The human epidermal growth factor receptor-2 (HER2) is a receptor tyrosine kinase, belonging to the ErbB family (EGFR/HER1, HER2, HER3, and HER4), involved in signal transduction pathways that mediate key cellular processes including cell proliferation, differentiation, and survival (1). HER2 deregulation via gene mutation, amplification, and post-transcriptional upregulation has been observed in a wide array of human cancers. Importantly, it has been well-established that its aberrant expression and activation is sufficient to drive cell transformation and oncogenesis in preclinical models; an observation consistent with clinical and etiological findings in certain cancer indications (2, 3).

Among the clinical evidence supporting its oncogenic driver role, HER2 antagonists have had demonstrable clinical activity in patients with HER2 overexpressing breast and gastric cancers. Clinical trials evaluating trastuzumab, a monoclonal antibody targeting HER2, when combined with chemotherapy in patients with HER2-positive breast and gastric cancers yielded objective response rates (ORR) of ~50% and median overall survivals (mOS) surpassing 1–2 years (4, 5). More recently it has been shown that addition of pertuzumab, an antibody targeting a different HER2 epitope, to trastuzumab/chemotherapy increased ORR to 80% and mOS to 56.5 months in patients with metastatic breast cancer (6, 7). Based on these encouraging results, HER2 therapies have been approved in the first-line setting for the treatment of metastatic breast and gastric cancer patients.

Similarly, impressive response rates and clinical benefit have been reported in clinical trials evaluating anti-HER2 agents in patients with heavily pretreated HER2-positive colorectal cancer (CRC). Among 33 patients enrolled in the phase II HERACLES trial testing trastuzumab and lapatinib, a small molecule HER2/EGFR tyrosine kinase inhibitor, in patients with ERBB2 (HER2) amplified/KRAS wild-type CRC, the ORR was 30% with disease stabilization achieved in 70% of patients (8). Importantly, the observed responses were durable; half of the patients remained on therapy for more than 6 months and one patient for over 3 years (9). Notably, patients with higher tissue ERBB2 copy numbers (CN; ≥10 copies) derived the most clinical benefit and had a significantly longer time to progression compared to those with lower copy number gains (53.1 vs. 34 weeks, respectively). These findings compare favorably to the response rates observed with immunotherapy (ORR 35%) in patients with treatment-refractory CRC and underscore ERBB2 gene amplification and overexpression as a targetable oncogenic driver in advanced CRC (10, 11).

Historically, ERBB2 gene amplification and protein overexpression in tumor biopsy material, as measured by in situ hybridization (ISH) or immunohistochemistry (IHC), has been used to select patients most likely to benefit from HER2-based therapeutic strategies. A major challenge with this approach is the described inter- and intratumoral heterogeneity in HER2 status, particularly in cancers of gastrointestinal (GI) origin and breast cancers, due to tissue sampling bias. ERBB2 gene amplification identified by next-generation sequencing (NGS) of tissue DNA or cell-free DNA (cfDNA) in the blood is emerging as a robust biomarker predictive of response to HER2 targeted agents and offers an alternative to IHC/ISH (12). We recently reported that among a cohort of HER2-positive gastric cancer (GC) patients treated with a combination of lapatinib and capecitabine/oxaliplatin, the response rate for those with detectable ERBB2 copy number amplification in their blood was 100% (12, 13). Moreover, it has been shown that cfDNA captures tumor heterogeneity in GC patients, allowing for the identification of the 10–20% HER2-negative primary tumors with synchronous HER2-positive metastatic lesions (12, 14–17).

Here, we examined the prevalence and genomic landscape of potentially actionable ERBB2 gene amplification-positive advanced solid tumors as detected by NGS of cfDNA in a large cohort of patients from Asia. We further present case reports of patients with metastatic and treatment-refractory GC that had meaningful clinical responses to HER2-based therapies initiated upon detection of ERBB2 gene amplification by cfDNA testing.

While it is well-appreciated that ERBB2 (HER2) gene amplification and protein expression is a bona fide therapeutic target in the treatment of patients with breast and gastric cancers, more recent trials have drawn attention to its actionability in other cancer types (23, 24). We report here that focal ERBB2 gene amplification, as identified by cfDNA analysis, is common in various solid tumor types in a large cohort of patients from Asia. Strikingly, the prevalence of ERBB2 gene amplification in Asian GC (21.4 vs. 11%) and CRC (11.1 vs. 5.5%) patients is higher than in non-Asian patient samples (Figure 3A). The former observation is consistent with a large study that identified HER2 overexpression, by IHC/ISH, in 9.7% of all Asian GC patients (n = 5,301) and 18.1% when China was excluded from the analysis, suggesting regional differences in prevalence (25). In the case of primary CRC, however, another study reported an ERBB2 amplification/overexpression frequency of only 3.3% in an Asian patient cohort (n = 4913). A potential explanation for the divergent findings is that the patient population reported here is biased toward heavily pre-treated patients such that they may have acquired ERBB2 gene amplification during the course of disease progression or as a mechanism of resistance to prior therapies (26). Interestingly, the absolute ERBB2 gene copy number gains as measured in plasma are highest in GI cancers, in particular in Asian CRC patients (median CN 6.6 vs. 2.7 in non-Asian patients; Figure 3B). While additional studies in larger Asian patient cohorts would be required to corroborate these findings and achieve statistical significance, they clearly demonstrate that a substantial portion of GC and CRC cancer patients present with or acquire high-level focal ERBB2 amplification.

Further supporting a pathogenic role for HER2 in patients with GC and CRC is the finding that ERBB2 gene amplification is rarely detected in conjunction with a canonical oncogenic mutation (Figure 5). Only 2 of 5 (40%) ERBB2 amplified CRC patients displayed concurrent oncogenic alterations; the first patient displayed multiple subclonal driver variants (BRAF V600E, KRAS Q61H, and a CCDC6-RET fusion) as well as EGFR ectodomain mutations (G465R, S464L) suggestive of the acquisition of several on- and off-target resistance mutations in response to standard-of-care cetuximab therapy while the second patient had a co-occurring subclonal KRAS G12R mutation (Figure 5, Supplementary Table 2). Recent cfDNA results from the HERACLES study (lapatinib plus trastuzumab in ERBB2-amplified metastatic CRC) found that the majority of non-responders (6/7) harbored concomitant clonal RAS/RAF gene mutations at baseline while subclonal RAS/RAF/PI3K-AKT pathway alterations emerged at progression after initial response or disease stabilization. The presence of subclonal RAS/RAF mutations co-occurring with ERBB2 amplification in this cohort of CRC patients begs the question if targeted HER2 therapy may still achieve benefit and/or if a combination approach might be warranted (27, 28). One ERBB2 amplified GC patient had a co-occurring subclonal PIK3CA E545G variant (1/6; 16.6%); one other had an atypical KRAS V14I loss-of-function mutation although its clinical significance remains to be elucidated. This is in stark contrast to non-GI cancers where half display co-occurring oncogenic alterations (Figure 5).

The most common co-occurring alterations observed in ERBB2 amplified GI patients were copy number gains in other genes, including other receptor tyrosine kinases (ex. MET, EGFR, FGFR1) and cell cycle genes (ex. CCNE1, CDK4/6). Many of these can, in principle, serve as bypass mechanisms of resistance to not only standard-of-care but HER2-directed therapies and are potentially targetable. Understanding the genomic landscape of various GI tumor types or in individual patients may facilitate the determination of optimal combination therapies in specific genetic contexts. For instance, two-thirds of ERBB2-positive patients in the CRC cohort had evidence of EGFR co-amplification providing potential rationale for combining a HER2 targeted agent with cetuximab/chemotherapy in these patients. Alternatively, treatment with a dual HER2/EGFR inhibitor, such as lapatinib, may be warranted in this scenario.

Here we presented two patient case reports of patients with metastatic GC with high-level focal ERBB2 amplifications as detected by cfDNA. These reports provide examples of how cfDNA testing can inform treatment decision both at diagnosis and at progression by (1) identifying the actionable alteration/capturing intertumoral heterogeneity and (2) confirming a lack of co-occurring mutations that can confound response to targeted therapy. Neither patient had evidence of alternative driver alterations in the plasma and both benefited from combined trastuzumab + chemotherapy therapeutic regimens (Figures 6A,B). Critically, ERBB2 amplification and overexpression was identified earlier in the first patient's clinical journey and hence he experienced a partial and a more durable response to therapy. Unfortunately, the second patient had already failed several lines of therapy but remarkably still had disease stabilization and symptom improvement when treated with trastuzumab. In addition to these case reports, several others have been published demonstrating that patients with advanced cancers treated based on a positive cfDNA test result derive clinical benefit from the corresponding molecularly-targeted therapy (summarized in Table 1). An important consideration is that Asian patients may have differential sensitivity to specific drugs and combination therapy approaches. It has been reported that a subset of Asian patients with GC derived benefit from treatment with lapatinib and trastuzumab in the LOGiC trial, while this combination was not effective in Caucasian patients (33). Hence optimal treatment strategies may need to be independently established for patients from different geographical regions and racial/ethnic backgrounds.

Nonetheless, an important advantage of cfDNA testing is the ability to capture both inter- and intratumoral heterogeneity to obtain a holistic view of the genomic landscape in primary and metastatic lesions. In the PANGEA trial of personalized antibodies for gastroesophageal patients, profiling of both primary and metastatic tumor tissue specimens led to treatment reassignment in almost a third of patients, many of which had HER2-positive metastases (14). Importantly, cfDNA and metastatic tissue genotyping results were 87.5% concordant while substantive discordance was seen between primary and metastatic tissue genotyping and between primary tissue and cfDNA genotyping results. Similarly, in a recent trial of pembrolizumab as salvage therapy in advanced GC patients, cfDNA analysis identified a tissue microsatellite stable (MSS)/tumor mutation burden (TMB)-low patient with pronounced intratumoral heterogeneity as being TMB-high. A second biopsy at the same site confirmed MSI/TMB-high status consistent with the cfDNA result, further highlighting that plasma testing allows for the enrichment and analysis of DNA shed from various heterogeneous tumor sites (32). Such inter- and intratumoral heterogeneity is certainly not limited to GC but is also observed in CRC and other cancer types. It has been reported that CNAs are the major source of tumor heterogeneity in CRC development (34). The same authors found that ERBB2 amplification is acquired in 6–7% of metastases. Given that distal metastases are a major cause of morbidity in cancer patients, targeted treatment of the secondary lesion(s) is strongly advisable if not recommended, as in the case of breast cancer. cfDNA genotyping provides a non-invasive solution that can drive therapy selection in metastatic patients where tissue sampling multiple metastatic lesions is not practical or in some cases possible.

A limitation to our study is that the relative patient numbers by cancer type are skewed based on standard-of-care ordering practices. For instance, lung cancer patients are over-represented in the Guardant Health patient cohort given that testing for actionable alterations at diagnosis, or progression on a targeted therapy, is routine. In the same vein, we observed a lower prevalence of ERBB2 amplification in this breast cancer patient cohort compared to that in a general population (3 vs. 10–15%, respectively) but a higher incidence of ESR1 mutations (Supplementary Figure 3) associated with progression on hormone-based therapies. Another important limitation is the lack of clinically annotated Asian patient cfDNA samples (i.e., disease stage, lines of therapy, tissue test results, and clinical outcomes). Establishing the genomic landscape of ERBB2 amplified Asian patients in more defined clinical contexts will be key in identifying and defining appropriate treatment strategies, and their sequencing, by disease stage and patient treatment history. While an impressively high degree of concordance (97.9%) has been described between cfDNA and tissue ERBB2 amplification status of CRC patients enrolled in the HERACLES trial (27), a head-to-head comparison of tissue and blood-based test results in larger patient cohorts will be necessary to define and develop HER2 positivity criteria. Indeed, “low” expressers have been described to derive some clinical benefit from anti-HER2 agents such that a more extensive study will be necessary to define cut-offs, particularly in the absence of alternative therapies.

Here we demonstrate, for the first time, that high level, focal ERBB2 gene amplification, as identified by cfDNA testing, is a common event in Asian patients with advanced cancers. In particular, higher ERBB2 incidence and CN gains were observed in GC and CRC patients in the absence of other oncogenic alterations, indicating that HER2 may be the dominant driver of tumor proliferation in those settings. Given its established role as an oncogene in certain contexts and the availability of HER2 antagonists, both approved and in clinical development, ERBB2 amplification is an attractive therapeutic target. Comprehensive cfDNA testing represents a non-invasive method of assessing HER2 status in the metastatic disease setting. It not only captures inter- and intratumoral heterogeneity but allows for assessment of co-occurring somatic mutations and the identification of patients most likely to benefit from HER2-based therapeutic strategies.

The datasets generated for this study are available on request to the corresponding author.

JL, AF, VR, and RL conceived and designed the study. AF and YS collected, assembled, and analyzed the data. AF, KB, VR, and RL interpreted the data. JL, SK, and K-MK treated the patients and provided clinical samples. AF, JL, and RL wrote the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.