Cancer genetics evaluation among individuals at risk for Lynch syndrome across all qualifying indications.

Abstract

10616 Background: < 20% of individuals suspicious for Lynch syndrome (LS) and other inherited cancer syndromes undergo genetic testing in the US. Undiagnosed individuals will not benefit from enhanced screening and prophylactic interventions. Thus, to dramatically increase the identification of at-risk individuals and respective indications, we set up the At-Risk Cancer Genetic Syndrome Identification Registry (ACAGEN-ID). Methods: NCCN/ACMG criteria for genetic testing were translated into three distinct rule-based conditional logic statements in the EHR. A total of 218 rules that serially evaluate each aspect of individual criteria and roll into a logic statement of “at-risk” for inherited cancer syndromes. The rules assess personal history (PH) and/or family history (FH) of cancers, determine age at onset, and categorize family relationships. Patient’s genetic evaluation status, sociodemographic, and clinical data were extracted. Descriptive statistics used for summary. Pearson chi-square used for comparison of categorical variables. Results: Out of 1.34 million individuals in Yale New Haven Health System, ARCAGEN-ID identified 5,190 at risk individuals for LS. Of those, 3,581 (69%) had not been previously evaluated. Accuracy was assessed through a manual review of 130 randomly selected individuals among the identified, which showed appropriate identification in 129 cases. Among the already evaluated, 509/1609 (31.6%) had a pathogenic variant (PV): 124 (24.6%) MSH2, 112 (22.2%) MSH6, 55 (10.9%) MLH1, 118 (23.4%) PMS2, 3 (0.6%) EPCAM, 141(28%) other PV. Newly identified individuals through ARCAGEN-ID more often had only a PH of cancer (39.99% vs 21.01%) or only FH of cancer (41.02% vs 38.41%), and less often both, PH and FH (18.99% vs 40.58%) (p <0.01). The great majority of individuals with only qualifying FH or PH had not been identified before (80.90% and 70.39% respectively), while half (51.01%) with both, PH and FH, had already been identified. Having an early onset (EO) LS-associated cancer was the most common reason for prior identification (22.30%), though EO endometrial cancer (EC) (107/1135, 9.42%) was much less recognized than EO Colorectal Cancer (CRC) (284/775, 36.64%, p < 0.01). The highest missed identification before ARCAGEN-ID implementation was individuals with ≥2 LS-related cancers: 69.85% (190/272); FH of EO-CRC: 66.67% (298/449); ≥3 FH of CRC. Even 29.73% (121/407) of individuals with FH of diagnosed LS had been missed. 263/385 (68.31%) Patients with Non-EC/CRC related LS cancers were not evaluated. Conclusions: Current practice misses most individuals at-risk for LS across all qualifying indications.A system that can leverage currently existing information in the EHR can dramatically improve the identification without any other added resources. An automated outreach pilot project is underway to assess feasibility and outcomes.