Re-evaluation and re-analysis of 152 research exomes five years after the initial report reveals clinically relevant changes in 18

Abstract

Iterative re-analysis of NGS results is not well investigated for published research cohorts of rare diseases. We revisited a cohort of 152 consanguineous families with developmental disorders (NDD) reported five years ago. We re-evaluated all reported variants according to diagnostic classification guidelines or our candidate gene scoring system (AutoCaSc) and systematically scored the validity of gene-disease associations (GDA). Sequencing data was re-processed using an up-to-date pipeline for case-level re-analysis. In 28/152 (18%) families, we identified a clinically relevant change. Ten previously reported (likely) pathogenic variants were re-classified as VUS/benign. In one case, the GDA (TSEN15) validity was judged as limited, and in five cases GDAs are meanwhile established. We identified 12 new disease causing variants. Two previously reported variants were missed by our updated pipeline due to alignment or reference issues. Our results support the need to re-evaluate screening studies, not only the negative cases but including supposedly solved ones. This also applies in a diagnostic setting. We highlight that the complexity of computational re-analysis for old data should be weighed against the decreasing re-testing costs. Since extensive re-analysis per case is beyond the resources of most institutions, we recommend a screening procedure that would quickly identify the majority (83%) of new variants.

Fig. 1. Infobox, re-evaluation and re-analysis flow diagram and recommendations.

Panel A shows an infobox summarizing the basic terms in this study (panel created with BioRender.com). B A schematic flow diagram of the previous analyses performed by Reuter et al. on the same cohort, which we re-analysed with three of the main possible outcome columns from exome screening, performed the described analyses, and finally evaluated recommendations for future pipelines incorporating automated iterative re-analysis and re-sequencing. C Infobox with our recommendations for re-evaluating and re-analysing NDD screening studies.

Fig. 2. Cohort and data characteristics with results of re-evaluation.

A Distribution of the 152 assessed families classified as simplex with one affected individual and multiplex with at least two affected individuals with sex distribution of the affected individuals in these groups (first panel). The age distribution by sex in the whole cohort (note, that not all individuals could be included in this plot since one index had no age and another had no sex data). The y-axis depicts age classes at 2-year intervals for pediatric cases, plus one class for adult index individuals. The x-axis shows the number of individuals, with females on the right (green) and males on the left (red) side (second panel). The mean coverage (third panel) and length of RoHs (fourth panel) were generally higher in samples analyzed by an Illumina platform than in samples sequenced with SOLiD technology, indicating quality differences. B The alluvial plot on the left side depicts the changes in clinical assessment conclusion of the families between 2017 and 2022 (Please note that this is distinct from variant classification changes. Reuter et al. identified, for instance, a variant in the candidate gene CLMN and a VUS in the TRAPPC9 gene in the MR333 family. We reclassified the VUS as benign and identified a pathogenic variant in ADNP as the cause of the affected individual’s symptoms. The alluvial plot thus displays a line from VUS to P.). This results in clinically relevant changes in 28/152 families (18%, red; bar plot on right side colored according to the alluvial connections on the left side). The dashed line displays the shift in families where a VUS or (likely) pathogenic variant was identified. LP likely pathogenic, P pathogenic, RoH run of homozygosity, VUS variant of uncertain significance, ♀ female individual, ♂ male individual.