Human epidermal growth factor receptor 2 expression loss after neoadjuvant chemotherapy in localized gastroesophageal adenocarcinoma: a retrospective study Authors: Jean-Baptiste Demigné, Marie Heinisch, Julia Gilhodes, Shems Kettani, Domitille Dano, Christelle de la Fouchardière Published: 2026 ABSTRACT AbstractBackgroundHuman epidermal growth factor receptor 2 (HER2) overexpression/ERBB2 amplification represents an established therapeutic target in advanced gastric and gastroesophageal junction (GEJ) adenocarcinoma. However, the reliability of HER2 assessment between diagnostic biopsies and surgical specimens (SS), and the potential impact of neoadjuvant therapy on HER2 status, remain incompletely characterized in localized disease. This study aimed to evaluate HER2 concordance between pre-treatment biopsies (PTB) and matched SS, and to describe HER2 changes following neoadjuvant treatment in localized gastric and GEJ adenocarcinoma.MethodsWe conducted a retrospective single-center cohort study including patients with localized gastric or GEJ adenocarcinoma who underwent curative-intent surgery between 2017 and 2021. HER2 status was assessed on PTB and matched SS using immunohistochemistry (IHC) with reflex in situ hybridization (ISH) for IHC-2+ cases, according to gastric cancer guidelines. Concordance between PTB and SS, the impact of biopsy sampling adequacy, and dynamic changes in HER2 status following neoadjuvant therapy were analyzed descriptively.ResultsFifty-seven patients were included. Median age was 67 years; 77% were male. Neoadjuvant treatment was administered in 89.5% of patients, while 10.5% underwent upfront surgery. HER2 positivity was observed in 21.1% of biopsies and 15.8% of SS. Overall concordance of HER2 status between PTB and SS was 87.7% [95% confidence interval (CI): 76.3–94.9%]. Seven discordant cases were identified, including five cases of HER2 loss (biopsy-positive/resection-negative) and two cases of HER2 gain. Among the 12 tumors initially classified as HER2-positive on biopsy, HER2 positivity was maintained in seven cases (58.3%) but lost in five (41.7%; 95% CI: 15.2–72.3%). When restricted to patients receiving neoadjuvant therapy, HER2 loss occurred in 5 of 10 HER2-positive cases (50.0%; 95% CI: 18.7–81.3%). Increasing the number of biopsy fragments did not significantly reduce discordance between biopsy and SS.ConclusionsHER2 status shows overall good concordance between diagnostic biopsies and SS in localized gastric and GEJ adenocarcinomas. However, a substantial proportion of initially HER2-positive tumors exhibited loss of HER2 expression following neoadjuvant chemotherapy. These findings suggest that HER2 expression/ERBB2 amplification may evolve during perioperative treatment and support further evaluation of HER2 reassessment on SS.Keywords: Gastric cancer, gastroesophageal junction adenocarcinoma (GEJ adenocarcinoma), biopsy-resection concordance, neoadjuvant chemotherapy, human epidermal growth factor receptor 2 loss (HER2 loss) Highlight box Key findings • Human epidermal growth factor receptor 2 (HER2) status showed high concordance between pre-treatment biopsies and surgical specimens (SS) [87.7%; 95% confidence interval (CI): 76.3–94.9%], yet clinically meaningful discordances persisted (12.3%). • HER2 loss was observed in 50% (95% CI: 18.7–81.3%) of evaluable HER2-positive cases tumors treated with neoadjuvant therapy. • Clinically meaningful HER2 discordance remained observable despite biopsy sampling generally meeting guideline-based expectations. What is known and what is new? • HER2 assessment in gastric/gastroesophageal junction adenocarcinoma is disturbed by spatial heterogeneity, with reported discordance rates of 5% to 20%. HER2 loss has been described in metastatic settings, particularly after HER2-targeted therapies. • Apparent HER2 loss may be observed following non-HER2-targeted neoadjuvant therapy in localized disease. HER2 discordance remained observable despite biopsy sampling generally meeting guideline-based expectations. It provides detailed paired-sample evidence suggesting dynamic HER2 changes during treatment. What is the implication, and what should change now? • A single pre-treatment biopsy assessment may not fully capture HER2 heterogeneity in localized disease. • The potential role of HER2 reassessment on SS warrants prospective evaluation. • Apparent HER2 loss may impact patient selection for perioperative HER2-targeted approaches. • Future strategies should incorporate dynamic biomarkers (e.g., circulating tumor DNA) and prospective evaluation of HER2 evolution. Introduction Human epidermal growth factor receptor 2 (HER2) protein overexpression or ERBB2 gene amplification occurs in approximately 10–20% of gastric and gastroesophageal junction (GEJ) adenocarcinomas and represents an established therapeutic target in advanced/metastatic disease (1). While HER2-targeted approaches have transformed management in metastatic settings, their role in localized disease remains under investigation (2,3). Several perioperative studies combining HER2-targeted therapies with chemotherapy and/or immune checkpoint inhibitors have reported encouraging pathological responses, although no definitive survival benefit or standard perioperative strategy has yet been demonstrated (4-10). Accurate HER2 assessment is nevertheless challenged by the marked spatial heterogeneity of gastroesophageal adenocarcinoma. HER2 expression may vary both within and between tumor regions, potentially compromising biomarker reliability (11). Paired-sample analyses have reported relevant discordance rates between diagnostic biopsies and matched surgical specimens (SS), with reported rates ranging from approximately 5% to 20% across retrospective series (12-15). These discrepancies are generally attributed to intratumoral heterogeneity, sampling variability, and methodological differences in HER2 interpretation (16-18). Although current guidelines recommend obtaining multiple tumor-containing biopsies (5–8 biopsies) (Table S1) to improve representativeness, the relationship between sampling adequacy and HER2 discordance remains incompletely characterized. Beyond baseline heterogeneity, HER2 expression may also evolve through time. Dynamic HER2 loss has been described following trastuzumab exposure in advanced disease, supporting the concept of HER2 plasticity and clonal evolution under therapeutic pressure (19,20). However, these observations derive almost exclusively from metastatic settings following HER2-targeted exposure. Whether HER2 status may change under neoadjuvant chemotherapy in localized disease—and whether such changes could have clinical implications—remains insufficiently explored. The potential modulation of HER2 expression through clonal selection in operable gastric cancer therefore represents an important unresolved question. These considerations also raise the need for improved sampling strategies, systematic reassessment of HER2 status, and potentially the integration of circulating tumor DNA (ctDNA) approaches to monitor ERBB2 amplification during disease evolution. In this single-center study, we aimed to: (I) assess the concordance of HER2 status between endoscopic biopsies and SS in localized gastric/GEJ adenocarcinoma; (II) explore whether biopsy fragment number is associated with HER2 discordance; and (III) describe the dynamics of HER2 expression under neoadjuvant therapy. We present this article in accordance with the STROBE reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0393/rc). Methods Study design and patient selection We conducted a retrospective observational single-center cohort study including all consecutive patients with localized gastric or GEJ adenocarcinoma who underwent curative-intent resection at a single tertiary cancer center between 2017 and 2021. The study was approved by the Institutional Review Board of Paoli-Calmettes Institute (IRB SHERPAGOL-IPC 2025-026). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Due to the retrospective nature of the study, individual informed consent was waived. Inclusion criteria were clinically and histologically confirmed localized gastric or GEJ adenocarcinoma, and availability of HER2 status assessment in both pre-treatment biopsies (PTB) and SS. Patients with metastatic disease at diagnosis or with incomplete medical records (defined as absence of HER2 testing on diagnostic biopsies) were excluded. Clinical, pathological, and treatment data were extracted from electronic medical records. Treatment characteristics Neoadjuvant chemotherapy regimens were recorded as categorical variables: FLOT (5-fluorouracil, leucovorin, oxaliplatin, docetaxel), FOLFOX (5-fluorouracil, leucovorin, oxaliplatin), cisplatin/5-fluorouracil ± pembrolizumab, CROSS regimen (chemoradiotherapy: 40.2 Gy plus weekly carboplatin-paclitaxel), or no neoadjuvant therapy (upfront surgery). Pathological assessment and HER2 testing Pathological data were extracted from PTB and SS pathology reports. HER2 status was primarily extracted from routine pathology reports. No systematic central pathology review was performed. Cases with missing HER2 status in the pathology report underwent expert pathological reassessment. Collected variables included histological subtype [according to the 5th World Health Organization (WHO) edition classification] (21), tumor differentiation grade, and pathological staging parameters including lymphovascular and perineural invasion. For each case, the number of biopsy fragments used for pathological assessment and HER2 testing was recorded as a surrogate for biopsy sampling adequacy. HER2 was assessed on formalin-fixed paraffin-embedded (FFPE) tissue using standardized immunohistochemistry (IHC) and scored as 0, 1+, 2+, or 3+ according to gastric-specific criteria, on both biopsy and SS (16). IHC analysis was performed using the Ventana anti-HER2/neu (4B5) antibody, and dual-color dual-hapten in situ hybridization (DDISH) was performed using the Ventana HER2 Dual ISH DNA Probe cocktail. All assays were performed on 3 µm tissue sections from FFPE blocks. In SS, a score of 0 was assigned if there was no reactivity or membranous reactivity in <10% of tumor cells. A score of 1+ was assigned if faint or barely perceptible membranous reactivity was seen in ≥10% of tumor cells, with cells reactive in part of their membrane only. A score of 2+ was assigned if weak to moderate complete, basolateral, or lateral membranous reactivity was seen in ≥10% of tumor cells. A score of 3+ was assigned when strong complete, basolateral, or lateral membranous reactivity was seen in ≥10% of tumor cells. Scores of 0 and 1+ were regarded as HER2-negative, score 2+ as equivocal, and score 3+ as HER2-positive. In biopsy specimens, the 10% threshold for tumor cell staining is not applied; HER2 positivity is defined by the presence of membranous reactivity in a cluster of at least five cohesive tumor cells, irrespective of the proportion of stained cells. In situ hybridization (ISH) was used in HER2-IHC-2+ cases to determine ERBB2 gene amplification status. HER2 status was ultimately classified as positive (IHC-3+ or IHC-2+ with amplification), negative, or low according to combined IHC/ISH interpretation. Concordance analysis Concordance between biopsy and surgical specimen was evaluated in binary form and by IHC score. For binary concordance analyses, HER2-IHC-2+/ISH-negative tumors (HER2-low/equivocal non-amplified) were classified as HER2-negative. Four concordance patterns were defined: (I) positive/positive (PTB+/SS+): HER2-positive both on biopsy and surgical specimen; (II) positive/negative, or HER2 loss (PTB+/SS−): HER2-positive biopsy, HER2-negative surgical specimen; (III) negative/positive, or HER2 gain (PTB−/SS+): HER2-negative biopsy, HER2-positive surgical specimen; and (IV) negative/negative (PTB−/SS−): HER2-negative both on biopsy and surgical specimen. HER2 loss and HER2 gain were grouped as discordances. For dynamic analyses, we primarily focused on patients with HER2-positive biopsies to assess treatment-associated changes, particularly HER2 loss. Statistical analysis Statistical analyses were primarily descriptive. Patient, tumor, and treatment characteristics were summarized using standard descriptive statistics. HER2 concordance between PTB and SS was evaluated using contingency tables, with concordance and discordance rates reported as percentages. The impact of biopsy sampling adequacy was explored by comparing HER2 concordance and discordance rates according to the number of biopsy fragments. Changes in HER2 status under neoadjuvant treatment were described by reporting the proportion of patients with HER2 status stability or variation between PTB and SS. Given the exploratory nature of the study and the limited sample size, all analyses were descriptive and hypothesis-generating, and no formal inferential testing was performed. The sample size was determined by the number of eligible patients during the study period; no formal sample size calculation was performed due to the exploratory nature of the study. No imputation for missing data was performed; analyses were conducted on available data. Results Patient, tumor, and treatment characteristics (Table 1) Table 1Baseline characteristics of the cohort (n=57)VariableValueAge (years)67 [55–72]Sex Male44 (77.2) Female13 (22.8)Tumour location Cardia (JEG)23 (40.4) Gastric34 (59.6)Histological subtype (OMS) Tubular adenocarcinoma27 (47.4) Mixed adenocarcinoma14 (24.6) Poorly-cohesive12 (21.1) Undifferentiated2 (3.5) Missing2 (3.5)Neoadjuvant regimen FLOT30 (52.6) FOLFOX12 (21.1) Cisplatin/5-FU ± pembrolizumab6 (10.5) CROSS3 (5.3) None (upfront surgery)6 (10.5)Pathological pT pT02 (3.5) pT11 (1.8) pT1b4 (7.0) pT1a3 (5.3) pT28 (14.0) pT335 (61.4) pT4a2 (3.5) pT4b2 (3.5)Pathological N pN021 (36.8) pN119 (33.3) pN210 (17.5) pN3a5 (8.8) pN3b2 (3.5)Data are presented as n (%) or median [interquartile range]. FLOT, 5-fluorouracil, leucovorin, oxaliplatin, and docetaxel. FOLFOX, 5-fluorouracil, leucovorin, and oxaliplatin. CROSS, chemoradiotherapy for oesophageal cancer followed by surgery study regimen. 5-FU, 5-fluorouracil; JEG, gastroesophageal junction; N, node; OMS, other morphological subtype; T, tumor. Baseline characteristics of the cohort (n=57) Data are presented as n (%) or median [interquartile range]. FLOT, 5-fluorouracil, leucovorin, oxaliplatin, and docetaxel. FOLFOX, 5-fluorouracil, leucovorin, and oxaliplatin. CROSS, chemoradiotherapy for oesophageal cancer followed by surgery study regimen. 5-FU, 5-fluorouracil; JEG, gastroesophageal junction; N, node; OMS, other morphological subtype; T, tumor. A total of 57 consecutive patients with localized gastric or GEJ adenocarcinoma and complete HER2 assessment on both biopsy and surgical specimen were included in the analysis. Median age was 67 years [interquartile range (IQR), 55–72 years], and 77% of patients were male. Tumors were predominantly located at the GEJ (cardia; 40%). According to the WHO classification, most tumors were tubular adenocarcinomas (47.4%), followed by mixed adenocarcinomas (24.6%), poorly cohesive carcinomas (21.1%), and undifferentiated carcinomas (3.5%). Six patients (10.5%) underwent upfront surgery. Neoadjuvant chemotherapy was administered to 51/57 patients (89%), with FLOT being the most frequently used regimen (n=30, 53%), followed by FOLFOX (n=12, 21%), cisplatin/5-fluorouracil ± pembrolizumab [n=6, 11%; Keynote-585 trial (22)], and CROSS chemoradiotherapy (n=3, 5%). Among FLOT-treated patients, 24/30 (80%) completed the standard four preoperative cycles. Adjuvant chemotherapy was administered to 42/57 patients (74%). A fully completed perioperative chemotherapy course was reported in 24/51 patients (47%). Pathological staging showed ypT3 stage in 61% and ypN1 in 33% of surgical samples. HER2 status and biopsy-surgical specimen concordance (Table 2, Figures 1,2) Table 2Concordance of binary HER2 status between pre-treatment biopsy and surgical specimenBiopsy HER2 statusHER2-negative specimenHER2-positive specimenTotalHER2-negative biopsy432†45HER2-positive biopsy5†712Total48957†, discordances between PTB and SS. HER2, human epidermal growth factor receptor 2; PTB, pre-treatment biopsy; SS, surgical specimen. Concordance of binary HER2 status between pre-treatment biopsy and surgical specimen †, discordances between PTB and SS. HER2, human epidermal growth factor receptor 2; PTB, pre-treatment biopsy; SS, surgical specimen. Distribution of global HER2 transition patterns between biopsy and surgical specimen (stable negative, stable positive, HER2 loss, HER2 gain). HER2, human epidermal growth factor receptor 2; Neg, negative; Pos, positive. Alluvial diagram showing transitions between HER2 IHC categories on biopsy (HER2 0+/1+/2+/3+) and surgical specimen. Band thickness represents the proportion of patients undergoing each transition. HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry. HER2 positivity was observed in 12/57 biopsies (21.1%) and 9/57 SS (15.8%). Overall, 50/57 pairs (87.7%; 95% CI: 76.3–94.9%) were concordant: PTB−/SS− in 43 patients and PTB+/SS+ in 7 patients. Discordant cases included PTB−/SS+ (HER2 gain) in 2 patients (3.5% of the entire cohort; 22.2% of HER2-positive specimens), and PTB+/SS− (HER2 loss) in 5 patients (8.8% of the cohort), yielding an overall discordance rate of 12.3%. When considering IHC score level, substantial within-category discordance was observed between PTB and SS (Table 2). HER2-IHC-3+ tumors accounted for 8.8% (5/57) of PTB and 10.5% (6/57) of SS. Among the five biopsy HER2-IHC-3+ cases, three remained 3+ on the corresponding surgical specimen. Regarding IHC-2+ tumors, 15 cases were classified as IHC-2+ on PTB, of which 46.7% (7/15) were ISH-positive. In contrast, 14 tumors were scored as HER2-IHC-2+ on SS, with only 21.4% (3/14) demonstrating ERBB2 amplification (Table S2). Lower IHC scores (IHC-1+) also showed reclassification between PTB and SS, highlighting variability at the individual score level beyond binary HER2 status. Impact of biopsy sampling adequacy on HER2 discordance (Table 3) Table 3Impact of number of tumour-containing biopsy fragments on HER2 discordanceNumber of fragmentsnDiscordant (n)Discordance rate (%)3†7004†16212.559222.268112.573133.3810094125.010–15400†, under recommendation threshold number (Table S1). HER2, human epidermal growth factor receptor 2. Impact of number of tumour-containing biopsy fragments on HER2 discordance †, under recommendation threshold number (Table S1). HER2, human epidermal growth factor receptor 2. Fragment counts were available for 52 of 57 patients (91%), with missing data in 5 cases. The median number of biopsy fragments was 5 per patient (IQR, 4–6), and no cases presented fewer than 3 fragments. There was no clear relationship between fragment number and HER2 concordance between PTB and SS. Discordance rates were 0/7 (0%) for 3 fragments, 2/16 (12.5%) for 4 fragments, 2/9 (22.2%) for 5 fragments, and 1/8 (12.5%) for 6 fragments. Discordant cases were also reported in higher fragment count (≥7) biopsies (Table 3). Dynamic changes in HER2 status with neoadjuvant therapy (Table 4, Figure 2; Tables S3,S4, Figure S1) Table 4Evolution of HER2 status between biopsy and surgical specimen according to neoadjuvant treatmentNo.Neoadjuvant regimenNumber of cyclesBiopsy IHCBiopsy DDISHBiopsy HER2Number of fragmentsSpecimen IHCSpecimen DDISHSpecimen HER2Evolution1FLOT53+–Positive40+–NegativeHER2 loss2FLOT42+AmplifiedPositive92+AmplifiedPositiveHER2 stable3FLOT42+AmplifiedPositive71+–NegativeHER2 loss4Cisplatin/5-FU ± Pembro33+–Positive61+–NegativeHER2 loss5Cisplatin/5-FU ± Pembro32+AmplifiedPositive32+AmplifiedPositiveHER2 stable6Cisplatin/5-FU ± Pembro32+AmplifiedPositive50+–NegativeHER2 loss7FLOT42+AmplifiedPositive92+Not amplifiedLowHER2 loss8None–3+–Positive43+–PositiveHER2 stable9FLOT43+–Positive33+–PositiveHER2 stable10FLOT42+AmplifiedPositive–3+–PositiveHER2 stable11FOLFOX63+–Positive103+–PositiveHER2 stable12None–2+AmplifiedPositive42+AmplifiedPositiveHER2 stable13FOLFOX + radiotherapy31+–Negative43+–PositiveHER2 gain14FLOT + radiotherapy41+–Negative53+–PositiveHER2 gainFLOT, 5-fluorouracil, leucovorin, oxaliplatin, and docetaxel. FOLFOX, 5-fluorouracil, leucovorin, and oxaliplatin. 5-FU, 5-fluorouracil; DDISH, dual-color dual-hapten in situ hybridization; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry. Evolution of HER2 status between biopsy and surgical specimen according to neoadjuvant treatment FLOT, 5-fluorouracil, leucovorin, oxaliplatin, and docetaxel. FOLFOX, 5-fluorouracil, leucovorin, and oxaliplatin. 5-FU, 5-fluorouracil; DDISH, dual-color dual-hapten in situ hybridization; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry. We focused on the 12 patients with HER2-positive PTB to assess changes in HER2 status following neoadjuvant therapy (Figure S1). Overall, 5/12 patients (41.7%; 95% CI: 15.2–72.3%) exhibited HER2 loss on the surgical specimen, while HER2 positivity was maintained in 7/12 cases. When restricted to patients who received neoadjuvant treatment (n=10), HER2 loss occurred in 5/10 cases (50.0%; 95% CI: 18.7–81.3%). Descriptive regimen-specific distributions are reported in Table 4. Patients who underwent upfront surgery without neoadjuvant therapy (n=2) showed no change in HER2-positive status and were not included in this treatment-related analysis. In all HER2-loss cases (n=5), HER2 expression decreased to IHC-0 or IHC-1+ without amplification in most cases, while one tumor shifted from IHC-2+ amplified to IHC-2+ non-amplified, consistent with true loss of ERBB2 amplification rather than minor IHC variability. Two patients demonstrated HER2 gain after neoadjuvant treatment, converting from HER2-negative biopsies to HER2-positive SS. In both cases, HER2 status shifted from biopsy IHC-1+/HER2-negative to surgical specimen IHC-3+/HER2-positive. Both patients had received neoadjuvant radiotherapy combined with systemic treatment, including FOLFOX-based chemoradiotherapy in one case and FLOT plus radiotherapy in the other. Among IHC-2+ tumors on baseline biopsy (n=15), ISH identified ERBB2 amplification in 7 cases (46.7%). In SS, ERBB2 amplification was detected in 3 of 14 HER2-IHC-2+ tumors (21.4%), illustrating variability in ERBB2 amplification status among HER2-IHC-2+ paired samples. Discussion In this single-center series of 57 localized gastric and GEJ adenocarcinomas with systematic HER2 assessment on both PTB and SS, we report several key findings. First, we confirmed 87.7% (95% CI: 76.3–94.9%) concordance of HER2 status between PTB and SS, consistent with prior reports describing discordance rates ranging from 5% to 20% and highlighting a small but existing risk of HER2 misclassification in gastric and GEJ adenocarcinomas (13). We also report a discordance in HER2 expression despite adequate biopsy sampling. In our cohort, the median number of biopsy fragments was five per case, consistent with European Society for Medical Oncology (ESMO) recommendations for HER2 assessment (Table S1). However, biopsy fragment number alone represents only an imperfect surrogate of sampling quality and does not necessarily ensure biological representativeness. Variations in fragment size, tumor cellularity, and sampling of non-representative areas such as normal mucosa, fibrosis, or necrosis may contribute to discordant HER2 results. Accordingly, given the exploratory nature of this analysis and the limited number of discordant cases, no firm conclusions can be drawn regarding the relationship between biopsy fragment number and HER2 discordance. Our study confirms the unstable nature of low and intermediate HER2 categories (IHC-1+ and 2+), which displayed extensive discordance between PTB and SS. This finding is consistent with prior reports showing that HER2-IHC-1+ and 2+ scores are the most heterogeneous and least reproducible categories in gastric cancer (11). Conversely, IHC-3+ status remained relatively stable between PTB and SS, reinforcing its strong positive predictive value. Beyond technical considerations, intrinsic tumor heterogeneity and treatment-related effects are also likely to play a major role in HER2 discordance. We observed apparent HER2 loss in 5 of 10 evaluable HER2-positive patients following neoadjuvant therapy. Given the limited number of HER2-positive cases within each treatment subgroup, these observations should be interpreted cautiously and regarded as descriptive only. Whether this apparent HER2 loss reflects true biological evolution or primarily baseline heterogeneity remains uncertain. One possible explanation is that HER2-positive subclones may be more sensitive to systemic treatment, leading to relative depletion of HER2-amplified populations in residual disease. Alternatively, discordance may simply reflect spatial heterogeneity and sampling variability rather than true biological evolution related to treatment. Similar HER2 dynamics have previously been described following trastuzumab exposure in advanced disease, supporting the broader concept of HER2 plasticity under therapeutic pressure (20). These observations may have implications for perioperative HER2-targeted strategies. If HER2 status changes during treatment, baseline assessment alone may not fully capture the biology of residual disease. Conversely, if discordance mainly reflects heterogeneity, excluding patients from postoperative HER2-targeted approaches solely on the basis of a HER2-negative surgical specimen could be problematic. Further studies integrating longitudinal tissue and circulating biomarkers are needed to better characterize HER2 evolution during perioperative treatment and to clarify the potential roles of baseline HER2 assessment, surgical specimen reassessment, and dynamic biomarker monitoring in future perioperative HER2-targeted strategies. Current evidence indicates that ERBB2 amplification can be detected in ctDNA in gastric cancer and may complement tissue-based assessment (23). However, most data derive from metastatic disease, while evidence in the perioperative or localized setting remains limited (24). Initial pilot studies showed that ERBB2 amplification can be identified in plasma using polymerase chain reaction (PCR)-based approaches, with variable concordance compared with tumor tissue and higher sensitivity in patients with greater tumor burden (25). Subsequent studies using digital PCR and next-generation sequencing confirmed that ERBB2 copy number alterations in ctDNA may correlate with response to anti-HER2 therapies and dynamically reflect treatment-associated clonal shifts, including loss of amplification under systemic therapy (26-28). These findings support the concept that plasma-based HER2 monitoring may capture temporal heterogeneity not fully reflected by single-site tissue sampling (23,29). More recent reports suggest that ctDNA analysis may occasionally detect ERBB2 amplification not identified in archival tissue, possibly reflecting spatial heterogeneity or clonal evolution (26,30). Longitudinal monitoring has also been proposed to identify emerging resistance or reappearance of ERBB2-amplified clones (28,31,32). Nevertheless, these observations predominantly concern advanced-stage patients treated with HER2-targeted therapies (27). In localized gastric cancer, available data remain exploratory, and the clinical role of ctDNA-based ERBB2 assessment in the perioperative setting warrants prospective evaluation. Strengths of this study include a well-characterized monocentric cohort with single-center pathology assessment, availability of HER2 status on both biopsies and SS, systematic recording of biopsy fragment number, and detailed documentation of neoadjuvant regimens enabling dynamic analyses of HER2 loss. Limitations include the retrospective design, which may introduce selection bias, and the single-center setting, which may limit generalizability. In addition, potential sampling bias related to intratumoral heterogeneity and variability in biopsy representativeness may have contributed to HER2 discordance. The limited sample size, particularly in the HER2-positive subgroup, precludes multivariable analyses and precise estimation of effect sizes. The observed pattern of treatment-associated HER2 loss, together with the limited impact of biopsy fragment number on discordance, highlights that HER2 discordance is likely multifactorial. Future studies in larger, multicenter cohorts are warranted to further explore HER2 dynamics in the perioperative setting. Integrating multi-site sampling, molecular profiling, and longitudinal ctDNA assessment may help better characterize HER2 heterogeneity and plasticity in operable gastroesophageal cancer. These findings should be interpreted in the context of a monocentric European cohort and may not be directly generalizable to other populations or healthcare settings. Conclusions In this single-center cohort of localized gastric and GEJ adenocarcinomas, HER2 status showed overall good concordance between PTB and SS, yet clinically meaningful discordances persisted despite biopsy sampling meeting current guideline-based expectations. We observed several cases of HER2 loss among evaluable HER2-positive cases after neoadjuvant therapy, supporting the concept that HER2 expression and ERBB2 amplification may also be dynamic under non-HER2 targeted treatment pressure. These findings underscore the limitations of relying solely on a single pre-treatment biopsy assessment to capture HER2 heterogeneity and suggest that reassessment on SS or on ctDNA may provide additional clinically relevant information in selected patients. While the clinical implications of perioperative HER2 dynamics remain to be defined, our results generate hypotheses that warrant prospective validation in larger multicenter cohorts, ideally integrating ctDNA and complementary molecular approaches to better characterize HER2 evolution under treatment. Supplementary The article’s supplementary files as Acknowledgments None. Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was approved by the Institutional Review Board of Paoli-Calmettes Institute (IRB SHERPAGOL-IPC 2025-026). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Due to the retrospective nature of the study, individual informed consent was waived. Footnotes Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0393/rc Funding: None. Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0393/coif). C.d.l.F. reports consulting or advisory role with Bristol Myers Squibb, Amgen, Servier, Pierre Fabre, MSD Oncology, Roche/Genentech, Daichi-Sankyo, Astellas Pharma, Gilead Sciences, BeiGene, Jazz Pharmaceuticals, Takeda, AbbVie, AstraZeneca. C.d.l.F. received research funding from Pierre Fabre (Inst), Servier (Inst), MSD (Inst), Agenus (Inst) and support for travel, accommodations, expenses from Pierre Fabre, Servier, MSD Oncology, Amgen, AstraZeneca. The other authors have no conflicts of interest to declare. Data Sharing Statement Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0393/dss References Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0393/prf Supplementary Materials The article’s supplementary files as Data Availability Statement Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0393/dss