Immunohistochemical staining for DNA mismatch repair proteins in intestinal tract carcinoma: how reliable are biopsy samples?

Background: In recent years, immunohistochemistry has emerged as an efficient tool in the detection of DNA mismatch repair protein abnormality in colorectal cancers. Currently, the immunohistochemical test is mainly applied to cancer resection specimens. Detection of mismatch repair abnormality in biopsies carries obvious clinical importance, as it would allow informed decision about the extent of surgery (segmental resection vs total colectomy, prophylactic hysterectomy or not). Moreover, in the case of treated rectal carcinoma with no residual tumor, it provides a means to evaluate the mismatch repair proteins. However, whether biopsy samples can be reliably used for mismatch repair protein detection remains to be determined. Materials and Methods: Paired biopsy and resection specimens of adenocarcinomas of the gastrointestinal tract, enriched for patients at increased risk for Lynch syndrome, were analyzed for immunohistochemical staining patterns for MLH1, MSH2, MSH6, and PMS2. Abnormal staining was defined as total loss of protein in the tumor with appropriate control. Cases with focal and weak staining, defined as staining of no more than moderate intensity present in < 10% of the tumor cells, were recorded. Correlation analysis with germline mutation data was in a subset of cases. Results: Among 70 gastrointestinal tract cancers (3 from the small bowel, 36 from the right colon, 15 from the left colon, and 16 from the anorectum), both the biopsy and resection specimens detected the same 29 cancers as having loss of staining for at least 1 protein, 14 affecting MLH1/PMS2 and 15 affecting MSH2/MSH6. Focal and weak staining was most commonly seen for MLH1 stain in biopsies (4 of 70, 6%), followed by MSH6 stain in biopsies (3 of 70, 4%). Concordant staining patterns between biopsies and resections were reached in all 70 cases for MSH2 and PMS2, whereas discordant patterns were identified in 3 cases (3 of 70, 4%) for MLH1 and in 2 cases (2 of 70, 3%) for MSH6. None of the discordant patterns affected the final interpretation of whether the immunohistochemistry test was normal or abnormal in either the biopsy or the resection. In 13 of the 13 cases that were known to have a pathogenic germline mutation (5 in MLH1 and 8 in MSH2), the stains were abnormal for the corresponding protein and/or its partner protein in both the biopsy and the resection specimens. Conclusions: This study provides data indicating that biopsy samples are as reliable as resections in the immunohistochemical detection of mismatch repair protein abnormality in intestinal cancers. Our study also shows that various staining variations can occur in both biopsies and resections. Awareness and further understanding of such variations will enhance the use of immunohistochemistry, a commonplace tool that is being increasingly used in the screening workup for Lynch syndrome.