A prospective trial comparing programmable targeted long-read sequencing and short-read genome sequencing for genetic diagnosis of cerebellar ataxia

Abstract

The cerebellar ataxias (CAs) are a heterogeneous group of disorders characterized by progressive incoordination. Seventeen repeat expansion (RE) loci have been identified as the primary genetic cause and account for >80% of genetic diagnoses. Despite this, diagnostic testing is limited and inefficient, often utilizing single gene assays. This study evaluates the effectiveness of long- and short-read sequencing as diagnostic tools for CA. We recruited 110 individuals (48 females, 62 males) with a clinical diagnosis of CA. Short-read genome sequencing (SR-GS) was performed to identify pathogenic RE and also non-RE variants in 356 genes associated with CA. Independently, long-read sequencing with adaptive sampling (LR-AS) was performed to identify pathogenic RE. SR-GS provided a genetic diagnosis for 38% of the cohort (40/110) including seven non-RE pathogenic variants. RE causes disease in 33 individuals, with the most common condition being SCA27B (n = 24). In comparison, LR-AS identified pathogenic RE in 29 individuals. RE identification for the two methods was concordant apart from four SCA27B cases not detected by LR-AS due to low read depth. For both technologies manual review of the RE alignment enhances diagnostic outcomes. Orthogonal testing for SCA27B revealed a 15% and 0% false positive rate for SR-GS and LR-AS, respectively. In conclusion, both technologies are powerful screening tools for CA. SR-GS is a mature technology currently used by diagnostic providers, requiring only minor changes in bioinformatic workflows to enable CA diagnostics. LR-AS offers considerable advantages in the context of RE detection and characterization but requires optimization before clinical implementation.

Figure 1.

Overview of the study and investigations performed. (SCA) spinocerebellar ataxia, (EH5) ExpansionHunter5, (RE) repeat expansion, (LP) likely pathogenic, (P) pathogenic, (VUS) variant of uncertain significance.

Figure 2.

Targeted RE screening in SR-GS identifies potential RE diagnoses. STR genotypes were determined for the 22 loci associated with CA (Table 1) with EH5. (A) Genotypes are shown for the longer allele in dominant RE disorders that cause ataxia (top 3 plots, loci separated based on maximum allele size) and those that cause other disorders (bottom). Blue circles indicate individuals with an expansion in the respective allele that is larger than the pathogenic threshold (solid red line), while yellow circles are individuals who exceed an EH5-specific threshold (dashed red line) and are candidates for further investigation. Benign AAAAT motifs do not have a threshold. (B) Genotypes for the shorter and longer alleles are shown for autosomal recessive RE ataxia disorders (top) and X-linked recessive disorders other than ataxia, split by sex (bottom). Individuals who are heterozygous carriers for an allele expanded beyond the pathogenic threshold (solid red line) or an allele larger than the EH5-specific threshold (dashed red line) are shown as brown squares. For FMR1, the blue dashed line indicates the threshold for FXTAS while the solid red line is the threshold for FXS. Individuals who carry two alleles expanded beyond either the pathogenic or EH5-specific threshold are shown as colored circles. (C) Concordance plot showing a comparison of the FGF14 STR EH5 genotypes from SR-GS compared to PCR sizing (longer allele only). The pathogenic (≥250 repeats) and VUS (≥180 repeats) thresholds are shown as dashed red lines. A 1:1 correlation is shown as a dashed blue line. The R² is a Pearson's correlation. Dark green circles are individuals with a confirmed SCA27B diagnosis and those in light green are SCA27B VUS. Circles indicate GAA motifs and squares are non-GAA motifs, determined by observation in IGV.

Figure 3.

Performance metrics for adaptive sequencing in a targeted panel of RE loci. (A) Mean read length in target regions significantly exceeds that of off-target regions. (B) Sequencing depth in on- versus off-target regions. (C) Genome-wide enrichment of on- versus off-target regions in individual experiments. (D) Number of reads aligning to each of the targeted CA repeat loci. (E) Number of reads aligning to each targeted clinically significant non-CA locus.

Figure 4.

Targeted RE screening panel in LR-AS and comparison to SR-GS. STR genotypes were determined for short-listed loci with Straglr. Green points indicate individuals with a confirmed clinical diagnosis for the respective locus. (A) Genotypes are shown for the longer allele in dominant RE disorders that cause ataxia (top 2 plots, loci separated based on maximum allele size) and those that cause other disorders (bottom). (B) Genotypes for the shorter and longer alleles are shown for autosomal recessive RE ataxia disorders (left) and X-linked recessive disorders other than ataxia, split by sex (right). Comparison of allele genotyping with SR-GS and LR-GS is shown for loci with a confirmed diagnosis for the longer allele in dominant disorders. (C) ATXN8OS (SCA8), (D) NOP56 (SCA36), and for both alleles in recessive disorders for (E). FXN (FRDA) and (F). RFC1 (CANVAS, AAGGG motif only). LR-AS with low coverage (≤7 reads) are shown as triangles, those with no coverage issues flagged (>7 reads) are shown as circles. Green circles indicate individuals with an expansion in the respective allele that is larger than the pathogenic threshold (solid or dashed red line). Benign AAAAT motifs do not have a threshold. For FMR1, the blue dashed line indicates the threshold for FXTAS while the solid red line is the threshold for FXS.

Figure 5.

Comparison of FGF14 STR sizing between LR-AS, SR-GS, and PCR identifies the strengths of LR-AS. Comparison of FGF14 STR sizing is shown as concordance plots for (A). LR-AS compared to PCR and (B). LR-AS compared to SR-GS. The R² is a Pearson's correlation. Dark green circles are individuals with a confirmed SCA27B diagnosis (P) and those in light green are SCA27B VUS. Triangles indicate low coverage (≤7 reads) on LR-AS, squares indicate non-GAA repeat motifs, and circles samples with coverage >7 reads and GAA motifs. The pathogenic (≥250 repeats) and VUS (≥180 repeats) thresholds are shown as dashed red lines. A 1:1 correlation is shown as a dashed blue line. The R² is a Pearson's correlation. Individuals with no genotype call from either LR-AS, SR-GS, or PCR are indicated separately to the numbered axis and are labeled nc (i.e., no genotype call).