Aims: To assess semiquantitative immunohistochemistry as used in the diagnosis of Lynch syndrome.
Methods and results: Tumour sections from 51 mutation carriers and 17 controls were stained with antibodies against MLH1, MSH2, MSH6 and PMS2. Intensity of immunoreactivity and percentage positivity were recorded on scales of 0-3 and 0-4, respectively. These scores were multiplied for a score of 0-12 per slide. Receiver-operator characteristic (ROC) curves of staining performance for the identification of mutation carriers were evaluated, and optimum cut-offs calculated. The area under the MLH1 ROC curve was 0.981 [95% confidence interval (CI) 0.952, 1.000]. The area under the MSH2 ROC curve was 0.899 (95% CI 0.796, 1.000). For MLH1 staining, a score<or=4 gives a sensitivity of 100.0% (95% CI 84.0, 100.0) and a specificity of 91.5% (95% CI 79.6, 97.6) for identifying MLH1 mutation carriers. For MSH2 staining, a score<or=4 gives a sensitivity of 87.5% (95% CI 61.7, 98.4) and specificity of 88.5% (95% CI 76.5, 95.6) for identifying MSH2 mutation carriers.
Conclusions: This study supports a semiquantitative slide assessment method. Protein expression may occur in the context of known pathogenic mutations, a potential pitfall in the screening process.