orantinib) in combination with or following TACE compared with TACE alone in terms of median objective response rate (mORR), median progression-free survival (mPFS) or mOS [92, 98–101].

SIRT is based on the injection of microspheres loaded with the pure beta emitter 90Y into the hepatic arterial circulation and has no or minimal ischaemic effect. SIRT with 90Y glass or resin microbeads produces tumour responses and high disease control rates with a safe profile in phase II studies and registries [102].

SIRT is not recommended as first-line therapy for HCC patients in intermediate and advanced stage [I, E]. Two recent phase III trials randomised patients free from extrahepatic metastasis and with preserved liver function to sorafenib or SIRT using resin microspheres. The SARAH trial in France (n = 459 patients) and the SIRveNIB trial in Asia-Pacific (360 patients) failed to meet the primary endpoint of improved OS compared with sorafenib; survival for the sorafenib arm ranged from 10.2 to 9.9 months compared with 8.8 to 8 months for 90Y [hazard ratio (HR) 1.12–1.15] [103, 104]. The applicability of 90Y was limited to 72%–77% of patients due to treatment contraindications. Also, the per-protocol subgroup analyses did not yield any survival advantages. The SORAMIC phase II trial additionally analysed whether the addition of SIRT to sorafenib improves OS in patients with advanced HCC. However, this study failed to meet the primary endpoint, and the addition of SIRT to sorafenib did not show an OS that was superior to sorafenib alone. Whether subgroups such as non-cirrhotic patients or non-alcoholic aetiology of the cirrhosis with high positive HR for SIRT addition to sorafenib hold promise must be further validated [105].

In the phase III studies, SIRT was associated with higher response rates, delayed tumour progression in the liver and fewer adverse events (AEs) compared with sorafenib. The observed delay in tumour progression was also observed in retrospective cohort studies with survival rates comparable to those reported for TACE and sorafenib [106–108]. Thus, in exceptional circumstances, for patients with liver-confined disease and preserved liver function in whom neither TACE nor systemic therapy is possible, SIRT may be considered. Additionally, SIRT may be considered instead of TACE for the treatment of small tumours in patients waiting for liver transplantation, in an attempt to avoid drop-out from the list due to tumour progression [106].

During the past 40 years, numerous RCTs testing treatments for advanced HCC have been published [109]. Sorafenib showed a survival benefit and it was established as the sole systemic treatment for patients with advanced HCC or those progressing from locoregional therapies. More recently, five additional drugs have shown positive clinical results in first- and second-line settings (see Figure 1).

Chemotherapy has not been shown to improve survival in randomised trials and is not recommended as a standard of care [II, C].

To date, four trials have been reported for which the experimental arms were: PIAF (cisplatin/interferon/doxorubicin/fluorouracil), the tubulin binding agent T138067, nolatrexed and FOLFOX (leucovorin/fluorouracil/oxaliplatin) [110–112]. None improved survival compared with doxorubicin, although response rates were higher with FOLFOX (8.2% versus 2.7%, P = 0.0233) and a small benefit in median survival was also seen on long-term follow up (6.4 versus 5.0 months, P = 0.0425). One trial has compared sorafenib with the combination of sorafenib and doxorubicin but did not demonstrate improved survival with combination therapy [113]. In summary, the clinical benefit of chemotherapy in the management of HCC has not been established.

Sorafenib is the standard of care for patients with advanced HCC and those with intermediate-stage (BCLC B) disease not eligible for, or progressing despite, locoregional therapies. It is recommended in patients with well-preserved liver function and Eastern Cooperative Oncology Group (ECOG) PS 0–2 [I, A].

Lenvatinib showed non-inferiority efficacy compared with sorafenib and can be considered in patients with advanced HCC without main portal vein invasion and with ECOG PS 0–1 as a front-line systemic treatment, pending European Medicines Agency (EMA) approval [I, A].

Sorafenib, a multikinase inhibitor blocking 40 kinases including vascular endothelial growth factor receptor 2 (VEGFR2) and BRAF, was established as the standard systemic therapy for HCC according to all international guidelines following the results reported a decade ago. It is indicated for patients with well-preserved liver function (Child-Pugh A class) and with advanced tumours (BCLC C) or those tumours at intermediate stage (BCLC B) progressing upon locoregional therapies. In the SHARP phase III trial, sorafenib improved survival compared with placebo (HR 0.69; P = 0.001; 7.9–10.7 months) [114]. The target population of this trial was mostly patients with advanced HCC (80%, including 35% with macrovascular invasion and 50% with extrahepatic spread). The results of the SHARP trial were subsequently confirmed in the Asia-Pacific phase III trial [115] and in 10 subsequent trials with an mOS in the range of 10–12 months. Objective responses are uncommon; 2% by Response Evaluation Criteria in Solid Tumours (RECIST) and ∼10% by modified RECIST (mRECIST) [116]. A recent meta-analysis of individual data of two RCTs testing sorafenib has shown that, although of benefit to all patients across the board, it provides better outcomes in patients with HCV-related HCC and those with liver-only disease [117]. No predictive biomarkers of responsiveness to sorafenib have been identified.

The recommended daily dose of sorafenib is 800 mg. Median treatment duration is estimated to be 5–6 months, but early prevention of toxicities can enhance tolerability. Treatment is associated with manageable AEs, such as diarrhoea, hand–foot skin reactions, fatigue and hypertension. Around 15% of patients are intolerant to sorafenib, and thus treatment needs to be withdrawn, while another 35% of patients require dose reduction. Treatment-related liver failure or life-threatening complications are marginal. Considering the restrictive indication of sorafenib in terms of liver failure (mostly Child-Pugh A class), it has been estimated that only half of patients at advanced stages can be suitable for this treatment. Clinically symptomatic vascular disease—either coronary or peripheral—is considered a formal contraindication.

Several phase III trials have been conducted to challenge sorafenib in front line (testing sunitinib, brivanib, erlotinib, linifanib or doxorubicin), but lenvatinib has only recently shown non-inferior clinical efficacy [118]. Lenvatinib is an oral multikinase inhibitor that targets VEGFR1–3 and fibroblast growth factor receptor (FGFR)1–4, among others. Lenvatinib demonstrated non-inferiority results compared with sorafenib in an open-label, phase III, multicentre, non-inferiority trial involving patients with advanced HCC (excluding main portal vein invasion, clear bile duct invasion and > 50% of tumour to total liver volume occupancy). The dose was adjusted to body weight. The study met its primary endpoint of non-inferiority in OS [HR 0.92; 95% confidence interval (CI) 0.79–1.06; mOS lenvatinib, 13.6 months versus sorafenib, 12.3 months]. Secondary endpoints such as PFS, time to progression and ORR (24% versus 9.2% for sorafenib, mRECIST ORR) were significantly better for lenvatinib. Lenvatinib-related most common any-grade AEs compared with sorafenib were as follows: hypertension (42% versus 30%), diarrhoea (39% versus 45%) and hand–foot skin reaction (27% versus 52%). Median time on lenvatinib was 5.7 months. Time to worsening in quality of life was similar in both treatment arms (HR 1.01). These results position lenvatinib as an option in first-line treatment for advanced HCC, once the drug is approved by regulatory agencies. No cost-effectiveness studies comparing both drugs are available.

Regorafenib is the standard of care for patients with advanced HCC who have tolerated sorafenib but progressed. It is recommended in patients with well-preserved liver function and ECOG PS 0–1 [I, A].

Cabozantinib can be considered for patients who had progressive disease on one or two systemic therapies with well-preserved liver function and ECOG PS 0–1, pending EMA approval [I, A].

Ramucirumab (RAM) can be considered for patients in second-line treatment with baseline AFP ≥ 400 ng/mL, well-preserved liver function and ECOG PS 0–1, pending EMA approval [I, A].

Recently, a phase III study comparing regorafenib (a multikinase inhibitor targeting similar kinases as sorafenib) with placebo in patients progressing despite sorafenib has reported a benefit in survival (HR 0.62; P < 0.0001, mOS 7.8–10.6 months) [119]. Treatment improved survival in all subgroups of patients. In this trial, 88% of patients were BCLC C and 12% BCLC B, with all of them tolerant to but progressing on sorafenib. Around 30% of patients presented with macrovascular invasion: 70% with extrahepatic spread and 45% with AFP > 400 ng/dL. The response rate was 10%, based upon mRECIST. Treatment was started at 160 mg/day (3 weeks on/1 week off). Median time on treatment was 3.5 months. AEs led to 51% dose reductions and 10% treatment discontinuation. Approval of regorafenib as a standard of care opens the field for third-line therapies. It should be kept in mind, however, that most patients at BCLC B-C stages not candidates to standard-of-care therapies (TACE, sorafenib, regorafenib) are generally unsuitable candidates to enter into clinical trials. These patients along with those at BCLC D stage should receive best supportive/palliative care, including management of pain, nutrition and psychological support.

Cabozantinib is a MET, VEGFR2, AXL and RET inhibitor approved for thyroid and renal cancer. The CELESTIAL trial, a randomised, global phase III trial, examined cabozantinib versus placebo in patients with advanced HCC who had been previously treated with sorafenib [120]. In contrast to regorafenib, this trial allowed the inclusion of patients that were intolerant to sorafenib and who had progressive disease on one or two systemic therapies. In this trial, 30% of patients presented with macrovascular invasion, 78% with extrahepatic spread and 42% with AFP > 400 ng/dL. Treatment was started at 60 mg/day, and median time on treatment was 3.8 months. OS results favoured cabozantinib compared with placebo (HR 0.76, 95% CI 0.63–0.92; P = 0.0049; mOS 10.2 versus 8.0 months). Response rate was 4% with cabozantinib based upon RECIST v1.1. The most common grade 3/4 AEs with cabozantinib versus placebo were palmar–plantar erythrodysesthaesia (17% versus 0%), hypertension (16% versus 2%), increased aspartate aminotransferase (AST) (12% versus 7%), fatigue (10% versus 4%) and diarrhoea (10% versus 2%) and led to 62% dose reductions and 16% treatment discontinuation.

RAM is a human immunoglobulin G1 (IgG1) monoclonal antibody (mAb) that inhibits ligand activation of VEGFR2. In the phase III REACH trial mOS in the overall population was not statistically significant, but a meaningful improvement was observed in a patient subgroup with baseline AFP ≥ 400 ng/mL. Based on these data, the REACH-2 phase III trial analysed the efficacy of RAM in patients with elevated baseline AFP following therapy with sorafenib. RAM treatment significantly improved mOS from 7.3 to 8.5 months (HR 0.710; 95% CI 0.531, 0.949; P = 0.0199) and mPFS from 1.6 to 2.8 months (HR 0.452; 95% CI 0.339, 0.603; P < 0.0001) compared with placebo [121]. ORR was 4.6% with RAM versus 1.1% with placebo (P = 0.1156) and ORR was 59.9% RAM versus 38.9% with placebo (P = 0.0006). The safety profile observed in the REACH-2 study was consistent with what has been previously observed, and the only grade ≥3 AEs occurring at a rate of ≥ 5% in the RAM arm were hypertension (12.2% versus 5.3%) and hyponatremia (5.6% versus 0%).

Immunotherapy with nivolumab and pembrolizumab can be considered in patients who are intolerant to, or have progressed under, approved tyrosine kinase inhibitors, pending EMA approval [III, B]. For a definitive recommendation, it is necessary to wait for the results of randomised trials.

To date, the most promising immunotherapeutic approach has been the use of immune checkpoint inhibitors. Initial results from a small single-arm phase II trial of tremelimumab [a fully humanised IgG2 anti-cytotoxic T-lymphocyte antigen 4 (CTLA-4) antibody] demonstrated a response rate of 17% and time to progression of 6.5 months [122]. More recently, a large single-arm phase I/II trial of the fully human IgG4 programmed cell death protein 1 (PD-1) inhibitor nivolumab (CheckMate 040), has been reported [123]. A total of 262 patients were treated of which 48 were in dose escalation and 214 in dose expansion. The dose of 3 mg/kg every 2 weeks was shown to be tolerable during dose escalation and was used in dose expansion (in which patients were required to be Child-Pugh A and ECOG PS ≤ 1). In dose expansion, there were no treatment-related deaths and grade 3/4 AST and alanine aminotransferase (ALT) increase occurred in 4% and 2%, respectively. The most common AEs of any grade were fatigue (23%), pruritus (21%) and rash (15%). The ORR was 20% (RECIST v1.1) and the PFS and 9-month OS were 4.0 months and 74%, respectively. Expression of programmed death-ligand 1 (PD-L1) on tumour cell membranes was not found to be predictive. Overall, 145 patients in the expansion cohort had received prior sorafenib and, after extended follow-up, the mOS was 15.6 months (13.2–18.9). This compares favourably with all of the previously reported phase III second-line trials in HCC, for which mOS has been between 7.6 and 10.6 months in the experimental arm. On this basis, the United States Food and Drug Administration (FDA) granted accelerated approval for the use of nivolumab in patients previously treated with sorafenib, on the condition that further trials were required to verify the clinical benefit of nivolumab in patients with HCC. The first-line phase III trial comparing sorafenib with nivolumab, CheckMate 459, is expected to report in 2018 and, if positive, will position nivolumab as a first-line treatment option.

Meanwhile, a phase II trial of the anti-PD-1 antibody pembrolizumab as second-line treatment (KEYNOTE-224) has recently been reported. The 16.3% response rate (RECIST v1.1) and 78% 6-month OS observed among the 104 patients included is in line with the results seen with nivolumab. Median time to progression was 4.9 months (95% CI: 3.9–8.0), mPFS was 4.9 months (95% CI 3.4–7.2) and mOS was 12.9 months (95% CI 9.7–15.5) [124].

Molecular profiling is not recommended as standard of practice since it currently has no direct implication for decision making. However, we recommend obtaining tissue in all research studies for exploring biomarkers of response.

There has been increasing interest in stratified trials driven by predictive biomarkers, and a number of earlier phase trials are exploring this strategy in HCC. Investigations into the molecular pathology of HCC have identified recurrent mutations of which the most common are in the TERT promotor, CTNNB1, TP53 and epigenetic regulators including ARID1A and ARID2 [125]. While these pathways provide a challenge for drug development, less common molecular aberrations are tractable and show promise. For example, overexpression of FGF19 is found in ≤ 20% of HCCs, and several compounds directed against its receptor FGFR4 are in development, including BLU-554 and FGF401. Despite the disappointing results of the tivantinib phase III trial [126], there are ongoing studies enriching for MET pathway activation or MET overexpression with INC280 and MSC2156119J. Activation of the transforming growth factor beta 1 (TGFβ1) pathway is associated with a more aggressive sub-class of HCC and is being targeted with galunisertib in combinations with sorafenib and nivolumab, although these trials are not currently enriched for pathway activation. Numerous other targets are being evaluated including androgen receptor, signal transducer and activator of transcription 3 (STAT3) inhibitor, histone deacetylase inhibitor (HDACi) and cyclin-dependent kinase 4/6 (CDK 4/6), but, while personalised therapy holds promise for the future, there is insufficient evidence for molecular stratification at the present time.

Many HCC treatments act by induction of tumour necrosis or reduction in vascularity, which is not necessarily accompanied by tumour shrinkage. Viable tumour should be assessed using dynamic CT or MRI studies and should be defined as uptake of contrast agent in the arterial phase [III, A]. mRECIST are recommended for assessment of response/progression to locoregional therapies [III, B]. RECIST were primarily designed for the evaluation of cytotoxic agents. Modifications of RECIST (mRECIST) are available and are based on the measurement of the diameter of the viable tumour component of target lesions (Table 5) [116]. mRECIST also include guidelines regarding evaluation of vascular invasion, lymph nodes, effusions and new lesions. In 2011, the first study reported a link between mRECIST, EASL (European Association for the Study of Liver) criteria and OS in patients treated with TACE in contrast to RECIST v1.1, which was subsequently confirmed and validated [127–130]. In contrast to locoregional therapies, the value of mRECIST in the evaluation of systemic therapy in HCC is not yet established. mRECIST were prospectively evaluated in the BRISK trial and responders had a better OS compared with non-responders [131]; however, a higher objective response by mRECIST does not correlate with an improved OS in subsequent phase III trials [118]. In addition, the prospective comparison between mRECIST and RECIST in two trials with nintedanib and one trial with regorafenib revealed a very similar outcome, with no clear advantage of mRECIST [119, 132]. Overall, mRECIST need further prospective validation but may be used in daily clinical practice to consider not only tumour diameters but also lesion viability in therapy decision making [III, B]. There is limited evidence that OS can be predicted more accurately by mRECIST than RECIST v1.1 [IV, B].

RECIST mRECIST
CR Disappearance of all target lesions Disappearance of any intratumoural arterial enhancement in all target lesions
PR At least a 30% decrease in the sum of diameters of target lesions, taking as reference the baseline sum of the diameters of target lesions At least a 30% decrease in the sum of diameters of viable (enhancement in the arterial phase) target lesions, taking as reference the baseline sum of the diameters of target lesions
SD Any cases that do not qualify for either partial response or PD Any cases that do not qualify for either partial response or PD
PD An increase of at least 20% in the sum of the diameters of target lesions (lymph nodes of 1.5 cm diameter), taking as reference the smallest sum of the diameters of target lesions recorded since treatment started An increase of at least 20% in the sum of the diameters of viable (enhancing) target lesions (lymph nodes of 2 cm diameter), taking as reference the smallest sum of the diameters of viable (enhancing) target lesions recorded since treatment started
Development of new ascites Development of new ascites with positive cytology

Response assessment following radioembolisation is challenging and should be carried out by multiple phase MRI or CT at ∼3–4 months intervals. Imaging carried out early after radioembolisation may show arterial enhancement (both rim and intratumoural) related to post-treatment inflammatory changes and may be erroneously labelled as infiltrative tumour. These findings usually resolve after 6 months [133]. Prospective radiological–pathological studies have shown that EASL criteria and mRECIST—and not World Health Organization (WHO) criteria or RECIST—may capture responses at 3 months after radioembolisation [134].

In the context of immunotherapy, response evaluation may also be very challenging as pseudoprogression (transient increase in tumour size and AFP, followed by response) has been described also in HCC [135]. Recent trials with immunotherapies reported response rates of up to 25% by RECIST v1.1, and mRECIST have not been validated in this setting. Serum tumour markers (such as AFP levels) may be helpful particularly in the case of not easily measurable disease but should not be used as the only determinant for treatment decisions [IV, B]. Pseudoprogression is incredibly rare but, in the future, immune RECIST (iRECIST) should be discussed in this context [136].

In summary, follow-up of patients who underwent radical treatments (resection or RFA) should consist of the clinical evaluation of liver decompensation and the early detection of recurrence by dynamic CT or MRI studies every 3 months during the first year and surveillance every 6 months thereafter [III, A] [66, 137, 138]. Patients with recurrence following radical therapies may still be candidates for curative therapies. Patients with more advanced stages of HCC who are treated with TACE or systemic agents (e.g. sorafenib) are evaluated clinically for signs of liver decompensation and for tumour progression by dynamic CT or MRI every 3 months to guide therapy decisions [III, A].

These Clinical Practice Guidelines were developed in accordance with the ESMO standard operating procedures for Clinical Practice Guidelines development http://www.esmo.org/Guidelines/ESMO-Guidelines-Methodology. The relevant literature has been selected by the expert authors. A summary of recommendations is shown in Table 6. Levels of evidence and grades of recommendation have been applied using the system shown in Table 7. Statements without grading were considered justified standard clinical practice by the experts and the ESMO Faculty. This manuscript has been subjected to an anonymous peer review process.

Incidence, epidemiology and surveillance
  • The association of chronic liver disease and HCC represents the basis for preventive strategies, including universal vaccination at birth against HBV [I, A] and early antiviral treatment of viral HBV and HCV [III, A]

  • Cost-effectiveness studies suggest surveillance of HCC is warranted in all cirrhotic patients irrespective of its aetiology, as long as liver function and comorbidities allow curative or palliative treatments [III, A]

  • Surveillance of non-cirrhotic, hepatitis-infected patients is also advocated, especially in HBV carriers with serum viral load > 10 000 copies/mL or HCV-infected patients with bridging fibrosis (F3) [III, A]

  • Surveillance of patients at risk for HCC should be carried out by abdominal US every 6 months with or without AFP [I, A]

Diagnosis and pathology/molecular biology
  • The diagnosis of HCC is based on histological analysis and/or contrast-enhanced imaging findings [III, A]

  • The diagnosis can be established if the typical vascular hallmarks of HCC (hypervascularity in the arterial phase with washout in the portal venous or delayed phase) are identified in a nodule of > 1 cm diameter using one of these two modalities in a cirrhotic patient [III, A]

  • Based on techniques such as diffusion-weighted imaging and the use of hepatobiliary contrast agents, MRI may allow identification and stratification of nodules as high-risk nodules (either HCC not displaying the typical imaging hallmarks features or high-grade dysplastic nodules) [IV, B]

  • For CEUS, an overlap between the vascular profile of HCC and CC has been described. However, recent data suggest CEUS as a suitable technique to diagnose HCC non-invasively in the setting of liver cirrhosis [IV, B]

  • When tumour biopsy fails to demonstrate a correlate for a focal lesion, a second tumour biopsy, a different contrast enhanced imaging modality or (if amenable) direct resection of the lesion may be considered according to tumour size [IV, B]

  • Histopathological diagnosis of tumour biopsies relies on standard (H&E) and special stains (e.g. reticulin), and—if required—IHC

  • It is important to distinguish combined HCC/CC from HCC due to the different therapeutic modalities; however, the mixed differentiation features might not be visible in the biopsy. In addition, significant expression of CK19 has been evaluated and is considered as a sign of poor prognosis in HCC [IV, B]

  • In highly differentiated HCC, definitive signs of malignancy (interstitial or vascular invasion) are frequently absent from biopsy. Further consented histological (trabecular alterations—more than two cell broad trabeculae, pseudoglands, reticulin loss, capsule formation) and cytological criteria (increased nuclear/cytoplasmic ratio, i.e. ‘nuclear crowding’, increased cytoplasmic basophilia) support HCC diagnosis [III, B]

  • IHC should be carried out in unclear cases: capillarisation of sinusoids could be assessed using CD34 IHC [IV, B]

  • Further immunohistochemical markers have been shown to improve the diagnosis of highly differentiated HCC, including glutamine synthetase, glypican 3, CTC, EZH2 and HSP70 [IV, B]

  • A combination of the three markers glutamine synthetase, glypican 3 and HSP70 has been consented as a diagnostic panel (2/3 marker positivity has 70% sensitivity and 100% specificity for HCC) and the use of further markers seems to increase the sensitivity [IV, B]

Staging and risk assessment
  • Staging of HCC is important to determine outcome and planning of optimal therapy and includes assessment of tumour extent, AFP level, liver function, portal pressure and clinical PS [III, A]

  • BCLC is the commonly accepted staging system for prognostic prediction and treatment allocation

  • There is no demonstrated clinical benefit of carrying out an FDG-PET scan as a staging modality, despite some evidence that there is a correlation of higher FDG uptake with poor differentiation, tumour size, serum AFP levels and microvascular invasion [IV, D]

  • Liver function is classically assessed by the Child-Pugh scoring system (serum bilirubin, serum albumin, ascites, prothrombin time and hepatic encephalopathy) [III, A]

  • Within the Child-Pugh A group, measurement of the ALBI score is able to split that group into best prognosis (ALBI 1) and lesser prognosis (ALBI 2), with median survivals of 26 versus 14 months, respectively [IV, B]

  • The finding of oesophageal varices and/or splenomegaly with blood platelet counts of 100 × 109 cells/L suggests clinically important portal hypertension, which can also be measured invasively by the transjugular route (hepatic-venous pressure gradient > 10 mmHg) [III, A]

Management of early and intermediate HCC
  • Child-Pugh A patients without immanent portal hypertension are considered good candidates for minor/major LRs [III, B]

  • Carefully selected patients with Child-Pugh B and/or portal hypertension may be candidates for minor surgical resection [III, A]

  • LR in cirrhosis should preferably be carried out as laparoscopic resection [IV, A]

  • The Milan criteria (one lesion < 5 cm; alternatively, up to three lesions, each < 3 cm; no extrahepatic manifestations; no evidence of macrovascular invasion) are currently the benchmark for the selection of patients with HCC for OLT. OLT is recommended for patients that fit the Milan criteria, for which < 10% recurrence and 70% 5-year survival are expected [II, A]

  • Among several more liberal proposals (up-to-seven, Toronto criteria, UCSF), only the UCSF criteria (one tumour ≤ 6.5 cm, three nodules at most with the largest ≤ 4.5 cm, and total tumour diameter ≤ 8 cm) were prospectively validated and showed similar outcome and, as such, may also be considered for OLT in patients with HCC beyond Milan criteria [III, B]

  • In the case of a long-anticipated waiting time (> 3 months), patients may be offered resection, local ablation or TACE in order to minimise the risk of tumour progression and to offer a ‘bridge’ to transplant [III, B]

  • Adjuvant therapy is not recommended for HCC patients after OLT, LR or local ablation [I, E]

  • Thermal ablation by means of RFA or MWA may be recommended as first-line treatment in very early-stage disease (BCLC 0) [II, A]

  • High conformal HDR radioablation and SBRT may be considered as alternatives for the ablation of tumours with a high risk of local failure after thermal ablation due to location [III, C]

  • Outside clinical trials, the use of therapeutic algorithms based on prognostic scores of unknown predictive values is currently not recommended for the selection of candidates to initial and repeated TACE [III, A]

  • Conventional lipiodol-based TACE is the standard of care for patients with intermediate HCC, although using DEB-TACE is an option to minimise systemic side effects of chemotherapy [I, C]

  • The combination of TACE with systemic agents such as sorafenib—either sequential or concomitant—is not recommended in clinical practice [I, E]

  • SIRT is not recommended as first-line therapy for patients in intermediate or advanced stage [I, E]

Management of advanced disease
  • Chemotherapy has not been shown to improve survival in randomised trials and is not recommended as a standard of care [II, C]

  • Sorafenib is the standard of care for patients with advanced HCC and those with intermediate-stage (BCLC B) disease not eligible for, or progressing despite, locoregional therapies. It is recommended in patients with well-preserved liver function and ECOG PS 0–2 [I, A]

  • Lenvatinib showed non-inferiority efficacy compared with sorafenib, and can be considered as first-line therapy in patients with advanced HCC without main portal vein invasion, clear bile duct invasion and ≥ 50% of tumour to total liver volume occupancy, pending EMA approval [I, A]

  • Regorafenib is the standard of care for patients with advanced HCC who have tolerated sorafenib but progressed. It is recommended in patients with well-preserved liver function and ECOG PS 0–1 [I, A]

  • Cabozantinib can be considered for patients who had progressive disease on one or two systemic therapies with well-preserved liver function and ECOG PS 0–1, pending EMA approval [I, A]

  • Ramucirumab can be considered for patients in second-line patients with baseline AFP ≥ 400 ng/mL, well-preserved liver function and ECOG PS 0–1, pending EMA approval [I, A]

  • Immunotherapy with nivolumab and pembrolizumab can be considered in patients who are intolerant to, or have progressed under, approved tyrosine kinase inhibitors, pending EMA approval [III, B]. For a definitive recommendation, it is necessary to wait for the results of randomised trials

Follow-up, long-term implications and survivorship
  • Viable tumour should be assessed using dynamic CT or MRI studies and be defined as uptake of contrast agent in the arterial phase [III, A]

  • mRECIST need further prospective validation but may be recommended in daily clinical practice to consider not only tumour diameters but also lesion viability in therapy decision making [III, B]

  • mRECIST are recommended for assessment of response/progression to locoregional therapies [III, B]

  • There is limited evidence that OS can be predicted more accurately by mRECIST than RECIST v1.1 [IV, B]

  • Follow-up of patients who underwent radical treatments (resection or RFA) should consist of the clinical evaluation of liver decompensation and the early detection of recurrence by dynamic CT or MRI studies every 3 months during the first 2 years and surveillance every 6 months thereafter [III, A]

  • Patients with more advanced stages of HCC who are treated with TACE or systemic agents are evaluated clinically for signs of liver decompensation and for tumour progression by dynamic CT or MRI every 3 months to guide therapy decisions [III, A]

Levels of evidence
I Evidence from at least one large randomised, controlled trial of good methodological quality (low potential for bias) or meta-analyses of well-conducted randomised trials without heterogeneity
II Small randomised trials or large randomised trials with a suspicion of bias (lower methodological quality) or meta-analyses of such trials or of trials with demonstrated heterogeneity
III Prospective cohort studies
IV Retrospective cohort studies or case–control studies
V Studies without control group, case reports, expert opinions
Grades of recommendation
A Strong evidence for efficacy with a substantial clinical benefit, strongly recommended
B Strong or moderate evidence for efficacy but with a limited clinical benefit, generally recommended
C Insufficient evidence for efficacy or benefit does not outweigh the risk or the disadvantages (AEs, costs, …), optional
D Moderate evidence against efficacy or for adverse outcome, generally not recommended
E Strong evidence against efficacy or for adverse outcome, never recommended

AV has received honoraria for talks and advisory boards from Bayer, Roche, Lilly, Bristol-Myers Squibb, Merck Sharp & Dohme, AstraZeneca, Eisai, Novartis and Ipsen; AC has provided consulting and advisory services for Merck Serono, Amgen, Servier, Roche, Lilly, Novartis, Takeda and Astellas and has received research support from Roche, Merck Serono, Servier, Beigene and Astellas; IC has reported being a member of the advisory boards of Eli-Lilly, Bristol-Myers Squibb, MSD, Bayer, Roche, Merck-Serono, Five Prime Therapeutics and Astra-Zeneca and has received research funding from Eli-Lilly, Janssen-Cilag, Sanofi Oncology, Merck-Serono and honorarium from Eli-Lilly; BD has received honoraria or consultation fees from Bayer, Bristol-Myers Squibb, MSD, Merck KGaA, Ipsen, Eisai and Lilly; JL has received consultancy fees from Bayer, Bristol-Myers Squibb, Incyte, Lilly, Eisai, Celsion, Glycotest, Ipsen, Merck and Exelixis and research support from Incyte, Bayer, Bristol-Myers Squibb and Eisai; TM has received consulting fees from Bristol-Myers Squibb, Bayer, Eisai, Ipsen, Merck, BTG and Beigene; UN has given presentations for Merck, Amgen, Roche, Grünenthal and Bayer on topics other than HCC; JR has received consulting fees and research grants from Bayer Healthcare and Sirtex Medical; BS has received consulting and/or lecture fees from Adaptimmune, AstraZeneca, Bayer, Bristol-Myers Squibb, BTG, Onxeo, Sirtex and Terumo; PS has reported being a member of advisory board and has received grants from Novartis and Bristol-Myers Squibb; CV has received consultancy fees and research funding from Bayer, Sirtex, Novartis and Ipsen; CJZ has received honoraria for lectures and advisory boards by Bayer Healthcare; DA has received honoraria for consultancy from Roche, Merck Serono, Bayer Healthcare, Servier, BTG, Terumo, Sanofi Oncology and Eli Lilly; EM has received honoraria for lecture and advisory boards for Roche, Amgen, Servier and Sanofi; JCN has reported no conflicts of interest.


Clinical Practice Guidelines

This update refers to the Cancer of the prostate: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up, Parker C, Gillessen S, Heidenreich A, Horwich A. Ann Oncol 2015; 26 (Suppl 5): v69–v77.

Management of advanced/metastatic disease

Two phase III trials have compared ADT alone versus ADT plus abiraterone and prednisolone in men with metastatic, hormone-naive disease. The LATITUDE trial included men with high risk metastatic disease. Based on 406 events, abiraterone/prednisone improved overall survival (HR: 0.62, 95% CI: 0.51–0.76) [1]. The STAMPEDE trial included men with metastatic and non-metastatic disease. Based on 446 events, abiraterone/prednisone improved overall survival (HR: 0.63, 95% CI: 0.52–0.76) [2].

Recommendations:

  • ADT plus abiraterone/prednisone may be considered as first-line treatment for metastatic, hormone-naive disease [I, A].


Clinical Practice Guidelines

This update refers to Thyroid cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up, Pacini F, Castagna MG, Brilli L, Pentheroudakis G. Ann Oncol 2012; 23 (Suppl 7): vii110–vii119.

Management of advanced/metastatic disease

In the randomised, double-blind, placebo-controlled, phase III study (ZETA), 331 patients with locally advanced or metastatic medullary thyroid cancer (MTC) were randomised in a 2:1 ratio to receive oral vandetanib or placebo until disease progression. A significant prolongation of PFS (primary endpoint) was observed for patients receiving vandetanib (hazard ratio [HR] 0.46, 95% confidence interval [CI] 0.31–0.69; P < 0.001). Median PFS was 19.3 months in the placebo group while it had not yet been reached for the vandetanib group (predicted median PFS by Weibull model: 30.5 months). Overall survival data were immature. Common adverse events (any grade) occurred more frequently with vandetanib.

In patients with unresectable locally advanced or metastatic, hereditary or sporadic MTC, vandetanib, as compared with placebo, is associated with a statistically and clinically significant improvement in PFS. In the absence of mature OS and quality of life data, the observed PFS benefit is associated with an ESMO-Magnitude of Clinical Benefit Scale (MCBS) score of 3.


Under the section “Adjuvant chemotherapy for early-stage disease”

Long-term follow-up of the ICON 1 trial confirms the benefit of adjuvant chemotherapy, particularly in those patients at higher risk of recurrence (stage 1B/C grade 2/3, any grade 3 or clear-cell histology) [34]. Therefore, adjuvant chemotherapy should be offered not only to suboptimally staged patients but also to those optimally staged at higher risk of recurrence [I, A].

Is replaced with:

Long-term follow-up of the ICON 1 trial confirms the benefit of adjuvant chemotherapy, particularly in those patients at higher risk of recurrence [34]. Therefore, adjuvant chemotherapy should be offered not only to suboptimally staged patients but also to stage IB grade 2/3, all stage IC, any grade 3 or clear-cell histology [I, A].


Related posts

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.