Allelic heterogeneity refers to instances when several causal alleles within one gene result in a single unifying disorder with shared mutational mechanisms and with phenotypes that lie across a continuum.1 Alternatively, in some instances, distinct alleles within a single gene can result in phenotypically nonoverlapping allelic disorders secondary to differential mutational mechanisms (e.g., loss of function [LOF] vs. gain of function).2 Human variants in the SOX10 gene represent an excellent example to study such genetic heterogeneity and phenotypic pleiotropy. The SOX10 gene encodes for the SOX10 (SRY [sex determining region Y]-box 10) protein, a transcription factor regulating neural crest development.3 To date, two separate syndromic phenotypes have been linked to SOX10 variants in humans: Waardenburg syndrome (WS [type 2 and 4]),3 an auditory-pigmentary disorder, and Kallmann syndrome (KS),4 a reproductive disorder. These observations raise the question whether WS and KS are examples of a phenotypic continuum or if they are two nonoverlapping allelic disorders.

KS is a rare genetic disorder defined clinically by a syndromic association of idiopathic hypogonadotropic hypogonadism (IHH) and olfactory dysfunction, i.e., hyposmia/anosmia.5 The candidacy of SOX10 as a KS gene emerged from the overlapping observation that agenesis of the olfactory bulb (a marker of anosmia) was seen in some WS individuals.6 Hence, Pingault et al. examined 17 KS subjects exhibiting one or more WS features (primarily hearing loss) for SOX10 variants and identified 5 KS subjects (~30%), all with hearing loss, harboring pathogenic SOX10 variants.4 Additional reports have subsequently implicated SOX10 variants in KS subjects with hearing loss.7,8,9 The cosegregation of KS and hearing loss (a WS feature) supports the hypothesis that SOX10-related WS and KS may actually represent phenotypically overlapping disorders. However, Pingault et al. also identified one KS individual with a SOX10 variant without overt WS features, hinting that disease-specific alleles suggestive of an allelic disorder may also exist.4 Hence, a systematic comparative genotypic and phenotypic analysis of a large cohort of SOX10-related IHH patients is required to fully validate either hypothesis (phenotypic continuum vs. distinct allelic disorders). Finally, some KS genes have also been known to contribute to normosmic forms of IHH (nIHH);5 however, apart from a single nIHH reported to harbor a SOX10 rare sequence variant (RSV),10 the contribution of SOX10 variants to nIHH is not known.

To address these questions, a large IHH cohort (with both KS and nIHH) was examined for SOX10 RSVs and their associated phenotypes defined. Additionally, previously reported SOX10 RSVs identified in KS and WS cases from the published literature were cataloged and their genotypic and phenotypic spectra compared. All disease-associated SOX10 RSVs were also compared against gnomAD,11 a publicly available data set, to identify critical domains of the SOX10 protein relating to human disease.


Study subjects

Study cohorts

Massachusetts General Hospital (MGH) study cohort

A total of 632 KS (IHH + anosmia/hyposmia) subjects from the Harvard Reproductive Endocrine Sciences Center’s IHH cohort were included in this study. We also included 677 nIHH subjects to determine any causal/contributory role for SOX10 pathogenic variants in nIHH. Diagnostic criteria for IHH were applied as previously reported5 and are described in Supplementary methods.

Previously published SOX10-associated WS and IHH patients

We cataloged the genotypes and phenotypes of SOX10 RSVs previously reported in IHH and WS in the literature and in the Leiden Open Variation Database (LOVD),12 a public repository of pathogenic SOX10 variants. This search yielded additional 101 WS probands with SOX10 RSVs and 17 IHH probands with SOX10 RSVs (Table S1A, B).

Phenotypic evaluation for MGH IHH cohort

Clinical charts were reviewed for phenotypic evaluation. In IHH patients with SOX10 RSVs, Waardenburg syndrome–like features (WS-like features) were deemed to be present if they exhibited at least one of the following WS-associated features: sensorineural hearing loss, atypical pigmentation, Hirschsprung disease, and/or features of peripheral demyelinating neuropathy, central demyelinating leukodystrophy, Waardenburg syndrome, and Hirschsprung disease (PCWH) syndrome.13

Phenotypic evaluation for WS/IHH cases reported in the literature

All text/tables of IHH subjects with SOX10 RSVs that were previously reported (LOVD and published literature) were reviewed for WS-like features (as described above) (Table S1A, B; Supplemental References). For previously reported WS patients with SOX10 RSVs, reproductive phenotypes suggestive of IHH was deemed present if they exhibited at least one of the following IHH-related phenotypes: micropenis or cryptorchidism (boys), absent puberty in both sexes, primary amenorrhea (girls), and/or a biochemical observation of hypogonadotropic hypogonadism.

Genetic analysis

Exome sequencing (ES) methodology and variant annotation for the MGH IHH cohort have been described previously.14 ES data was queried for SOX10 (RefSeq: NM_006941.3) rare sequence variants (RSVs, defined as: variants with <0.1% minor allele frequency [MAF] in gnomAD database) and for RSVs in other IHH genes (Supplemental Methods). All RSVs were confirmed by Sanger sequencing and pedigree segregation analysis was performed for determining mode of inheritance whenever possible. We also cataloged all SOX10 RSVs with <0.1% minor allele frequency (MAF) in gnomAD (Table S1C).

Functional annotation of SOX10 rare sequence variants

Functional effects of SOX10 RSVs identified in the MGH cohort was examined using the following in silico programs: PolyPhen-2,15 Sorting Intolerant from Tolerant (SIFT),16 Combined Annotation Dependent Depletion (CADD),17 and Rare Exome Variant Ensemble Learner (REVEL).18 All SOX10 RSVs were also categorized using the American College of Medical Genetics and Genomics (ACMG) Standards and Guidelines for interpretation of sequence variants.19 The amino acid sequence of the SOX10 high mobility group (HMG) protein domain has a very strong similarity (95% amino acid identity) to the sequence of the SOX9 HMG domain, which has a solved 3D structure (PDB ID 4s2q).20 Based on the pairwise sequence alignment of these two close homologs, the SOX10 variants were mapped on the structure of the SOX9 HMG domain.

Statistical analysis

Gene burden testing, including domain-based testing, was performed using Fisher’s exact test. Only filtered variant allele counts from ES data were considered for burden testing. Relative proportions of variant classes (loss-of-function [LOF] and missense variants) between IHH, WS, and gnomAD variants were compared using Chi-square test with Yates correction. P value of <0.05 was considered as significant.


SOX10 RSVs are associated with both KS and nIHH forms of idiopathic hypogonadotropic hypogonadism

In the MGH IHH cohort, we identified 18 heterozygous SOX10 RSVs: 5 LOF variants and 13 missense variants, in 20 unrelated (13 M:7 F) subjects (Fig. 1, Tables 1, 2, and S2). Of these 20 subjects, 16 IHH subjects manifested KS, while 4 subjects were normosmic (Fig. 1, Table S2). RSVs that were deemed likely pathogenic/pathogenic by ACMG criteria were inferred as pathogenic. RSVs that were deemed likely benign/variant of unknown significance by ACMG criteria were inferred to be pathogenic if they were assessed as deleterious in any one of the four in silico prediction programs (Table S2). Of the 18 RSVs only 1 variant (p.Ala28Ser) was deemed benign by the above criteria. Segregation analysis in informative pedigrees for the SOX10 RSVs was consistent with an autosomal dominant mode of inheritance with significant variable penetrance and/or expressivity (Fig. 1). In keeping with the known oligogenic basis of IHH pathogenesis,21 6/20 SOX10+ probands also harbored additional RSVs in known IHH genes (Table 1, Fig. 1). Review of the literature revealed 17 additional IHH subjects (16 KS; 1 nIHH) with published SOX10 RSVs (Fig. 2; Tables 1, 2, S1B and S2). In lieu of cellular studies, since there was a congregation of IHH-associated variants within the SOX10 HMG domain (see below), the functional impact of these variants was examined using SOX9-based homology modeling analysis. The majority of the IHH-associated SOX10 HMG domain mutated residues were predicted to be directly involved in DNA binding and/or helix formation (Supplemental Fig. 1). Overall, these results confirm the previously reported causal association of SOX10 RSVs in KS and, in addition, demonstrate a previously underappreciated role for SOX10 variants in nIHH.

Fig. 1: Family pedigrees of probands with SOX10 rare sequence variants (RSVs) identified in the Massachusetts General Hospital (MGH) idiopathic hypogonadotropic hypogonadism (IHH) cohort.
figure 1

Pedigrees are sorted by presence (a) or absence (b) of Waardenburg syndrome (WS)-associated features. Probands are identified by arrows. + wild-type (WT) allele; M mutant allele. KS Kallmann syndrome, nIHH normosmic idiopathic hypogonadotropic hypogonadism.

Table 1 SOX10 RSVs in IHH patients with Waardenburg syndrome–related features.
Table 2 SOX10 RSVs in IHH patients without overt Waardenburg syndrome–related features.
Fig. 2: SOX10 protein domains and positions of SOX10 rare sequence variants (RSVs) identified in idiopathic hypogonadotropic hypogonadism (IHH), Waardenburg syndrome (WS), and gnomAD.
figure 2

Heterozygous loss-of-function (LOF) alleles are shown in red circles; heterozygous missense alleles are shown in black circles; homozygous missense alleles are shown in green circles. *Only single-nucleotide variants associated with WS are shown and SOX10 structural variants are not depicted.

Phenotypic analyses show that SOX10-related IHH and WS represent phenotypic constituents of a unifying SOX10-related human syndrome

To determine if SOX10-related IHH and WS represented a phenotypic continuum versus distinct allelic disorders, we performed two complementary phenotypic analyses based on case ascertainment.

Phenotypic analysis of SOX10 cases ascertained by the IHH phenotype

We first reviewed the phenotypic spectrum of individuals harboring SOX10 RSVs in the MGH cohort. In the MGH cohort, 10/20 IHH subjects (all with the KS form of IHH) exhibited ≥1 WS-associated feature(s) (Table 1). Next, we reviewed the 17 previously reported IHH patients with SOX10 RSVs in the literature and found that 13/17 KS subjects also exhibited ≥1 WS-associated feature(s) (Table 1). Thus, in a significant majority of SOX10-IHH cases, WS-like features were present, suggesting that IHH represents a constituent phenotype that becomes evident through the natural history of an underlying unifying clinical syndrome. However, in a small number of SOX10-IHH pedigrees, WS-like phenotypes were not readily evident (Table 2). This observation raises three possibilities: (1) incomplete phenotypic information or cryptic phenotypes not readily evident, (2) variable expressivity of the constituent phenotypes within a singular syndrome, or (3) a small subset of patients may represent a distinct “allelic disorder” with no phenotypic overlap.

Phenotypic analysis of previously reported SOX10 cases ascertained by WS phenotypes

For a majority of reports, since subjects were typically prepubertal, reproductive phenotypes suggestive of IHH were not reported. However, we identified five individuals who had an initial diagnosis of WS, in whom at least one IHH-related phenotype was already documented (Table 3).

Table 3 SOX10 RSVs previously reported in WS patients in whom reproductive phenotypes were evident.

Taken together, the above data support the notion that IHH and WS represent phenotypic constituents positioned along a singular SOX10-associated human syndromic spectrum.

Genotypic analyses show SOX10-related IHH and WS represent phenotypic constituents of a unifying syndrome

To further examine the phenotypic continuum versus distinct allelic disorder hypothesis, we compared the variant spectrum and allelic overlap between SOX10 RSVs identified in subjects ascertained by an initial IHH diagnosis (MGH cohort and published literature) and those ascertained by an initial WS diagnosis (published literature). This analysis showed that the SOX10 allelic spectrum in IHH-ascertained cases (Tables 1, 2, & S1B) and WS-ascertained cases (Tables 3, S1A) included both LOF variants (frameshifting indels, nonsense, and splice-altering) and missense variants that spanned the entire length of the SOX10 protein. Furthermore, we noted that four SOX10 RSVs (p.Met112Ile, p.Trp114Arg, p.Leu145Pro, p.Val92Leu) showed allelic overlap—i.e., identical RSVs were identified in cases ascertained by IHH or WS independently. This latter observation strongly suggests that IHH and WS represent phenotypic tenets of a single disorder. However, cumulatively, SOX10 LOF alleles were proportionally more prevalent in WS-ascertained cases (79% LOF [n = 81]; 21% missense [n = 21]) while missense variants were proportionally more prevalent in IHH-ascertained cases (30% LOF [n = 11]; 70% missense [n = 26]) (p < 0.001) (Fig. 2). This observation suggests that, within this single disorder, the varian severity may partly account for the variable penetrance and/or variable phenotypic expressivity of IHH or WS features exhibited by individual patients.

We then performed a SOX10 protein domain-based mutational enrichment analysis to identify any specific domains that conferred phenotypic specificity. This approach revealed that there was no significant difference in the proportional prevalence of RSVs in any specific SOX10 domain between IHH- and WS-ascertained cases (p = 0.07). However, since a large proportion of SOX10 missense RSVs congregated within the HMG domain across both IHH (n = 15) and WS (n = 18), we then compared the specific mutated HMG domain residues in IHH- and WS-ascertained cases. This analysis showed that the mutated HMG domain residues in IHH-ascertained cases were largely nonoverlapping with the missense residues in WS-ascertained cases except for three overlapping residues (M112, W114, and L145) (Tables 1, 2, S1A, B). However, review of phenotypes among IHH cases harboring these distinct HMG residues showed that the vast majority of these distinct HMG residue variants resulted in WS features (Table 1). This observation strongly suggests that there may be no phenotypic specificity conferred by specific HMG domain residues and that the HMG domain is critical for pathogenesis of both phenotypes. This notion is further substantiated by the observation of a significant proportional enrichment for SOX10 missense variants within the HMG domain across both disease states when compared with their distribution in the gnomAD population database (70% in disease states vs. 6% in gnomAD; p < 0.001) (Fig. 2, Tables 1, 2, S1A, C). Intriguingly, a total of six gnomAD variants, including two HMG domain variants (R151 and R161) overlapped with disease-associated variants; however, no phenotypic details were available for the gnomAD subjects (Table S1D).


With the advent of next-generation sequencing, several pleiotropic phenotypes resulting from variants in a single gene are increasingly recognized.22 Phenotypic continuum versus distinct allelic disorders represent two dichotomous explanations for the observed pleiotropy. Among the SOX10-associated human phenotypes, WS-related features (e.g., hearing loss, Hirschsprung disease, skin pigmentation) are often recognizable early in childhood. In contrast, IHH is typically recognized later in adolescence. This study took advantage of this “temporal gap” in phenotypic manifestations (i.e., the natural history) and systematically examined SOX10-related IHH patients for the presence or absence of WS-like phenotypes that would have manifested in childhood. This approach allowed us to demonstrate that WS features cosegregated in a significant majority of cases ascertained by the IHH phenotype. These findings strongly suggest that IHH and WS represent constituent phenotypes within a single human SOX10-related continuum of developmental defects rather than representing distinct allelic disorders. These findings have important implications for clinical care of children harboring pathogenic SOX10 alleles as longitudinal follow up for incipient IHH during adolescence is imperative to allow early recognition of pubertal phenotypes and enable timely institution of hormonal therapy to avoid physical and psychological effects of IHH.

Three key observations provide validity to our conclusion that WS and IHH relate to a phenotypic continuum:

1. Overlapping phenotypes independent of case ascertainment. Among patients with SOX10 variants ascertained by the IHH phenotype, a majority (62%) exhibited one or more WS features and nearly 25% of subjects harbored two or more WS features (Table 1). Both in this report and the literature,4 hearing loss is almost ubiquitous in IHH individuals with pathogenic SOX10 RSVs. In this report, WS-like features such as pigmentation abnormalities and Hirschsprung disease also cosegregated with IHH further substantiating the IHH-WS phenotypic continuum hypothesis (Tables 1, 2, and S1B). Similarly, in a small number of previously reported SOX10 individuals ascertained by the WS phenotype, IHH-related phenotypes were present (Table 3). However, the prevalence of IHH features in WS-ascertained individuals is certainly an underestimate since a vast majority of these subjects were prepubertal children in whom reproductive phenotypes are typically quiescent. In addition, the majority of these reports were published before SOX10 was causally linked to reproductive phenotypes and hence these phenotypes were not sought.

2. Shared mutational spectrum, mutational mechanisms, and overlapping alleles in SOX10-related IHH and WS. We noted that the SOX10 variant spectrum of IHH and WS comprised both LOF and missense alleles. The LOF variants resulting in IHH or WS typically occurred in the heterozygous state and spanned the entire length of the SOX10 protein; hence it is likely that SOX10 haploinsufficiency is a shared underlying mechanism for these LOF alleles. In terms of missense alleles, no regional enrichment within the SOX10 protein was evidently different between IHH and WS. In contrast, there was a significant enrichment for missense alleles in the SOX10 HMG domain in disease cohorts compared with gnomAD, suggesting a shared importance of this domain for both phenotypes. Furthermore, functional cellular studies of WS and IHH-associated SOX10 HMG domain missense variants have been shown to be hypomorphic/LOF alleles3,23,24,25 with no discernible dominant-negative activity.3 Finally, in this study, three HMG domain missense RSVs (p.Met112Ile; p.Trp114Arg; p.Leu145Pro) and one non-HMG domain missense variant (p.Val92Leu) showed allelic overlap independent of the phenotypic ascertainment (IHH or WS). This finding adds further certitude to the shared mutational mechanisms–phenotypic continuum hypothesis.

3. Shared developmental biology of SOX10-related IHH and WS. The key clinical features of SOX10-related WS have been shown to result from distinct defects in neural crest derived cells such as melanocytes (skin pigmentation, deafness) and enteric nervous system (Hirschsprung disease).3 Studies in Sox10-mutant mice confirm a critical role for SOX10 protein in the development of a distinct subset of neural crest derived glial cells called olfactory ensheathing cells (OECs) and when OECs are absent, migration of gonadotropin-releasing hormone (GnRH) cells is impaired.4,26 Furthermore, emerging data using lineage tracing approaches in both zebrafish27 and mice28 support a dual origin of GnRH neuroendocrine cells, one cohort arising from olfactory placodal ectoderm and the other from neural crest progenitors. Thus, association of human SOX10 variants with IHH and their phenotypic overlap with WS provide human validation to the neural crest origins for GnRH cells and confirm that SOX10-related IHH also represents a neurocristopathy.29

The next question relates to the precise timing of targeted reproductive phenotyping in SOX10+ patients. Although puberty is initiated at adolescence, the reproductive axis is known to be active in the first few months of life in both sexes, a developmental window termed as the “minipuberty of infancy.”30 While the relevance of minipuberty in girls is poorly understood, hypogonadotropism during this period in boys can result in micropenis and/or cryptorchidism (undescended testes). Timely laboratory assessment of gonadotropin levels during this developmental window can permit a definitive diagnosis of congenital hypogonadotropic hypogonadism30 allowing early treatment in boys (e.g., testosterone/human chorionic gonadotropin (hCG) therapy to promote penile growth; hCG and follicle-stimulating hormone (FSH) therapy to promote sertoli cell proliferation and thereby improve future fertility potential31). Similarly, during adolescence, early identification of hypogonadotropism in both boys and girls who are SOX10+ will allow initiation of treatment to induce more timely development of secondary sexual characteristics, optimize bone maturation to allow height growth and skeletal maturation, and prevent psychological scars from delayed puberty. A longitudinal report of a female patient with WS? nicely illustrates the importance of longitudinal follow up in SOX10+ children.32 A female infant was initially diagnosed SOX10-related WS when she presented with iris hypopigmentation, bilateral sensorineural deafness, and a white forelock due to causal SOX10 LOF variant (SOX10 p.P169fsX117).33 Subsequently, during longitudinal follow up, she displayed delayed puberty at adolescence and endocrine evaluation revealed findings consistent with IHH necessitating hormonal therapy.32 This exemplary case report fully substantiates this study’s findings and argues for careful follow up of SOX10+ patients to discern the full phenotypic spectrum including incipient reproductive phenotypes. This is particularly important since SOX10+ patients may be encountered in multiple medical specialist domains (audiology, colorectal surgery, dermatology, endocrinology) and these providers must be cognizant of these cosegregating phenotypes to avoid a diagnostic odyssey and delayed recognition and treatment of other related phenotypes.

Despite the strengths of our observations presented herein suggesting that WS and IHH lie along a SOX10-related continuum of developmental phenotypes, there are some notable exceptions. First, a very rare, but severe SOX10-related neurological phenotype (PCWH) resulting from distinct SOX10 RSVs affecting the last coding exon has been described. These variants have been shown to result in a stable SOX10 mutant messenger RNA (mRNA) that escapes nonsense-mediated decay with resultant mutant protein that possesses a potent dominant-negative activity.25 In another PCWH patient, disruption of the SOX10 native stop codon extended the SOX10 protein into the 3’ untranslated region to create a SOX10 mutant fusion protein that was shown to exert a toxic gain-of-function activity.24 Similarly, Chaoui et al. have also demonstrated that specific SOX10 missense pathogenic variants may also exert a dominant-negative activity leading to a PCWH phenotype. Second, a biallelic SOX10 deletion34 resulting in a severe arthrogryposis syndrome in a stillborn infant at 32 weeks has also been described. In contrast, IHH- and WS-related SOX10 pathogenic LOF variants occur in the heterozygous state, spare the last coding exon, and do not exhibit dominant-negative or gain-of-function activity. Thus, PCWH phenotype resulting from dominant-negative or gain-of-function mechanisms and the arthrogryposis syndrome resulting from biallelic SOX10 deletions may represent distinct SOX10-related allelic disorders that have differential underlying mechanisms. Finally, some IHH subjects in this study did not manifest WS-like features, raising the possibility that this subset of patients may represent an allelic disorder. However, the variant types and locations were similar to those IHH subjects who exhibited WS-like features; hence, it is more likely that these cases represent examples of variable expressivity within a single disorder.

In this study, in lieu of cellular studies for determining pathogenicity, we utilized a combination of in silico prediction algorithms, variant classification using the ACMG criteria, and applied human population genetic data to help determine RSV pathogenicity. This composite analysis showed that almost all HMG domain missense variants in our cohort and the literature were likely to be pathogenic (Tables 1 and 2). Structural modeling analysis of the mutated HMG domain residues showed that almost all were predicted to disrupt DNA binding (Supplemental Fig. 1). Furthermore, previous functional cellular studies on HMG domain SOX10 missense variants have consistently shown that majority of these variants are deleterious.3 However, individual variants within the HMG domain may show differential gradients of transcriptional activity, ranging from null alleles to mild hypomorphic alleles, thus resulting in variable phenotypic expressivity. Domain-specific burden analyses showed significant enrichment for disease-associated missense variants within the HMG domain (Fig. 2) compared with gnomAD but no enrichment for other domains of the SOX10 protein. In silico analyses and ACMG variant classification of non-HMG variants in this study also showed variability in their predicted pathogenicity (Table S2). Furthermore, prior cellular studies have shown that non-HMG domain variants displayed preserved DNA binding and normal subcellular localization.3,8 These findings suggest that SOX10 variants affecting the non-HMG domains may not sufficient to be causal alleles on their own. Putative pathogenic variants in other IHH genes were seen in a subset of patients with SOX10 variants (Table S3) suggesting potential oligogenic inheritance mechanisms. Furthermore, murine studies have previously suggested genetic modifiers of SOX10-related neurocristopathies35 and similar mechanisms may also underlie human SOX10-related disorders. However, detailed phenomic, genomic, and transcriptomic studies in a large cohort of SOX10 variant–positive families will be required to uncover such modifiers.

In this study, surprisingly, in addition to KS, we identified four nIHH subjects harboring putatively pathogenic SOX10 RSVs: one LOF allele and three missense alleles, of which one was within the HMG domain (Table 2). Our results suggest that SOX10 variants appear to contribute to both KS and nIHH forms. The precise neurodevelopmental basis for the preservation of normal olfactory phenotype in these SOX10+ nIHH individuals remains unclear. Intriguingly, none of the nIHH subjects affected showed clear WS-like features (Table 2). This observation suggests that anosmia may be a key pathophysiologic link between WS and IHH and further studies will be required to corroborate this assertion.

This study has some limitations. First, for SOX10-WS variants, we relied on the previously reported limited phenotypic information and IHH-related phenotypes were only identifiable in five individuals. However, since most of the WS cases in the literature are prepubertal and have not been updated with longitudinal follow up through puberty, it is likely that this is a gross underestimate. Hence, future studies in older subjects with SOX10 variants will be needed to fully establish the prevalence of IHH in SOX10-related human disorders. Second, our genetic analysis focused primarily on single-nucleotide variant (SNVs); copy-number variants (CNVs) were not examined. Hence, the differential contribution of SOX10 CNVs to SOX10-related phenotypic pleiotropy relating to IHH/WS remains to be studied.

In conclusion, this study demonstrates that IHH and WS represent SOX10-associated developmental defects that lie along a phenotypic continuum. With the increasing use of genotype-first approaches to study rare genetic diseases, this study provides a framework for studying pleiotropic single-gene disorders and highlights the critical importance of using large well-phenotyped cohorts with natural history information to help (re)define Mendelian disease nosologies.