A retrospective cohort analysis of the Yale pediatric genomics discovery program

Abstract

The Pediatric Genomics Discovery Program (PGDP) at Yale uses next-generation sequencing (NGS) and translational research to evaluate complex patients with a wide range of phenotypes suspected to have rare genetic diseases. We conducted a retrospective cohort analysis of 356 PGDP probands evaluated between June 2015 and July 2020, querying our database for participant demographics, clinical characteristics, NGS results, and diagnostic and research findings. The three most common phenotypes among the entire studied cohort (n = 356) were immune system abnormalities (n = 105, 29%), syndromic or multisystem disease (n = 103, 29%), and cardiovascular system abnormalities (n = 62, 17%). Of 216 patients with final classifications, 77 (36%) received new diagnoses and 139 (64%) were undiagnosed; the remaining 140 patients were still actively being investigated. Monogenetic diagnoses were found in 67 (89%); the largest group had variants in known disease genes but with new contributions such as novel variants (n = 31, 40%) or expanded phenotypes (n = 14, 18%). Finally, five PGDP diagnoses (8%) were suggestive of novel gene-to-phenotype relationships. A broad range of patients can benefit from single subject studies combining NGS and functional molecular analyses. All pediatric providers should consider further genetics evaluations for patients lacking precise molecular diagnoses.

The best of both worlds: Blending cutting‐edge research with clinical processes for a productive exome clinic

“…The DRSC provides diagnostic services to patients with undiagnosed Mendelian disorders by combining standard clinical evaluations with cutting‐edge research activities. Our program is unique because it incorporates all the individual elements that have been implemented in prior published hybrid and clinical settings for ES 8,10–14,18 . These elements include: (a) specific inclusion criteria for patient selection and a clinical genetics evaluation, (b) a phenotype‐agnostic analysis of the ES data, followed by manual curation of variants by clinicians, with close collaboration with a bioinformatician, (c) collaborative pursuit of VUS via active and passive processes, (d) communication of diagnoses in well‐defined categories, to make clear the actionability of the results to the patients/families, (e) periodic time‐lapsed reanalysis, as well as “watching” genes via publicly available resources such as PubMed, and (f) capturing the change in diagnoses over time—although it is acknowledged that ES results can change over time, 13 the underlying reasons for these have not been systematically examined.…”

“…These processes included: (a) preselecting patients with specific criteria (2/4 programs 10,13 ), (b) utilizing a multidisciplinary team of clinicians, laboratory personnel, and researchers to determine patient selection (1/4 13 ) and to review results (3/4 12,13,18 ), (c) utilizing varying bioinformatics approaches, with two programs using commercial ES with research analysis only if the commercial ES was non‐diagnostic, 10,12 and two program pursuing ES analysis solely through a research pipeline 13,18 . One program also reanalyzed ES data from outside studies, 18 (d) pursuit of additional processes for diagnostic resolution: “dry laboratory” activities included case matching (two programs 12,18 ) and three programs pursuing extensive variant curation; “wet laboratory” activities included other OMICs, and functional assays 12,13,18 . These programs reported initial diagnosis rates of 25%–43%, with 2%–3% of diagnoses being novel gene discoveries 12,18 .…”

“…One program also reanalyzed ES data from outside studies, 18 (d) pursuit of additional processes for diagnostic resolution: “dry laboratory” activities included case matching (two programs 12,18 ) and three programs pursuing extensive variant curation; “wet laboratory” activities included other OMICs, and functional assays 12,13,18 . These programs reported initial diagnosis rates of 25%–43%, with 2%–3% of diagnoses being novel gene discoveries 12,18 . (e) Reanalysis of non‐diagnostic ES data: this occurred in all the four hybrid studies, with one program reporting an additional diagnosis rate of 17% and rescinding diagnoses in 1.9% 13 and (f) only one program communicated results to patients in specific categories, such as definitely causal, probably causal and uncertain 13 .…”