Context: Testing for HER-2 oncogene in breast cancer has increased because of its role as a prognostic and predictive factor. Some advocate gene testing by fluorescence in situ hybridization (FISH) vs protein testing by immunohistochemistry as the method which most accurately evaluates and predicts response to the anti-HER-2 antibody, trastuzumab. However, critical examination of FISH on a screening basis has yet to be performed.

Objectives: To determine the correlation between FISH and immunohistochemistry results by determining HER-2/neu gene status on tumor sections with indeterminate immunohistochemistry results (2+ score), confirm gene amplification on tumor sections with positive results (3+ score), and verify gene status on tumor sections with negative results (0 or 1+ score).

Design, setting, and patients: A quality control and quality assurance program for HER-2 testing by FISH, which used tumor specimens from 2963 patients (median age, 56 years) with breast cancer received from 135 hospitals and cancer centers in 29 states, was performed at a reference laboratory from January 1, 1999, to May 15, 2003. Every specimen evaluated by FISH was parallel tested with immunohistochemistry tests.

Main outcome measures: With FISH as the presumed standard testing method, the positive and negative predictive values and sensitivity and specificity of immunohistochemistry were calculated.

Results: A total of 3260 clinical HER-2 tests by FISH were performed on 2963 serially referred breast cancer specimens. Of these, 2933 tests were successful and 2913 breast cancer specimens had both FISH and immunohistochemistry results available. With FISH as the standard testing method, the positive predictive value of positive immunohistochemistry score (3+) was 91.6%, and the negative predictive value of negative immunohistochemistry score (0 or 1+) was 97.2%. The sensitivity of immunohistochemistry tests, including tumor sections with scores of 2+ or 3+, was 92.6% and the specificity of immunohistochemistry tests with scores of 3+ was 98.8%. The FISH test had a significantly higher failure rate (5% vs 0.08%) and reagent cost (140 dollars vs 10 dollars), and longer testing (36 hours vs 4 hours) and interpretation times (7 minutes vs 45 seconds) vs immunohistochemistry tests.

Conclusions: A testing algorithm for HER-2 determination is most efficient by using immunohistochemistry as the method of choice, with FISH performed for cancers with indeterminate results (2+ score). Successful quality control and quality assurance programs are a prerequisite for such approaches.