After the ToGA study, HER2-targeted drugs that are effective for breast cancer successively failed to show survival benefits in gastric cancer. There are clear differences between breast cancer and gastric cancer in HER2-targeted therapy, and hence there are several possible causes of the ineffectiveness of HER2-targeted therapy against gastric cancer.

The most likely reason for HER2 therapy ineffectiveness is the heterogeneity of HER2 overexpression and/or amplification, which many authors highlighted in the discussion sections of their papers [11,16,19]. Generally, one mechanism of resistance to targeted therapy is the heterogeneous expression of the therapeutic target within the tumor [21]. Even in breast cancer, heterogeneous HER2 expression is well known [22] and definitions for HER2 heterogeneity are proposed in the guidelines [23,24]. However, in gastric cancer, the HER2 expression pattern confirmed by IHC is fundamentally heterogeneous and that is why the IHC assessment of HER2 in gastric cancer was changed from breast cancer in the ToGA study and subsequent studies. The results of the quantitative proteomic analysis indicated that the expression levels of the HER2 protein in gastric cancer exhibited significant variation, and in some cases, the levels were undetectable even in tissue samples exhibiting IHC 3+ positivity [25].

HER2 signal dependency may differ between breast cancer and gastric cancer. Breast cancer is classified into six intrinsic molecular subtypes based on gene expression profiles, and the “HER2-enriched” subtype is independent as one of these subtypes [26,27]. Gastric cancer has also been classified into molecular subtypes via comprehensive analysis. Two representative classifications were reported by the Cancer Genome Atlas (TCGA) and the Asian Cancer Research Group (ACRG) [28,29]. TCGA proposed four principal molecular subgroups of gastric cancer: the microsatellite instable (MSI), Epstein–Barr virus (EBV)-associated, chromosomally instable (CIN), and genomically stable groups. Similarly, the ACRG divided gastric cancer into four subtypes: MSI, microsatellite stable (MSS)/TP53 active, MSS/TP53 inactive, and MSS/mesenchymal-like. Of note, neither classification independently categorized the “HER2-enriched” subtype. Interestingly, in TCGA classification, HER2 amplification is mainly associated with the CIN subgroup; however, other receptor tyrosine kinase alterations are also included in the same subgroup and partially overlap. Additionally, the EBV subgroup also includes HER2 amplification. These data indicated that in gastric cancer, HER2 alterations are molecularly heterogenous and might not be the independent oncogenic driver; that is, blocking the HER2 pathway might not be enough to kill cancer cells. Thus, HER2-targeted therapy for gastric cancer may require more than just blocking the HER2 pathway.

The “HER2-positive” definition is another discussion point. For some other biomarkers for targeted therapies, for instance “EGFR-positive” lung cancer, it is relatively easy to distinguish positive from negative according to whether there is an active EGFR gene mutation. Additionally, the effect of EGFR tyrosine kinase inhibitors is clearly different between “EGFR-positive” and “EGFR-negative” lung cancer [30].

However, HER2 is more complicated. HER2 expression has been used as a predictive biomarker of HER2-targeted therapy, but the degree of HER2 expression widely varies from no expression to high expression. In an early clinical trial of trastuzumab against breast cancer, HER2 expression was regarded as positive when more than 25% of tumor cells exhibited characteristic membrane staining for HER2 [31]. In the pivotal phase 2 single-agent and phase 3 combination studies of trastuzumab against breast cancer [8,32], the IHC scoring system was more organized in relation to the HER2 receptor number (no staining (score 0), <20,000 receptors; partial membrane staining with <10% of the cells showing complete membrane staining (score 1+), 100,000 receptors; light to moderate complete membrane staining in >10% of the cells (score 2+), 500,000 receptors; and complete membrane staining in >10% of the cells (score 3+), 2,300,000 receptors) [33]. In these two pivotal studies, HER2 IHC 2+ and 3+ were regarded as “HER2-positive”, but this cutoff was based on biological features, and the results of the studies indicated that patients with a HER2 score of 3+ would gain more benefit from trastuzumab [34]. Therefore, a subsequent study adopted HER2 IHC 3+ as the definition of “HER2-positive” breast cancer [35]. However, these and other data also indicated that trastuzumab had a slight effect against HER2 IHC 2+ breast cancer even though HER2 FISH was negative (HER2:CEP17 ratio < 2.0) [36]. Additionally, recent studies have shown that the intrinsic phenotypes of HER2 IHC 1+/2+ breast cancer overlap among luminal A/B, basal-like, normal-like, and HER2-enriched types [37,38]. The trastuzumab biomarkers of breast cancer would be continuous rather than categorical or binary, and this may be partially independent of genetic phenotypes. Namely, when HER2 expression/amplification is above the artificial cutoff regarding the clinically acceptable benefits of trastuzumab, it is described as “HER2-positive” breast cancer.

From this point of view, the “HER2-positive” definition should vary for each drug, as diverse HER2-targeted therapies necessitate distinct HER2 statuses to attain efficacy. Fortunately, in breast cancer, HER2-targeted therapies have the same “HER2-positive” definition as trastuzumab, but it might be different in gastric cancer. For instance, even though T-DM1 failed to show clinical benefits in ToGA study-based HER2-positive gastric cancer compared with taxane, it seemed that HER2 expression was a predictive biomarker for T-DM1 against gastric cancer [16]. If the cutoff value for HER2 positivity due to HER2 protein levels was set higher, T-DM1 might have demonstrated clinical benefits (however, this is not realistic because of the small number of applicable patients or technical/cost aspects regarding quantifying the amount of protein).

Regional differences in gastric cancer are not a major problem but should be noted. Gastric cancer in Asia and the West are biologically different with respect to incidence, etiology, tumor location, and clinicopathological characteristics [39,40]. Asia and the West also have clinical differences with regard to the screening protocols used or treatment modalities, such as surgical approaches or perioperative treatment [41]. Generally, Asian patients tend to have a better prognosis than Western patients [42,43], but it may be a chicken and egg scenario; Asia, especially Japan, tends to use more lines of chemotherapy than the West [44]. These differences complicate the interpretation of the results of global gastric cancer clinical trials.

The AVAGAST study was a global phase 3 study that aimed to evaluate the efficacy of another molecular targeted drug, bevacizumab, against gastric cancer following the ToGA study [45]. Although this study did not show an overall improvement in survival in the whole population, a subgroup analysis indicated that Western patients tended to have a prolonged median overall survival when bevacizumab was added to chemotherapy (11.5 months vs. 6.8 months, hazard ratio 0.63 in Pan America; and 11.1 months vs. 8.6 months, hazard ratio 0.85 in Europe). Conversely, in Asia, the addition of bevacizumab appeared to have no effect on the median overall survival (13.9 months vs. 12.1 months, hazard ratio 0.97). Notably, the numerical median overall survival of Asian patients without bevacizumab was longer than that of Western patients with bevacizumab. The authors pointed out that the difference in the use of the post-progression chemotherapy rate (68% in Asia and 23% other region) was one of the factors that contributed to the different survival outcomes between regions [46]. Even the ToGA study identified similar trends [47]. A subgroup analysis showed that the numerical median overall survival was not different with or without trastuzumab in Japanese patients (15.9 months for trastuzumab plus chemotherapy arm and 17.7 months for chemotherapy arm, hazard ratio 1.00) and that both were longer than the trastuzumab plus chemotherapy arm in the overall population (13·8 months). In the ToGA study, 80.4% in the trastuzumab plus chemotherapy arm and 82% in the chemotherapy arm in Japanese patients received second-line treatment, whereas only 42% in the trastuzumab plus chemotherapy arm and 45% in the chemotherapy arm in the overall population received second-line treatment. Intensive treatment and a long survival in Asia could mitigate the benefits of a new drug in clinical studies, especially in the first-line setting. These trends could also be seen in the latest global phase 3 studies of the PD-1-blocking antibody nivolumab for gastric cancer by comparing Checkmate 649, which was conducted mainly outside of Asia, with the ATTRACTION-4 study, which was conducted in East Asia [48,49,50]. In other words, very powerful HER2-targeted therapy is required to demonstrate a statistically significant overall survival in HER2-positive gastric cancer clinical studies, including in Asia.

In summary, to demonstrate the survival benefits of HER2-targeted therapy for gastric cancer, a drug is required that targets beyond the HER2 pathway-blocking mechanisms and has strong anti-tumor effects, especially for Asian patients, with an appropriate predictive biomarker (“HER2-positive” definition). Recently, a new drug, trastuzumab deruxtecan, has emerged that meets some of these requirements.