ESMO Scale for Clinical Actionability of Molecular Targets Driving Targeted Treatment in Patients with Cholangiocarcinoma

Cholangiocarcinomas are rare liver neoplasms that arise from the biliary epithelium. Biliary tract cancers (BTC) are traditionally subdivided according to site of origin in the biliary tree, as intrahepatic and extrahepatic cholangiocarcinoma. Since the randomized phase III ABC-02 study in advanced BTCs, chemotherapy with a combination of cisplatin plus gemcitabine has been widely adopted as first-line treatment (1). Advanced BTC is associated with overall survival (OS) of about 12 months.

Molecularly, cholangiocarcinoma is a highly heterogeneous disease, with genomic differences between intra and extrahepatic cholangiocarcinomas (2, 3). A genetic profiling study in a large BTC cohort reported divergent driver alterations varying by tumor location, with IDH and FGFR pathway alterations predominantly found in intrahepatic cases, along with RAS and ARID1A (4). Nearly 40% of cases harbored druggable alterations including kinases (FGFR1/2/3, PIK3CA, ALK, EGFR, ERBB2, BRAF, KRAS), other oncogenes (IDH1/2) and tumor-suppressor genes (BRCA1/2). A large international analysis using whole-genome and epigenomic analysis, uncovered new cholangiocarcinoma driver genes, noncoding promoter mutations, and structural variants (5). The authors highlighted the importance of selective rather than blinded treatment and proposed that integrating molecular subtypes into the cholangiocarcinoma treatment paradigm is a means of achieving this.

Several actionable targets have shown a degree of clinical success in patients with previously treated cholangiocarcinoma, although to date, two targeted therapies have been approved for advanced or metastatic cholangiocarcinoma. Pemigatinib received accelerated FDA approval in April 2020 following the promising results from the open-label, phase II FIGHT-202 study in previously treated patients with BTC harboring an FGFR2 fusion or rearrangement, which gave a 36% objective response rate (ORR), median progression-free survival (PFS) of 6.9 months [95% confidence interval (CI), 6.2–9.6] and a still immature median OS of 21.1 months (95% CI, 14·8–not reached) (6). In May 2021, infigratinib was approved following a phase II open-label single-arm trial in previously treated, unresectable locally advanced or metastatic cholangiocarcinoma with an FGFR2 fusion or rearrangement. ORR was 23% (95% CI, 16–32) and median duration of response (DoR) was 5.0 months (7). Other FGFR inhibitors have recently shown efficacy in phase II studies in this population, including derazantinib with a 21% ORR in a phase I/II study (8) and futibatinib (TAS-120) with a 36% ORR (9). IDH1 signaling is another promising actionable target in cholangiocarcinoma (10). It was evaluated in the randomized, double-blind, phase III ClarIDHy study in patients with advanced IDH1-mutant cholangiocarcinoma after progression on previous therapy. Benefit was seen with ivosidenib compared with placebo, with median PFS of 2.7 months versus 1.4 months (HR, 0.37; 95% CI, 0.25–0.54; P < 0.0001) and a 6-month PFS rate of 32% compared with 0% in the placebo group (11). In addition, the ROAR phase II study reported encouraging results with a dabrafenib/trametinib combination in pretreated patients with BRAF^V600E-mutated BTC, giving a 51% ORR, and median PFS and OS of 9 and 14 months, respectively (12). Most recently, the MyPathway trial, an open-label phase IIa multibasket study, has treated 39 patients with HER2-positive BTC with pertuzumab and trastuzumab, showing a 23% ORR, providing another promising actionable target (13).

Use of next-generation sequencing (NGS) to drive cancer treatments is gaining increasing momentum. The nonrandomized prospective MOSCATO-01 study used high-throughput genomic analysis in advanced cancers to successfully identify tumor subtypes most likely to benefit from treatment personalization (14). A post hoc analysis of the 43 BTC patients included in this study reported that 18 of them received targeted treatments selected by a multidisciplinary molecular board (15). Druggable alterations included IDH1/2 mutations, FGFR1/2 fusions or mutations, activating alterations in EGFR, ERBB2 or ERBB3, PTEN deletions or mutations, MDM2 amplifications or translocations, and PIK3CA mutations or amplifications. Median PFS in these 18 patients was 5.2 months (95% CI, 1.7–15.9), with a 33% ORR. OS was significantly better in these patients compared with those who did not receive matched targeted therapy (17.0 vs. 5.0 months; HR, 0.29; 95% CI, 0.11–0.76; P = 0.008). In light of this and other accumulating evidence that advanced BTCs are good candidates for molecular triage, the European Society for Medical Oncology (ESMO) recently recommended that NGS be performed routinely in patients with cholangiocarcinoma (16).

Given the rarity of cholangiocarcinoma and the wide molecular heterogeneity of these tumors, the integration of molecular profiling with clinical and molecular classifications may offer a means of addressing the treatment paradigm for BTCs (17). Several classification systems offering an infrastructure to select patients with cancer for targeted therapies have been developed. The ESMO Scale of Clinical Actionability for molecular Targets (ESCAT) defines six categories for the use of drugs targeting molecular alterations, based on level of evidence taking into account study rigor and disease context (Supplementary Table S1; ref. 18). Genomic alterations are assigned to the tier which best reflects their clinical utility for selecting patients to receive a matched targeted therapy on the strength of evidence from clinical trials, thereby offering clinicians a means of prioritizing and selecting treatments. We performed a single institutional retrospective review of patients with advanced cholangiocarcinoma who had undergone tumoral molecular analysis to evaluate the impact on survival of targeted treatments administered according to ESCAT categorization.

We performed a retrospective review of all consecutive patients diagnosed with cholangiocarcinoma at the Vall d'Hebron Hospital (Barcelona, Spain) between January 2011 and January 2020. Patients were eligible for the study if they had a histologically confirmed diagnosis of cholangiocarcinoma and had undergone molecular profiling. Since 2011, we have routinely performed NGS in BTC patients while on first-line chemotherapy as part of our molecular prescreening program for early clinical trials (19). Individual therapy options were determined according to available scientific and clinical experience, clinical trial availability and approved agents at the time. Molecular analyses were approved on a per patient basis and written informed consent was obtained. The study was performed in conformance with the Declaration of Helsinki. The study was approved by the ethics committee.

Data collected from patients' electronic records included demographics, clinical history, surgery, systemic therapy, targeted therapies, response to targeted treatment and survival status, as of June 30, 2020. Molecular alterations were classified according to the first four tiers of the ESCAT classification (18).

Molecular profiling was performed using tumor tissue obtained from tumor core biopsies or surgical samples. DNA was analyzed using in-house NGS panels (mutations in between 21 and 61 genes based on Amplicon-based assays, and fusions in 26 genes based on the NanoString nCounter assay, NanoString Technologies; ref. 19) or the FoundationOne CDx hybrid-capture NGS service platform (324 genes for both mutations and fusions, Foundation Medicine, Inc). In addition, data on defective mismatch repair (dMMR)/microsatellite instability (MSI) status were collected when reported.

Our primary objective was to assess OS of patients with ESCAT I-IV alterations matched to targeted drugs compared with those with nonactionable alterations. OS was defined as the time from the starting date of treatment with first-line systemic chemotherapy to the date of death; patients without documented death at the cut-off date were censored at the date the patient was last known to be alive. Secondary objectives were to assess OS in patients with ESCAT I-IV alterations who received unmatched second-line chemotherapy; to assess prognostic factors for OS and PFS in first-line systemic therapy; to assess time from progression to first-line chemotherapy until death or last follow-up (OS2); to determine ORR and clinical benefit rate (partial response or stable disease >4 months) stratified by ESCAT categories, and most frequent molecular subsets in patients receiving matched targeted agents. PFS was defined as the time from the initiation of treatment with chemotherapy or targeted therapy to the date of first documentation of disease progression or death due to any cause, whichever occurred first.

Descriptive statistics were used to summarize baseline data from patients with molecular test results. Logistic regression was used to estimate the odds ratio of receiving a targeted treatment according to ESCAT classification of the alteration. Analysis of survival prognostic factors was restricted to patients who received first-line palliative chemotherapy whereas to analyze the efficacy of target therapies only patients who received a second-line and, consequently, were eligible to receive targeted therapies, were included. Time-to-event endpoints were estimated using Kaplan–Meier methods and the log-rank test for statistical comparison. Cox proportional hazard models were used to obtain HRs with 95% CIs. Multivariate Cox models were calculated to obtain HR with 95% CIs after adjusting for potential confounders (age, sex, primary tumor location and surgery, stage at diagnosis, and metastatic location type). Median follow-up was calculated using Kaplan–Meier reverse method. Stage was defined as per TNM 8^th edition. P values are two-sided. Statistical analyses were done with R software (version 3.6.3).

The data generated in this study are available within the article and its supplementary data files. Data acquired and/or used in the study can be made available to readers on request.

Data were collected from 327 patients with cholangiocarcinoma treated at our institution between 2011 and 2020, and who had available molecular analyses. Figure 1 shows the study flowchart with different population subsets eligible for descriptive molecular analysis and survival outcomes. Demographics, disease characteristics and treatment are presented in Table 1. The majority of the cohort (78.9%) had intrahepatic cholangiocarcinoma. Characteristics were similar between patients with intra- and extrahepatic cholangiocarcinoma, although more patients had advanced disease at diagnosis in the intrahepatic cohort (68.2% vs. 42.0%). Almost all patients had metastatic disease (320 patients; 97.9%), the majority of whom had received systemic therapy. The most common chemotherapy was a gemcitabine plus platinum-based combination regimen (80.1% of the cohort). Median number of treatment lines in the metastatic setting was 2 (1–5).

The majority of patients (88.4%) had broad NGS panel covering driver mutations, copy number alterations and fusions, either in-house or commercial tests. Only 7% and 5% had mutation or fusion coverage only, respectively. Also, 35% had tumors analyzed for dMMR/MSI status with non-NGS methods. Molecular analyses in the 327 patients revealed that the most common alterations in the intrahepatic population included in mutations in IDH1/2 (23.1%/7.2% of patients), ARID1Amut (18.1%), BAP1mut (13.5%), KRASmut (11.5%), FGFR2 fusions (8.0%), and PIK3CAmut (7.7%; Fig. 2). This molecular profile differed from that of patients with extrahepatic cholangiocarcinoma, with the most common alterations being in KRASmut (33.3% of patients), ARID1Amut (12.5%), and BAP1mut (6.3%), whereas common, but less frequent, were BRAFmut (9.5%), BRCA2mut (6.3%), MSI (8.3%), and ERBB2 amplifications/mut (6.1%/4.8%). MET amplifications and BRCA1mut were found in <3% of cases.

Of the 327 patients, 184 (56.3%) had molecular alterations that could be classified according to ESCAT tiers I-IV, including 149 (57.8%) of the 258 patients with intrahepatic cholangiocarcinoma and 35 (50.7%) of the 69 patients with extrahepatic cholangiocarcinoma. Alterations classified as ESCAT tier I-II were more common in the intrahepatic population (75 patients, 50.3%) than in the extrahepatic population (11 patients, 31.4%). Supplementary Tables S2 and S3 present patient characteristics for intrahepatic and extrahepatic cholangiocarcinoma, respectively, stratified by ESCAT category.

Among the 184 patients with a tumor harboring an alteration with actionability according to ESCAT (tiers I-IV), 50 (27.2%) received corresponding targeted therapy, administered in the second or later lines. The large majority (48/50 patients) had intrahepatic cholangiocarcinoma, representing 32.2% (48 of 149) of patients with intrahepatic cholangiocarcinoma, compared with 5.7% (2 of 35 patients) of patients with extrahepatic cholangiocarcinoma. Table 2 presents identified alterations which could be classified according to ESCAT and any corresponding therapies administered. Most of the 184 patients had alterations categorized as IIB (N = 82, 44.6%) or IIIA (N = 54, 29.3%). The most common mutations for which patients received matched therapy were FGFR2 fusions and IDH1 mutations. Patients with alterations classified as ESCAT I-II were significantly more likely to receive corresponding treatment (44 of 86 patients; 51.2%) than those with alterations classified as ESCAT III-IV (6 of 98 patients; 6.1%), with an odds ratio of 15.3 (95% CI, 6.1–49; P < 0.001).

After a median follow-up of 45 months (95% CI, 40.1–not reached) in the 314 patients who received systemic treatment in the first-line metastatic setting, median OS was 16.5 months (95% CI, 14.5–18.8), and median PFS of palliative chemotherapy was 6.1 months (95% CI, 5.6–7.0). Among the 262 patients treated with a gemcitabine/platinum-based combination in the first-line, median OS was 15.5 months (95% CI, 14.2–18.7), and median PFS was 6.1 months (95% CI, 5.4–7.1).

Figure 3 shows prognostic factors for OS and PFS in first-line therapy setting in the overall population of 314 eligible patients who received systemic treatment. Median OS in the metastatic setting was significantly longer in patients with stage I-III at diagnosis (18.8 months, 95% CI, 16.6–25.3) compared with stage IV [14.6 months, 95% CI, 13.1–17.6; HR, 1.43 (95% CI, 1.09–1.86), P < 0.01]. Patients with alterations classified as ESCAT tier I-II (N = 86) had numerically longer OS (20.1 months, 95% CI, 17.4–26.0) than other patients (i.e., those with alterations classified as ESCAT tier III-IV or without ESCAT alterations (14.5 months, 95% CI, 12.7–17.7). Univariate analysis showed a significant difference (HR, 1.35; 95% CI, 1.01–1.80; P = 0.04), however significance was not maintained in multivariate analyses (P = 0.08; Fig. 3A). In terms of PFS under first-line treatment, outcomes were better in the ESCAT tier I-II population (7.1 months, 95% CI, 6.4–9.2) versus patients with alterations classified as ESCAT tier III-IV and those without actionable ESCAT alterations [5.5 months, 95% CI, 4.8–6.1; HR, 1.30 (95% CI, 1.00–1.69), P = 0.05; Figs. 3B and 4C].

Figure 4A shows Kaplan–Meier OS curves stratified by ESCAT tiers and exposure to matched targeted therapies. Median OS in 50 patients with ESCAT I-IV alterations who received matched therapy was 22.6 months (95% CI, 20.1–32.8), compared with 14.3 months (95%CI, 11.9–18.1) in 130 patients without actionable ESCAT alterations (HR, 0.58; 95% CI, 0.40–0.85; P = 0.005). After adjusting for age, sex, primary tumor location and surgery, stage at diagnosis, and metastatic location type the difference remains statistically significant (HR, 0.59; 95% CI, 0.40–0.87; P = 0.007). As a reference, median OS was 19.1 months (95% CI, 16.0–22.8) in 80 patients with ESCAT I-IV alterations who did not receive matched therapy but were eligible for second-line treatment (Fig. 4A). In addition, median OS2 (from progression to first-line chemotherapy until death or last follow-up) was 14.9 months (95% CI, 13.2–20.4) in ESCAT I-IV patients who received matched therapy, compared with 9.2 months (95% CI, 8.4–11.2) in ESCAT I-IV patients who did not receive matched therapy and 6.2 months (95% CI, 5.1–10.0) in patients without actionable ESCAT alterations (multivariate HR, 0.65; 95% CI, 0.43–0.95; P = 0.03; Fig. 4B).

Among patients who received targeted therapy, median PFS for patients with ESCAT tier I-II alterations was significantly longer than in patients with ESCAT tier III-IV alterations (5.0 months, 95% CI, 2.9–7.7 vs. 1.9 months, 95% CI, 1.2–not reached, respectively; HR, 0.36; 95% CI, 0.15–0.87; P = 0.02; Fig. 4D).

The 20.5% ORR in patients receiving targeted treatments for alterations classified as ESCAT tier I-II (8/39 evaluable patients) was similar to that in patients with alterations classified as ESCAT tier III-IV (20.0%, 1/5 evaluable patients). The clinical benefit rate was 71.8% (28/39 patients) in the former group compared with 40.0% (2/5 patients) in the latter (Table 1).

Survival was evaluated according to the main alterations identified in the 50 patients who received ESCAT-matched therapy. Median OS in the 23 patients treated with IDH inhibitors was 21.9 months (95% CI, 18.0–34.6), and median PFS was 3.4 months (95% CI, 2.5–5.2). More promising results were seen in the context of FGFR2 fusions or BRAF^V600E mut. Median OS for the 16 patients who had received targeted therapy for FGFR2 fusions was 32.0 months (95% CI, 21.9–not reached), and median PFS was 9.3 months, 95% CI, 3.7–not reached). Median OS in the 5 patients with BRAF mut exposed to targeted BRAF + MEK therapies was 33.7 months (95% CI, 15.4–not reached) and median PFS of this subpopulation was 9.3 months (95% CI, 7.5–not reached).

After failure on first-line chemotherapy for patients with advanced cholangiocarcinoma, second-line treatment options are limited. A recent phase III study showed a modest benefit when mFOLFOX was added to active symptom control in the second-line advanced BTC setting (20). Beyond these results, targeted therapies offer a glimmer of hope to patients with tumors harboring druggable targets. Several studies including patients with cholangiocarcinomas are ongoing, targeting FGFR fusions, IDH1 mutations, NTRK fusions, MSI, and other genetic aberrations (21). Personalized molecular characterization of cholangiocarcinomas using NGS offers a means of identifying therapeutic options to improve outcome in this rare disease with a dismal prognosis.

The ESCAT categorization was designed to provide a universal ‘language’ to assist precision cancer medicine in the care arena, overcoming some of the ambiguities of earlier proposals (22). In our analysis of a large cohort of advanced cholangiocarcinoma patients who had undergone molecular profiling, 26% had actionable alterations of high clinical relevance (categorized as ESCAT I/II), and 30% had potentially actionable alterations lacking proven therapies (categorized as ESCAT III-IV), whereas 44% of the population had no potentially actionable alterations. Our study reflected the value of NGS in guiding the choice of targeted therapies in cholangiocarcinoma. Among patients with alterations classified according to ESCAT, median OS was significantly increased for those who received matched targeted therapy compared with controls who did not receive matched therapy. As expected, this was notably the case for patients with alterations with the most robust clinical evidence (i.e., tiers I-II; tumors harboring MSI, FGFR2 fusions, BRAF^V600E mut, and IDH1mut). It should also be noted that patients with ESCAT tiers I-II in BTC may have more favorable outcome with standard-of-care therapies, indicating a better prognosis, which favors access to second-line targeted therapy approaches and improved OS, irrespective of matched therapy exposure.

The population from our single center analysis is representative of cholangiocarcinoma in the wider geographical setting both clinically and molecularly. The targetable alterations of IDH1 mutations and FGFR2 fusions were identified in 23% and 8% of patients with intrahepatic cholangiocarcinoma, respectively, but were rare in extrahepatic cholangiocarcinoma. The overall mutational profile in our population is largely coherent with published reports of divergent genetic profiles between intrahepatic and extrahepatic populations, and of an enrichment of alterations in patients with intrahepatic cholangiocarcinoma (17, 23, 24). However, the diversity of assays and constant evolution in test performance and biomarker coverage during our study period may limit precise molecular epidemiology assessment. The proportion of cholangiocarcinoma patients with druggable molecular alterations with the potential for administration of matched molecular targeted agents in our study was comparable with the 39% rate in a large BTC cohort (4), but lower than the 68% actionability rate reported in the MOSCATO-01 BTC sub-analysis (15). The authors of MOSCATO-01 suggest that the high rate of druggable alterations in the BTC population compared with other tumor types, supports that BTCs are particularly suitable candidates for molecular triage programs, and notably intrahepatic cholangiocarcinoma (15). The most common reasons for not treating patients with matched therapies in our cohort were lack of the appropriate trial or effective drug. Importantly, targeted agents and immunotherapies were not approved for clinical use in cholangiocarcinoma at our institution during the study period. The only alternative to access matched drugs was through the phase I–III clinical trial portfolio, and interest in targets and drugs changed significantly over the last decade. Other potential reasons for not offering patients matched drugs were clinical deterioration and strict inclusion/exclusion criteria for trial enrolment. Of note, patients with ESCAT alterations who ultimately did not receive matched therapies (only second-line chemotherapies and beyond) were clinically “fit” at the time of molecular profiling and were frequently “unmatched” to targeted drugs given the lack of slots in phase I trials. Therefore, the selection bias of “matched” versus “not-matched” cohorts is in part minimized.

Alterations identified with NGS were classified according to ESCAT in three recent studies in patients with advanced breast cancer (25–27) and a study in urothelial carcinoma (28). These studies highlight the importance of molecular profiling for identifying rare gene alterations and possible druggable cancer-driving pathways. The significantly higher likelihood in our study of patients with alterations in tiers I-II receiving matched treatment than those with alterations in tiers III-IV, reflects the respective ESCAT definitions, with tiers I-II for drug matches associated with improved outcome in clinical trials or with promising antitumor activity, albeit an unknown magnitude of benefit, whereas in tiers III-IV, the match is only suspected to benefit patients based on clinical trials in other tumor indications (or with similar molecular alterations) as along with preclinical evidence.

A number of limitations may restrict the interpretation of our data, namely the retrospective and monocentric nature which can introduce selection bias or immortal-time bias. In addition, the small number of patients with genetic alterations in ESCAT tiers III and IV who received matched treatment provided a limited statistical power to thoroughly study this patient subgroup. Some studies have also highlighted that divergent profiling between primary and metastatic tissue, and multiple alterations in the same patient are potential issues and merit investigation (26). Although obtaining tissue for NGS from patients with cholangiocarcinoma presents difficulties, the benefit suggested in this and previous analyses support the ESMO recommendation that NGS be implemented into the routine clinical workup for patients with advanced cholangiocarcinoma (16). We advocate for early access to comprehensive genomic profiling, which opens the door to promising molecularly guided therapies in the second-line setting and beyond. However, definitive evidence of the efficacy of ESCAT-matched targeted treatments can only be evaluated in the context of prospective clinical trials.

As technology associated with NGS platforms improves, combined with simplified logistics and decreased costs, sequencing multiple actionable genomic alterations susceptible to targeted treatments should become increasingly accessible in routine clinical practice. Integrating the application of ESCAT into treatment management for patients with cholangiocarcinoma, and notably in the case of molecularly rich intrahepatic cholangiocarcinoma, enriches the clinical picture and offers clinicians and patients a valuable evidence-based tool (29). It allows physicians to combine the tumor genetics and clinical experience to identify molecular alterations with higher impact for the patient's outcome, exploit effective matched treatments (including those not yet approved), and sidestep potentially toxic ineffective therapies for this disease with a poor prognosis and few therapeutic options. We conclude that implementing an evidence-based scale such as ESCAT into treatment management paradigms complements the assessment of the clinical picture in advanced cholangiocarcinoma, and may guide clinicians for prioritizing alterations and avoiding overestimating the potential benefits of tailored therapies, by adjusting expectations to the clinical reality.

H. Verdaguer reports other support from Eisai, Merck, and AstraZeneca outside the submitted work. D.A. Acosta reports nonfinancial support from Lilly, Roche, and Pfizer outside the submitted work. J. Hernando reports personal fees from EISAI, IPSEN, NOVARTIS, ADACAP, ROCHE, PFIZER, LEO, and ANGELINI outside the submitted work. P. Nuciforo reports personal fees from Novartis, Bayer, MSD Oncology, and Targos outside the submitted work. S. Peiró reports grants from Instituto de Salud Carlos III, Fundació Marató TV3; other support from AECC, MINECO, FERO Foundation, La Caixa Foundation, and Cellex Foundation outside the submitted work. G. Villacampa reports personal fees from MSD, AstraZeneca, and Pierre Fabre outside the submitted work. J. Tabernero reports personal fees from Array Biopharma, AstraZeneca, Avvinity, Bayer, Boehringer Ingelheim, Chugai, Daiichi Sankyo, F. Hoffmann-La Roche Ltd, Genentech Inc, HalioDX SAS, Hutchison MediPharma International, Ikena Oncology, IQVIA, Lilly, Menarini, Merck Serono, Merus, MSD, Mirati, Neophore, Novartis, Ona Therapeutics, Orion Biotechnology, Peptomyc, Pfizer, Pierre Fabre, Samsung Bioepis, Sanofi, Seattle Genetics, Scandion Oncology, Servier, Sotio Biotech, Taiho, Tessa Therapeutics, TheraMyc, and Imedex, Medscape Education, MJH Life Sciences, PeerView Institute for Medical Education and Physicians Education Resource (PER) outside the submitted work. R. Dienstmann reports personal fees from Roche, Amgen, Boehringer-Ingelheim, Ipsen, Servier, Sanofi, Bayer, and MSD; grants from Merck, Pierre Fabre, and BMS; and other support from Trialing outside the submitted work. T. Macarulla reports personal fees from Swedish Orphan Biovitrum (SOBI) AB, Ability Pharmaceuticals SL, Advance Medical HCMS, Aptitude Health, Basilea Pharma, Baxter, BioLineRX Ltd, Celgene, Eisai, Genzyme, Got It Consulting SL, IATTGI, Imedex, Ipsen Bioscience, Laboratorios Menarini, Lilly, Marketing Farmacéutico & Investigación Clínica SL, MDS, Medscape, Monte Verde SA, Novocure, Paraxel, PPD Development, QED Therapeutics, Roche Farma, TRANSWORLD EDITORS SL, and Zymeworks outside the submitted work, as well as personal fees and other support from AstraZeneca, Incyte, Sanofi-Aventis, and Servier. No disclosures were reported by the other authors.

H. Verdaguer: Conceptualization, supervision, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. T. Saurí: Investigation, writing–review and editing. D.A. Acosta: Investigation, writing–review and editing. M. Guardiola: Software, formal analysis, methodology, writing–original draft, writing–review and editing. A. Sierra: Investigation, writing–review and editing. J. Hernando: Investigation, writing–review and editing. P. Nuciforo: Investigation, writing–review and editing. J.M. Miquel: Investigation, writing–review and editing. C. Molero: Investigation, writing–review and editing. S. Peiró: Investigation, writing–review and editing. Q. Serra-Camprubí: Investigation, writing–review and editing. G. Villacampa: Conceptualization, supervision, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. S. Aguilar: Investigation, writing–review and editing. A. Vivancos: Investigation, writing–review and editing. J. Tabernero: Investigation, writing–review and editing. R. Dienstmann: Conceptualization, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. T. Macarulla: Conceptualization, supervision, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

The work was funded by the Vall d'Hebron Institute of Oncology (VHIO). We thank Dr. Sarah MacKenzie, PhD, for medical writing assistance (funded by VHIO).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.