Back from the Brink: EGFR Inhibition in Gastroesophageal Cancer

In this issue of Clinical Cancer Research, Corso and colleagues (1), present new evidence supporting the rationale to target the EGFR pathway in molecularly defined gastroesophageal adenocarcinomas (GEA). Through molecular profiling of four GEA cohorts, they show that EGFR amplification is present in a small but significant proportion of patients (∼7%), and correlates with adverse biology and poor outcome. Using patient-derived xenografts (PDX), the authors identify dual EGFR inhibition as a novel strategy to target EGFR-amplified tumors and define new mechanisms of primary resistance (1).

Despite some advances in the management of metastatic GEA, the outcome of these patients remains dismal with a median overall survival of approximately 12 months (2). Chemotherapy represents the backbone of treatment for GEA; immunotherapy has recently shown some positive signal while targeted-therapies have been largely disappointing so far with only two approved drugs, trastuzumab and ramucirumab, mAbs inhibiting HER-2 and VEGF-R2, respectively. The former is used as first line option in combination with chemotherapy in ERBB2-overexpressing/amplified metastatic tumors while the latter is used in the second-line setting in unselected patients (2).

EGFR inhibition has been tested in combination with chemotherapy or alone in large randomized trials for patients with first- and second-line GEA, respectively (reviewed in ref. 2). The lack of patient selection based on EGFR expression or amplification in these studies, has probably played an important role in the negative outcome of these trials. Indeed, retrospective analysis of EGFR amplification status in the COG trial has demonstrated that a subpopulation of patients that might benefit from the EGFR inhibitor gefitinib does exist, prompting to revisit the use of EGFRi in this context (3).

The study by Corso and colleagues (1) offers an opportunity to reflect on the promises and challenges ahead of us in the future re-development of EGFRi in patients with GEA. The results described in the present study reinforce the notion that EGFR amplification has a strong negative prognostic value in both patients with localized and advanced/metastatic GEA. These observations, coupled with the availability of effective EGFR inhibitors identify a potential group of patients with GEA that might draw a significant benefit from EGFR blockade. Although there is general consensus surrounding the negative prognostic value of EGFR amplification in GEA, many questions regarding the predictive value of the biomarker remain open. The short answer to these questions is that EGFR amplification alone is not sufficient to identify patients who are likely to benefit from EGFR inhibition. Using patient data and in vivo experiments, Corso and colleagues (1) provide strong evidence suggesting that cooccurrence of EGFR and other receptor tyrosine kinase amplifications is associated with resistance to EGFRi. Chromosomal instability characterizes the vast majority of GEAs and coamplification in RTKs such as EGFR, KRAS, MET, FGFR1&2, IGFR along with ERBB2 and VEGFA represents an intrinsic feature of these tumors (2) and a clear hurdle in the development of EGFR inhibitors in GEA (Fig. 1A). The data presented by Corso and colleagues (1) echo results from a recent retrospective analysis of the REAL3 trial, where co-amplifications in RTKs were observed in a significant proportion of patients and correlated with lack of benefit from the EGFR mAb panitumumab in combination with chemotherapy in advanced/metastatic GEA (4).

Taken together, these observations highlight the paramount importance of molecular ultra-selection in the definition of patients with GEA who might be candidate to EGFR inhibition. From a technical prospective, the use of targeted next-generation sequencing or low-pass whole-genome sequencing on tissue and/or liquid biopsies might help the simultaneous identification of EGFR amplification (positive biomarker) and lack of other RTK amplifications such as KRAS or MET (negative biomarkers), streamlining the selection of the right patient (4).

Precision oncology relies on the identification of the right treatment for the right patient. Besides shedding insights on patient selection, the results presented in the current article (1) also improve our understanding of EGFR targeting in GEA. Using randomized preclinical trials in PDXs, the authors show that dual EGFR inhibition with mAbs and small molecules targeting EGFR is superior to either of these therapeutics used alone (1). Given feasibility and safety data from phase I trials (5) support the preclinical results presented by Corso and colleagues (1), dual EGFR inhibition appears as a sound approach for the development of combinatorial approaches and might represent a better option compared to combining EGFR inhibitors with chemotherapy.

The authors did not include chemotherapy in their preclinical studies to avoid any confounding effect that the addition of chemotherapeutics might have had on the interpretation of their results. Although the combination of EGFRi with fluoropyrimidine and oxaliplatin chemotherapy has been associated to long-term responses in some patients with GEA (6), it is worth noticing that not all the chemotherapy agents commonly used in GEA represent a good therapeutic partner to combine with EGFRi. Indeed, clinical and preclinical data from patients with GEA and patient-derived organoids (PDO) suggest that combining EGFRi with epirubicin, a chemotherapy agent commonly used as standard of care in GEA, causes antagonisms and paradoxical increase in cell viability (4).

Using PDXs, Corso and colleagues also show that inactivating mutations in TSC2, a critical negative regulator of mTORC1, are responsible for resistance to EGFR inhibitors and suggest that the combination of everolimus (an mTOR inhibitor) with erlotinib might overcome primary resistance (1). Given the frequent co-occurrence of EGFR amplification and TSC2 inactivating mutation in GEA (https://www.cbioportal.org), these data have significant implications for patient selection and pathway inhibition in future studies.

Taken together, all these data highlight the importance of using patient-centered preclinical models (PDXs and PDOs) in the drug discovery trail, especially in those situations such as EGFR amplification in GEA, where commercially available cell lines are limited or not available.

In summary, the study by Corso and colleagues (1), along with other recent data in the field (3, 4, 6), broaden the reach of precision anticancer medicine to GEA, highlighting novel mechanisms of sensitivity to molecularly targeted therapeutics in selected populations and will contribute to reshape the selection and therapeutic exploitation of EGFR-amplified GEAs. Unfortunately, none of the tools and the preclinical models used for patient selection in these recent studies were available when several landmark trials of EGFR inhibitors in GEA were designed and conducted more than ten years ago. The study by Corso and colleagues (1) highlight the promise of forward and reverse translation in precision oncology but also underscores some of the challenges ahead. Molecular (ultra)selection of patients for EGFR inhibitors or other targeted therapies in GEA appears as a critical step for accurate patient selection (Fig. 1B) but will inevitably narrow down the number of patients eligible for such therapies; affecting the speed of recruitment and the sustainability of clinical trials in this setting. Strategic alliances between pharmaceutical companies and large international networks of academic investigators will represent a key for the feasibility and the success of molecularly driven trials in GEA.

D.J. Pinato reports personal fees from ViiV Healthcare, Bayer Healthcare, Roche; grants and personal fees from BMS and MSD; personal fees from MiNa therapeutics, EISAI, H3B, Astra Zeneca, and personal fees from DaVolterra outside the submitted work. N. Valeri reports personal fees from Bayer, Eli Lilly, Pfizer, and Merck outside the submitted work; and other from Menarini BioSystem. No disclosures were reported by the other author.

N. Valeri is supported by Cancer Research UK (grant numbers A18052 and A22909), the National Institute for Health Research Biomedical Research Centre at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research (grant numbers A62, A100, A101, A159), the European Union FP7 (grant number CIG 334261) and the Katherine and Douglas Longden Chair in Oncology at Imperial College London.