The data we presented in San Antonio described the initial results from patients screened for the Breast Cancer International Research Group clinical trials. In order to enter these clinical trials, the women needed to have tumors with HER2 gene amplification determined by FISH. Our laboratory and the laboratory of our collaborators in Switzerland are the central laboratories for screening all of the cases and determining HER2 gene amplification.

From the first 2,600 cases submitted, 2,543 had tissue samples that were acceptable for FISH characterization. We were able to obtain FISH results on 2,502 samples for a 98.4 percent success rate. Of those cases, 655 showed HER2 gene amplification, which was a 26 percent amplification rate.

Retrospectively, we requested the referring laboratories to indicate whether they had previously assessed the patients’ HER2 status by any means. IHC had been performed on 1,608 cases at an outside laboratory. We were interested in how those outside laboratory determinations correlated with our central laboratory FISH analyses.

Of the cases that were classified by IHC at an outside laboratory as 0 or 1+, four percent to six percent had HER2 gene amplification as determined at our central laboratory. Approximately 79 percent of the cases that scored 3+ by IHC at an outside laboratory actually had HER2 gene amplification. Cases that scored 2+ by IHC at an outside laboratory had a HER2 gene amplification rate of about 17 percent.

There was about a 92 percent agreement between the results obtained by FISH at an outside laboratory and FISH at our central laboratory. This was much higher than the agreement between IHC at an outside laboratory and FISH at our central laboratory.

Michael F Press, MD, PhD

IHC was all we initially had available for testing, but early on we saw that IHC was flawed. IHC has a false-negative rate of about 18 percent. In a good laboratory, the false-positive rate for IHC is probably a few percent; it goes up to 8 percent in general laboratories and was as high as 40 percent in some of the early reported trials.

Mike Press has data demonstrating a 52 percent concordance with the Dako HercepTest™ among Dako-approved pathologists. The College of American Pathologists has done its own study, evaluating the concordance between a central laboratory and pathologists in the community. They are seeing similar trends.

Dennis Slamon, MD, PhD