Tumours from MSH2 mutation carriers show loss of MSH2 expression but many tumours from MLH1 mutation carriers exhibit weak positive MLH1 staining

Abstract

Microsatellite analysis (MSA) in tumour tissue is useful for pre-selection of hereditary non-polyposis colorectal cancer (HNPCC) patients for mutation screening, but is time-consuming and cost-intensive. Immunohistochemistry (IHC) for expression of MLH1 and MSH2 proteins is simple, fast, and indicates the affected gene. IHC has therefore been proposed as an alternative pre-screening method. However, some authors report a lower sensitivity of IHC compared with MSA. The present study reports IHC results for MSH2 and MLH1 performed in 82 tumours with high microsatellite instability (MSI-H) from 81 carriers of pathogenic mutations in MSH2 or MLH1. One hundred per cent (38/38) of the tumours from MSH2 mutation carriers showed loss of MSH2 staining; in all cases, the affected MSH2 gene was predicted correctly by IHC. Complete loss of MLH1 expression was observed in 66% (29/44) of MLH1 mutation carriers. Weak positive MLH1 staining was observed in 14 (32%) cases and, in one case, normal MLH1 staining was seen. The pathologist was aware of the weak staining pattern as an indicator of an MLH1 mutation; 98% of the MLH1 mutations were predicted correctly. To evaluate whether weak positive MLH1 staining is observed more often with in-frame or missense mutations, IHC data from 23 MSI-H tumours from carriers of unspecified variants were added and mutations were grouped into truncating mutations, large non-truncating deletions, and small non-truncating mutations. Weak MLH1 staining was observed in all three categories and it is postulated that other factors, such as mutation of the second allele, also influence protein expression. In conclusion, IHC can be regarded as a very useful method for selecting HNPCC patients for mutation analysis, as long as it is interpreted by an experienced pathologist. The high specificity of IHC in terms of indicating the affected gene is useful for evaluating unspecified variants. However, the staining pattern does not predict whether the underlying germ-line mutation is truncating or not.