Award: Presidential Poster Award
Christian Davis1, Jack Harrigan2, Mahak Chauhan, MD3, Reza Hejazi, MD4, Daniel Buckles, MD5, Anwaar Saeed, MD6, Kevin Kennedy, MS2, Rashna Madan, MD5, Ajay Bansal, MD7
1University of Kansas Health System, Fairway, KS; 2University of Kansas Health System, Kansas City, MO; 3University of Missouri, Kansas City, MO; 4Kansas City University of Medicine and Biosciences, Kansas City, KS; 5University of Kansas Health System, Kansas City, KS; 6University of Kansas, Kansas City, KS; 7University of Kansas Health System / University of Kansas Cancer Center, Kansas City, KS
Introduction: The Prediction Model for Gene Mutations (PREMM) is a widely used tool for predicting the probability of Lynch syndrome but the predictive power of various cutoffs for the risk scores is unclear.
Methods: In this retrospective, single-center study, all patients with a diagnosis of colorectal or uterine cancer between January 1, 2016 and June 30th, 2018 were included. The patients were identified using specialized text-based search of the pathology lab information systems. Additionally, the search results were tallied with the local NCI-designated cancer center database to capture all patients with colon cancer. A structured RedCap electronic database was created to systematically abstract data about demographics, location and stage of colon cancer, PREMM5 scores, genetic testing and a detailed 3-generation family pedigree. All patients underwent screening for Lynch syndrome using immunohistochemical or microsatellite instability based testing. Final diagnoses in patients with abnormal screening was based on well-established commercial assays. The yield of Lynch syndrome was compared across three categories of PREMM5 scores: < 2.5%, 2.5-4.9%%, ≥ 5%. These scores reflect pre-test probability of Lynch syndrome. Statistical comparisons were performed using ANOVA.
Results: Of a total of 737 patients with either colon or uterine cancer, the final diagnosis of Lynch syndrome was made in 2.8% (21/737). The mean age at the time of cancer diagnosis across the three PREMM categories (< 2.5%, 2.5-4.9%%, ≥ 5%) was progressively lower: 68.9 ± 9.0 vs. 54.7 ± 9.5 vs. 48.2 ± 15.2, P < 0.001. Table 1 outlines the characteristics of the three groups. Table 2 describes the performance characteristics of different cutoffs of PREMM5 criteria.
Discussion: In patients with colorectal or uterine cancers, PREMM cutoff value of 2.5% identified the majority of patients with Lynch syndrome with a modest increase in the number needed to screen. Therefore, a PREMM score of 2.5% should trigger genetic testing.
Citation: Christian Davis; Jack Harrigan; Mahak Chauhan, MD; Reza Hejazi, MD; Daniel Buckles, MD; Anwaar Saeed, MD; Kevin Kennedy, MS; Rashna Madan, MD; Ajay Bansal, MD. P0228 - LOWERING THE PREMM5 MODEL THRESHOLD TO 2.5% IDENTIFIED THE MAJORITY OF PATIENTS WITH LYNCH SYNDROME. Program No. P0228. ACG 2019 Annual Scientific Meeting Abstracts. San Antonio, Texas: American College of Gastroenterology.