PURPOSE

Trastuzumab is an effective treatment for human epidermal growth factor receptor 2 (HER2)-amplified breast cancers. We sought to develop a simple protocol for HER2 image analysis of breast cancer specimens.

MATERIALS AND METHODS

In a preliminary test, we found that at least 1000 tumor cells need to be examined in the most strongly stained areas. Next, we evaluated the clinical usefulness of this established protocol of image analysis in 555 breast cancer patients. Results of the HER2 immunohistochemical (IHC) staining were compared between manual scoring and image analysis.

RESULTS

The HER2 IHC results obtained by the image analysis method correlated well with those obtained by the manual scoring method (Cohen's kappa=0.830). Using the HER2 silver hybridization (SISH) results as a gold standard, sensitivity values were 72.1% for manual scoring and 74.0% for image analysis; specificity values were 96.2% for manual scoring and 94.7% for image analysis; and accuracy values were 91.7% for manual scoring and 90.8% for image analysis. McNemar's test was applied to the results, and there were no statistically significant differences in sensitivity and specificity between the positive (=0.688) and negative (=0.118) SISH groups.

CONCLUSION

HER2 image analysis results were similar to those obtained via the manual scoring method, indicating that the use of image analysis can reduce assessment time and effort. We suggest that image analysis-based evaluation of 1000 tumor cells in the most strongly IHC-stained area, regardless of stroma content, is sufficient for determining HER2 expression levels in breast cancer specimens.