ERBB Signaling Interrupted: Targeting Ligand-Induced Pathway Activation

Members of the EGFR family of receptor tyrosine kinases (RTK) transduce cues from the extracellular environment into the cell to regulate proliferation and survival in response to ligand-induced homodimerization and/or heterodimerization. The four EGFR family members (EGFR/ERBB1, ERBB2/HER2, ERBB3, and ERBB4) are all single-chain transmembrane receptors with a tyrosine kinase domain but have distinct characteristics, ligands, and preferred dimerization partners (1). For example, ERBB3 has weak kinase activity and relies on heterodimerization with other RTKs for activation. In addition, EGFR, ERBB3, and ERBB4 each have several ligands, whereas ERBB2 has none. Interestingly, there is partial overlap in ligand specificity such that EGFR and ERBB4 share three ligands and the ERBB3 ligands NRG1 and NRG2 also bind ERBB4. The ligands induce receptor dimerization and activation of downstream signaling.

Several genetic alterations in the EGFR family are well-established cancer drivers (2). Most notable among these are ERBB2 overexpression in breast cancer and EGFR mutations in lung adenocarcinoma. These common alterations are paradigms for precision oncology such that breast cancers and lung adenocarcinomas are routinely tested for ERBB2 overexpression or EGFR mutations, respectively. Advanced tumors harboring these alterations are treated with molecularly targeted agents such as ERBB2-directed antibodies and EGFR tyrosine kinase inhibitors (TKI). However, through large-scale cancer sequencing efforts like The Cancer Genome Atlas (TCGA) and the increasingly widespread genomic profiling of tumors, it is becoming clear that additional rarer mutations in this family of RTKs occur in tumors (3). Moreover, mutations in ligands that regulate the function of these receptors are also being discovered (4, 5). Initial studies of these rarer mutations suggest that many are oncogenic and can be targeted using ERBB-directed therapies (6). However, our understanding of the oncogenic properties of many of these mutations and the therapeutic vulnerabilities of the tumors harboring them remains limited despite the availability of numerous agents targeting components of the EGFR signaling network.

In this issue, Drilon and colleagues report an exceptional therapeutic response in a patient with advanced NRG1-rearranged lung adenocarcinoma treated with an investigational anti-ERBB3 mAb (GSK2849330) as part of a phase I clinical trial (Fig. 1; ref. 7). The patient had previously received two lines of systemic chemotherapy and a third line of immunotherapy in the setting of disease recurrence following resection and radiation for early-stage disease. Prior to trial enrollment, comprehensive molecular profiling of the patient's tumor revealed a somatic, in-frame gene rearrangement fusing exons 1–6 of CD74 with exons 6–13 of NRG1 to generate the chimera CD74–NRG1, which retains the EGF-like domain responsible for receptor binding. This gene fusion was initially reported in 5 invasive mucinous lung adenocarcinomas from never-smokers in 2014 (4). Ectopic expression of CD74–NRG1 in cells leads to ERBB2–ERBB3 heterodimerization and promotes cell survival and proliferation via activation of the MAPK and PI3K pathways (4).

The patient described in this report experienced a 90% reduction in tumor volume with a durable response to GSK2849330 lasting 19 months. The patient was subsequently treated with the EGFR family TKI afatinib but experienced further disease progression after 8 weeks of therapy. Three additional targeted therapy–naïve patients with NRG1-rearranged lung adenocarcinoma also did not respond to afatinib. In contrast, responses to afatinib have been described in other isolated cases of NRG1-rearranged lung carcinoma (5, 8). The authors note that no other responses to GSK2849330 were observed among the 29 clinical trial participants, who were enrolled on the basis of increased ERBB3 expression. This observation hints at a distinction between target expression and target activation in predicting response to ERBB3-directed therapies. Similarly, activating EGFR mutations (but not increased EGFR levels) are predictive of response to EGFR-directed therapies in lung cancer. In contrast, ERBB2-directed antibodies have activity in breast, esophageal, and gastric cancers with ERBB2 overexpression; this suggests the relationships among expression, activation, and response to molecular therapies may be biologically complex.

Drilon and colleagues demonstrate that GSK2849330 inhibits phosphorylation of ERBB2, ERBB3, and AKT and impairs cell proliferation in MDA-MB-175-VII cells (a breast cancer cell line harboring a DOC4–NRG1 fusion) and in HCC-95 cells (a lung cancer cell line with NRG1 amplification). Similar findings were noted with the pan-ERBB inhibitors afatinib and neratinib. These cells were dependent on ERBB2 and ERBB3 as evidenced by a reduction in growth with genetic depletion of either receptor. Although these cell-based studies revealed comparable antiproliferative effects of GSK2849330 and pan-ERBB inhibitors, only GSK2849330 induced tumor regression in an ovarian cancer patient-derived xenograft (PDX) model with a CLU–NRG1 fusion. Treatment with afatinib only impaired tumor growth, consistent with partial inhibition of ERBB signaling (Fig. 1). Further studies will be required to reconcile these in vitro and in vivo findings with afatinib and to assess any clinical role for this agent in the treatment of NRG1-rearranged solid tumors. Moreover, additional studies to investigate the specific molecular mechanisms underlying the differences in sensitivity to afatinib and GSK2849330 in NRG1-driven tumors will be valuable. This would include examination of the dimerization patterns, activation, and total protein levels of all EGFR family receptors (including EGFR and ERBB4) in the presence of different NRG1 fusions exposed to the therapies.

Importantly, the authors note that among over 17,000 solid tumors subjected to comprehensive DNA-based molecular profiling at their institution, NRG1 rearrangements were identified in 0.14% of non–small cell lung cancers, 0.13% of pancreatic adenocarcinomas, and 0.04% of breast cancers. Additional NRG1 rearrangements were identified from targeted RNA sequencing as well as whole-transcriptome sequencing of tumors from the TCGA. Other smaller studies suggest that NRG1 rearrangements in lung cancer may be enriched in never-smoker populations of Asian descent with invasive mucinous adenocarcinoma (4). Although larger patient cohorts will be necessary to explore these potential associations, the possibility of a NRG1 fusion should be considered in the context of these pathologic and demographic features.

The findings of this study suggest that NRG1 rearrangements may represent an actionable genetic alteration in lung and other solid tumors. Given the low frequency of these molecular alterations, innovative approaches to clinical trial design will be essential to characterize the activity of emerging therapeutics in these rare patient populations. Although clinical trials have historically focused on homogeneous patient populations with a single tumor type, it may prove difficult to identify sufficient patients with NRG1-rearranged tumors of a single tumor type to conduct prospective clinical trials. To address similar challenges associated with the study of another rare genetic driver, a recent phase I–II study demonstrating activity of the selective TRK inhibitor larotrectinib in solid tumors harboring TRK fusions incorporated 55 adults and children with various TRK fusions encompassing 17 distinct tumor types (9). This follows on the heels of multiple studies demonstrating activity of the anti–PD-1 antibody pembrolizumab across 15 microsatellite instability–high or mismatch repair–deficient metastatic solid tumor types—findings that led to the first FDA approval of a cancer therapeutic in conjunction with a biomarker and agnostic to tissue of origin. These “basket” trials incorporating distinct tumor types harboring a common molecular alteration or biomarker have the potential to expedite clinical evaluation of promising genome-directed therapies. However, caution must be exercised in generalizing findings across tissue types due to tissue-specific differences in response to targeted therapies in some contexts (10).

Should formal clinical studies confirm activity of ERBB3-directed therapies in NRG1-rearranged lung tumors, NRG1 fusions would add to a growing number of targetable genetic alterations in lung cancer. Currently, targeted therapies are approved for use in the treatment of advanced lung cancers harboring activating EGFR mutations, BRAF^V600E mutations, and rearrangements involving ALK or ROS1. Additional emerging targets include activating HER2 mutations, MET exon 14 splice alterations, and rearrangements involving RET or NTRK. As the number of actionable genetic alterations in lung and other cancers continues to rise, it will be essential to streamline molecular characterization of tumor specimens and circulating tumor DNA to achieve precision oncology patient care. With only limited tissue from advanced solid tumors generally available for analysis, it will be impractical to assay for individual genetic alterations in a sequential manner. With the growing recognition of diverse structural variants (including targetable chimeric proteins with multiple potential fusion partners), it may become increasingly important to incorporate RNA-sequencing platforms into comprehensive tumor genomic profiling to permit detection of fusion transcripts.

Members of the EGFR family of RTKs are critical therapeutic targets in multiple tumor types. This study highlights the potential for aberrant ligand expression to promote a dependency on ERBB3 signaling that could potentially be exploited for therapeutic purposes. Although the low frequency of NRG1 rearrangements may pose challenges in the clinical development of agents to overcome the effects of NRG1-mediated ERBB3 activation, it is essential to pursue promising therapies that may provide meaningful clinical benefit for individuals whose tumors harbor NRG1 fusions.

F.H. Wilson reports receiving a commercial research grant from Agios and is a consultant/advisory board member for Loxo Oncology. K. Politi reports receiving commercial research grants from AstraZeneca, Kolltan, Roche, and Symphogen, is a consultant/advisory board member for AstraZeneca, Merck, NCCN, Novartis, and Tocagen, and has received IP royalties from MSKCC/Molecular MD.

This work was supported by Mentored Clinical Scientist Research Career Development Award K08CA204732 (to F.H. Wilson) and Yale SPORE in Lung Cancer P50 CA196530 (to K. Politi).