Problems with HER2 Testing

Abstract & Commentary

Synopsis: The overall performance of HER2 testing by local laboratories in a sample of 119 patients reportedly positive for overexpression of HER2 protein or amplification of the HER2/neu gene was examined by retesting samples in a central reference laboratory. Of these, 26% were found to be not positive. This poor concordance between local and central laboratories has led to modifications in the intergroup adjuvant breast cancer trial and also has implications for community practitioners who rely on these test results in their treatment recommendations.

Source: Roche PC, et al. J Natl Cancer Inst. 2002;94: 855-857.

Traztuzumab (antiher therapy, herceptin®) is a monoclonal antibody targeted against the extracellular domain of the HER2 protein, which is overexpressed by neoplastic cells of some patients with breast cancer. The overexpression of HER2 confers a negative prognosis, but the inclusion of trastuzumab in treatment schemes (either concordant with or following chemotherapy) has improved outcomes for patients with metastatic breast cancer in which HER2 is overexpressed. Accordingly, trials have now shifted to the adjuvant setting. One such trial, N9831 is a large, intergroup study (including patients enrolled from Cancer and Leukemia Group B, Eastern Cooperative Oncology Group, Southwest Oncology Group and North Central Cancer Treatment Group), the latter of which is serving as the coordinating center. Women with node-positive breast cancer that strongly overexpresses HER2 protein or displays HER2/neu gene amplification as determined in local laboratories have been enrolled. As was the original intention, after 119 patients were enrolled, central testing (as performed in reference laboratories at the Mayo Clinic) was compared with the results from the local laboratories. Only 74% (all of whom tested positively in local laboratories) were found to be HER2 positive in the central lab and only 6 of 9 (67%) that were demonstrated to have the HER2/neu gene amplified by the local laboratory were with similar results when tested in the reference laboratory. Incidentally, the concordance rate for the immunohistochemistry (by HercepTest™) and fluorescence in situ hybridization (FISH) assays, when both tests were performed centrally, was 92%. N9831 has been, as the result of this analysis, modified such that all future enrolled patients will have confirmation of their HER 2 status performed at the reference laboratory.

Comment by William B. Ershler, MD

The findings are a bit disconcerting. Most often community physicians rely on local laboratories for tissue and cellular analysis, and it appears from this study, when the tests are repeated in a reference laboratory, which presumably has greater experience with the assays, a quarter or more are false positives. Of additional concern is that there may be an equal or even greater number of false negatives (but this was obviously not examined in the current report). For the purposes of the intergroup adjuvant breast cancer trial in which trastuzumab was a component treatment, it appears that at least some women, perhaps as many as 26% of the initial recruitment in N9831, were not HER2 positive. This, of course has implications for reaching the study goal, which is to test the capacity for trastuzumab to influence clinical outcomes for HER2-positive patients. More importantly, HER2-negative patients may be unnecessarily subjected to any potential toxicity of the monoclonal antibody treatment. It is a credit to the study designers that this interim check was included, and that appropriate adjustments to future enrollment were made accordingly.

The take-home message for community oncologists is that we must exercise caution when interpreting laboratory tests such as the immunohistochemistry and FISH analyses performed on patient samples from which treatment programs are based. With time, it is likely that local laboratories will have gained sufficient experience such that these levels of discordance (between their interpretation and that of central reference laboratories) will be diminished. However, in the meantime, if there is any question (equivocal results or discordance between immunohistochemistry and FISH analysis), sending samples to a reference laboratory may be of critical value in assuring optimal patient care.

Dr. Ershler of INOVA Fairfax Hospital Cancer Center, Fairfax, VA; Director, Institute for Advanced Studies in Aging, Washington, D.C.


1. Cobleigh MA, et al. J Clin Oncol. 1999;17:2639-2648.