Aims Variability in determining HER2 status has been reported, especially, differences in sensitivity and specificity among commercially available antibodies, with false positive and false negative results. We compared the sensitivity and specificity of five anti-HER2 antibodies by immunohistochemistry (IHC), using the new dual colour brightfield in situ hybridisation (DDISH) as the gold standard, on invasive breast carcinomas (IBC) arrays.

Material and methods Serial sections from tissue microarrays (TMA) containing 200 preselected primary IBC were submitted to DDISH (VENTANA INFORM HER2 Dual ISH assay), and immunohistochemistry, using Dako A0485 and HercepTest (polyclonal), Novocastra CB11 (mouse monoclonal), NeoMarkers SP3 and Ventana 4B5 (rabbit monoclonal).

Results From the initial 200 cases, 184 were assessed by DDISH and IHC. The concordance among the antibodies was considered very good (kappa statistics varied from 0.82 to 0.9). The overall concordance between IHC and DDISH ranged from 94.1% for CB11 to 96.6% for A0485. The antibodies A0485, HercepTest, SP3 and 4B5 were over 95% sensitive and specific. CB11 was the most specific antibody (97.1%). 60% (CB11) to 83.3% (SP3) of the 2+ cases showed no gene amplification by DDISH. False negative cases varied from 0.5% (A0485) to 3.8% (CB11) of the cases, and false positive from 1.6% (CB11) to 2.7% (HercepTest, SP3 and 4B5) of the 184 cases.

Conclusions There was very good agreement among the five anti-HER2 antibodies. CB11 was the most specific antibody, but showed more false negative cases. A0485, SP3, 4B5 and HercepTest were highly sensitive and specific, but showed more false positive cases.