Abstract
INTRODUCTION: Accurate determination of human epidermal growth factor receptor 2 (HER2) status is essential for optimal patient management with trastuzumab (Herceptin). However, standard guidelines do not specify a particular commercial kit, antibody or probe for testing, and discrepancies arise from variability between kits. The aim of this study was to compare the accuracy of four commercially available fluorescence/chromogenic in situ hybridisation (FISH/CISH) kits and validate one for the resolution of borderline immunohistochemistry (IHC) cases. The interpretation pitfalls, optimal threshold values, assay duration and complexity of each kit were also considered. METHODS: The Food and Drug Administration (FDA)-approved dual-probe FISH assay PathVysion was chosen as the 'gold standard' against which pharmDx (dual-probe) and INFORM (mono-probe) FDA-approved FISH kits and the SPoT-Light CISH kit were compared. Tumours were also evaluated by IHC with the FDA-approved HercepTest kit and a validated in-house IHC protocol. Fifty-five patients with invasive breast carcinoma were selected as a representative proportion of HER2 IHC 2+ cases. RESULTS: HER2 amplification was observed in 31% of tumours by PathVysion compared with 33% with pharmDx. The number of amplified tumours detected by INFORM and CISH varied with the threshold applied. Agreement was excellent between PathVysion and pharmDx (100%), good with SPoT-Light (89%; cutoff at least five signals per nucleus) and moderate with INFORM (76%; cutoff more than four signals per nucleus). Agreement with INFORM improved to 98% with a cutoff of at least six signals per nucleus. CONCLUSION: With an appropriate cutoff, the INFORM kit was comparable to dual-probe FISH kits for evaluating HER2 status. We validate and recommend CISH as an appropriate assay for HER2 scoring that is easy to interpret and requires equipment readily found in, or that can be adapted to, all pathology laboratories. For borderline IHC cases, dual-probe FISH analysis remains the most useful protocol to apply.