A validation of models for prediction of pathogenic variants in mismatch repair genes

Leveraging Electronic Health Record Data to Understand Gaps Underlying the Underdiagnosis of Lynch Syndrome

“…Clinical prediction models, such as PREMM5 and MMRpro, have helped simplify this, in that they can quantify the probability that an individual has LS on the basis of relatively simple personal/family history data. 12,13…”

“…Clinical prediction models, such as PREMM5 and MMRpro, have helped simplify this, in that they can quantify the probability that an individual has LS on the basis of relatively simple personal/family history data. 12,13 In parallel, it was recognized that LS cancers almost universally demonstrate a specific molecular biology of frameshift mutations at DNA tandem repeat sequences known as microsatellite instability (MSI), which can be assessed directly from polymerase chain reaction or through immunohistochemical staining for deficient MMR protein expression (MMR-D). This popularized an approach of using tumor testing for MSI/MMR-D as a means of screening patients with cancer to determine who should undergo GGT for LS.…”

“…32,36,37 Even once appropriate data are collected, however, there needs to be improved clinical decision support to help busy clinicians understand who, how, and when to refer for GGT; the EHR may help facilitate these decisions. 34,38 Clinical prediction models such as PREMM5 and MMRpro use relatively simple clinical data to numerically quantify an individual's likelihood of LS, 12,39,40 and the National Comprehensive Cancer Network guidelines recommend GGT for those with ≥2.5%-5.0% predicted probability of LS. 13 Our group has recently developed a patient-facing version of PREMM5 embedded within the EHR, through which patients submit the required data points (age, personal and family history of specific LSrelated cancers) before a clinical visit.…”