Are Immunohistochemical Markers Useful in Phenotypic Gastric Cancer Classification?

To identify useful markers for prognostic and therapeutic purposes, The Cancer Genome Atlas (TCGA) provided a molecular classification of gastric cancers (GCs). Previous studies have used immunohistochemistry (IHC) and chromogenic in situ hybridization (CISH) to define immunophenotypic surrogate markers of the molecular alterations. Some critical issues concerning the correct definition of immunophenotypic groups have emerged in these studies that employed tissue microarrays (TMAs). We performed an immunophenotypic classification by evaluating MLH1, p53, HER2, E-cadherin, and Epstein-Barr virus (EBV) on the whole section of the surgical GC samples compared to most of the studies conducted on TMAs. We also investigated the immunohistochemical expression of PD-L1, a known therapeutic target. We identified the following immunophenotypic groups: EBV (2.9%); mismatch repair deficient (MMR-D) (7.2%); overexpressed p53 and/or HER2+ (61.4%); aberrant E-cadherin (11.4%); and normal pattern (17.1%). The use of surgical samples emphasized that some immunohistochemical markers were not useful for properly classifying the GC specimens. We can state that EBV (significantly correlated to PD-L1 expression) and MMR-D GCs are well-defined groups, mutually exclusive, and easily assessable with IHC and CISH, and could be candidates for immunotherapy with PD-1/PD-L1 inhibitors. As regards p53, our findings suggest that IHC assessment may be responsible for a misclassification of GC groups. Immunohistochemical evaluation of E-cadherin needs to be standardized, particularly in terms of the heterogeneous cytoplasmic/membranous staining pattern. Whether to consider the normal-pattern group as a separate category remains to be clarified. Because GC specimens with known therapeutic targets account for only 40%, we suggest reviewing the immunophenotypic classification to find new therapeutic targets, such as PD-L1, MLH1, and HER2.