Based on our results we propose three possible mechanisms
Based on our results, we propose three possible mechanisms of HER2 loss after trastuzumab-based therapy (Fig. 3). The first possibility is a result of the differential effects of treatments on clonal subsets. In breast cancer, such selective pressure has been suggested to preferentially eradicate HER2-positive cancer clones, whereas the HER2-negative ones might drive tumour progression , , . Therefore, first, it is conceivable that in gastric cancer, residual HER2-negative 777 symbolism re-grow after trastuzumab selectively kills the HER2-positive cells. Second, heterogeneity of HER2 expression in gastric cancer could affect HER2 status if evaluated using biopsy samples. Heterogeneity of HER2 expression and/or HER2 amplification exists in a subset of HER2-positive breast cancers . This phenomenon can be observed in gastric cancer, but it may be even more relevant in this context than in breast cancer because of the marked heterogeneity of gastric cancer . An assessment of the percentage of the HER2-stained tumour cells before and after trastuzumab-based therapy would be helpful to answer the question of whether clonal selection or heterogeneity is the main mechanism underlying HER2 loss. However, unfortunately, the HER2-positive status before the first-line, trastuzumab-based therapy was determined by each institution in this study, meaning that it was impractical to collect the biopsied samples used for the initial diagnosis of HER2 positivity and to evaluate the percentage of the HER2-stained tumour cells before and after trastuzumab-based therapy. In a recent study , among HER2 positive patients, about 70% of them were reported to have HER2 heterogeneity, and the percentage of HER2 overexpressing cells ranged from 10% to 90%. This evidence suggests that both clonal selection and heterogeneity could still be possible mechanisms that would explain HER2 loss. The third possibility is due to the effect of fixation conditions for the biopsy samples. In this study, 10% NBF was not used for formalin fixation of re-biopsied samples in 16 of 33 cases (48.5%), although the guidelines recommend the use of 10% NBF for HER2 testing . Ten percent NBF was used significantly less often in the HER2-negative group than in the HER2-positive group, suggesting that the use of fixation methods other than 10% NBF may have reduced the detection rate of HER2 positivity. Reduced IHC staining intensity of HER2 has been experimentally observed when 10% NBF was not used for fixing . Because ours was an observational study based on biopsy samples obtained from clinical practice at several locations, the fixation conditions unexpectedly varied. Our results highlight that it is of critical importance to optimise sample preparation conditions for HER2 testing in gastric cancer. Preclinical studies have suggested several genomic mechanisms of resistance to trastuzumab in HER2-positive gastric cancer. Notably, retrospective in silico analyses demonstrated that the HER2 gene might be coamplified with EGFR or c-met , , . Receptor co-amplification and some other putative mechanisms of resistance to trastuzumab, such as PIK3CA/PTEN mutations, have been observed in samples from patients with HER2-positive gastric cancer and were preclinically validated . In this study, acquired gene aberrations after trastuzumab-based therapy were found only in one patient, who had EGFR amplification, although 9 of 33 patients (27.3%) carried some gene aberrations in the initial biopsy samples. On the other hand, interestingly, 7 EGFR/c-met amplifications were apparently lost after trastuzumab-based treatment, regardless of HER2 loss. These results suggest that trastuzumab-based therapy may have eradicated EGFR/c-met–amplified cancer clones in those seven patients. Our work has some limitations. First, the small sample size of this observational study limits the accuracy of results. In particular, the large variation in formalin fixation conditions could considerably affect the evaluation of HER2 loss and gene aberrations. HER2 loss was identified in 7 of 17 (41.2%) samples in the cases whose biopsy samples were fixed with 10% NBF. Taking into account results from previous studies , , we assume that HER2 loss occurs at a rate of 30%–40% after the development of trastuzumab resistance in HER2-positive gastric cancer. Although it would be ideal to determine the usefulness of re-biopsies through the use of large-scale trials, this is thought to be impractical because re-biopsies in recurrent/advanced gastric cancer patients are often difficult in practice. Therefore, we believe that the findings from the present study offer some useful information for determining the significance of the re-assessment of the HER2 status. Second, other rare genomic (such as KRAS, HER3 and PTEN aberrations) or even nongenomic mechanisms of resistance might have been missed in this study. Future comprehensive genomic analyses to explore the biological mechanisms of resistance to trastuzumab are needed.