Reanalysis of Next-Generation Sequencing Data to Detect Tandem Repeat Expansions in 1,106 Czech Probands With Neurologic Disease

Abstract

Background and Objectives

Tandem repeats (TRs) are DNA regions of tandemly repeated nucleotide motifs. Their pathogenic expansions cause various, mainly neurologic, diseases.

Methods

We analyzed 65 TR loci using ExpansionHunter in individuals who underwent short-read whole-exome sequencing (WES) or whole-genome sequencing (WGS) for the diagnosis of a rare neurologic condition.

Results

Of 1,106 proband samples (1,053 WES, 53 WGS), we detected 232 TR expansions in the intermediate or pathogenic range in 18.7% (207/1,106). However, 51 TR expansions were revised as false positives (FPs) and 83 as nondisease-causing. Of the 98 disease-causing TR expansions, 5 were classified as causal hemizygous or heterozygous TR expansions associated with X-linked recessive (XLR) or autosomal dominant (AD) neurologic disorders in 5 probands (0.5%). The low incidence is due to the fact that individuals with typical clinical symptoms (spinocerebellar ataxia) were tested for TR expansion by conventional laboratory methods. Only 1 proband with clinical suspicion of spinal and bulbar muscular atrophy was fully explained by TR expansion in the AR gene, and in 4 others, we hypothesize the possible involvement of 2 different neurologic diseases. Another 82 causal hemizygous or heterozygous TR expansions associated with XLR or AD non-neurologic diseases (secondary findings) were identified in 81 probands (7.3%), of which 70 expansions in TCF4 were associated with Fuchs endothelial corneal dystrophy, a common eye disease in older patients. Finally, we detected 11 heterozygous TR expansions for XLR and autosomal recessive (AR) diseases in 11 probands who had no clinical symptoms of the associated TR disease.

Discussion

The unexpectedly high detection rate (18.7%) of TR expansions necessitates the filtration of FPs and nondisease-causing expansions, thereby underscoring the necessity of visual inspection of ExpansionHunter results. The study demonstrated that both WES and WGS diagnostics can benefit from TR expansion analysis. The secondary findings indicate that the previously published pathogenic ranges of TR expansions in RUNX2 and ZIC3 warrant further investigation.

Introduction

Tandem repeat (TR) is a DNA sequence where a specific nucleotide motif is consecutively repeated. TRs are widely distributed in the human genome, and they are prone to variation in their length due to DNA polymerase slippage.^1,2 TR expansions can become pathogenic and cause various diseases. To date, 65 known pathogenic TR loci in 62 genes were published to be disease-causing (eTable 1). Most of the TR diseases are neurologic, including various forms of spinocerebellar and other ataxias, developmental and epileptic encephalopathies, myotonic and muscular dystrophies, myotonic epilepsies, Huntington disease (HD), and others. The pathogenicity of TR expansions involves many different mechanisms such as RNA toxicity, alternative splicing, repeat-associated non-AUG transcription, and others as reviewed elsewhere.^1,3

TR expansions cannot be effectively detected from next-generation sequencing (NGS) data using variant calling tools for single-nucleotide variants, short insertions, and deletions. Therefore, the use of specific tools is required. ExpansionHunter is a bioinformatic tool for detecting of TR expansions from NGS data such as whole-exome sequencing (WES) or whole-genome sequencing (WGS). It uses sequence graphs to genotype even complex TR regions with multiple or imperfect repeats.⁴

Neurologic diseases have a notable adverse influence on the quality of life of affected individuals and their families. The discovery of the causal aberration is important not only for the correct diagnosis and compilation of a treatment protocol but also for the prevention of the disease recurrence in the family. Because most TR diseases are neurologic, the detection of abnormal TR expansions represents an essential part of the genetic diagnostic process in individuals with neurologic conditions.

In our study, we used ExpansionHunter⁴ to analyze the length of 65 known disease-causing TR loci in WES (n = 1,053) and WGS (n = 53) data of 1,106 Czech probands investigated in our laboratory for the clinical diagnosis of a rare neurologic disease. The conditions included epilepsy (EPI), peripheral neuropathy (PN), hereditary spastic paraparesis (HSP), hearing loss (HL), malformations of cortical development (MCD), and other rare neurologic conditions (OTHs). In this study, we aimed to expand the scope of routine genetic diagnostics by detecting the TR expansions, which were not identified by standard NGS analysis, and to determine the proportion of pathogenic TR expansions in our cohort.

Methods

Proband and Healthy Relative's Inclusion Criteria

We included 1,106 Czech probands for whom WES or WGS data were available in our laboratory. This included samples analyzed in the period of 2015–2023. We included probands with clinical diagnoses of EPI, PN, HSP, HL, malformations of cortical development, and OTHs. Only 1 proband per family was included in the study.

To analyze the presence of TR expansions also in healthy individuals, we further performed TR genotyping in a cohort of available healthy relatives of our probands regardless of relationship.

Standard Protocol Approvals, Registrations, and Proband Consents

The study was approved by the Institutional Ethics Committee of University Hospital Motol under the numbers EK-1240.11/21 and EK-602.3/22. All probands or their legal guardians provided the informed consent for the genetic examination, which included NGS, and underwent genetic counseling with a clinical geneticist.

WES and WGS and Data Processing

Sample whole-blood DNA was extracted using standardized protocols. Short-read WES libraries were prepared with Agilent SureSelect v.4 (n = 13, 2015), v.5 (n = 33, 2015–2017), v.6-Post (n = 572, 2017–2022), and v.8-Post (n = 332, 2022–2023) kits (Agilent, CA) and with the Twist Human Comprehensive Exome Kit (n = 103, 2023) (Twist Bioscience, CA). Short-read WGS libraries were prepared with a TruSeq PCR free library kit (Illumina, CA). Samples were sequenced as 150-bp paired-end in an outsourced laboratory (Macrogen, Amsterdam, Netherlands) or in the Neurogenetic Laboratory (Charles University and Motol University Hospital, Prague, Czechia) using the Illumina HiSeq or NovaSeq platform (Illumina, CA). The aimed coverage was 150x for WES and 30x for WGS.

WES and WGS fastq files were processed using fastp v0.20.1⁵ to trim out adapters and low-quality bases. Then, the data were aligned to the human reference genome GRCh38 using DRAGMAP v.1.2.1.⁶ PCR and optical duplicates were removed with picard MarkDuplicates.⁷

TR Detection, Visualization, and Detection of False Positives

The length of 65 published disease-causing TR loci was genotyped by ExpansionHunter v5.0.0⁴ using a custom-made catalog (eTable 1). Most of them are located in exons or untranslated regions (exons = 32, unstranslated region = 14; in total 46/65) and thus are sufficiently covered in both WES and WGS (eFigures 1 and 2). The intermediate and pathogenic TR expansion ranges were sourced from the literature. TR alleles with the ExpansionHunter “LowDepth” filter were excluded from the analysis.

TR length visualization was performed using REViewer v0.2.7⁸ or STRipy v.2.2.⁹ All intermediate and pathogenic TR expansions were checked visually. The TR expansions with multiple mismatches and insertions/deletions were marked as false positives (FPs) and reanalyzed by STRipy.

TR Length Verification by Conventional Laboratory Methods

Selected TR expansions up to 150 bp were verified by Sanger sequencing (AR, HOXD13, PHOX2B, RUNX2, ZIC3). Longer TR expansions were verified by PCR and fragment analysis (FXN) and repeat-primed PCR (DMPK, CNBP) by an accredited diagnostic laboratory. HTT expansions were not verified because the diagnosis of HD is performed under a special regime regarding the guidelines by the Czech Society of Medical Genetics and Genomics.

Data Availability

Anonymized data will be shared on reasonable request.

Results

Characterization of the Proband Cohort

We used ExpansionHunter to genotype 65 known disease-causing TR loci in 1,106 Czech probands (WES = 1,053, WGS = 53) with a rare neurologic condition. The number of individuals within neurologic condition groups is as follows: EPI = 263, PN = 262, HSP = 194, HL = 126, MCD = 216, and OTHs = 45. The respective diagnoses were made by the referring neurologists and medical geneticists. The proband cohort consisted of 562 male patients and 544 female patients ranging from 0 to 70 years of age at referral (eFigure 3).

Characterization of All Detected TR Expansions in the Intermediate and Pathogenic Range

In total, we identified 232 TR expansions in the intermediate (n = 83) and pathogenic (n = 149) length range in 207 probands (207/1,106, 18.7%), with 196 being detected in WES and the remaining 36 in WGS. The intermediate and pathogenic expansions were found in 22 TR loci located in 20 genes. The genes most frequently affected by TR expansions were TCF4 (74 pathogenic), HTT (55 intermediate, 5 pathogenic), ATXN1 (19 intermediate, 3 pathogenic), and RUNX2 (18 pathogenic). Other TR expansions were detected in fewer than 10 probands (Figure 1A).

Figure 1. Overview of Detected TR Expansions With and Without False Positives.

(A) Overview of all 232 detected intermediate (yellow) and pathogenic (red) TR expansions, which were discovered in WES and WGS data of 1,106 probands. (B) 51 false-positive TR expansions genotyped using ExpansionHunter in WES (light blue) and WGS (dark blue). (C) One hundred eighty-one intermediate (yellow) and pathogenic (red) TR expansions without FP results divided into 2 groups—disease-causing and nondisease-causing. FP = false positive; TR = tandem repeat.

Detection of False-Positive Results

We visually inspected the read alignments of all 232 identified intermediate and pathogenic TR expansions, which revealed that 51 (51/232, 22.0%) of them were FPs. All of them fell within the pathogenic length range (Figure 1B). The length of all ExpansionHunter false-positive TR expansions is shown in eFigure 4. One of the FP results, CACNA1A TR genotype of 24/11 repeats, was also excluded by Sanger sequencing with the result of 12/13 repeats.

The main FP “hotspots” were exonic TR loci with a GCN repeat motif in ARX, HOXA13, and RUNX2. These constituted 86.3% (44/51) of all FP results. All FP RUNX2 results came only from WGS (Figure 1B), although this exonic region is well covered in WES (eFigure 1). Furthermore, 10 of these FP expansions were identified simultaneously in 4 proband samples in various combinations of TR loci (WGS: “ARX_2”, “HOXA13_2”; WGS: “HOXA13_1”, “HOXA13_2”, “RUNX2”; WES: “HOXA13_1”, “HOXA13_2”, “HOXA13_3”; WES: “HOXA13_2”, “HOXA13_3”).

The FP discovery rate for genotyped intermediate and pathogenic TR expansions was 14.3% (28/196) and 64% (23/36) for WES and WGS data, respectively. A comparison of the mean coverage of TR loci revealed that most of the FP results were located in the lower third quartile. However, some FPs had very high coverage (eFigure 5).

All FP pathogenic results were reanalyzed by STRipy,⁹ which applies additional automatic filters on ExpansionHunter TR genotypes. Only 4 WES and 2 WGS FP results (6/51, 11.7%) were flagged in the TR loci “ARX_2” (n = 1), “CACNA1A” (n = 1), “HOXA13_1” (n = 1), “HOXA13_3” (n = 2), and “PRDM12” (n = 1).

Characterization of Disease-Causing TR Expansions

After the filtration of FP results, the final set of 181 TR expansions was obtained in 169 probands (169/1,106, 15.3%). Of the 181 TR expansions, 98 were classified as disease-causing—comprising 2 in the intermediate range and 96 in the pathogenic range (Figure 1C). Figure 2 presents the length of all ExpansionHunter-genotyped alleles of TR loci in reported genes in our cohort of 1,106 probands.

Figure 2. Length of All Genotyped TR Alleles in TR Loci With Any Pathogenic Expansion.

Length of all genotyped TR alleles in 1,106 proband samples without FP results. Only TR loci with an intermediate or pathogenic TR expansion classified as disease-causing are depicted. Respective normal (light blue), intermediate (light yellow), and pathogenic (light red) ranges are highlighted for every TR locus. The TR alleles (dots) are nonpathogenic (gray) or expanded into a pathogenic range—pathogenic for a neurologic condition (dark red), pathogenic for a non-neurologic condition (orange), and recessive unaffected carriers (purple). FP = false positive, TR = tandem repeat.

The ExpansionHunter genotype for each TR allele includes an assigned genotype and a predicted length range, which can vary significantly from the genotype in the case of long alleles.⁴ Two TR expansions in CNBP and FXN had their assigned genotype length in the intermediate or very near pathogenic range—43 repeats for CNBP (pathogenic ≥55) and 62 repeats for FXN (pathogenic ≥65). However, their predicted length range extended into pathogenic range for a given TR locus. Subsequent verification by conventional laboratory methods (Methods) confirmed that these expansions were indeed disease-causing, as their lengths were found to be within the pathogenic range (Table).

Table.

Disease-Causing TR Expansions

Proband	Gene symbol	Associated TR disease; inheritance	Pathogenic range start	Publication	NGS type	Sex	Age at symptom onset	Age at NGS examination	Clinical neurologic diagnosis group	Genotype range	Predicted genotype range	Carrier	Allele 1	Allele 1 range	Allele 2	Allele 2 range	Verified	Allele 1 - HGVS variant identifier
TR expansions for neurologic disease
P01	AR (HGNC:644)	Spinal and bulbar muscular atrophy (OMIM 313200); XLR	38	Spada et al. 199110	WES	M	47	67	HSP	Pathogenic	Pathogenic		45	45–46	—	—	45	NM_000044.6:c.172CAG[45]
P02	CNBP (HGNC:13164)	Myotonic dystrophy type 2 (OMIM 602668); AD	55	Liquori et al.11 2001	WES	F	20	55	HSP	Intermediate	Pathogenic		43	38–81	16	16–16	Verified pathogenic exp	NM_003418.5:c.-14-885CCTG[43]
P03	DMPK (HGNC:2933)	Myotonic dystrophy type 1 (OMIM 160900); AD	50	Brook et al.12 1992	WES	F	5	7	PN	Pathogenic	Pathogenic		65	56–105	27	27–27	72/27	NM_004409.5:c.*224CTG[65]
P04	DMPK (HGNC:2933)	Myotonic dystrophy type 1 (OMIM 160900); AD	50	Brook et al.12 1992	WES	F	10	15	EPI	Pathogenic	Pathogenic		53	50–75	49	33–63	54/5	NM_004409.5:c.*224CTG[53]
P05	HTT (HGNC:4851)	Huntington disease (OMIM 143100); AD	36	Gatto et al.13 2020	WES	F	2	21	EPI	Pathogenic	Pathogenic		41	41–41	17	17–17		NM_001388492.1:c.52CAG[41]
Secondary findings
P06	HOXD13 (HGNC:5136)	Synpolydactyly (OMIM 186000); AD	22	Albrecht et al.14 2004	WES	F	10	40	PN	Pathogenic	Pathogenic		22	22–22	15	15–15	22/15	NM_000523.4:c.169GCN[22]
P07	HTT (HGNC:4851)	Huntington disease (OMIM 143100); AD	36	Gatto et al.13 2020	WES	F	39	57	HSP	Pathogenic	Pathogenic		37	37–37	17	17–17		NM_001388492.1:c.52CAG[37]
P08	HTT (HGNC:4851)	Huntington disease (OMIM 143100); AD	36	Gatto et al.13 2020	WES	M	—	69	HSP	Pathogenic	Pathogenic		37	37–37	17	17–17		NM_001388492.1:c.52CAG[37]
P09	HTT (HGNC:4851)	Huntington disease (OMIM 143100); AD	36	Gatto et al.13 2020	WES	F	4	17	MCD	Pathogenic	Pathogenic		37	37–37	16	16–16		NM_001388492.1:c.52CAG[37]
P10	HTT (HGNC:4851)	Huntington disease (OMIM 143100); AD	36	Gatto et al.13 2020	WES	M	59	64	PN	Pathogenic	Pathogenic		36	36–36	17	17–17		NM_001388492.1:c.52CAG[36]
P11	NIPA1 (HGNC:17043)	Amyotrophic lateral sclerosis (OMIM 105400); AD	11	Tazelaar et al.15 2019	WES	M	3	15	EPI	Pathogenic	Pathogenic		12	12–12	8	8–8		NM_144599.5:c.22GCG[12]
P12	PHOX2B (HGNC:9143)	Central hypoventilation syndrome (OMIM 209880); AD	24	Sivan et al.16 2019	WES	M	1	10	EPI	Pathogenic	Pathogenic		24	24–24	20	20–20	24/20	NM_003924.4:c.721GCN[24]
P13	RUNX2 (HGNC:10472)	Cleidocranial dysplasia (OMIM 119600); AD	20	Shibata et al.17 2016	WES	M	2	29	HL	Pathogenic	Pathogenic		23	23–23	17	17–17	23/17	NM_001024630.4:c.217GCN[23]
P14	RUNX2 (HGNC:10472)	Cleidocranial dysplasia (OMIM 119600); AD	20	Shibata et al.17 2016	WES	F	0	8	EPI	Pathogenic	Pathogenic		21	21–21	17	17–17	21/17	NM_001024630.4:c.217GCN[21]
P15	RUNX2 (HGNC:10472)	Cleidocranial dysplasia (OMIM 119600); AD	20	Shibata et al.17 2016	WES	M	20	52	MCD	Pathogenic	Pathogenic		20	20–20	11	11–11		NM_001024630.4:c.217GCN[20]
P16	ZIC3 (HGNC:12874)	X-linked VACTERL (OMIM 314390); XLR	12	Wessels et al.18 2010	WGS	M	2	36	HL	Pathogenic	Pathogenic		12	12–12	—	—	12	NM_003413.4:c.136GCN[12]
P17	ZIC3 (HGNC:12874)	OAVS (OMIM 164210); XLR	11	Trimouille et al.19 2020	WES	M	0	17	EPI	Pathogenic	Pathogenic		11	11–11	—	—		NM_003413.4:c.136GCN[11]
Unaffected carriers of disease-causing TR expansions
P18	DMD (HGNC:2928)	Duchenne muscular dystrophy (OMIM 310200); XLR	59	Kekou et al.20 2016	WGS	F	0	2	EPI	Pathogenic	Pathogenic	Yes	64	56–90	15	15–15		NM_004006.3:c.9225-23589GAA[64]
P19	FXN (HGNC:3951)	Friedreich ataxia (OMIM 229300); AR	65	Epplen et al.21 1997	WGS	F	0	30	HL	Intermediate	Pathogenic	Yes	62	55–81	8	8–8	>65/7	NM_000144.5:c.165 + 1340GAA[62]
P20	PABPN1 (HGNC:8565)	Oculopharyngeal muscular dystrophy (OMIM 164300); AR (7 repeats)/AD (≥8 repeats)	7	Brais et al.22 1998	WGS	M	2	14	HL	Pathogenic	Pathogenic	Yes	7	7–7	6	6–6		NM_004643.4:c.4GCG[7]
P21	PABPN1 (HGNC:8565)	Oculopharyngeal muscular dystrophy (OMIM 164300); AR (7 repeats)/AD (≥8 repeats)	7	Brais et al.22 1998	WES	M	9	13	HL	Pathogenic	Pathogenic	Yes	7	7–7	6	6–6		NM_004643.4:c.4GCG[7]
P22	PABPN1 (HGNC:8565)	Oculopharyngeal muscular dystrophy (OMIM 164300); AR (7 repeats)/AD (≥8 repeats)	7	Brais et al.22 1998	WES	F	8	10	MCD	Pathogenic	Pathogenic	Yes	7	7–7	6	6–6		NM_004643.4:c.4GCG[7]
P23	PABPN1 (HGNC:8565)	Oculopharyngeal muscular dystrophy (OMIM 164300); AR (7 repeats)/AD (≥8 repeats)	7	Brais et al.22 1998	WES	F	0	7	MCD	Pathogenic	Pathogenic	Yes	7	7–7	6	6–6		NM_004643.4:c.4GCG[7]
P24	PABPN1 (HGNC:8565)	Oculopharyngeal muscular dystrophy (OMIM 164300); AR (7 repeats)/AD (≥8 repeats)	7	Brais et al.22 1998	WES	F	2	19	HL	Pathogenic	Pathogenic	Yes	7	7–7	6	6–6		NM_004643.4:c.4GCG[7]
P25	VWA1 (HGNC:30910)	Hereditary motor neuropathy (OMIM 619216); AR	3	Pagnamenta et al.23 2021	WES	F	7	9	MCD	Pathogenic	Pathogenic	Yes	3	3–3	2	2–2		NM_022834.5:c.51GGCGCGGAGC[3]
P26	ZIC3 (HGNC:12874)	X-linked VACTERL (OMIM 314390); XLR	12	Wessels et al.18 2010	WES	F	0	3	MCD	Pathogenic	Pathogenic	Yes	13	13–13	10	10–10		NM_003413.4:c.136GCN[13]
P14	ZIC3 (HGNC:12874)	X-linked VACTERL (OMIM 314390); XLR	12	Wessels et al.18 2010	WES	F	0	8	EPI	Pathogenic	Pathogenic	Yes	12	12–12	10	10–10		NM_003413.4:c.136GCN[12]
P27	ZIC3 (HGNC:12874)	OAVS (OMIM 164210); XLR	11	Trimouille et al.19 2020	WES	F	0	8	MCD	Pathogenic	Pathogenic	Yes	11	11–11	10	10		NM_003413.4:c.136GCN[11]

Abbreviations: AD = autosomal dominant; AR = autosomal recessive; EPI = epilepsy; F = female; FECD = Fuchs endothelial corneal dystrophy; HGNC = HUGO (Human Genome Organization) Gene Nomenclature Committee; HGVS = Human Genome Variation Society; HL = hearing loss; HSP = hereditary spastic paraparesis; M = male; MCD = malformations of cortical development; NGS = next-generation sequencing; OMIM = Online Mendelian Inheritance in Man; PN = peripheral neuropathy; TR = tandem repeat; WES = whole-exome sequencing; WGS = whole-genome sequencing; XLR = X-linked recessive inheritance.

All probands with disease-causing TR expansions divided to 3 groups—pathogenic TR expansions for neurologic disease, pathogenic TR expansions for non-neurologic disease (secondary findings), and unaffected carriers of TR expansion. Seventy pathogenic TCF4 TR expansions are not included, because we did not inspect the presence of FECD in our probands because of a very late onset and non-neurologic background of the disease.

A total of 98 disease-causing TR expansions were identified in 94 probands (94/1,106, 8.5%) (83 WES, 11 WGS). Four individuals were found to harbor 2 disease-causing TR expansions in 2 different TR loci simultaneously. Of these, 3 cases had pathogenic TR expansion in TCF4 with another pathogenic TR expansion in AR (proband P01), PABPN1 (P21), and VWA1 (P25). The remaining case had pathogenic TR expansions in RUNX2 and ZIC3 (P14). The disease-causing TR expansions were classified into 3 categories: (1) causal hemizygous or heterozygous TR expansions associated with X-linked recessive (XLR) or autosomal dominant (AD) neurologic diseases, (2) causal hemizygous or heterozygous TR expansions associated with XLR or AD non-neurologic diseases (secondary findings), and (3) heterozygous TR expansions for XLR and autosomal recessive (AR) diseases (unaffected carriers) (Figure 3).

Figure 3. Summary of TR Expansion Analysis in 1,106 Czech Probands With a Rare Neurologic Condition.

Overview of the whole process of finding 5 causal disease-causing TR expansions for neurologic conditions, 11 carriers of recessive disease-causing TR expansions, and 82 secondary findings. TR = tandem repeat.

Hemizygous/Heterozygous TR Expansions Associated With XLR/AD Neurologic Diseases

Five causal pathogenic TR expansions for a neurologic condition were found in 5 probands (5/1,106, 0.45%) in AR (n = 1), CNBP (n = 1), DMPK (n = 2), and HTT (n = 1) (Table).

In proband P01, the hemizygous pathogenic AR expansion was the genetic cause and explained the clinical symptoms. CAG expansions exceeding 40 repeats in AR cause X-linked spinal and bulbar muscular atrophy (SBMA; XLR).¹⁰ The proband had a clinical suspicion of SBMA because of peripheral paresis, occasional fasciculation on upper and lower limb muscles, and gynecomastia. However, no issues with swallowing were present.

In the remaining 4 probands, we assume the possibility of co-occurrence of 2 distinct neurologic conditions because the respective TR diseases do not explain all the neurologic symptoms that were recorded. The second condition may be either acquired or genetic. We identified a heterozygous pathogenic expansion in CNBP, which causes myotonic dystrophy type 2 (DM2; AD)¹¹ in proband P02. A review of the clinical documentation revealed that the proband had been indeed under observation for DM2 since 20 years of age. However, at the age of 50, she developed lower limb spasticity, which is not a typical clinical manifestation of DM2. Therefore, the HSP genetic diagnostics were requested; however, the results were negative.

The proband P03 harbored heterozygous pathogenic expansion of 72 CTG repeats in DMPK, which is the known cause of myotonic dystrophy type 1 (DM1; AD).¹² Individuals with a repeat length of 50–100 repeats usually develop only mild symptoms with a late onset of the disease or remain asymptomatic.²⁴ This 7-year-old proband with demyelinating PN confirmed by EMG had the first indications of medial foot twisting observed at the age of 5. Genetic counseling revealed that the proband's uncle and his children were diagnosed with DM1. The clinical symptoms of PN in the proband P03 did not correlate with those of DM1, which may manifest later in life.

The 15-year-old proband P04 had the heterozygous pathogenic DMPK expansion with the verified length of 54 repeats. The proband was referred for drug-resistant focal EPI presumably caused by MRI-negative focal cortical dysplasia. Owing to the length of the expansion at the lower end of the pathogenic range, we expect the proband to remain asymptomatic or develop only mild symptoms of DM1 with a very late onset.

The proband P05 was found out to have heterozygous 41 CAG repeat long pathogenic expansion in the HTT gene (eFigure 6). Expansion alleles of 40 or more repeats cause HD (AD) with a full penetrance and clinical expression.¹³ The 21-year-old proband P05 has severe drug-resistant EPI since the age of 2 and mild intellectual disability. We hypothesize that the expansion was inherited maternally, given the proband's maternal grandmother's diagnosis of HD. Seizures and developmental delay are common symptoms of juvenile-onset HD, which is caused by CAG expansions exceeding 50 repeats.²⁵ However, their incidence in adult-onset HD is infrequent.²⁶

Secondary Findings

In our proband cohort, we identified 82 causal hemizygous/heterozygous pathogenic TR expansions which are associated with XLR/AD non-neurologic conditions. Additionally, the HTT alleles that fell within a reduced penetrance range are also addressed in this section. These 82 pathogenic TR expansions were found in 81 probands (81/1,106, 7.3%) in HOXD13 (n = 1), HTT (n = 4), NIPA1 (n = 1), PHOX2B (n = 1), RUNX2 (n = 3), TCF4 (n = 70), and ZIC3 (n = 2).

The heterozygous expansion was found in proband P06 with PN resulting in 22/15 GCN repeats in the first exon of HOXD13. Twenty-two and more GCN expansions in HOXD13 are known to cause synpolydactyly (AD).¹⁴ Indeed, the proband underwent surgical removal of supernumerary digits on the feet at the age of 8 years.

We identified 4 probands with the heterozygous CAG pathogenic expansion within 36–39-repeat range in HTT. This range is reported to consist of reduced penetrance alleles with a very late onset or a complete absence of HD manifestation.^13,27 Probands P07 and P08, both with the clinical signs of HSP, had 37/17 repeats. Proband P09 with MCD harbored 37/16 repeats, and proband P10 with PN had 36/17 repeats. The loss of CAA interruption at the end of expanded CAG repeat tract is associated with significantly earlier onset of HD, especially for reduced penetrance alleles.²⁸ However, all 4 probands had the CAA interruption. The clinical phenotype of described probands (HSP, MCD, and PN), therefore, cannot be directly linked to the found pathogenic HTT expansions with a reduced penetrance.

Proband P11 with EPI carried a heterozygous +4 GCN expansion in the first exon of NIPA1. The expanded allele composed of GCA(1)GCT(1)GCG(1)GCA(1)GCG(3)GCA(1)GCG(8)GCC (1). The heterozygous expansion of polyalanines in NIPA1 above 12 repeats was associated with an increased susceptibility to amyotrophic lateral sclerosis, but it also occurred in a healthy population.¹⁵

The heterozygous pathogenic expansion of polyalanine tract above 23 repeats in PHOX2B is known to cause congenital central hypoventilation syndrome ([CCHS]; AD).¹⁶ The 24/20 GCN repeats in PHOX2B were found in proband P12 with drug-resistant EPI after encephalitis of unknown origin at the age of 1 year. The seizures occurred only during sleep, but no hyposaturation was present. The monoallelic 24 GCN expansion was verified by Sanger sequencing in the proband and his healthy father. Neither the proband nor his father had any clinical symptoms of CCHS. This is in accordance with previous findings that the asymptomatic phenotype is often observed in individuals with 24/20 genotype,¹⁶ but the symptoms could be triggered under environmental stress in some individuals.²⁹

Pathogenic variants in the RUNX2 are a known cause of cleidocranial dysplasia (CCD; AD).³⁰ CCD is a congenital condition characterized by hypoplastic or aplastic clavicles, midface hypoplasia, absent or delayed closure of cranial sutures, and dental anomalies. The heterozygous expansion of the polyalanine tract to 20 and more repeats in the third exon of RUNX2 was reported to cause CCD.^17,30 Three probands in our cohort were discovered to have heterozygous pathogenic GCN expansion in RUNX2. Proband P13 with HL had 23/17, proband P14 with EPI had 21/17, and proband P15 with MCD had 20/11 GCN repeats. The heterozygous GCN expansions were confirmed through Sanger sequencing in proband P13 (23 repeats) and proband P14 and her healthy mother (21 repeats). Neither of the probands nor the healthy mother had reported hypoplastic or aplastic clavicles, midface hypoplasia, or other CCD-related phenotypes by their referring geneticist or medical reports.

The most prevalent pathogenic TR expansions found in our proband cohort were located in TCF4 (n = 70). The heterozygous CTG expansions above 50 repeats in TCF4 possess >76-fold risk of Fuchs endothelial corneal dystrophy (FECD; AD). FECD is a late-onset degenerative disease of the posterior cornea with a high prevalence in the White population.³¹ Because of a very late onset and non-neurologic background of the disease, we did not inspect the presence of FECD in our probands.

Loss-of-function pathogenic variants in ZIC3 are known to cause XLR heterotaxy.³² A de novo hemizygous addition of 2 alanines by GCC expansion in the first exon of ZIC3 was reported to be a possible contributor of overlapping XLR VACTERL association/heterotaxy features in 1 male deceased newborn.¹⁸ We identified 1 male adult proband P16 with HL to harbor hemizygous 12 GCC repeats in ZIC3. In addition, we analyzed TR expansions in available 77 healthy relative's WES (n = 66) and WGS (n = 11) data and found hemizygous 12 GCC repeats in ZIC3 in 1 male adult healthy relative who is unrelated to proband P16. The expansion was verified by Sanger sequencing in both the proband and the healthy relative. None of them had any clinical report describing heterotaxy or congenital defects themselves or present in their family. We also looked at the family background of female carriers P26 (13 repeats) and P14 (12 repeats) and did not find any reports of presence of heterotaxy or congenital defects in their families.

Recently, it was reported that +1Ala expansion (11 GCC repeats) in ZIC3 could be responsible for X-linked oculo-auriculo-vertebral spectrum ([OAVS]; XLR) defects in 1 family. The affected male individuals had variable manifestations of microtia, with or without hearing impairment.¹⁹ We found 11 GCC repeats in ZIC3 as a hemizygous expansion in 1 male proband P17 with EPI and as a heterozygous expansion in 1 female carrier P27 with MCD. Both had no signs of OAVS defects or hearing impairment themselves or in the family background.

TR Expansions Observed in Unaffected Carriers

We identified 11 probands (11/1,106, 0.9%) to be heterozygous for pathogenic TR expansion for a recessive TR disease in the genes DMD (n = 1), FXN (n = 1), PABPN1 (n = 5), VWA1 (n = 1), and ZIC3 (n = 3). The length of revealed pathogenic TR expansions in carriers and TR diseases associated with these genes, altogether with proband-specific information, are listed in the Table.

None of the described carriers had clinical symptoms of the disease associated with their respective TR expansion. This was confirmed by the referring clinical geneticist and available medical records. None of the carriers harbored another pathogenic variant in the coding region of the gene with the pathogenic TR expansion.

Characterization of Nondisease-Causing TR Expansions

Eighty-three TR expansions in 79 probands (79/1,106, 7.1%) were classified as nondisease-causing. Eighty-one intermediate TR expansions were in ATXN1 (n = 19), ATXN2 (n = 1), DMPK (n = 6), and HTT (n = 55) (Figure 1C). The length of all TR alleles is presented in Figure 2 (DMPK, HTT) and eFigure 7 (ATXN1, ATXN2).

The CAG expansion in ATXN1 above 39 repeats causes spinocerebellar ataxia type 1 (SCA1; AD) unless interrupted by CAT.³³ Two probands harbored heterozygous pathogenic TR expansions in ATXN1 (TR genotypes 40/29 and 39/30). Nevertheless, visual inspection revealed that CAT interruptions were also incorporated into the final TR length. None of the probands manifested clinical signs of SCA1.

Discussion

We used ExpansionHunter to analyze the length of 65 known disease-causing TR loci in 1,106 Czech probands with a wide spectrum of rare neurologic conditions. The analysis proved to be beneficial for both WES and WGS data because most of the previously published TR loci were also encompassed by WES. A total of 232 TR expansions in the intermediate or pathogenic length range were identified in 18.7% of the probands. The unexpectedly high prevalence was largely attributable to 3 factors: (1) 51 FP results, (2) 83 nondisease-causing TR expansions in 7.1% of probands, and (3) 70 TR expansions in TCF4, which is associated with a common disease—FECD—in 6.3% of probands.

We observed a high incidence of FPs among ExpansionHunter results. Overall, 22.2% of intermediate and pathogenic TR expansions were eventually rendered false positive by visual inspection of read alignment. The FP discovery rate was 14.3% for WES and 64% for WGS. A recent study reported a substantially higher FP ratio for WES data (mean coverage ∼20×), at 69.8% (81/116). However, the study was constrained to only 20 disease-causing TR loci.³⁴ The discrepancy is likely attributable to the higher coverage of our data (overall mean coverage 97.2× for WES and 44.7× for WGS), as the precision of TR detection directly correlates with coverage depth.³⁵ In fact, most of the FP results were identified in the lower third quartile of mean coverage for a specific TR locus. However, we also observed some FPs with a very high coverage, indicating that coverage is not the sole influencing factor.

The 86.3% of FP results were exonic GCN repetitions in ARX, HOXA13, and RUNX2. The RUNX2 locus was problematic for ExpansionHunter TR genotyping, particularly in WGS data. In the STRipy publication, ARX and HOXA13 TR loci were also identified as sources of FP results, prompting the development of dedicated analysis settings for these sites in STRipy.⁹ However, the STRipy reanalysis of FP results only filtered out 11.7% of FPs because of low coverage. We emphasize the importance of visual inspection of aligned reads of intermediate and pathogenic TR expansions to filter FPs before verification by conventional laboratory methods.

We did not anticipate a high incidence of causal TR expansions for neurologic diseases in our cohort, given that individuals presenting with typical clinical symptoms (spinocerebellar ataxia) had undergone TR expansion testing by conventional laboratory methods before WES. Most notably, 5 causal hemizygous/heterozygous pathogenic TR expansions for a XLR/AD neurologic condition were identified in 5 probands (0.45%). The phenotype matched the TR expansion best in the proband with a pathogenic AR expansion, who was clinically suspected of having SBMA. In the remaining 4 probands, the identified TR expansions did not fully explain the clinical phenotype, suggesting either atypical phenotypes or the co-occurrence of 2 distinct neurologic conditions. The latter is rare but has been documented.^36,37

After excluding 70 TR expansions in the TCF4 gene, 12 causal hemizygous/heterozygous pathogenic TR expansions for a XLR/AD non-neurologic condition were identified in 12 probands (1.1%). These findings suggest that not all TR expansions are disease-causing within the published pathogenic range. First, the polyalanine expansions in RUNX2 were reported to cause CCD. One study reported a family with 16 affected individuals exhibiting mild CCD features who carried 27 (+10Ala) GCN repeats³⁰ while another article reported a case of 1 proband with 20 GCN repeat expansions (+3Ala) who manifested a severe CCD phenotype. In addition, researchers demonstrated that the +10Ala variant exhibited markedly reduced transcription activity while the +3Ala variant demonstrated normal activity.¹⁷ We identified 3 probands with polyalanine expansions in RUNX2 with the lengths of 23, 21, and 20 repeats. The 21-repeat expansion was also verified in the proband's healthy mother. None of the probands or the healthy mother had reported CCD-related phenotype by the referring geneticist or medical records. This suggests the potential for incomplete penetrance within the low pathogenic range or the possibility that the pathogenic range may be longer than 23 repeats.

Second, hemizygous polyalanine expansions in ZIC3 were published to cause OAVS (+1Ala, 11 GCC)¹⁹ and heterotaxy/VACTERL (+2Ala, 12 GCC).¹⁸ We found 1 male proband and 1 female carrier with +1Ala who did not exhibit any reported OAVS-related phenotype themselves or in their family members. In addition, 1 male proband, 1 unrelated male healthy control, and 1 female carrier were identified as having +2Ala in ZIC3. Furthermore, another female carrier had even +3Ala. No clinical features of heterotaxy or congenital defects were reported in the patients themselves or in their family members. These findings are consistent with the functional study, which indicated that both variants were of uncertain significance, because the mutated ZIC3 exhibited normal cytoplasmic localization and normal reporter gene transactivation.³⁸ The STRipy⁹ and TR-gnomAD³⁹ databases both contain a small number of samples with 11 and 12 GCC repeats, but there is a paucity of information regarding suspect phenotypes or their carrier status.

In addition, we identified 11 individuals (0.9%) with 1 recessive disease-causing TR expansion. None of the carriers exhibited any clinical manifestations of the associated TR disease. However, all the detected recessive disease-causing TR expansions are known to cause serious diseases. Therefore, the detection of carriers is very important for disease prevention, especially in conditions with a high carrier frequency. For instance, we identified 1 carrier of a pathogenic expansion in FXN, which causes Friedreich ataxia (FRDA). FRDA is the most common recessively inherited childhood-onset ataxia,²¹ with an estimated carrier frequency of 1 in 85 in the White population.⁴⁰ The FXN intronic locus is not covered by WES probes of Agilent SureSelect v4-v8 or Twist Comprehensive Exome. Consequently, only a single carrier was detected in our cohort (1/53 WGS). Given that WES remains the prevailing NGS diagnostic method because of its price, the incorporation of the FXN locus into updated WES versions could expand the accessibility of FRDA screening.

Conventional laboratory methods for TR detection, such as Sanger sequencing, repeat-primed PCR, and Southern blot are laborious procedures in comparison with WES.¹ WES encompasses most of the known disease-causing TR loci that can be genotyped in a single analysis using ExpansionHunter. Another benefit of ExpansionHunter is that it facilitates the rapid reanalysis of NGS data after the publication of a novel disease-causing locus. Consequently, WES emerges as an effective first-choice method for TR detection, particularly in cases involving ataxia, given its capacity to simultaneously investigate numerous TR loci.

There are several limitations of ExpansionHunter TR analysis of NGS data. Primarily, the ExpansionHunter is constrained to analyzing only those TR loci that have been previously identified. Second, the read length is insufficient for precise analysis of long expansions, potentially leading to the failure to detect all expansions. Furthermore, the detection of most intronic or intergenic TRs is limited with WES. These limitations can be overcome by leveraging long-read sequencing in the future, when the technology becomes more economical.¹

In summary, we demonstrated that despite the methodology limitations, both WES and WGS diagnostics are enhanced by the incorporation of TR expansion analysis, as has been corroborated by previous studies.^34,35,41 The study also demonstrates the necessity of visual inspection of ExpansionHunter-detected TR expansions to filter FPs and nondisease-causing results. We have proposed the potential to extend WES by new probes for already known intronic TR loci to increase WES diagnostics yield, given that it is currently the predominant genetic diagnostics method. Our secondary findings indicate that TR expansions in the genes RUNX2 and ZIC3 may not always be disease-causing in the previously published length range. These findings are important for the scientific community to further evaluate the pathogenicity of these expansions.

Glossary

AD

autosomal dominant

CCHS

congenital central hypoventilation syndrome

EPI

epilepsy

FECD

Fuchs endothelial corneal dystrophy

FPs

false positives

XLR

X-linked recessive

FRDA

Friedreich ataxia

HD

Huntington disease

HL

hearing loss

HSP

hereditary spastic paraparesis

MCD

malformations of cortical development

NGS

next-generation sequencing

OAVS

oculo-auriculo-vertebral spectrum

OTHs

other rare neurologic conditions

PN

peripheral neuropathy

TR

tandem repeat

WES

whole-exome sequencing

WGS

whole-genome sequencing

Author Contributions

A. Musilova: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; analysis or interpretation of data. P. Lassuthova: major role in the acquisition of data; study concept or design; analysis or interpretation of data. A. Uhrova Meszarosova: major role in the acquisition of data; analysis or interpretation of data. B. Straka: drafting/revision of the manuscript for content, including medical writing for content; study concept or design. J. Krejcikova: major role in the acquisition of data. A. Berounska: major role in the acquisition of data. M. Vlckova: major role in the acquisition of data. Z. Musova: major role in the acquisition of data. D. Safka Brozkova: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data.

Study Funding

This study was supported by the Ministry of Health of the Czech Republic AZV NU22-04-00097 and NU23-04-00209, and the Ministry of Health, Czech Republic - conceptual development of research organization, Motol University Hospital, Prague, Czech Republic 00064203.

Disclosure

The authors report no relevant disclosures. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/NG.