Summary: Anti-EGFR therapies have failed to improve survival for unselected patients with metastatic gastroesophageal cancer, but in a subset of patients, EGFR amplification may predict treatment benefit. Maron and colleagues report the clinical activity of anti-EGFR therapies in a cohort of patients with EGFR-amplified metastatic gastroesophageal cancer and utilize serial blood and tumor tissue collection to identify molecular drivers of treatment sensitivity and resistance. Their insights offer a path to overcome technical limitations associated with EGFR amplification and facilitate molecularly targeted therapeutic strategies. Cancer Discov; 8(6); 679–81. ©2018 AACR.
See related article by Maron et al., p. 696.
Gastric and esophagogastric junction adenocarcinomas (GEA) are a leading cause of cancer-related death worldwide. Despite modest survival improvements from molecularly targeted therapies, survival outcomes remain poor. Improved biomarkers are needed to identify patients most likely to benefit from existing therapies, and novel immunotherapies and targeted therapies are needed to more broadly improve survival.
The epidermal growth factor receptor (EGFR; ERBB1) is a member of the ERBB receptor tyrosine kinase superfamily. Multiple ligands—including epidermal growth factor (EGF), TGFα, amphiregulin, and epiregulin—bind EGFR, thereby inducing tyrosine phosphorylation and receptor homodimerization and heterodimerization. EGFR activation stimulates multiple downstream signaling cascades, including the RAS–RAF–MAPK pathway, the PI3K pathway, the phospholipase C gamma protein pathway, and the STAT signaling pathway. Activation of these pathways promotes cell proliferation, angiogenesis, migration, survival, and adhesion (1). Deregulation of EGFR signaling can occur due to receptor overexpression, activating mutations, and gene copy number (GCN) gain, which promotes malignant transformation and metastasis. Activating EGFR mutations are associated with response to EGFR tyrosine kinase inhibitors (TKI) and are most prominent in metastatic non–small cell lung cancer (NSCLC). EGFR monoclonal antibodies, on the other hand, are active in tumors that overexpress EGFR, independent of EGFR mutations. EGFR monoclonal antibodies are thought to have multiple mechanisms of antitumor activity, including competitive inhibition of ligand binding, receptor endocytosis, antibody-dependent cell-mediated cytotoxicity (ADCC), and complement mediated cytotoxicity (1). EGFR inhibitors have survival benefit in patients with metastatic colorectal cancer, head and neck cancer, NSCLC, and pancreatic cancer, but results for patients with metastatic GEA have so far been disappointing (2, 3). Despite negative results for anti-EGFR therapies in multiple randomized trials, survival benefit has been observed in a minority of patients with GEA, suggesting that a molecular predictor may yet identify patients likely to benefit (4, 5).
In this issue of Cancer Discovery, Maron and colleagues (i) describe the frequency of EGFR amplification and GCN gain in a cohort of patients with GEA; (ii) evaluate the correlation between GCN and protein expression; (iii) prospectively screen patients with EGFR-amplified GEA for treatment with EGFR antibody therapies; (iv) report clinical outcomes from patients treated with EGFR antibody therapies; (v) perform molecular characterization of pretreatment and posttreatment tumor biopsies and serial circulating tumor DNA (ctDNA) to identify mechanisms of treatment resistance; and (vi) evaluate the contribution of ADCC to antitumoral effects (6). This analysis is strengthened by the use of orthogonal methods of EGFR assessment, molecular profiling of primary and metastatic sites of disease, pretreatment and posttreatment tumor biopsies, and serial ctDNA assessments. The dataset presented by Maron and colleagues offers compelling evidence that EGFR amplification is a therapeutically actionable target, albeit ideally in the context of a prospective randomized clinical trial.
To better understand the clinical relevance of EGFR amplification for patients with metastatic GEA, the authors describe the prevalence of EGFR amplification in a large single-institution cohort. In this University of Chicago cohort (N = 363), the frequency of EGFR amplification (5%) was comparable with that observed in an independent commercial database of 4,645 GEA cases (6%) and The Cancer Genome Atlas (4%). After accounting for variables associated with EGFR amplification—including advanced tumor stage and proximal tumor location—the three databases are strikingly similar. Past efforts to characterize EGFR amplification and GCN have been confounded by the absence of a universal diagnostic scoring criterion, interlaboratory variability, and intratumoral and intertumoral heterogeneity (7). Indeed, intratumoral and intertumoral heterogeneity are a common feature of metastatic GEA, and these same authors have shown that discordant profiling results between primary and metastatic tissue may lead to treatment reassignment in up to a third of patients (8). Benchmarking results from the University of Chicago dataset against independent datasets are critical. The reproducibility of EGFR amplification results allays concerns about interlaboratory variability and the lack of standardized scoring criteria. Additionally, the statistically significant linear correlation between EGFR GCN and protein expression by mass spectrometry further supports EGFR GCN as a valid biomarker and an actionable therapeutic target.
Several studies have examined EGFR GCN as a biomarker to predict sensitivity to anti-EGFR therapy; however, thus far results have been inconsistent. A meta-analysis of 19 studies by Yang and colleagues concluded that—in patients with metastatic colorectal cancer—EGFR GCN gain is generally associated with benefit from anti-EGFR antibodies (7). However, these same authors concluded that technical limitations of measuring EGFR GCN limit clinical utility. Because KRAS and NRAS mutations are strong negative predictors of EGFR antibody therapy in metastatic colorectal cancer—and patients without EGFR amplification may still benefit from anti-EGFR therapies—additional studies of the biomarker have not been pursued. In this study, the authors report outcomes from a cohort of 7 patients who were prospectively screened for EGFR amplification and treated with anti-EGFR monoclonal antibodies. Exceptional clinical benefit was observed in several patients, with an objective response rate of 57% (4 of 7 patients). Although some of these treatment responses are confounded by the addition of chemotherapy to anti-EGFR therapy, the depth and duration of treatment response exceeds expectations for conventional chemotherapy alone and provides supporting evidence that the addition of anti-EGFR therapy was a significant contributor. The complete response observed in a subject receiving third-line single-agent cetuximab further supports EGFR inhibition as a primary driver of clinical benefit. Because EGFR inhibitors are not approved for the treatment of metastatic GEA, identification of predictive biomarkers for the therapeutic class is of critical importance.
Even with the impressive benefit observed in this small cohort, the optimal EGFR targeting strategy is unknown. Prior post hoc analyses from randomized clinical trials suggested that patients with EGFR-amplified metastatic GEA benefit from both anti-EGFR antibodies and TKIs (4, 5). In this study, patients received either cetuximab, a chimeric IgG1 monoclonal antibody that binds the EGFR ectodomain, or ABT-806, an anti-EGFR antibody that binds mutant EGFRvIII with high affinity. The authors suggest that the optimal therapeutic approach may include an anti-EGFR monoclonal antibody, as this therapeutic class has an immune-mediated mechanism of action (ADCC). Descriptive analysis from this study found interval increase of CD3+ infiltrate in patients treated with EGFR antibodies, suggesting that cell death may in part be immune mediated. These descriptive findings merit independent confirmation in a larger dataset. Whether ADCC is critical to the activity of anti-EGFR antibodies is unknown. The anti-EGFR monoclonal antibodies cetuximab and panitumumab have important structural differences that affect ADCC, and yet a head-to-head comparison in patients with metastatic colorectal cancer indicates that these antibodies have identical therapeutic activity (9). Unfortunately, the limited availability of EGFR-amplified GEA cell lines and xenografts hinders efforts to perform additional preclinical studies and identify the optimal therapeutic approach.
Even after an appropriate therapeutic approach is identified, conducting a prospective clinical trial for patients with EGFR-amplified metastatic GEA may face technical limitations. Given the substantial heterogeneity of metastatic GEA tumors, generating a reliable ROC-based cutoff point to distinguish between responders and nonresponders is challenging (10). Additionally, application of that cutoff point in a prospective clinical trial may face similar hurdles. Studies have reported significant heterogeneity in different sections within a tumor, leading to potential misclassification of EGFR GCN status in over a third of patients (10). The cases presented in this report had exceptionally high EGFR GCN by next-generation sequencing, ranging from 54 to 167 copies. There were no patients with low- or intermediate-grade amplification (8–53 copies). Those tumors with lower GCN may be particularly susceptible to misclassification. Blood-based “liquid biopsies,” which offer an anatomically unbiased profile of total mutational burden, may provide an important point of reference for tumor tissue results and could, in some cases, be used to identify patients for inclusion or exclusion from EGFR-directed therapies (8). In this study, responders had substantially higher EGFR copy number in blood (33.9) than nonresponders (2.5), suggesting that EGFR amplification in blood can complement tissue analyses of EGFR GCN and potentially prevent misdirected therapy. The authors should be congratulated for recognizing the nuanced relationship between tumor and blood-based profiling results and clinical actionability. Tumor heterogeneity and other predictive factors are captured in the “genogram” (see Fig. 4G), a framework to detail those predictive factors that may be used to identify patients appropriate for molecularly targeted therapy.
In summary, Maron and colleagues should be commended for their thorough and well-executed study. Through their comprehensive investigation of a large clinically annotated institutional dataset, they were able to identify a potentially actionable therapeutic target, prospectively treat patients with rational targeted therapies, and then identify mechanisms of primary and acquired resistance. This investigation offers rare insight into EGFR amplification, a poorly understood but clinically important molecular driver for patients with metastatic GEA. Their efforts to better characterize the clinical actionability of this molecular driver—coupled with serial blood and tumor tissue collection to identify drivers of treatment resistance—will contribute to ongoing drug-discovery efforts. It is through these efforts that investigators will better understand the foundations of GEA tumor biology and advance precision medicine initiatives.
Disclosure of Potential Conflicts of Interest
J.H. Strickler reports receiving commercial research support from AbbVie, Exelixis Inc., and Amgen, and is a consultant/advisory board member for Amgen, Genentech/Roche, and Bayer.