Introduction

Gastric cancer (GC) is the fifth most commonly diagnosed cancer and the fourth leading cause of cancer-related death worldwide [1]. The incidence rates are highest in Eastern Asia including Korea; thus, GC is the most common cancer in men (fourth in women) and the fourth most common cause of cancer-related death in Korea [2].

Human epidermal growth factor receptor 2 (HER2), a member of a family of receptors associated with tumor cell proliferation, apoptosis, adhesion, migration, and differentiation, is involved in the pathogenesis and poor outcomes of several types of cancer, including advanced gastric cancer (AGC) and gastroesophageal junction (GEJ) cancer [3, 4]. HER2 overexpression or amplification is observed in 7%–34% of patients with GC [4, 5]. Trastuzumab is a monoclonal antibody targeting HER2, which exerts an antitumor effect by inducing antibody-dependent cellular cytotoxicity and inhibiting HER2-mediated signaling [6]. The phase 3 ToGA study demonstrated the clinical benefits of combining trastuzumab with chemotherapy in patients with previously untreated and unresectable or metastatic HER2-positive (HER2+) GC or GEJ cancer [7], making trastuzumab-based chemotherapy the standard treatment in the clinical setting. Improvement in the overall survival (OS) after trastuzumab-based chemotherapy was also confirmed in previous real-world analyses, highlighting its effectiveness as a first-line treatment [8, 9].

One of the factors known to affect the outcomes of patients treated with trastuzumab is the expression levels of HER2 assessed by immunohistochemistry (IHC): IHC 3+ tumors reportedly exhibit favorable outcomes than IHC 2+ tumors with in situ hybridization (ISH) positivity [7, 10,11,12]. Given that HER2 expression in GC frequently shows incomplete membrane staining and heterogeneity [13], the different treatment outcomes appear to be attributable to the heterogeneity of HER expression. HER2 heterogeneity in GC was commonly recognized in IHC studies with reported frequencies of HER2-stained cells ranging from 16.8 to 60.9% [12, 14, 15].

In our previous GASTHER1 study, extra-gain of HER2-positivity was noted in 8.7% of patients with initially HER2-negative GC via repeated endoscopic biopsy (rescued HER2-positive) [16]. Rescued HER2-positive patients who received trastuzumab-based first-line chemotherapy had similar treatment outcomes to those of initially HER2-positive patients [16]; however, this finding needs to be evaluated further in a larger cohort with longer follow-up duration. Moreover, rescued HER2 positivity (i.e., initially HER2 negative and HER2 positive by subsequent evaluation) per se may reflect heterogeneous HER2 expression, which is associated with reduced trastuzumab efficacy [11, 12, 17].

In this study, we aimed to confirm whether rescued HER2-positive patients exhibit comparable treatment outcomes to those of initially HER2-positive patients in the GASTHER1 study, using a larger patient cohort with longer follow-up duration. We also aimed to analyze the treatment outcomes of GC patients in terms of HER2 heterogeneity status with rescued HER2 positivity and H-score.

Materials and methods

Patients

Patients aged > 18 years with histologically confirmed unresectable, initially metastatic, or recurred HER2-positive GC or GEJ cancer who started receiving trastuzumab-based first-line chemotherapy at Asan Medical Center, South Korea, between May 2011 and November 2018 were eligible for the study (n = 162). After the GASTHER1 study, HER2 assessment based on repeated endoscopic biopsy was routinely performed if the initial HER2 status was negative. As a result, 11 rescued HER2-positive patients in the GASTHER1 study conducted between May 2011 and April 2013 and 16 additional rescued HER2-positive patients were included in this study. After excluding patients whose tumor tissue was unavailable for additional evaluation (n = 6) and those who were lost to follow-up (n = 3), 153 patients were included in the final analysis (Supplementary Fig. 1).

Clinical and laboratory data were retrospectively obtained by reviewing the patients’ medical records. This study was approved by the Institutional Review Board of Asan Medical Center and was performed in accordance with the ethical standards of the institutional research and the Declaration of Helsinki. The need for obtaining an informed consent was waived due to the retrospective nature of the study.

Evaluation of HER2 status

HER2 IHC was performed using a Ventana anti-Her2/neu (4B5) rabbit monoclonal primary antibody (Ventana Medical System, Tucson, AZ), and HER2 protein expression was scored on a scale of 0 to 3 using the GC consensus panel recommendations [5, 18]. ISH was performed only in patients with IHC 2+ at the time of diagnosis. HER2 gene amplification was evaluated by counting the signals in the 20 non-overlap** tumor cells with the highest gene counts. The ISH results were interpreted as positive if the HER2/CEP17 ratio was > 2.2 and negative if it was < 1.8. When the results were equivocal (1.8–2.2), 20 additional tumor cells were counted, and an HER2/CEP17 ratio ≥ 2 was considered positive [19].

HER2 positivity was defined as IHC 3+ or IHC 2+ with positive ISH results (ISH +). HER2 status was classified as initially HER2-positive and rescued HER2-positive. Rescued HER2 positivity was defined if HER2 positivity was obtained on repeated endoscopic biopsy before initiation of the first-line chemotherapy in patients with initially HER2-negative GC. The median interval between the initial biopsy and re-biopsy was 14 days (range, 2–74 days).

H-score

To quantitatively assess the HER2 heterogeneity, which cannot be fully recapitulated by HER2 IHC score, we utilized the histoscore or H-score calculated based on the staining intensity and proportion of HER2-expressing tumor cells. The score can range from 0 (case with absent staining) to 300 (case with 100% 3+ staining) and was calculated based on the following formula [20, 21]:

$${\text{H - score}} = (\% \,{\text{tumor}}\,{\text{IHC}}\,1 + ) + 2\,(\% \,{\text{tumor}}\,{\text{IHC}}\,2 + ) + 3\,(\% \,{\text{tumor}}\,{\text{IHC}}\,3 + ).$$

H-score was calculated by a board-certified gastrointestinal pathologist (YSP) with more than 10 years of experience in gastrointestinal pathology.

Statistical analysis

Objective response rate (ORR), progression-free survival (PFS), and OS were analyzed. The ORR was evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1. The PFS was defined as the time from the initiation of trastuzumab-based first-line chemotherapy to the time of disease progression or death, whichever occurred first. The OS was defined as the time from the initiation of trastuzumab-based first-line chemotherapy to the time of death from any cause. Survival outcomes were estimated using the Kaplan–Meier method. A two-sided p value < 0.05 was considered significant. Cox proportional hazards model was used to estimate the hazard ratio (HR) for survival outcomes. The maximal chi-square method was used to determine the optimal cut-off value of H-score that best segregated survival outcomes. All statistical analyses were performed using the Statistical Package for the Social Sciences version 23.0 (IBM, Armonk, NY, USA) ad statistical software package R version 3.0.2.

Results

Patients’ characteristics

The patients’ baseline characteristics are summarized in Table 1. Their (n = 153) median age was 60 years (range, 22–89 years), and majority of them (n = 122, 79.7%) were male. Most patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0–1 (n = 133, 86.9%) and were initially diagnosed with metastatic disease (n = 143, 93.5%). The patients were grouped based on a previously reported prognostic model for metastatic or recurrent GC [22].

Table 1 Baseline characteristics
Full size table

Of the 153 patients, 126 (82.4%) had initially HER2-positive GC, while 27 (17.6%) had rescued HER2-positive GC. The baseline characteristics were comparable between the subgroups divided by HER2 status (Table 1, Supplementary Table 1).

IHC score according to HER2 status

The baseline and final IHC results of initially and rescued HER2-positive patients are summarized in Table 2. Among 126 initially HER2-positive patients, the majority (n = 105, 83.3%) had an ICH score of 3+ , while 21 (16.7%) had IHC 2+ /ISH+ , whereas rescued HER2-positive patients had a relatively higher frequency of IHC 2+ /ISH+ (n = 10, 37.0%) tumors. The baseline HER2 scores in the rescued HER2-positive group were IHC 0 in 17 (63.0%), IHC 1+ in eight (29.6%), and IHC2+ /ISH− in two (7.4%) patients, respectively. Among 17 patients with HER2 IHC 0 at baseline, the scores of six (35.3%) and 11 patients (64.7%) changed to IHC 2+ /ISH+ and IHC 3+ on repeated biopsy, respectively. The scores of three (37.5%) and five patients (62.5%) with initially IHC 1+ changed to IHC 2+ /ISH+ and IHC 3+ on repeated biopsy, respectively. Meanwhile, two patients had IHC 2+ and ISH− on initial biopsy, but was ISH+ (n = 1, 50.0%) and IHC 3+ (n = 1, 50.0%) on repeated biopsy.

Table 2 IHC score of the initially and rescued HER2-positive patients
Full size table

Survival outcomes according to IHC score and HER2 status

According to RECIST v1.1, among the 108 patients with measurable and evaluable lesions who received trastuzumab-based chemotherapy, 5 (4.6%), 83 (76.9%), 12 (11.1%), and 8 (7.4%) patients achieved complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD), respectively. The ORR was not significantly different according to HER2 status (initially vs rescued HER2 positive, p = 0.143) (Supplementary Table 2).

With a median follow-up duration of 47.6 months (range 18.9–76.3 months), the median PFS and OS of all patients were 7.2 months [95% confidence interval (CI) 5.6–8.8 months] and 15.0 months (95% CI 11.5–18.5 months), respectively. When the study population was stratified according to the IHC score, the patients with IHC 2+ /ISH+ had significantly shorter PFS and OS than those with IHC3+ (PFS: median 5.7 months [95% CI 5.0–6.4 months] vs 8.4 months [95% CI 6.4–10.4 months], p = 0.034; OS: median 11.3 months [95% CI 7.6–15.1 months] vs 16.7 months [95% CI 12.5–20.9 months], p = 0.02) (Fig. 1a). According to the HER2 status (initially vs rescued HER2 positive), both PFS and OS were worse in rescued HER2-positive patients than in initially HER2-positive patients (PFS, 5.4 months [95% CI 3.6–7.1 months] vs 7.8 months [95% CI 6.0–9.7 months], p = 0.017; OS, 10.4 months [95% CI 1.9–18.9 months] vs 16.3 months [95% CI 11.5–21.2 months], p = 0.036) (Fig. 1b).

Fig. 1
figure 1

Survival outcomes according to the HER2 status and IHC score. PFS and OS according to a the HER2 status (initially vs rescued HER2-positive GC) and b IHC score (IHC 3+ vs IHC 2+ /ISH+) in all patients. PFS progression-free survival; OS overall survival

Full size image

We evaluated the survival outcomes according to the HER2 status (initially vs rescued HER2 positive) in the same IHC category (IHC 2+ /ISH+ or IHC 3+). In IHC 2+ /ISH+ patients, PFS and OS were not significantly different according to the HER2 status (PFS, 5.6 months [95% CI 4.5–6.7 months] vs 6.3 months [95% CI 5.3–7.2 months], p = 0.659; OS, 11.3 months [95% CI 8.3–14.3 months] vs 11.9 months [95% CI 2.8–18.0 months], p = 0.605) (Fig. 2a). However, among IHC 3 + patients, rescued HER2-positive patients had worse survival outcomes than those of initially HER2-positive patients [PFS, 5.0 months (95% CI 3.9–6.1 months) vs 9.0 months (95% CI 3.5–11.4 months), p = 0.012; OS, 8.6 months (95% CI 0.0–20.4 months) vs 16.8 months (95% CI 12.2–21.5 months), p = 0.031] (Fig. 2b).

Fig. 2
figure 2

Survival outcomes according to the HER2 status in the same IHC score. PFS and OS according to (a) the HER2 status (initially vs rescued HER2-positive GC) in patients with IHC 2+ /ISH+ and (b) in patients with IHC 3+ . PFS, progression-free survival; OS, overall survival

Full size image

H-score according to IHC score and HER2 status

We assessed the H-score of all patients and divided them into four subgroups according to HER2 status (initially vs rescued HER2 positive) and IHC score (IHC 2+ /ISH+ vs IHC 3+). The initially IHC 3+ subgroup had the highest H-score (median 250, range 15–300) (Fig. 3a). The results of staining intensity and proportion of HER2-expressing tumor cells in each patient are shown in the Supplementary Fig. 2. The results in initially IHC 3+ patients showed the highest frequency of homogenous patterns than those of other three subgroups. Furthermore, the results of rescued IHC 2+ /ISH+ patients showed the lowest HER2 staining in each patient’s tumor cells and the lowest portion of homogenous pattern than those of other three subgroups.

Fig. 3
figure 3

H-score distribution and survival outcomes by H-score. a H-score in the four subgroups with different HER2 status (initially vs rescued HER2-positive GC) and the IHC score (IHC 3 + vs IHC 2 + /ISH +) and b PFS and OS according to the H-score; high (H-score > 210) vs low (H-score ≤ 210). PFS progression-free survival; OS overall survival

Full size image

The optimal cut-off value of H-score for survival outcome was 210, which was consistent with the median. With H-score of 210, the patients were divided into the high H-score group (H-score > 210, n = 74) and low H-score group (H-score ≤ 210, n = 79). 56.6% (n = 70) of initially HER2-positive patients and 14.8% (n = 4) of rescued HER2-positive patients had a high H-score. Meanwhile, 56 (44.4%) initially HER2-positive patients and 23 (85.2%) rescued HER2-positive patients had a low H-score. The low H-score group had significantly worse survival outcomes than those with the high H-score group [PFS, 5.7 months (95% CI 5.2–6.2 months) vs 10.4 months (95% CI 7.3–13.4 months), p = 0.001; OS, 12.9 months (95% CI 8.7–17.0 months) vs 17.6 months (95% CI 13.6–21.6 months), p = 0.004] (Fig. 3b).

Factors associated with survival outcomes

A multivariate analysis of PFS and OS was performed according to age, sex, HER2 status, IHC score, H-score, and previously identified risk group (Table 3). We utilized different models considering the potential multicollinearity in the final IHC score, HER2 status and H-score. The risk group, H-score, and final IHC score were included in model 1 (Table 3); HER2 status instead of the final IHC score was included in model 2 (Supplementary Table 3). In both models, poorer risk scores and H-score ≤ 210 were independently associated with shorter OS (model 1, HR = 3.18, 95% CI 1.88–5.37, p < 0.001 and HR = 1.54, 95% CI 1.02–2.31, p = 0.040, respectively; model 2, HR = 3.23, 95% CI 1.91–5.47, p < 0.001 and HR = 1.55, 95% CI 1.06–2.28, p = 0.025, respectively).

Table 3 Factors associated with the survival outcomes (model 1)
Full size table

Discussion

In this retrospective analysis of patients with HER2-positive AGC treated with trastuzumab-based first-line therapy, we investigated the clinical implications of the HER2 heterogeneity with different HER2 IHC scores, HER2 status, and HER2 H-scores. Unlike our previous GASTHER1 study, rescued HER2-positive patients showed unfavorable survival outcomes than initially HER2-positive patients, despite having same IHC score of 3+ . H-score was highest in the subgroup with initial IHC score of 3+ . Importantly, a lower H-score was independently associated with worse survival outcomes. These results highlight the impact of HER2 heterogeneity on the treatment outcomes, and H-score may be an indirect indicator of heterogenous HER2 expression levels by reflecting both proportion and intensity of HER2 expression and a predictor of trastuzumab-based first-line chemotherapy efficacy.

Although our previous GATHER1 study showed comparable treatment outcomes to those with initially and rescued HER2-positive diseases, the current study which was based on a larger number of patients and longer follow-up period revealed different survival outcomes according to the HER2 status (i.e., initially vs rescued HER2 positive), especially those with an IHC score of 3+ . Considering that the discrepancy between HER2 assessments in rescued HER2-positive patients (i.e., from initially HER2-negative tumor to HER2 + tumor on subsequent biopsy) may represent heterogenous HER2 expression; the poor survival outcomes of these patients could be due to HER2 heterogeneity. Indeed, HER2 heterogeneity has been considered as an important issue affecting the clinical outcomes of trastuzumab-based chemotherapy. Previous studies reported a limited clinical benefit of trastuzumab-based chemotherapy in patients with intermediate and heterogenous HER2-expressing GC [11, 12, 23]. Therefore, our findings accord well with those of previous studies highlighting the clinical value of heterogeneity. In particular, despite having the same HER2 IHC score of 3+ , the difference in survival outcomes between initially and rescued HER2-positive patients highlights the importance of HER2 heterogeneity, even in those who demonstrated complete and strong staining patterns (IHC 3+). Although the clinical outcomes of rescued HER2-positive tumors were worse than those of initially HER2-positive tumors due to HER2 heterogeneity, the survival outcomes between initially and rescued IHC 2+ /ISH + patients were comparable. Moreover, the survival outcomes of those with rescued IHC 3+ were non-inferior to those with initially IHC 2+ /ISH+ . Therefore, these results support the necessity of performing repeat endoscopic biopsy for rescuing patients with HER2-negative GC on initial biopsy and the use of anti-HER2 treatment in patients with rescued HER2-positive GC, albeit the suboptimal treatment outcomes due to HER2 heterogeneity.

One of the unique aspects of this study is that it comprehensively evaluated the H-score, which accounts for both intensity and proportion of staining that could reflect HER2 heterogeneity indirectly. Considering that initially IHC 3+ tumors expectedly show most homogeneous HER2 expression, our finding that H-score was highest in this subgroup suggests that it may appropriately reflect the heterogeneity of HER2 expression (homogeneity in the initially IHC 3+ subgroup). Given the relatively lower H-score of the subgroup with rescued IHC 3+ compared with the subgroup with initially IHC 3+ and its association with worse survival outcomes, H-score appears to provide additional value to the HER2 IHC scores. Importantly, the homogeneous HER2 expression represented by a high H-score (> 210) indicated a favorable survival outcome, which was confirmed in the multivariate analysis. As the H-score was associated with heterogeneity and survival outcomes, we analyzed the baseline characteristics according to the H-score. The results showed no significant differences between the subgroups divided by an H-score of 210 (Supplementary Table 4).

Our results suggest that HER2 H-score can be used as a simple tool for evaluating HER2 heterogeneity indirectly in further studies. HER2 IHC score has been classically utilized as one of the stratification factors in HER2 + GC clinical trials. However, given the substantially different H-scores and survival outcomes according to the HER2 status (initially vs rescued HER2-positive tumors) even within the same HER2 IHC category (IHC 3+), HER-2 H-score deserves to be investigated for its potential use as a stratification factor in clinical trials. It will also be interesting to investigate the value of HER2 heterogeneity with H-score in the different clinical context of HER2-directed therapies (i.e., additional pembrolizumab to first-line trastuzumab-based chemotherapy [24] and trastuzumab-deruxtecan as third-line therapy [25]).

Our study was limited by its retrospective nature, which is subject to unintended biases.

In conclusion, with more patients and long-term follow up, the survival outcomes of rescued HER2-positive patients treated with trastuzumab-based first-line chemotherapy were effective but worse than those of initially HER2-positive patients, even though the IHC score was the same as IHC 3+ in GC. Given the substantially different HER2 H-score between these patients and its association with distinct clinical outcomes, the H-score may reflect not only the intensity and proportion of HER2 expression but also the HER2 heterogeneity indirectly, which is clinically relevant in GC patients.