Background: The aim of this study was to assess the performance of the Revised Bethesda Guidelines (RBG) and the accuracy of the Amsterdam II criteria (AM II) in identifying possible Lynch syndrome (LS) compared with the results of molecular tumour testing.
Methods: Tumours from 336 unselected colorectal cancer patients were analysed by three molecular tests (namely microsatellite instability (MSI), BRAF mutation and methylation of mismatch-repair genes), and patients were classified according to the RBG and AM II criteria.
Results: A total of 87 (25.9%) patients fulfilled the RBG for molecular tumour analyses (MSI and/or immunohistochemistry), and the AM II identified 8 (2.4%) patients as having possible LS. Molecular tests identified 12 tumours (3.6%) as probable LS. The RBG identified 6 of the 12 patients (sensitivity 50%), whereas 5 of the 8 patients who fulfilled the AM II criteria were not likely to be LS, based on molecular tests (predictive value of positive test, 38%).
Interpretation: Assuming a fairly high accuracy of molecular testing, the performance of the RBG in identifying patients with possible LS was poor, and the AM II criteria falsely identified a large proportion as having possible LS. This favours the use of molecular testing in the diagnosis of possible LS.