Mutation prediction models in Lynch Syndrome: external validation in a clinical genetics setting

D. Ramsoekh1,2, M.E. van Leerdam1, A. Wagner3 , E. J. Kuipers1, E.W. Steyerberg2
Departments of Gastroenterology and Hepatology1, Public Health2 and Clinical Genetics3, Erasmus University Medical Center, Rotterdam

Background: The diagnosis of Lynch Syndrome is hampered by the absence of specific diagnostic features, requiring new diagnostic strategies. In the recent years, several models have been developed to predict the likelihood of carrying a germline mutation. The aim of the present study was to validate recently developed prediction models for presence of mutations in the mismatch repair genes.

Methods: We collected data of 321 families who were referred to the Department of Clinical Genetics of the Erasmus Medical Center between 1995 and 2006 because of a family history of CRC. These data were used as input for 5 different, previously published models i.e. the PREMM1,2, Leiden, Edinburgh, UK-Ams and UK-Alt model. External validity was assessed by discriminative ability (Area under the receiver operating characteristics curve, AUC) and calibration (Hosmer-Lemeshow goodness-of-fit test). For further insight, predicted probabilities were categorized as 5% or less, 5.1% to 10%, 10.1% to 20%, 20.1% to 40% and more than 40%.

Results: Of the 321 families, 66 (21%) were diagnosed with a germline mutation (25 MLH1, 23 MSH2 and 18 MSH6). All models discriminated well between high risk and low risk families (AUC: 0.82-0.84). However, calibration of the Leiden, Uk-Ams and Uk-Alt models was relatively poor, reflecting that predicted probabilities were systematically too high or too low. Using a 5% probability cut off for all the models, the sensitivity of the models ranged from 73% to 100% and the specificities ranged from 9% to 91%.

Conclusions: The Edinburgh and Premm1,2 model were the models with the best performance for an intermediate to high-risk setting and these models may be of use in clinical practice. The poor calibration of the Leiden, Uk-Ams and UK-Alt model hampers direct application of these mutation prediction models in a clinical genetics setting. Further evaluation of mutation prediction models across different settings is needed.