Interpretation of HER2 tests in gastric cancer: confirmation of interobserver differences and validation of a QA/QC educational program.

摘要

Accurate testing for human epidermal growth factor 2 (HER2)-positive status is now mandatory to identify gastric cancer patients that will respond to trastuzumab treatment. Immunohistochemistry testing is the primary method used in hospitals. We performed a study of diagnostic accuracy by assessing interobserver variability in immunohistochemistry scoring of HER2 and determined the effectiveness of an educational program for general pathologists that used full sections of gastric cancer specimens. A first ring study (Japanese gastric cancer [JGC] ring study) was performed by five expert pathologists, using 50 whole surgical sections selected by a coordinator, to confirm interobserver discrepancies. A second study (quality assurance/quality control program) involved administration of an educational program to 49 general pathologists that consisted of (i) comments and explanation for a set of pre-educational program cases, (ii) a lecture, and (iii) presentation of typical and special cases for discussion. Effectiveness was measured by comparing indices of the difference between scores before and after the program. The JGC ring study demonstrated good agreement in the interpretation of HER2-immunohistochemistry. Kappa coefficients among the five observers were 0.73 (substantial) and 0.84 (almost perfect) in 4?×?4 and 3?×?3 cross tests, respectively. In the second study, the concordance rate and kappa coefficients improved from pre-educational program levels of 78.6 % and 0.68, respectively, to post-educational program levels of 87.1 % and 0.79, respectively. The present results suggest that effective educational programs reduce interobserver differences between pathologists and provide optimal information regarding patient selection for treatment.