HER2 assessment in gastric cancer: proposal of a work-flow for practical routine use

Background. In gastric cancer (GC) the expression of HER2 is known as a marker of prognosis and recently it has been confirmed as a predictive marker of response to trastuzumab. Methods. GC specimens of 42 patients were collected. Representative samples from both primary tumors (42 samples) and lymph node metastases (23 samples), were selected. In each case, 4B5 (Ventana), CB11 (kit Oracle Menarini), HercepTest (Dako) antibodies were tested in immunohistochemistry (IHC) and scored as proposed for GC. HER2 gene status was studied by double probe fluorescence in situ hybridization (FISH) in all cases. Concordance among IHC scoring results of the 3 antibodies and between FISH results and IHC (0/1+ and 2+/3+), independently from the percentage of positive cells, were evaluated using the Cohen-Fleiss’ kappa statistic (K). Then, the number of specimens needed to be tested in cases with < 10% of HER2 overexpression was assessed. Finally, influence of gain of CEP17 (copies number > 3) on the results of FISH ratio was considered. Results. The 3 antibodies showed a K of 0,76 (p < 0,05). In the primary tumor, the overall concordance of FISH/IHC, taking into account the results of at least one antibody, was of 0,95 (p < 0,05). FISH/IHC concordance decreased to 0,82 (p < 0,05) when correlated with lymph node metastases. Five cases showed 2+/3+ score values in < 10% of cells. In 4 of these cases the percentage increased to > 10% adding 2 more sections from different tissue blocks of the primary tumor. In our results, the gain of CEP17 did not influence the final score ratio of FISH analysis. In conclusions, the HER2 analysis in GC needs a specific protocol avoiding working over-load and to solve equivocal cases.