Clinical RNA sequencing clarifies variants of uncertain significance identified by prior testing

Abstract

Purpose: Sequencing-based genetic testing often identifies variants of uncertain significance (VUS) or fails to detect pathogenic variants altogether. We evaluated the utility of RNA sequencing (RNA-seq) to clarify VUS or identify missing variants in a clinical setting.

Methods: Over a 2-year period, genetics providers at a single institution referred 26 cases for clinical RNA-seq. Cases had either no candidate variant identified by prior testing or a VUS suspected to impact splicing or expression. A committee reviewed each submission to ensure it met study criteria.

Results: Among 26 cases, 8 could not be sequenced because of poor expression in an accessible tissue, 2 did not meet inclusion criteria, 3 were solved prior to collection, and 4 families declined participation or did not complete sample collection. For the 9 cases sequenced, the clinical laboratory reported two positive, four negative, and three "indeterminate." For all three indeterminate cases, original RNA-seq data was manually evaluated and deemed explanatory.

Conclusion: Clinical RNA-seq can clarify VUS, especially splice variants, but laboratory-specific interpretation guidelines may lead to indeterminate results. Identifying individuals likely to benefit from RNA-seq and providing appropriate counseling poses unique challenges.

Keywords: RNA sequencing; clinical genetic testing; germline testing; splice variants; variant of uncertain significance.

Figure 1

RNA sequencing workflow and examples of indeterminate variants felt to be pathogenic after re-evaluation. A. For each case submitted, the clinical laboratory was contacted to see if the specific gene could be sequenced. For those that could be sequenced, a committee reviewed the case to ensure that it met the inclusion criteria. After approval, families were contacted and consented to the study. Three cases were withdrawn before the family was contacted, and 4 families either could not be reached for consent or did not complete the sample collection process. Sequencing was completed on 9 samples, and original sequencing data were obtained to better understand the laboratory interpretation. Review of 3 cases reported as indeterminate by the laboratory resulted in the variant being treated as likely pathogenic by the clinical team. B. RNA-seq analysis for an intronic ZEB2 variant was reported as indeterminate by the clinical laboratory. Review of the data revealed evidence of altered splicing in approximately 25% of reads that were split between an exon and the position of the intronic variant (split reads are colored yellow). C. A synonymous variant in DICER1 predicted to alter splicing and result in a 29-amino acid in-frame deletion within exon 16 was reported as indeterminate by the clinical laboratory. Review of the RNA sequencing data revealed that of the 61 reads mapping to the c.2523A>G polymorphism, 12 did not contain the first 29 amino acids (highlighted in yellow), suggesting altered splicing. A further 3 reads contained the c.2523A>G polymorphism, suggesting that they were processed normally (dashed circle); 2 reads contained the c.2523A>G, but at the third-to-last base pair position of the read (dashed box), which is also contained on the neighboring exon; thus, it is unclear if these reads should be split or if this is evidence of normal processing of the transcript. D. To assess if nonsense-mediated decay resulted in underrepresentation of abnormally spliced transcripts in the DICER1 case, we identified a heterozygous polymorphism in the 3′ untranslated region that was present in a 50/50 ratio in the RNA-seq data. This suggests that both alleles were equally represented in the data but that difficulties with mapping short reads masked the true impact of the splice variant.