One hundred and four studies were included in the review: 3 studies assessed analytical validity, 40 assessed clinical validity and 61 assessed the benefits and harms of testing.
Analytical validity (3 studies).
One study compared conversion analysis with DNA sequencing alone for the detection of mutations in MLH1, MSH2 and MSH6. This was estimated to increase yield by 56% (35 of 63 cases). A second study (n=20) assessed immunoperoxidase assays for MLH1 and MSH2 in known positive patients; the sensitivity ranged from 84 to 100% and the specificity from 45 to 100%. The third study (n=46) assessed denaturing high-performance liquid chromatography compared with DNA sequencing in colorectal cancer patients from HNPCC families; the chromatographic technique detected 19 DNA-identified mutations and 16 novel ones with no false negatives.
Clinical validity (40 studies).
The proportions of patients with mutations of MLH1, MSH2 and other MMR genes, and hence their predictive value, were reported. Pooled estimates ranged from 39 to 51%. The proportions of patients with MSI-H were 71% and 68% for Amsterdam 1 and 2 criteria, respectively. Abnormal IHC had a prevalence of 40%. For all clinical predictors, the sensitivity ranged from 28 to 100% and the specificity from 77 to 99%. Overall, the presence of one of the following predictors was as effective as more complex approaches: age younger than 50 years at diagnosis; history of colorectal or endometrial cancer in a first-degree relative; multiple synchronous or metachronous colorectal or endometrial cancers in a proband combined with IHC or MSI testing of tumour tissue.
Benefits and harms (61 studies).
There was no evidence of long-term harm caused by testing colorectal cancer patients for MMR mutations (3 studies). Indirect evidence of the impact of HNPCC diagnosis on prognosis in colorectal cancer patients was inconclusive (7 studies). Cumulative risk of colorectal cancer in family members of probands with HNPCC diagnosed by different methods ranged from 23% for women based on the Bethesda criteria to 100% for men with MLH1/MSH2 mutation (2 studies). In each case the risk was higher in men than in women. The efficacy of genetic counseling of family members before testing was considered good (16 studies). The identification of HNPCC mutations was associated with improved screening uptake (primarily colonoscopy) and consequent lower incidence of HNPCC-related cancers and mortality in family members (8 studies). There was little evidence of harm consequent upon screening or associated interventions (9 studies).