16 Background: False-negative HER2 testing in GEC may deny patients the benefit of HER2-targeted therapy. The aim of this study is to estimate the false-negative rate for HER2 testing in GEC and to evaluate predictive factors that might be used to select patients for re-testing. Methods: Two clinical trials conducted in patients with HER2-negative GEC were conducted by Genentech (NCT01662869 and NCT01590719). Information on local HER2 results was collected and testing was repeated centrally by immunohistochemistry (IHC; Ventana, 4B5 antibody) and FISH. HER2 positivity was determined using published criteria (Mod Pathol 2012;25:637). Results: Of 1,189 patients screened, 738 GECs were HER2 negative by local testing and had central results available. Twenty-nine patients (3.9%, 95% confidence interval [CI] 2.5–5.3) were locally HER2 negative, but centrally HER2 positive. The cases showed a wide geographic distribution, a mixture of biopsies (n=19) and resections (n=10), and a mixture of gastric cancer (n=20) and GEC (n=9). All false-negative cases were centrally confirmed to be Lauren’s intestinal (n=25) or mixed (n=4) morphology; amongst intestinal/mixed GEC, the false-negative rate was 6.4% (95% CI 4.1–8.6). Detailed local testing results were available for 25 cases, of which 21 cases showed local IHC scores of 0 or 1+ but were centrally IHC 2+/FISH positive (n=17) or IHC 3+ (n=4). The remaining 4 cases were locally HER2 IHC 2+, FISH negative, but centrally IHC 2+, FISH positive. Conclusions: A clinically significant false-negative rate of 6.4% was observed for HER2 testing in GEC (intestinal/mixed type). Several measures might improve the accuracy of HER2 testing, including running FISH and IHC on all cases, carefully evaluating internal positive controls, improving the education of pathologists, and retesting cases with mixed or intestinal morphology. Clinical trial information: NCT01590719 and NCT01662869.