Background/Introduction

With an estimated prevalence of 1:250-400 in the general population dilated cardiomyopathy (DCM) is one of the most frequent causes of heart failure and heart transplantation as ultimate treatment option. The underlying etiology of DCM is diverse, but genetic factors are estimated to play a significant role in 30-40% of all cases. Today, guidelines for variant classification developed by the American College of Medical Genetics (ACMG) have been established to classify the growing amount of novel variants as“pathogenic”(P), “likely pathogenic” (LP), “uncertain significance” (VUS), “likely benign” (LB) and “benign”(B). Advanced computational algorithms enable an automated prediction for most of the 28 different criteria, but the majority of the variants are still classified as VUS. 

Aim

Here, we sought to perform a detailed per variant workup of all ACMG-criteria in a cohort of DCM patients tested by panel sequencing. By including a manual pedigree/co-segregation analysis, we aim to utilize the criteria PP1 and BS4 and hence raise the confidence of a variant (re-)classification.

Methods

From 304 consecutively recruited DCM patients, 94 index patients with additional 103 family members have been subjected to panel-based genetic testing of in total 78 genes using SureSelect XT enrichment (Agilent) and short-read sequencing (Illumina HiSeq, MiniSeq and NovaSeq). Variants were detected using GATK HaplotypeCaller and annotated with the variant effect predictor (VEP; ENSEMBL).

Results

In our cohort of 94 index patients, 67% were male and aged between 20-83 years (mean age =51.78; ±14.55) and had a mean LVEF of 41.94 (±13.55). Further, 33 patients received an ICD and 12 heart transplantation. After filtering for allele frequency (AF) and functional impact of the variants, in the mean 9 variants could be detected per patient in our panel of 78 genes. The most affected genes was TTN, followed by SYNE2 and DSG2. For variants prioritization, we first followed a non-ACMG-guideline based analysis pipeline and compared the variants to the Clinvar database only and find 7 known pathogenic variants in 7 patients. Further, 61 truncating variants in 93 patients and 299 missense variants in 94 patients with a possible pathogenic prediction (CADD >20) could be detected. When applying ACMG-guideline-based annotation, we identified 52 pathogenic variants in 29 patients, 27 LP in 20 patients, 118 VUS in 66, 85 LB in 76 and 189 B in 94 patients. In total, 79 P/LP variants were detected in 42 patients (45%) in 32 genes. Among the detected VUS, 18 variants were found among 23 index patients with family members, enabling us to add information of the ACMG-category PP1 and BS4. By this approach, we could upgrade the prediction for two variants from VUS to LP and downgrade the prediction for six variants from VUS to LB. When extending our analysis to all DCM patients in our cohort (index + family members), we could further reclassify 3 VUS to B and 18 to LB and 8 LB to B. 

Conclusions

By applying ACMG-guidelines on variant interpretation, compared to non-ACMG-based classification of variants, we are able to detect 35 (37.23%) more patients with a pathogenic or likely pathogenic variant. Further, we show the value of an individual workup of the patient familes in case of a VUS, enabling the reclassification of such a variant in 5 (21.74%) more families of the index patients with family information.