Skip to main content
NCBI home page

This is a preprint.

It has not yet been peer reviewed by a journal.

The National Library of Medicine is running a pilot to include preprints that result from research funded by NIH in PMC and PubMed.

medRxiv logoLink to medRxiv
[Preprint]. 2026 Jul 10:2026.07.06.26355672. [Version 1] doi: 10.64898/2026.07.06.26355672

Genomic basis of developmental defects of enamel and sex-specific effects

P Shrestha 1,2, M Graff 2, Y Gu 3, Y Wang 2, HS Ahn 1, KN Nguyen 1, A Khanna 1, CL Avery 2, HM Highland 2, J Ginnis 1, MA Simancas-Pallares 1, AG Ferreira Zandoná 4, RN Alotaibi 5, DY Lin 6, JS Preisser 6, GD Slade 1, ML Marazita 7, KE North 2,8,9, K Divaris 1,2
PMCID: PMC13370533  PMID: 42465942

Abstract

We conducted a multi-ancestry genome-wide association study (GWAS) of developmental defects of enamel (DDE) in the primary dentition among 6,061 U.S. preschool-aged children (3–5 years). We investigated four DDE phenotypes (demarcated opacities, diffuse opacities, hypoplastic defects, and a combined DDE trait) leveraging main-effect models, joint gene-sex interaction testing (2df), and sex-stratified analyses. SNP-based heritability for the combined DDE trait was estimated at 20%, with concordance analyses robustly supporting a genetic etiology. We identified 39 unique genome-wide significant loci (P<5×10 ), with five surpassing a study-wide Bonferroni-corrected statistical significance criterion (P<1.25×10 9), including Y RNA and ALDH1A1. The main-effect GWAS identified 20 loci, including HBS1L and MYB, genes regulating hematopoiesis with plausible roles in amelogenesis. Joint test and sex-stratified analyses revealed 19 additional loci, including ALDH1A1, TENM2, and DLGAP2, demonstrating sex-specific heterogeneity. Nineteen loci exhibited sex-specific differences after Bonferroni correction (P<2x10−3), including genes involved in retinoic acid signaling (ALDH1A1), odontogenesis (TENM2), and neurodevelopment (DLGAP2, CDH10). Pathway enrichment highlighted ectodermal and synapse organization networks, suggesting shared etiological mechanisms between DDE and systemic conditions like neurofibromatosis and autism spectrum disorder. Notably, no locus generalized in an external GWAS of permanent dentition DDE, underscoring fundamental biological differences in the genetic architectures governing primary versus permanent enamel formation. Crucially, a comprehensive cross-trait pleiotropy lookup against early childhood caries (ECC) revealed no shared genetic architecture, supporting the notion that the established clinical and epidemiological association between DDE and ECC is likely driven by structural defects increasing caries lesion susceptibility rather than genetic pleiotropy. By integrating gene-sex interaction testing, this study offers novel insights into the complex, sexually dimorphic genetic etiology of DDE and augments the biological evidence base that can support the development of precision pediatric dentistry.

Keywords: enamel defects, enamel hypoplasia, developmental defects of enamel, genomics, gene-environment interaction, heritability, genetics, genetic variants

Introduction

Developmental defects of enamel (DDE) are clinically manifest non-carious conditions resulting from perturbations during amelogenesis. Affecting 25% to 40% of the primary dentition globally, DDE encompass a heterogeneous group of conditions ranging from qualitative hypomineralization (e.g., demarcated and diffuse opacities) to quantitative hypoplastic defects (Seow et al. 2011; Robles et al. 2013). Beyond aesthetic concerns, these defects present a profound clinical burden (Jälevik and Klingberg 2002). Conditions such as molar-incisor hypomineralization (MIH) frequently cause severe hypersensitivity, rapid post-eruptive breakdown, and elevated caries risk. Consequently, DDE exert a substantial psychosocial and economic toll (Marshman and Rodd 2015), underscoring the urgency of elucidating their complex etiology.

While environmental factors such as early childhood illnesses and nutritional deficiencies are established as modifiers of enamel formation, amelogenesis is fundamentally governed by a highly conserved genetic program. Human candidate gene studies have reported associations between SNPs in enamel forming genes (Gerreth et al. 2018) and DDE. Moreover, animal models have demonstrated that disruptions in core enamel matrix genes (e.g., ENAM, AMTN, KLK4) result in characteristic enamel malformations (Hu et al. 2008; Núñez et al. 2016). Furthermore, because the enamel organ shares an ectodermal embryonic origin with the central nervous system, and tooth mineralization overlaps with systemic developmental windows, enamel defects frequently exhibit pleiotropy with broader systemic conditions, including neurodevelopmental and hematopoietic disorders

Despite these insights, our understanding of the polygenic architecture of DDE, particularly in the primary dentition, remains incomplete. To-date, genome-wide association studies (GWAS) have primarily evaluated DDEs in permanent teeth and older cohorts (Kühnisch et al. 2014) (Alotaibi et al. 2022). Crucially, recent GWAS of complex dental phenotypes have highlighted the critical need to consider sex as a biological variable (Freitag-Wolf et al. 2021; Shrestha et al. 2025), as well as sex-based prevalence differences in DDE (Lunardelli et al. 2024). To address this knowledge gap and to add to the knowledge base of DDE genomic susceptibility, we sought to systematically identify genetic risk loci associated with DDE using a GWAS that leverages gene-sex interactions within a multi-ancestry population-based cohort of preschool-aged children.

Methods

Study design and population.

The analysis sample comprised 6,061 participants with clinical DDE data from the ZOE 2.0 study, a community-based, multi-ethnic genetic-epidemiologic study of early childhood oral health (Divaris et al. 2020). Participants were 3–5-year-olds sampled from public preschools (Head Start) in North Carolina, United States, between August 2016 and February 2019. Details on study design, recruitment, exclusions, and the genotyped population are presented in the Appendix (Appendix Figure 1).

Phenotype ascertainment.

Trained clinical examiners used the modified DDE index to ascertain developmental defects of the enamel at the tooth level (Ginnis et al. 2019). Specifically, the facial/buccal surface of each was assessed with respect to the presence or absence of demarcated opacities (dema), diffuse opacities (diff), and hypoplastic defects (hypo). Based on these assessments, quantitative (count, with range=0-20) and binary (presence versus absence) traits were defined at the person level, including a combined DDE trait encompassing ‘any DDE’ (Appendix Figure 2 and Appendix Table 1). To rigorously account for the right-skewness and zero-inflation inherent to count data (Appendix Figure 3), traits were regressed on age, sex, and race/ethnicity utilizing zero-inflated negative binomial models. The resulting Pearson residuals were carried forward to association testing.

Genotyping, trait heritability, and concordance.

Among the 6,262 genotyped saliva samples, 6,144 met strict quality control criteria. High-density genotyping and subsequent imputation via the Trans-Omics for Precision Medicine (TOPMed) server yielded ~14.3 million high-quality single nucleotide polymorphisms (SNPs) (MAF >1%, R2 >0.3). Details on quality control have been reported previously (Shrestha et al. 2025). The heritable variance (h2) of the DDE traits attributable to human genome was estimated among 5,580 unrelated participants using Genome-wide Complex Trait Analysis (GCTA) using high-quality imputed SNPs (R2>0.7); excluding SNPs with MAF<5%; and adjusting for age, sex, eight ancestry principal components and self-reported race/ethnicity (Yang et al. 2013). To further evaluate DDE’s genetic underpinning, we additionally estimated trait concordance among 682 related pairs utilizing Cohen’s kappa and intraclass correlation coefficients (ICC).

Genome-wide association analyses.

To investigate the genetic basis of DDE and interrogate potential sex dimorphism, we employed SAIGE (Zhou et al. 2018) to efficiently account for sample relatedness and case-control imbalance in a three-tiered approach (Figure 1): a) Approach 1 (main effects). Discovery GWAS (n=6,061) using linear mixed models for the quantitative traits (utilizing the aforementioned zero-inflated negative binomial Pearson residuals) and logistic mixed models for the binary traits. b) Approach 2 (joint gene-sex testing). Joint 2-degree-of-freedom (2df) testing to simultaneously evaluate genetic main effects and gene-sex (GxS) interactions, a method optimizing power for discovering loci with heterogeneous effects between sexes (Kraft et al. 2007; Manning et al. 2011). C) Approach 3 (sex-stratified models). The sample was split into male and female strata, to fit separate sex-stratified main-effects models. We subsequently calculated the P-value for difference (Pdiff) between the stratum-specific beta coefficients to formally screen for GxS interactions.

Figure 1:

Figure 1:

Open in a new tab

Summary of study design and results for primary count traits. Approach 1 conducted a genome-wide association study (GWAS) in the entire sample. Approach 2 conducted a joint 2-degree-of-freedom test to test the main and interaction effects jointly accounting for sex. Approach 3 conducted sex-specific GWAS of the developmental defects of enamel primary quantitative traits. The approaches have been detailed in the Methods section. The total number of loci is less than the sum of the boxes owing to the identification of the same loci in different traits or analyses. *Covariates include age, sex, race/ethnicity, and the first 8 principal components of ancestry. ¥ Analysis of secondary binary traits identified further 5 independent genomic risk loci. Abbreviations: DDE: Developmental Defects of Enamel, DEMA: Demarcated Opacities, DIFF: Diffuse Opacities, HYPO: Hypoplastic defects, SNP: Single Nucleotide Polymorphism, 2df: 2-degree-of-freedom, GWAS: Genome-wide association study, P: p-value.

EasyStrata was used to carry out QC, generate Manhattan and quantile-quantile (Q-Q) plots, and conduct the 1df (sex-specific), 2df (Pjoint) and the Pdiff tests (Winkler et al. 2015)

Statistical testing, annotation, and generalization.

We applied a conventional genome-wide significance threshold (P<5×10 ) to detect novel loci, and a stringent study-wide Bonferroni-corrected genome-wide statistical significance threshold (P<1.25×10 ) to account for multiple testing (i.e., total 40 models emanating from 8 phenotypes interrogated in 5 models). We excluded SNPs with effective allele sample size (effN) <100 from reporting genome-wide significant association results. We utilized FUMA GWAS for functional annotation (Watanabe et al. 2017), alongside MAGMA v1.6 for gene-centric and pathway enrichment analyses.

To evaluate the generalization of our findings, we examined the directional consistency and Bonferroni-corrected statistical significance (P<1.3×10 3) of our top SNPs against summary statistics from external cohorts: a multi-ethnic GWAS of enamel hypoplasia (Alotaibi et al. 2022; n=7,159), a childhood caries meta-analysis (Haworth et al. 2018; n=19,000), and an adult caries meta-analysis (Shungin et al. 2019; n=487,000). Finally, to investigate whether the established clinical association between DDE and early childhood caries (ECC) is driven by shared genetics, we conducted a comprehensive cross-trait pleiotropy look-up. We examined 939 trait-variant associations against primary (cavitated lesions; ICDAS≥3) and secondary (clinically-manifest; ICDAS≥1) ECC phenotypes using summary statistics from a recent multi-ancestry GWAS of ECC conducted within the same ZOE 2.0 cohort (Shrestha et al. 2025; n=6,103). This look-up systematically evaluated both main effects and gene-environment interactions (i.e., interactions with fluoride and sugary beverage consumption frequency) to estimate the proportion of loci reaching nominal or genome-wide significance.

Results

The study's analytical sample (n=6,061) included 50% females with a mean age of 53 months. Approximately half were non-Hispanic African Americans (48%) with 20% Hispanic Americans, 18% Non-Hispanic Whites, and 2% American Indian/Alaskan Natives (Appendix Table 3). To evaluate the genetic underpinning of DDE, we estimated SNP-based heritability (h2) and familial concordance. Common SNPs explained 20% of the phenotypic variance for the combined DDE quantitative trait (h2=0.20, se=0.06, P=7.9×10 ). Among distinct subtypes, heritability was highest for diffuse opacities (h2=0.16), with estimates for other traits <10% (Appendix Table 4). As expected for heritable traits, familial concordance increased with the degree of relatedness. Concordance for the combined DDE quantitative trait was highest among monozygotic twins (ICC=0.43, 95% CI: 0.15–0.65), intermediate for first-degree relatives (ICC=0.41, 95% CI: 0.30–0.50), and lowest for second- and third-degree relatives (ICC=0.19, 95% CI: 0.09–0.28) (Appendix Table 4).

Across the three analytical approaches, genomic inflation factors (λGC) ranged between 0.93 and 1.01, indicating good control of population stratification (Appendix Table 5; Appendix Figure 4). We identified 39 unique genome-wide significant loci (P<5×10 ) across all models, with five (Y RNA, RP11-157E21.2, AC007950.1, KRTAP7-1, ALDH1A1) surpassing a stringent study-wide Bonferroni-corrected threshold (P<1.25×100 ) (Figure 1). The main-effect GWAS identified 20 unique loci reaching genome-wide significance (Table 1, Figure 2). Nineteen of these loci were associated with quantitative traits, representing 20 distinct trait-locus associations (i.e., 4 for any DDE, 2 for demarcated opacities, 2 for diffuse opacities, and 12 for hypoplastic defects) (Table 1, Table 2). Notably, the Y RNA locus provided a strong association signal for both diffuse opacities and overall DDE. Moreover, RP1-281H8.3 was significantly associated with the diffuse opacities binary trait.

Table 1.

Novel loci emerging from the genome-wide association analyses conducted for developmental defects of enamel in the full sample of the ZOE 2.0 study (Approach 1)

Locus Lead SNP
(rsid)		 EA/OA EAF effN b SE P Imputed
Quantitative traits (count of affected teeth)
Developmental defects of enamel
Y_RNA * rs143210152 G/C 0.02 174 0.45 0.08 4.4E-09 Yes
DEK rs567243779 A/C 0.06 619 0.23 0.04 8.1E-09 Yes
RP4-724E13.2 rs150075871 T/C 0.01 119 0.53 0.09 2.0E-08 Yes
HBS1L rs116989668 A/G 0.01 120 0.52 0.10 4.0E-08 Yes
Demarcated opacities
SLCO4A1 rs75880235 T/C 0.01 132 0.51 0.09 1.9E-08 No
C21orf54 rs895438538 CA/C 0.03 298 0.33 0.06 4.2E-08 Yes
Diffuse opacities
YRNA * rs77669438 A/G 0.01 142 0.56 0.09 9.1E-11 Yes
CTA-21C21.1 * rs1658697761 CA/C 0.41 2914 0.11 0.02 2.0E-09 Yes
Hypoplastic defects
RP11-157E21.2 * rs192067416 G/A 0.02 205 0.46 0.07 2.4E-10 Yes
AC007950.1 * rs79505735 C/T 0.02 189 0.46 0.07 5.1E-10 Yes
RP1-167G20.1 * rs149507176 C/T 0.01 143 0.52 0.09 1.5E-09 Yes
ADAMTSL1 rs115373518 G/A 0.03 342 0.32 0.06 7.5E-09 Yes
RAVER2 chr1:64830234 T/TG 0.01 131 0.50 0.09 7.6E-09 Yes
FHIT rs73835238 G/A 0.01 129 0.52 0.09 9.1E-09 No
UBE2G2 rs115617269 A/G 0.02 172 0.43 0.08 2.0E-08 Yes
MPV17L2 rs79028494 A/G 0.04 408 0.28 0.05 2.9E-08 Yes
PPAPDC3 rs75061854 T/C 0.02 196 0.41 0.07 2.9E-08 Yes
CLUL1 rs114989924 G/A 0.02 232 0.37 0.07 3.6E-08 Yes
MAML2 rs150489138 A/G 0.01 147 0.45 0.08 4.3E-08 Yes
MTHFD1L rs11963290 T/C 0.11 1160 0.17 0.03 4.8E-08 Yes
Binary traits (case status)
Diffuse opacities
RP1-281H8.3 rs184135739 C/T 0.01 153 0.98 0.17 3.0E-09 Yes
Open in a new tab

Boldface signifies meeting the stringent study-wide Bonferroni-corrected genome-wide statistical significance threshold (P<1.25×10 )

*

Denotes SNP is also genome-wide statistically significant for the joint test (main genetic effect + gene x environment effect)

Denotes SNP is genome-wide significant for more than one trait (e.g., diffuse opacities and overall DDE) Abbreviations: Chr - chromosome; EA: Effect allele; effN: effective allele count; OA: Other allele; EAF: effect allele frequency; SNP - Single nucleotide polymorphism

Figure 2:

Figure 2:

Open in a new tab

Manhattan plots for the primary quantitative traits with genome-wide statistically significant loci (Approach 1). A) DDE, B) Demarcated opacities, C) Diffuse Opacities, D) Hypoplastic defects

Table 2.

Summary of association results for the novel genetic risk loci identified in Approaches 2 and 3

Main Discovery Male stratum Female stratum
Locus Lead SNP
(rsid)		 EAF b SE P b SE P b SE P p-Joint Pdiff¥
Quantitative traits
Developmental defects of enamel
ALDH1A1 rs1417532O5 0.02 0.29 0.07 1.2E-05 0.03 0.09 7.7E-01 0.58 0.10 9.6E-10 7.3E-09 2.5E-05
RP11-384F7.1 rs202021524 0.03 0.26 0.06 7.0E-05 0.02 0.09 8.5E-01 0.51 0.09 4.8E-08 3.4E-07 1.2E-04
TMPRSS13 rs117786109 0.02 0.31 0.07 1.2E-05 0.49 0.09 4.8E-08 0.04 0.11 7.1E-01 3.2E-07 1.2E-03
Y_RNA * rs143210152 0.02 0.45 0.08 4.4E-09 0.42 0.11 1.0E-04 0.50 0.11 7.0E-06 2.1E-08 6.0E-01
Demarcated opacities
JRKL:JRKL-AS1 rs141989840 0.02 0.30 0.06 1.2E-06 0.07 0.09 4.5E-01 0.49 0.09 1.1E-08 6.0E-08 4.5E-04
DLGAP2 rs2906574 0.98 −0.19 0.07 5.7E-03 0.17 0.10 8.8E-02 −0.53 0.10 3.4E-08 5.7E-08 3.1E-07
RP11-57H14.3 rs77466522 0.02 0.23 0.07 4.8E-04 0.58 0.10 3.2E-09 −0.07 0.09 4.1E-01 1.7E-08 7.9E-07
Diffuse opacities
GBE1 rs57583981 0.02 0.36 0.07 4.6E-07 0.10 0.10 3.2E-01 0.59 0.10 4.5E-09 2.1E-08 6.0E-04
TENM2 rs141872488 0.02 0.14 0.07 3.5E-02 −0.22 0.09 1.8E-02 0.58 0.10 1.6E-08 7.2E-09 6.4E-09
RP11-384F7.1 rs202021524 0.03 0.21 0.06 1.3E-03 −0.08 0.09 3.8E-01 0.51 0.09 3.8E-08 1.8E-07 4.4E-06
Y RNA * rs77669438 0.01 0.56 0.09 9.1E-11 0.56 0.12 1.4E-06 0.54 0.13 2.7E-05 1.3E-09 9.1E-01
CTA-21C21.1 * rs1658697761 0.41 0.11 0.02 2.0E-09 0.12 0.03 4.2E-06 0.11 0.03 1.1E-04 1.5E-08 6.5E-01
Hypoplastic defects
KRTAP7-1 rs111896551 0.02 0.29 0.06 2.6E-06 0.01 0.09 8.8E-01 0.59 0.09 5.4E-10 4.2E-09 8.2E-06
CCSER1 rs139393106 0.02 0.30 0.06 3.3E-06 0.07 0.09 4.1E-01 0.55 0.10 8.0E-09 4.3E-08 2.7E-04
PLEKHD1 rs147589787 0.02 0.34 0.07 8.5E-07 0.10 0.10 3.2E-01 0.56 0.10 2.3E-08 1.0E-07 9.4E-04
MAST4 rs1753671813 0.02 0.28 0.07 6.5E-05 −0.02 0.10 8.1E-01 0.55 0.10 4.5E-08 3.1E-07 4.8E-05
NF1 rs147836184 0.02 0.27 0.07 3.0E-04 −0.10 0.11 3.7E-01 0.58 0.11 4.6E-08 2.2E-07 9.0E-06
CCDC18 rs187188755 0.02 0.24 0.07 8.2E-04 0.55 0.10 2.4E-08 −0.14 0.11 2.2E-01 7.9E-08 4.1E-06
TENM2 rs62383285 0.02 0.38 0.07 4.2E-07 0.56 0.10 4.9E-08 0.15 0.11 2.0E-01 1.5E-07 6.6E-03
AC007950.1 * rs79505735 0.02 0.46 0.07 5.1E-10 0.34 0.11 2.1E-03 0.56 0.11 1.4E-07 8.1E-09 1.4E-01
RP1-167G20.1 * rs149507176 0.01 0.52 0.09 1.5E-09 0.69 0.13 1.4E-07 0.36 0.12 3.2E-03 1.2E-08 6.5E-02
RP11-157E21.2 * rs192067416 0.02 0.46 0.07 2.4E-10 0.43 0.11 7.6E-05 0.47 0.10 5.5E-06 1.3E-08 8.1E-01
Binary traits
Developmental defects of enamel
SLC19A1 rs62214342 0.32 0.12 0.04 5.6E-03 0.34 0.06 3.9E-08 −0.08 0.06 1.6E-01 1.0E-07 6.5E-07
Demarcated opacities
CTD-2644I21.1 rs2776544 0.37 0.18 0.04 1.3E-05 0.34 0.06 3.2E-09 0.01 0.06 8.1E-01 2.4E-08 5.5E-05
RP11-203H2.1 rs17143504 0.12 −0.25 0.06 1.0E-05 −0.44 0.08 3.9E-08 −0.06 0.08 4.3E-01 2.0E-07 1.0E-03
Diffuse opacities
LINC00670 rs11078084 0.56 −0.16 0.04 1.1E-05 −0.29 0.05 2.1E-08 −0.04 0.05 4.4E-01 1.1E-07 7.8E-04
Genome-wide significant Loci for p-difference test
Quantitative trait (Demarcated opacities)
SLC2A3 rs61921622 0.09 0.05 0.03 1.8E-01 −0.14 0.05 3.2E-03 0.25 0.05 8.4E-07 7.1E-08 1.8E-08
Binary trait (Developmental defects of enamel)
snoU13 rs7228955 0.52 −0.04 0.04 3.0E-01 −0.28 0.06 3.2E-06 0.17 0.06 3.0E-03 2.4E-07 4.0E-08
Open in a new tab

Underlined loci meet the stringent study-wide Bonferroni-corrected genome-wide statistical significance threshold (P<1.25×10 ).

*

Denotes SNP is also genome-wide statistically significant for the main discovery analysis (Approach 1) The genome-wide statistically significant P-values (P<5x10−8) are in bold.

Locus is genome-wide significant for more than one trait (e.g., diffuse opacities and overall DDE)

Locus emerged as genome-wide significant for two distinct traits (i.e., diffuse opacities-females and hypoplastic defects-males)

¥

Bonferroni corrected p-value for the Pdiff = 0.05/26=1.9x10−3

Abbreviations: Chr - chromosome; EA: Effect allele; effN: effective allele count; OA: Other allele; EAF: effect allele frequency; 2df: 2-degree-of-freedom, Pdiff: p-value for difference between betas or interaction effect; Pjoint: Joint 2-degree-of-freedom test

The joint 2-degree-of-freedom tests (Approach 2) and sex-stratified models (Approach 3) provided additional biological sex-informed insights, nominating 19 additional unique loci (e.g., ALDH1A1, TENM2, and DLGAP2). Sex-stratified analyses illuminated signals driven by male or female strata; e.g., RP11-384F7.1 reached genome-wide significance exclusively in the female stratum for both the combined DDE and diffuse opacities traits (Table 2). Nineteen loci exhibited statistically different effect sizes between males and females using a Bonferroni correction (Pdiff<2×10 3) (Figure 3), underscoring a sexual dimorphism in DDE genetics. Strikingly, three loci demonstrated significant sex-specific heterogeneity at a strict genome-wide association threshold: SLC2A3 (Pdiff=1.8×10 ), TENM2 (Pdiff=6.4×10 ), and snoU13 (Pdiff=4.0×10 ).

Figure 3:

Figure 3:

Open in a new tab

Forest plots demonstrating heterogeneity of genetic effect in different sexes for the 4 primary traits. (A) Developmental defects of enamel count, (B) Demarcated opacities, (C) Diffuse opacities, (D) Hypoplastic defects. The triangles and circles represent effect allele estimates of association and the error bars represent the corresponding 95% confidence intervals.

To prioritize variants, we annotated significant loci using CADD and RegulomeDB (RDB) scores (Appendix Tables 6 and 7) and tissue-specific gene expression (Research Data Table 1). The biological roles of nearby (~250Kb) protein-coding genes are detailed in Research Data Tables 2 and 3. We identified multiple loci harboring highly deleterious, non-coding regulatory variants that serve as prime candidates for future in vitro functional validation assays. For diffuse opacities, variants in high LD with CTA-21C21.1 had high CADD scores (e.g., rs566606 CADD: 21.4). Similarly, for hypoplastic defects, potentially functional regulatory variants were identified at CLUL1 (rs114989924; CADD: 13.0) and MAML2 (rs142941402; CADD: 17.3).

Gene-based testing via MAGMA mapped SNPs to 18,273 protein-coding genes and identified CDH10 as genome-wide significant. CDH10 encodes a cadherin protein, plays a crucial role in synaptic adhesion, axon guidance, and maintaining tissue structure, and has been previously associated with autism spectrum disorder (ASD) (Redies et al. 2012). Further pathway enrichment revealed links to neurodevelopmental conditions, with the DLGAP2 locus (synapse organization) emerging for demarcated opacities, and the NF1 locus (neurofibromatosis type-1) emerging for hypoplastic defects. Gene-set analyses identified three sets reaching Bonferroni-corrected significance for hypoplastic defects "GO_bp: go_elastic_fiber_assembly" and "Curated_gene_sets: valk_aml_with_t_8_21_translocation" from Approach 1, and "GO_cc: go_stereocilium_membrane", and enrichment of the prioritized genes from Approach 1 for hypoplastic defects within the facial emotion recognition gene set (Appendix Table 10).

No signal generalized to the external cohorts with GWAS of enamel hypoplasia and dental caries (Appendix Tables 11-14) using strict multiple-testing-corrected criteria, likely reflecting phenotypic and demographic differences between the cohorts. However, one locus, SLC19A1, was nominally significant and directionally consistent with a previous GLIDE meta-analysis of primary teeth. While exact variants did not broadly generalize, a regional examination (±120 Kb) of dental caries GWAS data identified strong, co-localizing signals at two loci on chromosome 15 (AC007950.1, GLIDE adult caries, P=1.3x10−26) and chromosome 6 (DEK, clinical decay, P=1.8x10−7) (Appendix Figure 6). Cross-trait look-ups were carried out to illustrate the contribution of GWAS signals across multiple distinct DDE clinical subtypes and the overall DDE trait (Appendix Table 15).

To evaluate possibly shared genetic susceptibility underlying the established DDE-ECC clinical association, we conducted a cross-trait pleiotropy look-ups evaluating 939 DDE trait-SNP associations against ECC phenotypes (Shrestha et al. 2025). Examining main effects and gene-environment interactions, we found no greater-than-chance associations, with only 40 SNP look-ups (4.3%) reaching nominal significance (P<0.05) and none reaching genome-wide significance, as expected under the null hypothesis of no association. This finding is supportive of the notion that a DDE-ECC clinical and epidemiologic association is most likely structural (i.e., tooth surface susceptibility) rather than driven by pleiotropy (i.e., shared genetics).

Discussion

To further our understanding of the polygenic underpinning of developmental defects of the enamel, we conducted the largest multi-ancestry GWAS of primary dentition DDE. Our findings suggest a considerable genetic basis, with common SNPs explaining approximately 20% of the phenotypic variance of DDE. By integrating main-effect models with joint gene-sex interaction testing and sex-stratified analyses, we identified a total of 39 unique genome-wide significant risk loci. Some of the newly identified loci (e.g., in CTA-21C21.1 and MAML2) harbor SNPs with exceptionally high CADD scores, representing highly deleterious, non-coding regulatory variants, which we propose as prime candidates for future functional validation assays. Crucially, the results support shared biological mechanisms with odontogenesis, hematopoiesis, and neurodevelopment, a profound sexual dimorphism in DDE susceptibility, and demonstrate that the established clinical association between DDE and childhood caries is likely driven by structural enamel defects rather than genetic pleiotropy.

The main discovery analysis in the analysis sample (Approach 1) identified 19 loci with several harboring genes with highly plausible biological roles in amelogenesis and systemic health. For the combined DDE trait, the HBS1L locus emerged as highly significant. HBS1L encodes a GTP-binding elongation factor associated with sickle cell disease (Safran et al. 2021), aligning with epidemiological reports of a higher DDE prevalence among patients with sickle cell disease (Lopes et al. 2018). Located ~170 Kb from the lead variant near HBS1L is MYB, a transcription regulator that is strongly expressed in odontoblasts and ameloblasts during embryonic tooth development, with a well-documented role in tooth mineralization (Matalová et al. 2011). Among the loci identified for hypoplastic defects, FHIT is a tumor suppressor previously associated with odontogenic tumors and non-syndromic cleft lip/palate with hypodontia (Malčić et al. 2008; Ghazali et al. 2021). We also identified signals near TSPEAR, which regulates the Notch signaling pathway; the latter is a critical driver of ameloblast derivation from the ectoderm and overall enamel development (Cai et al. 2011; Peled et al. 2016). Furthermore, the diffuse opacities-associated LATS1 encodes a kinase regulating the Hippo signaling pathway, which is essential for the development of the enamel organ (Wang et al. 2022). Additional loci included IL10RB (near C21orf54), previously linked to MIH (Elzein et al. 2022), and NTSR1 (near SLCO4A1), a neurotensin mediator associated with dental care-related anxiety (Zhou et al. 2022). For the secondary binary traits, COL18A1 (near SLC19A1) and DLK1 have been implicated in enamel matrix maturation and ameloblast proliferation, respectively (Väänänen et al. 2004; Qi et al. 2018).

Differences in genetic effects between sexes have been reported for several common-complex dental, oral, and craniofacial diseases including ECC and early-onset periodontitis (Freitag-Wolf et al. 2021; Shrestha et al. 2025). By leveraging joint 2-degree-of-freedom tests and sex-stratified GWAS, we identified 19 additional loci marked by sex-specific heterogeneity. For example, ALDH1A1, which encodes an enzyme responsible for retinoic acid synthesis, strongly emerged in female-specific and joint models. ALDH1A1 has been shown to be downregulated during the mid-late maturation stage of enamel formation relative to the secretory stage, highlighting a sex-mediated biological effect in amelogenesis (Lacruz et al. 2012). Similarly, TENM2 exhibited distinct risk profiles between males and females and has been previously shown to be associated with periodontal pathogen colonization and moderate periodontitis (Divaris et al. 2012); (Hong et al. 2015). The finding of 19 loci exhibiting statistically different effect sizes between males and females (e.g., SLC2A3, snoU13, and TENM2 reaching genome-wide significance for interaction) underscores the notion that analyzing sexes in aggregate may mask potentially informative genetic heterogeneity and supports the necessity of treating sex as a biological variable in dental genomics (Shrestha et al. 2025).

The enamel organ and the central nervous system share an ectodermal embryonic origin, and it is thus unsurprising that our results demonstrate a genetic overlap of DDE with neurodevelopmental disorders, with gene-based association testing and pathway enrichment analyses strongly supporting this overlap. For example, we identified CDH10 (diffuse opacities) and DLGAP2 (demarcated opacities), both of which play crucial roles in synapse organization and are established candidate genes for ASD (Redies et al. 2012) (Rasmussen et al. 2017). This finding provides a genomic basis for clinical observations showing a higher incidence of enamel defects in children with intellectual disabilities and ASD (Modrić et al. 2016). Similarly, we identified the NF1 locus for hypoplastic defects; neurofibromatosis type-1 is a condition known to present with an elevated prevalence of DDE (Santoro et al. 2020). While additional research is required to substantiate and elucidate these associations, these shared genetic pathways support the notion that disruptions during overlapping developmental windows may manifest pleiotropically across multiple ectoderm-derived tissues.

Clinically, individuals and teeth with DDE are known to be at a high risk of developing ECC (Caufield et al. 2012; AAPD 2025). As demonstrated by our cross-trait pleiotropy look-up against ECC phenotypes (Shrestha et al. 2025), this established clinical association is almost certainly driven by structural vulnerability rather than shared genetics. Because DDE constitutes altered enamel formation primarily in utero or during early infancy, affected teeth erupt with pre-existing structural deficits. This pre-existing structural susceptibility rather than a shared genetic susceptibility, is what is likely driving ECC susceptibility when faced with a cariogenic, post-eruptive, dysbiotic biofilms.

None of our 39 genome-wide significant loci generalized in an external multi-ethnic GWAS of enamel hypoplasia (Alotaibi et al. 2022). The lack of replication is likely a consequence of a fundamental difference in phenotypes and underlying biology: Alotaibi et al. investigated permanent teeth in an older cohort (7–82 years), whereas the present report exclusively focused on primary teeth of 3-5-year-olds. Because primary enamel forms in utero under maternal physiological influences, whereas permanent enamel mineralizes post-natally, their genetic underpinnings likely differ significantly. We did, however, observe nominal significance and directional consistency for SLC19A1 in a GLIDE meta-analysis of primary teeth, as well as a nominal overlap for rs13058467 near SCUBE1, a suggestive MIH locus previously reported by (Kühnisch et al. 2014) (Appendix Table 16).

Strengths of this study include its multi-ancestry, community-based design, which improves statistical power, refines signals across populations, and enhances public health relevance (Li and Keating 2014). The rigorous modeling of zero-inflated quantitative traits and the large sample size represent substantial methodological improvements over prior GWAS of DDE. The innovative incorporation of joint 2df testing maximized our power to uncover latent heritability driven by gene-sex interactions (Manning et al. 2011). However, several limitations must be acknowledged. The scarcity of comparable primary dentition DDE GWAS data prevented validation in an independent, age-matched replication cohort. Additionally, because severe forms of DDE (e.g., widespread and severe hypoplastic defects of the enamel) can rapidly undergo post-eruptive breakdown and serve as the basis of extensive caries lesion development, it is conceivable that some DDE phenotypes were masked by ECC during clinical examinations, leading to an under-estimation of the phenotype. Finally, some of our identified loci are driven by less common variants (MAF between 1–5%), increasing the potential for false positives. While we identified non-coding regulatory variants with exceptionally high CADD scores indicative of deleterious regulatory effects (e.g., in CTA-21C21.1, CLUL1, and MAML2), functional validation via in vitro and in vivo mechanistic experiments is warranted to confirm their roles in amelogenesis.

In conclusion, this study reports valuable evidence on the genomic architecture of developmental defects of enamel in the primary dentition. It highlights 39 novel candidate loci, addressing a knowledge gap in the genomic basis of amelogenesis, and offers several plausible links between systemic neurodevelopmental and hematopoietic pathways and ECC. At the same time, our data support the notion that the clinical and epidemiological link between DDE and ECC is likely driven by structural enamel defects rather than genetic pleiotropy. Importantly, the profound sex-specific genetic effects reported in this study underscore the importance of treating sex as a biological variable in dental genomics investigations. Upon further functional validation, these discoveries have the potential to enhance our biological understanding of enamel development, augment ECC risk assessment tools, and advance future applications in precision pediatric oral health.

Supplementary Material

Supplement 1
media-1.docx (4.3MB, docx)
Supplement 2
media-2.docx (821.1KB, docx)
Supplement 3
media-3.xlsx (319.8KB, xlsx)

ACKNOWLEDGEMENTS

This work was supported by research grants from the National Institutes of Health: National Institute for Dental and Craniofacial Research U01DE025046 (K.D., A.G.F.Z, D.Y.L, J.S.P, G.D.S., K.E.N) and National Human Genome Research Institute X01HG010871 (K.D.). K.D. and P.S. are supported by the National Institute of Dental and Craniofacial Research (NIDCR), National Institutes of Health (NIH), under award R01DE034410. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders.

The authors thank CIDR investigators and staff at Johns Hopkins University for carrying out genotyping and imputation for the project with support from a resource-allocation grant NIH/NIHGR X01-HG010871; Dr. Patricia V. Basta and her team at the UNC-Chapel Hill Biospecimen Processing facility for the accessioning, storage, and disbursement of the saliva and extracted nucleic acid samples in the ZOE studies; all study participants and their families for their contributions.

Footnotes

Competing Interests

During the preparation of this manuscript, John Preisser served on a data safety and monitoring board of a study funded by NIDCR. The remaining authors declare no competing interests.

Data Availability

The genotype and phenotype datasets from ZOE 2.0 are publicly available through phs002232.v1.p1 “TOPDECC-Trans-omics for Precision Dentistry and Early Childhood Caries: Genome-Wide Genotyping (CIDR) and Microbiome in the ZOE 2.0 Study”. Genomic summary results of the GLID-adults study are available at: https://data.bris.ac.uk/data/dataset/2j2rqgzedxlq02oqbb4vmycnc2 and GLIDE-children at: https://data.bris.ac.uk/data/dataset/108de7fd6f8ff188ebb9ac07fe77bfb5.

References

  1. AAPD. 2025. Caries-risk assessment and management for infants, children, and adolescents. In: The Reference Manual of Pediatric Dentistry. American Academy of Pediatric Dentistry. p. 325–31. [Google Scholar]
  2. Alotaibi RN, Howe BJ, Moreno Uribe LM, Sanchez C, Deleyiannis FWB, Padilla C, Poletta FA, Orioli IM, Buxó CJ, Wehby GL, et al. 2022. Genetic Analyses of Enamel Hypoplasia in Multiethnic Cohorts. Hum Hered. 87(2):34–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Caufield PW, Li Y, Bromage TG. 2012. Hypoplasia-associated severe early childhood caries-a proposed definition. J Dent Res. 91(6):544–550. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Divaris K, Monda KL, North KE, Olshan AF, Lange EM, Moss K, Barros SP, Beck JD, Offenbacher S. 2012. Genome-wide Association Study of Periodontal Pathogen Colonization. J Dent Res. 91(7 Suppl):S21–S28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Divaris K, Slade GD, Ferreira Zandona AG, Preisser JS, Ginnis J, Simancas-Pallares MA, Agler CS, Shrestha P, Karhade DS, Ribeiro A de A, et al. 2020. Cohort profile: Zoe 2.0—a community-based genetic epidemiologic study of early childhood oral health. Int J Environ Res Public Health. 17(21):1–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Freitag-Wolf S, Munz M, Junge O, Graetz C, Jockel-Schneider Y, Staufenbiel I, Bruckmann C, Lieb W, Franke A, Loos BG, et al. 2021. Sex-specific genetic factors affect the risk of early-onset periodontitis in Europeans. J Clin Periodontol. 48(11):1404–1413. [DOI] [PubMed] [Google Scholar]
  7. Gerreth K, Zaorska K, Zabel M, Nowicki M, Borysewicz-Lewicka M. 2018. Significance of genetic variations in developmental enamel defects of primary dentition in Polish children. Clin Oral Investig. 22(1):321–329. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Ginnis J, Ferreira Zandoná AG, Slade GD, Cantrell J, Antonio ME, Pahel BT, Meyer BD, Shrestha P, Simancas-Pallares MA, Joshi AR, et al. 2019. Measurement of Early Childhood Oral Health for Research Purposes: Dental Caries Experience and Developmental Defects of the Enamel in the Primary Dentition. Methods Mol Biol. 1922:511–523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Hong KW, Shin MS, Ahn YB, Lee HJ, Kim HD. 2015. Genomewide association study on chronic periodontitis in Korean population: Results from the Yangpyeong health cohort. J Clin Periodontol. 42(8):703–710. [DOI] [PubMed] [Google Scholar]
  10. Hu JCC, Hu Y, Smith CE, McKee MD, Wright JT, Yamakoshi Y, Papagerakis P, Hunter GK, Feng JQ, Yamakoshi F, et al. 2008. Enamel defects and ameloblast-specific expression in Enam knock-out/lacZ knock-in mice. J Biol Chem. 283(16):10858–10871. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Jälevik B, Klingberg GA. 2002. Dental treatment, dental fear and behaviour management problems in children with severe enamel hypomineralization of their permanent first molars. Int J Paediatr Dent. 12(1):24–32. [PubMed] [Google Scholar]
  12. Kraft P, Yen YC, Stram DO, Morrison J, Gauderman WJ. 2007. Exploiting gene-environment interaction to detect genetic associations. Hum Hered. 63(2):111–119. [DOI] [PubMed] [Google Scholar]
  13. Kühnisch J, Thiering E, Heitmüller D, Tiesler CMT, Grallert H, Heinrich-Weltzien R, Hickel R, Heinrich J. 2014. Genome-wide association study (GWAS) for molar-incisor hypomineralization (MIH). Clin Oral Investig. 18(2):677–682. [DOI] [PubMed] [Google Scholar]
  14. Lacruz RS, Smith CE, Bringas P, Chen YB, Smith SM, Snead ML, Kurtz I, Hacia JG, Hubbard MJ, Paine ML. 2012. Identification of novel candidate genes involved in mineralization of dental enamel by genome-wide transcript profiling. J Cell Physiol. 227(5):2264–2275. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Li YR, Keating BJ. 2014. Trans-ethnic genome-wide association studies: Advantages and challenges of mapping in diverse populations. Genome Med. 6(10). [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Lunardelli SE, Lunardelli AN, Martins LGT, Traebert E, Traebert J. 2024. Prenatal, neonatal and postnatal factors and the developmental defects of dental enamel. Rev Paul Pediatr. 42. [accessed 2025 Nov 4]. https://pubmed.ncbi.nlm.nih.gov/37646751/. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Manning AK, LaValley M, Liu CT, Rice K, An P, Liu Y, Miljkovic I, Rasmussen-Torvik L, Harris TB, Province MA, et al. 2011. Meta-analysis of gene-environment interaction: Joint estimation of SNP and SNP × environment regression coefficients. Genet Epidemiol. 35(1):11–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Marshman Z, Rodd HD. 2015. The Psychosocial Impacts of Developmental Enamel Defects in Children and Young People. In: Planning and Care for Children and Adolescents with Dental Enamel Defects: Etiology, Research and Contemporary Management. Springer, Berlin, Heidelberg. p. 85–97. [Google Scholar]
  19. Modrić VE, Verzak Ž, Karlović Z. 2016. Developmental Defects of Enamel in Children with Intellectual Disability. Acta Stomatol Croat. 50(1):65–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Núñez SM, Chun YHP, Ganss B, Hu Y, Richardson AS, Schmitz JE, Fajardo R, Yang J, Hu JC-C, Simmer JP. 2016. Maturation stage enamel malformations in Amtn and Klk4 null mice. Matrix Biol. 52–54:219–233. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Rasmussen AH, Rasmussen HB, Silahtaroglu A. 2017. The DLGAP family: Neuronal expression, function and role in brain disorders. Mol Brain. 10(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Redies C, Hertel N, Hübner CA. 2012. Cadherins and neuropsychiatric disorders. Brain Res. 1470:130–144. [DOI] [PubMed] [Google Scholar]
  23. Robles MJ, Ruiz M, Bravo-Perez M, González E, Peñalver MA. 2013. Prevalence of enamel defects in primary and permanent teeth in a group of schoolchildren from Granada (Spain). Med Oral Patol Oral Cir Bucal. 18(2):e187. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Santoro R, Santoro C, Loffredo F, Romano A, Perrotta S, Serpico R, Lauritano D, Lucchese A. 2020. Oral clinical manifestations of neurofibromatosis type 1 in children and adolescents. Appl Sci. 10(14):4687. [Google Scholar]
  25. Seow WK, Ford D, Kazoullis S, Newman B, Holcombe T. 2011. Comparison of Enamel Defects in the Primary and Permanent Dentitions of Children from a Low-fluoride District in Australia. Pediatr Dent. 33(3):207–212. [PubMed] [Google Scholar]
  26. Shrestha P, Graff M, Gu Y, Wang Y, Avery CL, Ginnis J, Simancas-Pallares MA, Zandoná FAG, Ahn HS, Nguyen KN, et al. 2025. Multi-ancestry Genome-Wide Association Study of Early Childhood Caries. J Dent Res. 10(3):280–289. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Wang T, Li K, Liu H, Luo E. 2022. Focusing on Hippo Pathway in Stem Cells of Oral Origin, Enamel Formation and Periodontium Regeneration. Organogenesis. 18(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Watanabe K, Taskesen E, Van Bochoven A, Posthuma D. 2017. Functional mapping and annotation of genetic associations with FUMA. Nat Commun. 8(1):1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Winkler TW, Kutalik Z, Gorski M, Lottaz C, Kronenberg F, Heid IM. 2015. EasyStrata: Evaluation and visualization of stratified genome-wide association meta-analysis data. Bioinformatics. 31(2):259–261. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Yang J, Lee SH, Goddard ME, Visscher PM. 2013. Genome-wide complex trait analysis (GCTA): Methods, data analyses, and interpretations. Methods Mol Biol. 1019:215–236. [DOI] [PubMed] [Google Scholar]
  31. Zhou W, Nielsen JB, Fritsche LG, Dey R, Gabrielsen ME, Wolford BN, LeFaive J, VandeHaar P, Gagliano SA, Gifford A, et al. 2018. Efficiently controlling for case-control imbalance and sample relatedness in large-scale genetic association studies. Nat Genet. 50(9):1335–1341. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1
media-1.docx (4.3MB, docx)
Supplement 2
media-2.docx (821.1KB, docx)
Supplement 3
media-3.xlsx (319.8KB, xlsx)

Data Availability Statement

The genotype and phenotype datasets from ZOE 2.0 are publicly available through phs002232.v1.p1 “TOPDECC-Trans-omics for Precision Dentistry and Early Childhood Caries: Genome-Wide Genotyping (CIDR) and Microbiome in the ZOE 2.0 Study”. Genomic summary results of the GLID-adults study are available at: https://data.bris.ac.uk/data/dataset/2j2rqgzedxlq02oqbb4vmycnc2 and GLIDE-children at: https://data.bris.ac.uk/data/dataset/108de7fd6f8ff188ebb9ac07fe77bfb5.

Articles from medRxiv are provided here courtesy of Cold Spring Harbor Laboratory Preprints

RESOURCES