ABSTRACT
BACKGROUND: Frontotemporal lobar degeneration (FTLD) is a leading cause of dementia, and elucidating its genetic underpinnings is critical. FTLD research centers typically recruit patient cohorts that are limited by the center's specialty and the ways in which its geographic location affects the ethnic makeup of research participants. Novel sources of data are needed to get population estimates of the contribution of variants in known FTLD-associated genes. METHODS: We compared FLTD-associated genetic variants in microtubule-associated protein tau (MAPT), progranulin (GRN), and chromosome nine open reading frame 72 (C9ORF72) from an academic research cohort and a commercial clinical genetics laboratory. Pathogenicity was assessed using guidelines of the American College of Medical Genetics and Genomics and a rule-based DNA variant assessment system. We conducted chart reviews on patients with novel or rare disease-associated variants. RESULTS: A total of 387 cases with FTLD-associated variants from the commercial (n=2,082) and 78 cases from the academic cohort (n=2,089) were included for analysis. In the academic cohort, the most frequent pathogenic variants were C9ORF72 expansions (63%, n=49), followed by GRN (26%, n=20) and MAPT (11%, n=9). Each gene's contribution to disease was similarly ranked in the commercial laboratory but differed in magnitude: C9ORF72 (89%, n=345), GRN (6%, n=24), and MAPT (5%, n=19). Of the 37 unique GRN/MAPT variants identified, only six were found in both cohorts. Clinicopathological data from patients in the academic cohort strengthened classification of two novel GRN variant as pathogenic (p.Pro166Leufs*2, p.Gln406*) and one GRN variant of unknown significance as a possible rare risk variant (p.Cys139Arg). CONCLUSION: Differences in gene frequencies and identification of unique pathogenic alleles in each cohort demonstrate the importance of data sharing between academia and community laboratories. Using shared data sources with well-characterized clinical phenotypes for individual variants can enhance interpretation of variant pathogenicity and inform clinical management of at-risk patients and families.
ABSTRACT
We developed a rules-based scoring system to classify DNA variants into five categories including pathogenic, likely pathogenic, variant of uncertain significance (VUS), likely benign, and benign. Over 16,500 pathogenicity assessments on 11,894 variants from 338 genes were analyzed for pathogenicity based on prediction tools, population frequency, co-occurrence, segregation, and functional studies collected from internal and external sources. Scores were calculated by trained scientists using a quantitative framework that assigned differential weighting to these five types of data. We performed descriptive and comparative statistics on the dataset and tested interobserver concordance among the trained scientists. Private variants defined as variants found within single families (n = 5,182), were either VUS (80.5%; n = 4,169) or likely pathogenic (19.5%; n = 1,013). The remaining variants (n = 6,712) were VUS (38.4%; n = 2,577) or likely benign/benign (34.7%; n = 2,327) or likely pathogenic/pathogenic (26.9%, n = 1,808). Exact agreement between the trained scientists on the final variant score was 98.5% [95% confidence interval (CI) (98.0, 98.9)] with an interobserver consistency of 97% [95% CI (91.5, 99.4)]. Variant scores were stable and showed increasing odds of being in agreement with new data when re-evaluated periodically. This carefully curated, standardized variant pathogenicity scoring system provides reliable pathogenicity scores for DNA variants encountered in a clinical laboratory setting.
