HER2 overexpression or amplification has been reported in approximately 20% of GC cases[22-25]. In contrast to breast cancer, in which HER2-positivity is significantly correlated with a poor prognosis, the prognostic significance of HER2 overexpression or its amplification in GC has been the subject of controversy[23,26]. This issue might be attributable to the unique characteristics of HER2 expression patterns in GC, such as discrepancies in the frequency of HER2 overexpression (30% in intestinal type vs 6% in diffuse type) according to the Lauren’s classification, a well-known prognostic factor in GC, and variability in IHC staining that can indicate tumor heterogeneity in HER2 expression, particularly in IHC 2+ cases[25].

Based on the efficacy of trastuzumab in preclinical models of GC[27] and its marked success in HER2-positive breast cancer, a randomized phase 3 ToGA trial that compared chemotherapy with or without trastuzumab was conducted[15]. When the ToGA trial was planned, validated test methods and scoring systems for HER2 status were widely available for breast cancer, but not GC. Therefore, a validation study was performed to establish a HER2 scoring system for GC to identify suitable patients for enrollment in the trial[23]. In that study, which used the HercepTest^TM and FISH PharmDx^TM Kit, it was noted that tumor heterogeneity and basolateral membrane staining were more common in GC than in breast cancer. That study concluded that the scoring system of HER2 IHC assessment using the HercepTest^TM in breast cancer should be applied to GC with some modifications because of incomplete reactivity in tumor cell membranes and tumor heterogeneity, which are both frequently observed in GC (Table 1). It was also recommended that both IHC and FISH testing should be used to select patients for the ToGA trial because of differences in results between GC and breast cancer. Based on this validation study of the HER2 scoring system in GC, patients were enrolled in the trial if their tumor samples were scored as 3+ by IHC or as FISH-positive (a HER2: CEP17 [centromeric probe 17] ratio ≥ 2)[15].

In the ToGA trial that enrolled 594 chemotherapy-naive patients with GC, the addition of trastuzumab significantly improved the efficacy of chemotherapy with 2.7 mo of benefit in median overall survival (13.8 mo for chemotherapy with trastuzumab vs 11.1 mo for chemotherapy alone)[15]. After the success of the ToGA trial, trastuzumab became the first biological agent approved for the treatment of GC, and the combination of trastuzumab and cytotoxic chemotherapy is now considered as a standard treatment for metastatic or recurrent HER2-positive GC. Although the results of the FISH and IHC assays have been shown to be highly correlated[24], HER2 overexpression assessed by IHC was more significantly correlated with the efficacy of trastuzumab, irrespective of the FISH results. Indeed, the median overall survival was 10.0 mo in the IHC 0 or 1+/FISH-positive subgroup, whereas it was 16.0 mo in the IHC 2+ or 3+/FISH-positive subgroup[15]. This improved efficacy of trastuzumab in HER2 IHC 2+/FISH-positive or IHC 3+ GC patients was supported by the findings of subsequent phase 2 trials that used trastuzumab-containing regimens against GC[28,29]. Ongoing trials for novel HER2-targeted therapies, such as pertuzumab[30] and T-DM1 (NCT01641939), now use this selection criterion (i.e., IHC 2+/FISH-positive or IHC 3+) for the inclusion of patients. Additionally, IHC staining is recommended as the initial testing modality for all GC patients to define HER2 positivity in daily clinical practice[24,31].

Lapatinib is a tyrosine kinase inhibitor that blocks both epidermal growth factor receptor (ErbB1) and HER2, which has been approved for the treatment of HER2-positive breast cancer after progression to trastuzumab[32]. The efficacy of lapatinib in GC was investigated in two large randomized phase 3 trials[33,34]. The TyTAN trial compared the combination of lapatinib plus paclitaxel with paclitaxel alone in a second-line setting for 281 patients with HER2-amplified GC that was assessed by FISH (HER2: CEP17 ratio ≥ 2)[33]. That trial failed to show significant improvements in overall survival with the addition of lapatinib in a second-line setting. Despite the negative results in an intent-to-treat population, subgroup analyses revealed that lapatinib was significantly associated with better overall survival in patients with HER2 IHC 3+ GC. This finding suggested that the efficacy of lapatinib might correlate with HER2 overexpression as assessed by IHC, which was also shown in a ToGA trial for trastuzumab. As this trial only used FISH for patient selection, more patients with HER2 IHC 0/1+ (35%) were included than in the ToGA trial in which 22% of patients were HER2 IHC 0/1+[15,33]. Therefore, improper selection of the target patient population might represent a potential reason for negative results, although the lack of efficacy of lapatinib against HER2-positive GC should also be considered. The TRIO-013/LOGiC trial tested first-line capecitabine plus oxaliplatin with or without lapatinib in patients with HER2-amplified or overexpressed GC, which was defined as IHC2+ and FISH amplified, or IHC 3+, or FISH, CISH, or SISH amplified[34]. In this trial of 487 patients, the addition of lapatinib did not improve overall survival, consistent with the results of the TyTAN trial. Moreover, subgroup analyses of this study did not show a correlation between IHC and outcomes with lapatinib, which contradicted the TyTAN trial findings. The lack of efficacy with lapatinib in these trials might be a consequence of negative interactions between lapatinib and partner cytotoxic chemotherapy agents, insufficient activity of lapatinib on HER2-positive GC, or improper selection of the patient population.

In addition to the IHC HercepTest^TM and FISH PharmDx^TM used in the ToGA trial, alternative assays or techniques to assess HER2 status have been evaluated. In the subset analysis of the TRIO/LOGiC trial, the results of the PathVysion HER2 FISH probe (Abbott Molecular Inc., IL) were highly correlated with those of HER2 FISH PharmDx^TM in the central laboratory, with rates of positive agreement of 97.9% and negative agreement of 99.1%[35]. In a retrospective analysis, there was a high concordance of IHC staining results between the HercepTest^TM (polyclonal antibody) and Pathway (monoclonal antibody; Ventana Medical System, Tucson, AZ) for GC patients[36]. Among various methods used to assess HER2 gene amplification, silver in situ hybridization (SISH), a bright field in situ hybridization (ISH) method, has been suggested to be a valid alternate option for GC, as previous studies showed 94-100% concordance with FISH findings[36-38].

Although there are no approved companion diagnostics for detecting activation of the hepatocyte growth factor receptor (MET) signaling pathway, it represents one of the most widely investigated biomarkers in GC[39-41]. The role of the MET signaling pathway in tumorigenesis and metastasis has been well documented[42] and MET overexpression or amplification has been suggested to be a negative prognostic marker in GC patients[40]. The preclinical activity of MET inhibitors against MET-amplified or overexpressed GC has also been well established[39]. Multiple drugs that target MET signaling pathways are now in the early and late phases of clinical trials for GC patients. Along with the early clinical development of these agents, efforts to define biomarkers predictive of the efficacy of these agents have been ongoing. Onartuzumab (MetMab) is an anti-c-MET monoclonal antibody and, as MET overexpression has been associated with increased efficacy of onartuzumab in a randomized phase 2 trial for patients with lung cancer[43], MET overexpression assessed by IHC was selected as a marker to enrich patients with MET-positive tumors in trials for GC. A randomized phase 2 trial was conducted for patients with HER2-negative GC to compare modified FOLFOX6 plus onartuzumab with modified FOLFOX6 alone[44]. In that trial, the addition of onartuzumab did not improve progression-free survival in either an unselected population or in MET-positive patients defined by IHC (≥ 50% of a tumor with moderate to strong intensity staining on central review). Moreover, there was no correlation between the efficacy of onartuzumab and the intensity of MET expression or different definitions for MET positivity (≥ 90%). Rilotumumab, a hepatocyte growth factor (HGF)-targeted monoclonal antibody, was investigated in combination with epirubicin, cisplatin, and capecitabine (ECX) in a randomized phase 2 trial for GC[45]. In that trial, two different doses of rilotumumab (7.5 and 15 mg/kg) were tested and the addition of rilotumumab was significantly associated with improved progression-free survival [5.7 mo in both rilotumumab groups (pooled) vs 4.2 mo in the placebo group; HR = 0.60; P = 0.0116]. In an exploratory analysis of this trial, MET-positivity was defined as ≥ 25% tumor membrane staining and this MET-positive subgroup appeared to have a benefit in overall survival with rilotumumab. Meanwhile, in early phase trials of ABT-700, an anti-c-MET monoclonal antibody, and AMG 337, an oral MET kinase inhibitor, promising efficacy was shown in subgroups of patients with MET-amplified tumors as assessed by FISH[46,47].

Well-defined MET positivity appears to be critical for the success of MET inhibitors in GC. This may depend upon the tumor characteristics, properties of the IHC assay, and characteristics of the therapeutic agents. As suggested previously, assessing MET overexpression by detecting the extracellular domain of MET is more likely to predict the efficacy of a monoclonal antibody, such as rilotumumab[45]. However, further validation studies that are based on larger sample sizes are needed to bolster this conclusion.

The fibroblast growth factor signaling pathway is also considered to be a potential target for the treatment of GC[39,41,48-50]. AZD4547[51] and dovitinib[52] are fibroblast growth factor receptor 2 (FGFR2) tyrosine kinase inhibitors that are currently under investigation in phase 2 trials for GC (NCT01457846 and NCT01719549). Both trials include patients with FGFR2-amplified GC, but use different detection methods (FISH in the AZD4547 trial and quantitative real-time PCR in the dovitinib trial). In a study that compared the result obtained using quantitative real-time PCR, IHC, and FISH in GC tissue samples, robust correlations of both quantitative real-time PCR and IHC data were shown with the FISH findings, which represents the most commonly used technique and is considered to be a standard method for FGFR2[53]. However, this correlation should be validated in future prospective clinical trials.

Olaparib, an oral small molecule inhibitor of poly (ADP-ribose) polymerase (PARP), was tested in a randomized phase 2 trial[54]. For inclusion in that study, ataxia telangiectasia mutated (ATM)-negative tumors assessed using an IHC assay were necessary considering the preclinical finding that low ATM protein expression correlates with olaparib sensitivity in GC cell lines[55]. Although progression-free survival, the primary endpoint of that study, did not differ between the olaparib and placebo groups, overall survival was significantly improved in the olaparib group compared with the placebo group[54]. The mechanism underlying the benefit in overall survival without causing a difference in progression-free survival was unclear based on that study. However, a phase 3 trial of olaparib for GC is currently ongoing (NCT01924533) and its results will be helpful for determining whether the absence of ATM expression is a predictive biomarker for PARP inhibitors.

As ramucirumab, an anti-VEGFR-2 antibody, has been approved for the treatment of GC based on the success of the REGARD and RAINBOW trials[16,17], targeting the VEGF pathway is now considered to be a valid strategy for treating GC. However, no biomarker has been established that can predict the efficacy of ramucirumab. Further studies are urgently required to identify potential biomarkers for VEGF-targeted therapy, including ramucirumab.