Comprehensive genomic landscape of ERBB2 in Chinese GI tumors: mutation-centered landscapes and precision treatment opportunities

Abstract

Background:

ERBB2 aberrations are established oncogenic drivers with validated therapeutic relevance in breast cancer and emerging indications across solid tumors. In gastrointestinal malignancies, the prevalence, molecular contexts, and therapeutic implications of ERBB2 alterations remain incompletely defined, particularly for mutation-driven (non-amplified) disease.

Objectives:

This study aimed to delineate the genomic landscape of ERBB2 alterations across colorectal cancer (CRC) and gastric cancer (GC), with emphasis on mutation versus amplification subtypes and their associated molecular contexts.

Design:

This was a retrospective observational genomic cohort study based on next-generation sequencing (NGS) data from patients with gastrointestinal malignancies.

Methods:

Tumors harboring ERBB2 alterations were identified through targeted NGS. Variants were annotated using InterVar and classified according to oncogenic significance. Alterations were mapped to HER2 functional domains and integrated with co-mutation patterns, copy number profiles, tumor mutational burden (TMB), and microsatellite instability (MSI) status to characterize subtype-specific genomic features.

Results:

Across CRC and GC, ERBB2 mutations predominantly clustered within the HER2 kinase domain, with recurrent hotspots (R678Q, S310F/Y, L755S, V842I) largely classified as oncogenic or likely oncogenic. In CRC, ERBB2-mutant tumors frequently co-harbored alterations in APC, TP53, PIK3CA, ARID1A, and SMAD4, whereas ERBB2-amplified tumors showed co-gains in RARA, TOP2A, and SMARCE1. In GC, mutation-positive cases were enriched for APC, TP53, ARID1A, MUC16, and LRP1B alterations, while amplification was associated with EGFR and cell cycle regulators. Oncogenic ERBB2 mutation subgroups exhibited higher TMB and MSI-H enrichment than amplification-positive counterparts, with no material differences in overall copy-number burden between subtypes. These patterns indicate that non-amplified ERBB2-mutant tumors form a genomically distinct subset with potential immunogenic features.

Conclusion:

ERBB2 alterations in CRC and GC converge on recurrent kinase domain hotspots but arise within tumor type-specific genomic milieus that likely influence therapeutic response. Compared with amplification, oncogenic ERBB2 mutations are preferentially associated with higher TMB/MSI-H and characteristic co-mutation signatures, supporting the clinical evaluation of mutation-selective HER2 inhibitors and rational combinations with immune checkpoint blockade. Our findings expand the molecular epidemiology of ERBB2 in Chinese GI cohorts, suggesting potential implications for resistance to standard chemotherapy or anti-EGFR strategies in select settings.

Plain language summary

Insights into ERBB2 gene alterations in Chinese GI tumors and their impact on precision treatment

HER2, also called ERBB2, is a gene that can drive the growth of several cancers. While HER2-targeted therapies are well established in breast cancer, the role of HER2 alterations in gastrointestinal cancers such as gastric and colorectal cancer is less clear, particularly when HER2 is altered by mutation rather than amplification. Understanding these differences is important because different types of HER2 alterations may respond differently to treatment.

In this study, we analyzed genetic testing data from Chinese patients with gastric and colorectal cancers to better understand the patterns of HER2 mutations and their potential clinical significance. We found that HER2 mutations often occur in key functional regions of the protein and frequently coexist with other genetic changes. Compared with HER2 amplification, tumors with HER2 mutations were more likely to show high tumor mutation burden or MSI-H, features that may increase the likelihood of responding to immunotherapy. These findings suggest that HER2-mutated tumors may require different treatment strategies and could benefit from mutation-specific HER2-targeted therapies or combination approaches with immunotherapy.

Introduction

ERBB2, also known as human epidermal growth factor receptor 2 (HER2), is a well-recognized oncogenic driver that plays a critical role in tumorigenesis and therapeutic response.^1,2 HER2 aberrations can arise from gene amplification, protein overexpression, or activating mutations, all of which promote malignant progression.^3,4

The clinical significance of ERBB2 has been most extensively demonstrated in breast cancer, where HER2 amplification/overexpression occurs in approximately 15%–20% of cases^5,6 and anti-HER2 targeted therapy has dramatically improved patient survival, leading to its establishment as a standard of care in international guidelines. ⁷ Similarly, in non-small cell lung cancer (NSCLC), recurrent ERBB2 mutations, particularly the exon 20 insertion p.Y772_A775dup, have been identified as oncogenic drivers,^8–10 and in 2025, HER2-targeted therapy received regulatory approval for this indication, expanding its clinical impact beyond breast cancer. In gastric cancer (GC), HER2 overexpression or amplification is observed in 6.1%–23% of patients. ¹¹ Trastuzumab in combination with chemotherapy remains the backbone of first-line therapy, and the integration of PD-1 inhibitors further improves outcomes in HER2-positive, PD-L1-expressing disease.^12–14

In contrast, in colorectal cancer (CRC), HER2 alterations are less common, with amplification or overexpression present in approximately 3%–5% of patients. ¹⁵ Although the prevalence is relatively low compared with breast and GC, the absolute patient number is substantial given the high global incidence of CRC. HER2-positive CRC is frequently associated with poor prognosis, early recurrence, and resistance to standard therapies.^16–20 Importantly, emerging evidence suggests that ERBB2 alterations—including mutations, amplification, and protein overexpression—may provide novel therapeutic opportunities.^21–23 However, HER2-directed therapy has not yet been incorporated into all international guidelines for CRC, reflecting the need for further clinical validation.

The biological and clinical significance of ERBB2 alterations varies across tumor types. For example, kinase domain mutations involving exons 19–21 are enriched in NSCLC, while extracellular domain mutations such as S310F/Y are common in biliary tract cancer, and the L755 substitution is frequently detected in breast cancer. ²⁴ These observations suggest tumor type–specific effects of HER2 mutations and highlight the necessity of tailored treatment strategies.

To better understand the molecular heterogeneity and therapeutic relevance of ERBB2 alterations in gastrointestinal malignancies, we conducted a next-generation sequencing (NGS)-based retrospective analysis of ERBB2-mutant gastrointestinal tumors. This study aimed to delineate the clinical and molecular landscape of ERBB2 alterations, thereby providing valuable insights into the optimization of HER2-targeted precision medicine strategies for gastrointestinal cancers.

Methods

Patient cohort

A total of 6823 patients with gastrointestinal malignancies were included in this study. All patients underwent targeted NGS between November 2017 and December 2024.

Inclusion criteria were: (1) age ⩾ 18 years; (2) diagnosed with stage I–IV gastrointestinal tumors; (3) availability of tumor tissue or peripheral blood plasma for sequencing; (4) samples passing predefined laboratory quality control metrics.

For tumor tissue samples, a minimum tumor cellularity of ⩾10% was required. For plasma samples, ctDNA libraries required a mean sequencing depth ⩾1000×; for tissue samples, DNA libraries required a mean sequencing depth ⩾500×. Sequencing performed using the validated 539- or 551-gene targeted panel.

Exclusion criteria included sequencing failure, insufficient DNA input, poor quality control metrics, or incomplete genomic annotation.

For patients with multiple sequencing records (tissue and plasma or repeated testing), only one sample per patient was included in the final analysis to avoid duplication. When multiple eligible samples were available, the earliest successfully sequenced sample was selected.

Among the total cohort of N = 6823 patients, 5705 (83.6%) samples were tumor tissue while 1118 (16.4%) were peripheral blood ctDNA. Tissue specimens were processed according to standard pathology criteria. Plasma ctDNA samples were collected prior to treatment initiation. We also took into account the influence of sample type on the results. Therefore, we only analyzed the MSI results of the tissues. For TMB, we conducted separate analyses based on the sample type (tissue or liquid).

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Simcere Diagnostics. Informed consent was waived according to institutional ethical guidelines for retrospective analyses of de-identified data. The reporting of this study conforms to the STROBE statement for observational cohort studies.

Library preparation and NGS sequencing

Genomic DNA was extracted from tumor tissue or peripheral blood samples using the DNeasy Blood & Tissue Kit (Qiagen). Sequencing libraries were prepared with the KAPA Library Preparation Kit (Roche), and library concentrations were quantified using the Qubit 4.0 fluorometer (Invitrogen). High-depth sequencing was performed on the Illumina NextSeq 550 or NovaSeq 6000 platforms, with an average coverage depth of 1500×.

Variant calling and annotation

Raw sequencing reads were pre-processed using Fastp (v2.20.0) to remove adapters and filter low-quality reads. ²⁵ Clean paired-end reads were aligned to the human reference genome (hg19) with BWA-MEM (v0.7.17). Single-nucleotide variants (SNVs) and insertions/deletions (indels) were identified using VarDict (v1.5.7) and annotated with InterVar, classifying variants as oncogenic, likely oncogenic, inconclusive, or unknown. The “ERBB2-unknown” category included tumors harboring ERBB2 variants classified as variants of uncertain significance (VUS) based on InterVar annotation according to ACMG/AMP criteria. ²⁶ Copy number variations (CNVs) were detected with CNVkit (v1.1), while FACTERA (v1.4.4) was used for fusion detection. High-level amplification was defined as ⩾10 copies, consistent with thresholds used in prior NGS-based clinical genomic profiling studies to identify therapeutically relevant ERBB2 amplification.^27–30

MSI status was determined using 334 homopolymeric repeat loci, with MSI-high (MSI-H) defined as an instability score ⩾0.15. TMB was calculated as the total number of somatic SNVs and indels per mega-base in the targeted coding regions.

Statistical analysis

All analyses were conducted using R software (v4.3.2). Genomic landscapes were visualized with the ComplexHeatmap package (v2.18.0), while lollipop plots were generated using maftools (v2.18.0). Bar plots and pie charts were generated with ggplot2 (v3.5.2). Continuous variables were compared using the Wilcoxon rank-sum test or Kruskal–Wallis test, whereas categorical variables were assessed with the chi-square test or Fisher’s exact test. A p-value < 0.05 was considered statistically significant. Given the high-dimensional nature of the genomic data, multiple-testing correction was applied using the Benjamini–Hochberg false discovery rate (FDR) procedure. Both nominal p-values and FDR-adjusted q-values are reported.

Results

Mutation spectrum of ERBB2 in colorectal cancer

The present study characterized the ERBB2 alteration spectrum in the Simcere CRC cohort. Among 4508 CRC patients, ERBB2 alterations were identified in 238 cases, corresponding to an overall prevalence of 5.3%. Annotation using the InterVar database revealed that 129 cases harbored oncogenic ERBB2 mutations, yielding an overall frequency of 2.9% for ERBB2 oncogenic mutations or insertions. Based on InterVar annotation, multiple ERBB2 mutations were identified and mapped to functional domains of the HER2 protein (Figure 1(a)). Oncogenic hotspots were predominantly located within the tyrosine kinase domain, including R678Q (15.8%), V842I (10.8%), and S310Y/F (7.4%), alongside recurrent variants such as L755S (3.9%), D277T (3.4%), and T798I (2.5%) (Figure 1(b)). Exon distribution analysis revealed a marked enrichment of mutations in exon 17 (17.2%), exon 20 (13.8%), and exon 19 (11.3%), collectively representing the majority of kinase domain variants (Figure 1(c)). Functional annotation indicated that more than half of these variants were classified as oncogenic or likely oncogenic, whereas only a minority were inconclusive or of unknown significance (Figure 1(d)). Notably, recurrent activating substitutions, including Y296N, S310F, A640T, and L755Sdup, were concentrated within functionally critical regions of HER2 (Figure 1(e)), underscoring their potential biological and therapeutic relevance in CRC. The proportions of oncogenic, likely oncogenic, uncertain/unknown variants, and amplifications were comparable between RAS/RAF mutant and RAS/RAF wild-type subgroups. Somatic mutations similarly clustered within key functional domains of the ERBB2 protein (Figure S1).

Figure 1.

Genomic landscape and distribution of ERBB2 alterations in colorectal cancer. (a) Schematic representation of ERBB2 protein domains with mapped mutation hotspots observed in CRC. (b, c) Frequency distribution of ERBB2 mutations and exon distribution across the CRC cohort. (d) The overall composition of ERBB2 alteration types, including oncogenic mutations, amplifications, and variants of unknown significance. (e) The positional distribution, functional domain mapping, and recurrence of ERBB2 mutations across the entire gene in CRC.

Co-mutation patterns and genomic instability in ERBB2-mutant colorectal cancer

Integrated genomic profiling revealed that ERBB2 mutations frequently co-occurred with typical CRC driver alterations, including APC, TP53, PIK3CA, ARID1A, and SMAD4 (Figure 2(a)). Among these, APC, TP53, and MUC16 mutations showed a higher co-occurrence with ERBB2 mutations (Figure 2(b)). In contrast, ERBB2 amplification was more often accompanied by copy number gains in RARA, TOP2A, and SMARCE1 (Figure 2(c)). Importantly, tumors harboring oncogenic ERBB2 mutations exhibited a significantly lower TMB compared with tumors carrying ERBB2 amplification or with unknown ERBB2 status (median TMB: oncogenic vs amplified vs unknown, p < 0.001; Figure 2(d)). The median TMB was 8.2 mutations/Mb (IQR 5.7–52.5) in the oncogenic mutation group, 4.3 mutations/Mb (IQR 2.9–5.7) in the amplification group, and 54.6 mutations/Mb (IQR 7.5–97.9) in the ERBB2 unknown group. Moreover, the prevalence of MSI-H was markedly enriched in the oncogenic mutation subgroup (34.8%) compared to the amplification subgroup (0.7%) and the unknown subgroup (52.7%, p < 0.001; Figure 2(e)). Conversely, CNV burden did not differ significantly between oncogenic mutation and amplification subgroups (Figure 2(f)). Collectively, these results demonstrate that ERBB2 mutations are associated with distinct co-mutation patterns and genomic instability features, indicating biological behaviors that differ from those observed in ERBB2 amplification in CRC.

Figure 2.

Co-mutation patterns, clinicogenomic associations, and immunogenomic features of ERBB2-mutant and ERBB2-amplified colorectal cancer. (a) The co-alteration landscape of ERBB2-oncogenic mutation, ERBB2-amplification, and ERBB2-unknown subgroups within the CRC cohort. (b) The co-mutations in ERBB2-oncogenic mutants versus ERBB2–amplified-type cases in CRC. (c) The co-amplifications in ERBB2-amplified tumors versus ERBB2–oncogenic-type cases. (d) Comparison of TMB across ERBB2-oncogenic mutation, ERBB2-amplification and ERBB2-unknown subgroups. (e) Proportion of MSI-H versus MSS cases across ERBB2 subtypes. (f) Genome-wide CNV burden across ERBB2 subtypes.

Distinct oncogenic spectrum of ERBB2 mutations in gastric cancer

We characterized the spectrum of ERBB2 alterations in a GC cohort comprising 2412 patients. A total of 338 cases harbored ERBB2 alterations, including 95 with ERBB2 mutations, 257 with ERBB2 amplification, and 14 with concurrent mutations and amplification. Distinct ERBB2 mutational profiles were observed in GC (Figure 3(a)). Oncogenic variants predominantly clustered within the tyrosine kinase domain, with R678Q (26.9%) representing the most frequent hotspots, followed by S310F/Y (10.4%), L755S (9.7%), and V842I (8.2%) (Figure 3(b)). Exon distribution analysis demonstrated that mutations were largely enriched in exon 17 (28.4%), exon 8 (15.7%), and exon 19 (14.9%), which together accounted for the majority of kinase domain alterations (Figure 3(c)). Functional annotation indicated that the vast majority of ERBB2 variants were classified as oncogenic or likely oncogenic, with only a minority remaining of uncertain significance (Figure 3(d)). The distribution of recurrent activating substitutions, including D277N, S310F/Y, and L755S, was observed to be concentrated in critical HER2 functional motifs (Figure 3(e)), thereby underscoring their potential relevance as therapeutic targets in GC. Collectively, these results indicate that ERBB2 mutations in GC are characterized by recurrent kinase domain hotspots and strong oncogenic potential, highlighting distinct biological and therapeutic implications compared with other gastrointestinal malignancies.

Figure 3.

Genomic distribution, mutation spectrum, and alteration composition of ERBB2 in GC. (a) Schematic representation of the ERBB2 protein structure with mapped mutations detected in the GC cohort. (b) Frequency distribution of ERBB2 mutations across the GC cohort, showing the most common hotspot variants and low-frequency alterations. (c) Frequency distribution of ERBB2 amplifications in GC, illustrating the leading amplification events and their relative proportions. (d) The distribution of ERBB2 alteration classes in GC, including oncogenic mutations, amplifications, and variants of unknown significance. (e) The positional distribution and recurrence levels of ERBB2 mutations across the entire coding sequence within the GC cohort.

Genomic landscape and instability features of ERBB2-mutant gastric cancer

Comprehensive genomic profiling further demonstrated that ERBB2 mutations in GC frequently co-occurred with alterations in canonical driver genes, including APC, TP53, ARID1A, MUC16, and LRP1B (Figure 4(a) and (b)). Meanwhile, co-mutations in cell cycle regulators (e.g., CCNE1, CDKN2B, and CDKN2A) as well as EGFR were seen in ERBB2 amplification (Figure 4(c)). TMB analysis showed that TMB was significantly higher in the unknown ERBB2 variant group compared to oncogenic mutations and amplifications (p < 0.001, Figure 4(d)), and the corresponding median TMB values were 7.8 mutations/Mb (IQR 5.0–23.8), 5.4 mutations/Mb (IQR 3.6–8.5), and 9.9 mutations/Mb (IQR 5.7–70.8). Similarly, MSI-H was significantly enriched in the unknown group (43.3%) and oncogenic mutations (27.0%), whereas it was almost absent in amplified cases (1.2%, p < 0.001, Figure 4(e)). In contrast, CNV levels were not significantly different between ERBB2 subgroups (Figures 4(f) and S2). Taken together, these findings emphasize that ERBB2-mutant GC exhibits a distinct genomic background, especially in the non-amplified subtype, with unique co-mutation tags and enrichment for MSI-H and high TMB, which may influence treatment response and precision targeting strategies.

Figure 4.

Co-mutation patterns, clinicogenomic associations, and immunogenomic features of ERBB2-mutant and ERBB2-amplified GC. (a) The co-alteration landscape of ERBB2-oncogenic mutation, ERBB2-amplification, and ERBB2-unknown subgroups within the GC cohort. (b) The co-mutations in ERBB2-oncogenic mutants versus ERBB2–amplified-type cases in GC. (c) The co-amplifications in ERBB2-amplified tumors versus ERBB2–oncogenic-type cases. (d) Comparison of TMB across ERBB2-oncogenic mutation, ERBB2-amplification, and ERBB2-unknown subgroups. (e) Proportion of MSI-H versus MSS cases across ERBB2 subtypes. (f) Genome-wide CNV burden across ERBB2 subtypes, demonstrating no substantial difference in CNV load.

Discussion

In this study, the distribution characteristics, co-mutation patterns and genomic instability differences of ERBB2 mutations in CRC and GC were systematically depicted. Overall, ERBB2 mutations showed several common molecular features in both gastrointestinal tumors. Firstly, the mutations were predominantly enriched in the HER2 tyrosine kinase structural domain (exons 17, 19, and 20) and recurred in key functional motifs (e.g., R678Q, S310F/Y, L755S, and V842I). These mutations not only drive aberrant downstream signaling pathway activation, but also partially overlap with HER2 hotspot variants in breast, cholangiocarcinomas and lung cancers, suggesting the existence of conserved oncogenic mechanisms across tumor types.^31–34 Functional annotation results showed that more than half of the mutations were classified as oncogenic or possibly oncogenic, emphasizing their biological and clinical relevance in gastrointestinal malignancies. Secondly, in both GC and CRC, ERBB2 mutations frequently co-occur with classical driver genes such as TP53, APC and ARID1A, and ERBB2 amplification is observed concurrently with FGFR1, MYC, or BCL2L1 alterations.^35–38 This pattern of co-mutations suggests that ERBB2 mutations may represent independent oncogenic pathways and not merely alternative events to HER2 amplification. Previous studies have identified that functional mutations in BRAF, PI3KCA, PTEN, PDGFRA, or MEP2K1, as well as amplification of the c-MET, ERBB2, and FGFR1 genes, may be associated with primary intrinsic resistance to anti-EGFR-targeted therapies, ³⁹ suggesting the selection of an anti-EGFR regimen for clinical care. In addition, oncogenic ERBB2 mutations were found to be significantly associated with higher TMB and MSI-H status types compared to the ERBB2 amplification group in both tumors,^40,41 suggesting differences in immunogenicity from the amplified type. This feature provides a rational basis for future exploration of combination therapy with HER2 inhibitors and immune checkpoint inhibitors.^42,43 MSI-H status and elevated tumor mutational burden (TMB) have been associated with increased tumor immunogenicity and favorable responses to immune checkpoint inhibitors in prior clinical studies; however, immunotherapy exposure and clinical response were not evaluated in the present cohort. Therefore, the enrichment of MSI-H and high TMB observed in specific ERBB2 subgroups should be interpreted as a biological association rather than evidence of therapeutic benefit.^44,45

Interestingly, tumors in the ERBB2 unknown group exhibited higher TMB and increased MSI-H prevalence. Previous studies have suggested that ERBB2 mutations may occur more frequently in hypermutated CRCs, particularly in tumors with mismatch repair deficiency and MSI-high molecular features.^19,46 In our cohort, further genomic profiling revealed that this subgroup also harbored frequent alterations in genes involved in the DNA damage response (DDR) pathway, as well as components of the SWI/SNF chromatin remodeling complex. Dysfunction in these pathways has been linked to impaired DNA repair and genomic instability, which can contribute to the accumulation of somatic mutations and elevated TMB.

These findings suggest that the enrichment of MSI-H and higher TMB in the ERBB2 unknown group may reflect an underlying hypermutated genomic background associated with defects in DNA repair and chromatin remodeling processes. In this context, ERBB2 VUS may arise as passenger alterations rather than primary oncogenic drivers. Nevertheless, the biological and potential therapeutic relevance of these variants remains to be clarified in future studies with functional validation and clinical outcome data.

In summary, our findings suggest that gastric and CRCs harboring specific ERBB2 mutations, particularly those occurring in tumors with high TMB or MSI-H, may represent a biologically distinct subgroup. These observations raise the possibility that ERBB2-altered tumors could exhibit differential responses to conventional chemotherapy or anti-EGFR–based therapies and may warrant further investigation in the context of HER2-targeted therapies or immunotherapy. However, as treatment outcomes were not evaluated in the present cohort, these findings should be considered hypothesis-generating and require validation in prospective studies incorporating clinical response data.

However, ERBB2 mutations in GC and CRC also exhibit several differential features. In GC, specific mutations such as R678Q have been associated with poor response to oxaliplatin-based chemotherapy, ⁴⁷ suggesting that ERBB2 variants may serve as molecular markers for predicting chemotherapy resistance. In comparison, the frequency of ERBB2 mutations in CRC was lower, yet the hotspot mutation patterns were also concentrated in functional regions. Furthermore, the results of mutation site and occurrence frequency demonstrate that the mutation frequency in our CRC cohort is highly analogous to that reported in other studies of Asian populations. This suggests that the population background (genetic background, environmental exposure, microbiota, etc.) may influence the distribution and function of ERBB2 mutations. ⁴⁸ This may explain why the same ERBB2 mutation can have different clinical effects, across tumor types and among individual patients. Clinically, ERBB2 amplification has been well established as a mechanism of both primary and acquired resistance to anti-EGFR therapy in gastrointestinal tumor. Preclinical studies have demonstrated that HER2 amplification leads to increased receptor expression and constitutive activation of downstream MAPK and PI3K-AKT signaling pathways, thereby bypassing anti-EGFR blockade. Clinical analyses have further shown enrichment of ERBB2 amplification in tumors refractory to cetuximab-based therapy,^19,49,50 supporting its role as a bona fide resistance mechanism. Clinically, dual HER2 blockade has demonstrated meaningful activity in HER2-amplified gastrointestinal tumor (e.g., HERACLES).^22,51

In contrast, the relationship between ERBB2 mutations and anti-EGFR resistance is less clearly defined. ERBB2 mutations are molecularly heterogeneous, encompassing kinase domain activating alterations, extracellular domain mutations, and variants of uncertain functional significance. Unlike gene amplification, which uniformly increases receptor abundance and signaling output, the functional consequences of ERBB2 mutations vary depending on mutation type and genomic context.^24,52 Although certain activating mutations have been shown in preclinical systems to enhance HER2 signaling and potentially reduce EGFR dependency,^53,54 clinical data specifically linking ERBB2 mutations to anti-EGFR resistance in cancer remain limited. Importantly, most available evidence regarding ERBB2-mutant tumors originates from small retrospective cohorts or basket trials evaluating irreversible HER2 TKIs such as neratinib and poziotinib rather than to interrogate mechanisms of anti-EGFR resistance.^24,55 These studies support the oncogenic potential of selected ERBB2 mutations but do not establish a definitive role in mediating anti-EGFR resistance. Prospective studies specifically assessing anti-EGFR treatment outcomes in ERBB2-mutant gastrointestinal tumor are lacking.

Therefore, although ERBB2-mutant tumors were identified in our cohort, our findings should not be interpreted as definitive evidence of resistance to anti-EGFR therapy. Rather, they suggest a biologically plausible but clinically unconfirmed association that warrants prospective validation in treatment-annotated cohorts. Our study refined the types of ERBB2 mutations, with particular attention to the frequent mutations and further exploration. This is conducive to revealing the correlation between ERBB2 mutations and resistance to anti-EGFR treatment. The direct evidence requires future prospective validation in treatment cohorts.

In summary, this study reveals the unique molecular profiling features of ERBB2 mutations in GC and CRC, the presence of mutational hotspots clustering in both cancers, co-mutations with classical driver genes, and immunogenic potential associated with MSI/TMB, all of which provide new evidence for understanding their biological behaviors and optimizing precision therapeutic strategies. These findings not only expand our understanding of GC and CRC heterogeneity but also provide potential translational value for the clinical management of HER2 mutant subtypes. The results of this study not only complement the molecular epidemiological data on the ERBB2 mutational spectrum in GC and CRC in Chinese cohorts, but also suggest that precision therapeutic strategies targeting mutations in the structural domains of these recurrent kinases should be further explored in the future, especially in mutant patient populations in the absence of amplification background.

While this study is not without its limitations. Firstly, the genomic testing cohort was used as the basis for the study, rather than a randomized clinical sample, which may be clinically biased. Secondly, the study was based only on the genetic test results and lack of longitudinal treatment response and survival data, and we were unable to confirm whether patients received anti-HER2 therapy or experienced clinically documented resistance to anti-EGFR treatment. Importantly, robust prospective, tumor-specific clinical trials evaluating HER2-targeted therapies in ERBB2-mutant gastrointestinal cancers remain limited. Future studies should integrate clinical endpoints such as objective response rates, progression-free and overall survival, and correlate these outcomes with individual ERBB2 alteration subtypes and co-mutational contexts. There is also a lack of functional validation of the mutations, with the pathogenicity and functional impact of some low-frequency mutations remaining unclear, which may result in the clinical significance of some variants remaining uncertain. Therefore, the therapeutic implications of ERBB2 alterations in this cohort should be interpreted as hypothesis-generating and require validation in prospective clinical studies. And our findings should not be interpreted as definitive evidence supporting routine HER2-targeted therapy in this population. Based on several testable hypotheses proposed in this study, clinical trials and mechanistic validation studies containing stratified biomarker endpoints should be prioritized as the next step in order to truly translate these molecular findings into individualized therapeutic decisions. This will be critical for assessing the predictive value of ERBB2 mutations and amplifications in guiding therapy choices.

Conclusion

In this large-scale genomic landscape analysis of ERBB2 alterations across gastrointestinal malignancies, we systematically characterized the distribution of ERBB2 amplification, pathogenic mutations, and VUS. Across gastric and CRCs, ERBB2 alterations converged on recurrent kinase domain hotspots yet arose within tumor type–specific genomic contexts marked by distinct co-mutation patterns and differential associations with MSI status and TMB. While ERBB2 amplification remains a clinically validated therapeutic target in selected gastrointestinal cancers, the biological and clinical significance of ERBB2 mutations requires further functional investigation and prospective clinical validation.

Overall, these findings expand the molecular epidemiology of ERBB2 alterations in gastrointestinal cancers and provide a genomic framework to guide future biomarker-driven and treatment-annotated clinical studies.

Supplemental Material

sj-docx-1-tam-10.1177_17588359261445706 – Supplemental material for Comprehensive genomic landscape of ERBB2 in Chinese GI tumors: mutation-centered landscapes and precision treatment opportunities

Supplemental material, sj-docx-1-tam-10.1177_17588359261445706 for Comprehensive genomic landscape of ERBB2 in Chinese GI tumors: mutation-centered landscapes and precision treatment opportunities by Yinan Shi, Xiaoxuan Wang, Xiaotong Xi, Mengxiao Wang, Jinfang Guo, Xing Zhang, Dongsheng Chen and Wenhui Yang in Therapeutic Advances in Medical Oncology