The Diverse Genetic Landscape of Hearing Impairment in South African Families

ABSTRACT

To elucidate the genetic etiology of hearing impairment (HI) in South Africa, 45 nonsyndromic HI (NSHI) and syndromic HI (SHI) families with ≥ 2 affected members were analyzed. Exome and sanger sequencing were used to identify causal genes. For NSHI, 14 of 24 families segregated variants in NSHI genes, that is, CDH23, GJB2, MITF, MYO7A, MYO15A, PCDH15, POU3F4, REST, SLC26A4, TMPRSS3, and WFS1. For the 21 SHI families, 14 have Waardenburg syndrome, two Branchio‐Oto‐Renal syndromes, and one each with Bartter, Chudley‐McCullough, Deafness‐Albinism, MYH9‐related disorder, and Pendred syndromes. The cause of SHI was determined for 14 families, with EDN3, EDNRB, GPSM2, MITF, MYH9, SLC12A1, and SLC26A4 underlying the syndrome in a single family, EYA1 in two families, and PAX3 in five families. For the NSHI and SHI genes, 52.9% and 35.7% of the variants, respectively, have not been reported in disease etiology. Additionally, two Waardenburg families segregated variants in NSHI genes, BDP1 and MYO6, but these findings need to be validated. This study enhances the understanding of the genetic landscape of HI in South Africa, revealing a high level of locus and allelic heterogeneity. Studying diverse populations provides new insights into HI etiology that, in turn, can improve genetic diagnosis and personalized management.

South African Families with Nonsyndromic (N = 24) and Syndromic Hearing Impairment (N = 21) with ≥ 2 affected members were analyzed. The underlying etiology was uncovered using exome and Sanger sequencing for 31 of these families.

1. Introduction

South Africa has a multi‐ancestry population of 62 million people of which 81.7% are Black African, 8.5% mixed, 7.2% European, and 2.6% Indian/Asian [1]. In South Africa, hearing impairment (HI) is the most common disorder after visual impairment and cognitive disability [2] with a prevalence of 5.5 in 1000 live births [3].

The genetic landscape of South Africa has been structured by geographic isolation, migration, population bottleneck, admixture, and natural selection [4, 5, 6, 7]. Despite the cultural, linguistic, and genetic diversity within South Africa, this region remains underrepresented in population genetic studies [8, 9, 10, 11] as well as genetic studies on HI [12].

Addressing the underrepresentation of African populations in genetic studies is critical for improving molecular diagnostic accuracy and personalized treatment for HI. Expanding genetic research in diverse populations such as South Africa will provide additional insights into the underlying etiology of HI, contributing to better healthcare outcomes and more equitable access to genetic diagnostics.

In this study, we analyzed South African families with nonsyndromic (NS) (N = 24) and syndromic (S) (N = 21) HI, with at least two affected members per family, to identify the underlying genetic etiology. We show that the genetic etiology of HI is heterogeneous in the South African population. This work highlights the complex genetic architecture of HI in South Africa, driven by its rich cultural and ancestral diversity.

2. Methods

2.1. Informed Consent and Recruitment

This study was conducted with full compliance of the declaration of Helsinki. Institutional ethical committees/institutional review board (IRB) approvals were obtained from Columbia University (IRB‐AAAS2343) and the University of Cape Town [Human Research Ethics Committee (HREC): 104/2018]. Prior to recruitment, approval was obtained from the relevant provincial education departments and school principals. Written consent was obtained from all participants over the age of 18, and parents and legal guardians provided written consent for their children. If the child was at least 14 years of age, verbal assent was also obtained.

Patients were primarily recruited from schools for the deaf in seven provinces in South Africa (Table S1, Figure S1). Family history, medical records, and pure tone audiometry were obtained and otological examinations performed. Individuals with HI, due to infections, ototoxic medications, or trauma, were excluded from the study. The families included in this study are of Black African, mixed (“colored”), and White African ancestry (Table S1).

2.2. Sample Collection

Forty‐five families with NSHI (N = 24) and SHI (N = 21) with at least two hearing‐impaired members were ascertained. Peripheral blood samples from 108 affected and 57 unaffected family members were obtained and genomic DNA was extracted (Figures S2–S46).

2.3. Exome Sequencing and Data Analysis

Genomic DNA from 138 study participants underwent exome sequencing (Figures S2–S46). Exome sequence data were generated for two affected members (N = 39 families), three affected members (N = 4 families), and four affected members (N = 2 families). Additionally, DNA samples from 40 unaffected family members underwent exome sequencing (Table S2).

Exome sequencing was performed using the following capture kits: Nextera Rapid Capture Exome kit (37 Mb) (N = 98 DNA samples); Agilent SureSelect Human All Exon V4 + UTR 71 Mb All Exon Capture kit (N = 34 DNA samples); and Agilent SureSelect Human All Exon V6 (N = 6 DNA samples). Table S2 provides information on the kit used for each family. Pair‐end sequencing (100 bp run format) was performed on an Illumina HiSeq 2500 (Illumina, San Diego, CA). Sequence reads were aligned to the human reference genome (hg19/hg38) using the Burrows‐Wheeler Aligner. Duplicate reads were marked with Picard. The Genome Analysis Toolkit (GATK) [13] was used to call single nucleotide variants and small insertions/deletions for each sample, and then joint variant calling and recalibration were performed. Variants were annotated using ANNOVAR [14], dbNSFP35a [15], and dbscSNV1 [16].

Exome sequence data were first examined for likely pathogenic/pathogenic HI variants obtained from ClinVar. Next, analysis via filtering was performed by retaining variants with a population‐specific minor allele frequency (MAF) < 0.005 in the Genome Aggregation database (gnomAD) [17] for pedigrees with an autosomal recessive (AR) or X‐linked (XL) mode of inheritance and MAF < 0.0005 for pedigrees with an autosomal dominant (AD) mode of inheritance. Variants predicted to affect protein function, pre‐mRNA splicing (nonsense, frameshift, missense, start‐loss, splice region, etc.) and synonymous (known NSHI/SHI genes) variants that were shared among affected family members and either heterozygous or absent in unaffected members depending on the mode of inheritance were selected for validation and testing for segregation using Sanger sequencing. Copy number variant calls were generated using CONiFER (v0.2.2) [18] and annotated using the BioMart Database [19].

3. Results

3.1. Family Demographics

The 45 families that were included in the study were ascertained from seven of the nine provinces in South Africa, that is, Eastern Cape (N = 5), Free State (N = 6), Gauteng (N = 5), KwaZulu‐Natal (N = 2), Limpopo (N = 4), Northern Cape (N = 4), and Western Cape (N = 19). The families are of Black (N = 28), mixed (N = 16), or White (N = 1) ancestry (Table S1, Figure S1).

3.2. Clinical Findings

Clinical evaluations determined that 24 families presented with NSHI and 21 families with SHI (Table S1, Figures S2–S46). The majority of SHI families had Waardenburg syndrome (OMIM: 277580) [14/21 (66.7%), 13 families AD and 1 AR]. Two families segregated Branchio‐oto‐renal (BOR) syndrome (OMIM: 113650) [2/21 (9.5%), AD]. Bartter (OMIM: 607364) (AR), Chudley‐McCullough (OMIM: 604213, AR), MYH9‐related disorder (OMIM: 155100, AD), Pendred (OMIM: 274600, AR), and Tietz albinism‐deafness (OMIM: 103500, AD) syndromes were each observed in one (4.8%) family (Table S2).

3.3. Pedigree Analysis

The average number of family members with an available DNA sample is 3.7. The maximum number of affected individuals per family with DNA samples was 6 (n = 1), with most families having two affected members with a DNA sample (n = 33 families) (Table S2).

3.4. Nonsyndromic Hearing Impairment

Among the 24 NSHI families, the modes of inheritance are AR (N = 12), AD (N = 10), XLR (N = 1), and AR or XLR (N = 1) (Figures S2–S25). The families with ARNSHI (Families 4, 6, 60, 63, 316, 327, 343, 350, 396, 406, 407, and 412) all had severe to profound bilateral sensorineural HI affecting all frequencies, which was either reported to be congenital or prelingual. While the AD NSHI families had both congenital (Families 73, 351, 365, 367, 390, 391, 400, 401, and 402) and prelingual (Family 408) onset of severe to profound bilateral sensorineural HI. Family 3 had XLR NSHI, which was congenital severe to profound bilateral sensorineural HI. Family 5, with a mode of inheritance that could be either AR or XLR, had congenital bilateral profound sensorineural HI (Table S3).

The underlying cause of HI was uncovered in 14 [58.3%, 7 AR (Families 60, 63, 316, 343, 396, 406, and 412), 6 ad (Families 73, 351, 367, 390, 391, and 408), 1 XLR (Family 3)] families (Table S4).

The 11 NSHI genes underlying HI are CDH23 (OMIM: 605516), GJB2 (OMIM: 121011), MITF (OMIM: 156845), MYO7A (OMIM: 276903), MYO15A (OMIM: 602666), PCDH15 (OMIM: 605514), POU3F4 (OMIM: 300039), REST (OMIM: 600571), SLC26A4 (OMIM: 605646), TMPRSS3 (OMIM: 605511), and WFS1 (OMIM: 606201). These genes were all observed in only one family, except for WFS1, which was observed in four families (Families 73, 351, 367, and 390) with an AD mode of inheritance. These 11 genes contained 17 rare variants (Tables 1 and S4). Ten of these variants were not previously associated with NSHI. Variants in ARNSHI genes CDH23, GJB2, MYO7A, SLC26A4, and TMPRSS3 were observed in a homozygous state. The known pathogenic GJB2 founder variant p.(Gly12Valfs*2) was identified in Family 406, which is of mixed ancestry. Novel variants were observed in CDH23 [p.(Tyr3195*)] and TMPRSS3 [c.205 + 6 T>A]. For ARNSHI Family 63, three MYO15A variants were observed in the compound heterozygous state, with two affected half‐siblings sharing the c.9303 + 5G>A variant and both having a different second variant, that is, novel pathogenic stop gain variant p.(Gln460*) and p.(Glu2212Lys). For ARNSHI Family 60, variants in PCDH15, a pathogenic stop gain variant p.(Glu1677*) and a synonymous variant [c.3285G>T] were observed in a compound heterozygous state. A novel pathogenic stop gain variant in POU3F4 [p.(Ser261*)] was observed segregating with an XLR mode of inheritance in Family 3. For the ADNSHI families, heterozygous variants in MITF [p.(Asn328Lys)], REST [p.(Cys415Ser)], and WFS1 [p.(Ala684Val), p.(Glu864Lys), p.(Asp797Asn) and p.(Asn714Ser)] were observed (Tables 1 and S4, Figures S2–S15).

TABLE 1.

Variants segregating in 14 families with non‐syndromic HI.

Family ID	Inh	GT	Phenotype	Gene	cDNA change	Protein change	Function	Reported	Classification a
Family 3	XLR	Hemi	NSHI	POU3F4	c.782C>A	p.(Ser261*)	Stopgain		Pathogenic
Family 60	AR	Comp Het	NSHI	PCDH15	c.5029G>T	p.(Glu1677*)	Stopgain		Pathogenic
					c.3285G>T	p.(=)	Synonymous		VUS
Family 63	AR	Comp Het	NSHI	MYO15A	c.1378C>T	p.(Gln460*)	Stopgain		Pathogenic
					c.6634G>A	p.(Glu2212Lys)	Missense	Yes	Likely Pathogenic
					c.9303 + 5G>A	p.?	Splicing	Yes	Likely Pathogenic b
Family 73	AD	Het	NSHI	WFS1	c.2051C>T	p.(Ala684Val)	Missense	Yes	Pathogenic
Family 316	AR	Hom	NSHI	SLC26A4	c.2174_2177dup	p.(Leu727Tyrfs*28)	Frameshift insertion	Yes	Pathogenic
Family 343	AR	Hom	NSHI	CDH23	c.9585 T>G	p.(Tyr3195*)	Stopgain		Pathogenic
Family 351	AD	Het	NSHI	WFS1	c.2590G>A	p.(Glu864Lys)	Missense	Yes	Likely Pathogenic
Family 367	AD	Het	NSHI	WFS1	c.2389G>A	p.(Asp797Asn)	Missense	Yes	Likely Pathogenic
Family 390	AD	Het	NSHI	WFS1	c.2141A>G	p.(Asn714Ser)	Missense		Likely Pathogenic
Family 391	AD	Het	NSHI	MITF	c.984 T>A	p.(Asn328Lys)	Missense		Likely Pathogenic
Family 396	AR	Hom	NSHI	TMPRSS3	c.205 + 6 T>A	p.?	Splicing		VUS
Family 406	AR	Hom	NSHI	GJB2	c.35del	p.(Gly12Valfs*2)	Frameshift deletion	Yes	Pathogenic
Family 408	AD	Het	NSHI	REST	c.1244G>C	p.(Cys415Ser)	Missense		Likely Pathogenic
Family 412	AR	Hom	NSHI	MYO7A	c.6377del	p.(Pro2126Leufs*5)	Frameshift deletion	Yes	Pathogenic

Abbreviations: AD, Autosomal Dominant; AR, Autosomal Recessive; Comp Het, Compound Heterozygous; GT, Genotype; Hemi, Hemizygous; Het, heterozygous; Hom, homozygous; Inh, Inheritance; VUS, variant of unknown significance; XLR, X linked Recessive.

^a Classification based on American College of Medical Genetics and Genomics guidelines.

^b Reclassified from VUS to likely pathogenic.

The variants observed include missense (N = 7), insertion/deletion (N = 3), stop gain (N = 4), splice site (N = 2), and synonymous (N = 1). Eight of the observed 17 variants (47%) were not previously reported to be associated with NSHI and were not observed in gnomAD, All of Us Bravo [20], TOPMed [21] or ClinVar [22] databases. Each variant was only observed in one South African NSHI family. Of these variants, 15 were classified as either pathogenic or likely pathogenic, while the TMPRSS3 splice site [c.205 + 6 T>A] and the synonymous PCDH15 [c.3285G>T] variants were classified as a VUS (variant of unknown significance) according to the American College of Medical Genetics and Genomics (ACMG) guidelines [23] (Tables 1 and S4 and S5).

3.5. Syndromic Hearing Impairment

Understudy there were 21 [17 AD (Families 8, 9, 10, 11, 15, 315, 326, 328, 330, 340, 347, 348, 363, 364, 369, 370, and 377) and 4 AR (Families 19, 79, 329, and 405)] SHI families (Figures S26–S46).

Families 8, 9, 10, 11, 15, 326, 328, 330, 348, 363, 364, 369, and 370 presented with AD Waardenburg syndrome, and family 329 had AR Waardenburg syndrome. Families 8, 9, 11, 15, 326, 348, 363, 364, 369, and 370 had congenital mild to profound bilateral sensorineural HI. Three Waardenburg families have a different HI presentation: Family 10 prelingual bilateral profound mixed HI; Family 328 unilateral profound sensorineural HI; and Family 330 congenital bilateral moderate conductive HI (Table S3).

Additional AD syndromes included BOR, MYH9 related disorder, and Tietz albinism deafness syndrome. Families 315 and 377 have BOR syndrome with congenital bilateral HI, with Family 377 presenting with sensorineural HI and Family 315 with profound mixed HI. Family 347 has MYH9 related disorder with congenital bilateral HI. Tietz albinism deafness syndrome was observed in Family 340, whose members presented with congenital profound bilateral sensorineural HI (Table S3).

In addition to Family 329 presenting with AR Waardenburg syndrome, three additional families had ARSHI: Bartter syndrome (Family 19 with congenital profound unilateral sensorineural HI); Chudley‐McCullough syndrome (Family 79 has congenital profound bilateral sensorineural HI); and Pendred syndrome (Family 405 has congenital moderate to profound bilateral sensorineural HI) (Table S3).

The cause of the syndrome was identified in 14 [11 AD (Families 9, 10, 11, 315, 326, 340, 347, 348, 363, 364, and 377) and 3 AR (Families 19, 79, and 405)] families with 14 variants segregating in nine SHI genes, that is, EDN3 (OMIM: 131242), EDNRB (OMIM: 131244), EYA1 (OMIM: 601653), GPSM2 (OMIM: 609245), MITF (OMIM: 156845), MYH9 (OMIM: 160775), PAX3 (OMIM: 606597), SLC12A1 (OMIM: 600839), and SLC26A4 (OMIM: 605646) (Tables 2 and S6, Figures S26–S39). Seven of the AD families for which the underlying causal variant was identified had Waardenburg syndrome, with five families (Families 10, 11, 348, 363, and 364) having a variant in PAX3 and one family each with a variant in EDN3 (Family 326) and EDNRB (Family 9). For the four additional AD SHI families for which the causal variant was identified, two had BOR syndrome with variants in EYA1 (Families 315 and 377), one had MYH9‐related disorder with a variant in MYH9 (Family 347), and one presented with Tietz Albinism‐Deafness syndrome with a variant in MITF (Family 340). For the AR SHI families, affected members had homozygous variants in SLC12A1 (Bartter syndrome Family 19) and GPSM2 (Chudley‐McCullough syndrome Family 79) and compound heterozygous variants in SLC26A4 (Pendred syndrome Family 405).

TABLE 2.

Variants segregating in 14 families with syndromic HI.

Family ID	Inh	GT	Phenotype	Gene	cDNA change	Protein change	Function	Reported	Classification a
Family 9	AD	Het	Waardenburg	EDNRB	c.764G>A	p.(Cys255Tyr)	Missense		VUS
Family 10	AD	Het	Waardenburg	PAX3	c.811C>T	p.(Arg271Cys)	Missense	Yes	Pathogenic
Family 11	AD	Het	Waardenburg	PAX3	c.811C>T	p.(Arg271Cys)	Missense	Yes	Pathogenic
Family 19	AR	Hom	Bartter	SLC12A1	c.949G>A	p.(Val317Ile)	Missense	Yes	VUS
Family 79	AR	Hom	Chudley‐McCullough	GPSM2	c.1492C>T	p.(Arg498*)	Stopgain	Yes	Pathogenic
Family 315	AD	Het	Branchio‐Oto‐Renal	EYA1	c.1319G>A	p.(Arg440Gln)	Missense	Yes	Pathogenic
Family 326	AD	Het	Waardenburg/Skin pigmentation	EDN3	c.52 + 7G>A	p.?	Splicing		VUS
Family 340	AD	Het	Tietz Albinism‐Deafness	MITF	c.957 T>G	p.(Ile319Met)	Missense		Likely Pathogenic
Family 347	AD	Het	MYH9 related disorder	MYH9	c.2105G>A	p.(Arg702His)	Missense	Yes	Pathogenic b
Family 348	AD	Het	Waardenburg	PAX3	c.185 T>C	p.(Met62Thr)	Missense	Yes	Likely Pathogenic
Family 363	AD	Het	Waardenburg	PAX3	c.218C>T	p.(Ser73Leu)	Missense	Yes	Pathogenic
Family 364	AD	Het	Waardenburg	PAX3	c.797_798inv	p.(Trp266Ser)	Missense		Likely Pathogenic
Family 377	AD	Het	Branchio‐Oto‐Renal	EYA1	c.966 + 5G>A	p.?	Splicing	Yes	Pathogenic
Family 405	AR	Comp Het	Pendred	SLC26A4	c.94 T>C	p.(Phe32Leu)	Missense		Likely Pathogenic
					c.716 T>A	p.(Val239Asp)	Missense	Yes	Pathogenic

Abbreviations: AD, Autosomal Dominant; AR, Autosomal Recessive; Comp Het, Compound Heterozygous; GT, Genotype; Het, heterozygous; Hom, homozygous; Inh, Inheritance; VUS, variant of unknown significance.

^a Classification based on the American College of Medical Genetics and Genomics guidelines.

^b Reclassified from VUS to likely pathogenic.

For the SHI genes, the identified variants include missense (N = 11), splice site (N = 2), and stop gain (N = 1). These variants were classified as either pathogenic or likely pathogenic, with the exception of the splice site EDN3 [c.52 + 7G>A], missense EDNRB [p.(Cys255Tyr)], and SLC12A1 [p.(Val317Ile)] variants that were classified as VUS. Five variants were not previously reported to be associated with SHI. Of these, three variants, EDNRB [p.(Cys255Tyr), MITF [p.(Ile319Met)], and PAX3 p.(Trp266Ser)] were also not observed in gnomAD, All of Us Bravo, TOPMed, or ClinVar databases. Pathogenic variant p.(Arg271Cys) in PAX3 was observed in two families, the only variant observed in more than one family. The remaining 13 variants were unique to each family (Tables 2 and S5, S6).

3.6. Potentially Solved and Unsolved Families

Three families were classified as potentially solved (Tables S5 and S7, Figures S16 and S40, S41). For Family 402 with ADNSHI, four missense LOXHD1 (OMIM: 613072) [p.(Val2112Ile), p.(Arg1800Gln), p.(Cys616Ser), and p.(Gly760Asp)] variants were found on the same haplotype in the two affected members. These four variants are all rare with MAF < 0.0001 in databases and were all classified as VUS. It is not possible to confirm which of these variants is underlying the NSHI phenotype segregating in this family (Table S7, Figure S16).

For family 329, that presented with AR Waardenburg syndrome, the only variant found to segregate was a missense variant [p.(Val2204Phe)] in BDP1 that was homozygous in the two affected family members. BDP1 (OMIM: 607012) is known to underlie ARNSHI but not Waardenburg syndrome. Additionally, for family 330 with AD Waardenburg syndrome, a MYO6 splice site variant [c.1674 + 12C>T] with a gnomAD v4.1.0 MAF = 3.0 × 10⁻⁶ was identified. MYO6 (OMIM: 600970) is a known ADNSHI gene but has not been implicated in Waardenburg syndrome (Table S7, Figures S40 and S41).

Nine NSHI (5 AR, 3 ad and 1 AR or XLR) and 5 ad SHI families remain unsolved (Table S8, Figures S17–S25 and S42–S46).

4. Discussion

There has only been limited investigation of HI in the South African population [12] despite its high incidence [3]. In general, HI is understudied in sub‐Saharan populations, and there is only limited knowledge of the genes and variants which play a role in its etiology in these populations [24]. This study investigated the underlying genetic cause of HI in 45 (24 NSHI and 21 SHI) South African families of varied ancestry. Known genes were identified as the underlying cause of HI for all solved families (62.2%). With 45.2% (14/31) of the observed variants not previously reported to cause HI. These findings suggest that, similar to previous studies in the Cameroonian, Ghana, and Nigerian populations, the degree of allelic and locus heterogeneity is also high in the South African population [25, 26, 27].

GJB2 [c.35del] was observed in only one family of mixed ancestry (Tables S1 and S4). The variant is of European origin and is also common in North Africa [28]. The observation of this variant in South Africa is most likely due to European admixture. MYO15A, a gene commonly reported to be associated with NSHI in Northern and Central Africa [28], was only identified as the causative gene in Family 63 in this study. One of the observed MYO15A variants that underlies ARNSHI, [p.(Glu2212Lys)] was previously observed in Ghanian, Middle Eastern, and Chinese populations [26, 29, 30].

A novel variant in TMPRSS3 [c.205 + 6 T>A] which is predicted to affect splicing was observed in the homozygous state. Nearly 100 variants spread across all functional domains of TMPRSS3 have been identified and shown to disrupt its proteolytic activity, leading to ARNSHI [31]. The genetic contribution of TMPRSS3 to ARNSHI has been observed in North African, Asian, European, and Mediterranean populations [32, 33, 34], with TMPRSS3 being the eighth most common cause of ARNSHI [35].

POU3F4 is the most common gene associated with XLR (DFNX2), accounting for ~50% of XL NSHI. In Africa, variants in POU3F4 linked to NSHI have been identified in South African [36] and Nigerian populations [37]. A novel stop gain variant for POU3F4 [p.(Ser261*)] was identified in this study and is classified as pathogenic.

One of the families presented in this study, Family 408, with progressive ADNSHI due to a heterozygous REST [p.(Cys415Ser)] variant was previously reported in a case report [38] (Table S4). The case report provides additional clinical data on Family 408.

Some of the genes identified in our cohort of NSHI families have been implicated in both NSHI and SHI, that is, CDH23 [Usher syndrome type ID (OMIM: 601067) and ARNSHI (DNB12)]; MYO7A [Usher syndrome type 1B (OMIM: 276900) and ADNSHI (DNA11)]; PCDH15 [Usher syndrome, type 1F (OMIM: 602083), Usher syndrome type 1D/F (OMIM: 601067), and ARNSHI (DFNB23)]; WFS1 [AD Wolfram‐like syndrome (OMIM: 614296), Wolfram syndrome (OMIM: 222300), and ADNSHI (DFNA6/14/38)]; SLC26A4 [Pendred syndrome (OMIM: 274600) and ARNSHI (DFNB4)]; and MITF [Waardenburg syndrome type II (OMIM: 193510), Tietz albinism‐deafness syndrome (OMIM: 103500), and ADNSHI [39, 40].

A novel likely pathogenic missense MITF variant [p.(Asn328Lys)] was observed in an ADNSHI Family 391 (Table S4). Since MITF is involved in the regulation of TYR transcription, a key enzyme in melanin synthesis, the phenotype associated with variants in this gene can also include hypopigmentation of the skin, hair, and eyes. None of the six affected members of Family 391 presented with any other phenotypes in addition to HI, similar to the previously reported Latino family with ADNSHI [39]. For Family 60 with ARNSHI, a compound heterozygous variant was identified in the ARNSHI gene PCDH15 (DFNB23) although one of the variants was a stop‐gain variant and the other was synonymous. There are reports of synonymous variants, for example, in TECTA [41] and MYO15A [42], underlying ARNSHI.

Waardenburg was the most frequent syndrome found in our cohort, representing 66.7% (Table S2) of the SHI cases. This is similar to previously reported studies of Waardenburg in sub‐Saharan cohorts, for example, 62% among Black children with SHI in southern Africa [43] and 50% among Cameroonians with SHI [44]. PAX3 was found to be the most common gene implicated for Waardenburg syndrome in our cohort. More than 150 PAX3 variants underlying Waardenburg syndrome have been reported [45, 46]. In our study, four PAX3 variants were observed in five families, of which one was novel, that is, [p.(Trp266Ser)] in exon 6. Although the [p.(Trp266Ser)] is a novel variant, previously at the same residue, that is, [p.(Trp266Cys)] was also reported to underlie Waardenburg syndrome [47]. These variants are in regions that interact with cell signaling proteins (Tables S5 and S6). In AD Waardenburg syndrome Family 326, a heterozygous splice EDN3 variant [c.52 + 7G>A] was identified. Heterozygous and homozygous mutations of EDN3 have been identified in Waardenburg syndrome type 4 families. Although Waardenburg syndrome typically follows an AD inheritance pattern, cases with incomplete dominance or recessive inheritance have also been observed. AD Waardenburg tends to show one or more features of the disease with low or incomplete penetrance, while AR Waardenburg is more penetrant and severe compared to AD forms [48]. Waardenburg syndrome is classified into four subtypes based on associated phenotypic features [49], that is, types 1–4. Type 1 has an AD mode of inheritance, while for types 2, 3, and 4, although the mode of inheritance is predominantly AD, in rare cases, an AR mode of inheritance has been observed. In this study, AR Waardenburg syndrome was observed only in Family 329. Family 9, with AD Waardenburg syndrome that segregates a missense EDNRB [p.(Cys255Tyr)] variant, has reduced penetrance with the variant being present in the unaffected mother. Incomplete penetrance has previously been reported for Waardenburg families with variants in EDNRB [50].

Family 79 that presents with Chudley‐McCullough syndrome has a previously reported pathogenic homozygous GPSM2 [p.(Arg498*)] variant. A family of Western European descent with Chudley‐McCullough syndrome presented with the GPSM2 [p.(Arg498*)] variant in compound heterozygous state [51].

The variants identified in Waardenburg families which may underlie the syndrome are BDP1[p.(Val2204Phe)] and MYO6 [c.1674 + 12C>T] (Figures S40 and S41). Although variants in MYO6 are known to cause ADNSHI (DFNA22) and ARNSHI (DFNB37) with the ARNSHI likely to be semi‐dominant [52] and variants in BDP1 cause ARNSHI (DFNB49/112) [53], there have been no reported associations with Waardenburg syndrome. There needs to be additional Waardenburg families identified with variants in BDP1 and MYO6 to determine if these genes are involved in the syndrome.

For 37.5% (9/24) of the NSHI families (Figure 1C) and 23.8% (5/21) of the SHI families (Figure 1D) no underlying genetic cause was found. NSHI/SHI in these families could be due to intrafamilial genetic heterogeneity, undetected structural variants, and variants that were not observed in the exome sequence data. Additionally, for families with NSHI, although each family has at least two hearing impaired family members, HI could be due to environmental causes.

FIGURE 1.

Distribution of genes for nonsyndromic and syndromic HI families. The pie charts display: Observed NSHI genes and the percent of NSHI families which has HI due to each gene (A); observed SHI genes and the percent of SHI families which has HI due to each gene (B); percent of solved, potentially solved, and unsolved NSHI families (C); and percent of solved, potentially solved, and unsolved syndromic HI families (D). [Colour figure can be viewed at wileyonlinelibrary.com]

For this study, family ascertainment was performed in seven of the nine South African provinces (Figure S1 and Table S1). There is an over‐representation of families of mixed ancestry since the largest proportion of families was ascertained in the West Cape.

We observed in our study a high level of heterogeneity for the variants and genes involved in NSHI. This may be partially explained by the long history of admixture within South Africa. The number of generations of admixture in South Africa varies by population. Admixture within the Afrikaner population is estimated to have occurred ~9.3 generations ago between the Northwestern Europeans and Khoe‐san groups. Although most admixture events have taken place < 10 generations ago occurring with European colonization, there has been admixture occurring as far back as 50 generations mainly occurring between the Khoe‐San and East‐Africans [11, 54, 55].

In conclusion, although our study represents a small subset of the population, it underscores the significant heterogeneity of both variants and genes associated with NSHI and SHI in the South African population. We also identified several novel variants in known genes for both syndromic and non‐syndromic forms of HI. Importantly, our study calls attention to the critical need for more comprehensive, large‐scale genetic investigations to fully understand the underlying causes of HI in this region. Expanding such research will not only enhance diagnostic accuracy but also guide the development of targeted interventions and improve access to genetic testing for South African individuals.

Conflicts of Interest

The authors declare no conflicts of interest.

Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/cge.14765.

Supporting information

Data S1. Supporting Figures.

Data S2. Supporting Tables.

Bharadwaj T., Acharya A., Manyisa N. R., et al., “The Diverse Genetic Landscape of Hearing Impairment in South African Families,” Clinical Genetics 108, no. 5 (2025): 511–520, 10.1111/cge.14765.

Data Availability Statement

The variants reported in this study were submitted to ClinVar (accession numbers SCV002520420, SCV005415623‐SCV005415639, SCV005431518‐SCV005431531, SCV005431533, and SCV000853303.1). Sanger sequencing results are available upon request from the corresponding author.