Hyperpigmentation is the most common skin change during pregnancy, observed in 90% of women (Barnawi et al., Reference Barnawi, Barnawi and Alamri2021). Linea nigra (LN) is a hyper-pigmented vertical line running down the middle of the abdomen that appears in the first trimester in 54−87% of pregnant women, predominantly in women with darker compared to fair-skinned complexions (Barnawi et al., Reference Barnawi, Barnawi and Alamri2021). LN can also occur in men, children, and non-pregnant women, and may reflect profound changes of endocrine origin (George et al., Reference George, Shittu, Enwerem, Wachtel and Kuti2005).
Pigmentation traits are highly heritable, with common and rare genetic polymorphisms contributing to variation in skin, eye, and hair color (Hysi et al., Reference Hysi, Valdes, Liu, Furlotte, Evans, Bataille, Visconti, Hemani, McMahon, Ring, Smith, Duffy, Zhu, Gordon, Medland, Lin, Willemsen, Jan Hottenga, Vuckovic and Spector2018; Lui et al., Reference Lin, Mbarek, Willemsen, Dolan, Fedko, Abdellaoui, de Geus, Boomsma and Hottenga2015; Simcoe et al., Reference Simcoe, Valdes, Liu, Furlotte, Evans, Hemani, Ring, Smith, Duffy, Zhu, Gordon, Medland, Vuckovic, Girotto, Sala, Catamo, Concas, Brumat, Gasparini and Hysi2021). However, heritability estimates for pregnancy-related LN (PLN) have not been established and the identity of genes possibly underpinning the occurrence of PLN are not known. Several advances have been made in understanding the differences in gene expression and pigmentation among individuals of Asian, European and African ancestry (Yin et al., Reference Yin, Coelho, Ebsen, Smuda, Mahns, Miller, Beer, Kolbe and Hearing2014). In African genomes, variants in or near SLC24A5, MFSD12, DDB1, HERC2/OCA2 and TMEM138 have been associated with dark epidermal pigmentation (Crawford et al., Reference Crawford, Kelly, Hansen, Beltrame, Fan, Bowman, Jewett, Ranciaro, Thompson, Lo, Pfeifer, Jensen, Campbell, Beggs, Hormozdiari, Mpoloka, Mokone, Nyambo, Meskel and Tishkoff2017), with QPCT, TRIM63 and MITF underlying hyper-pigmented skin (Yin et al., Reference Yin, Coelho, Valencia, Ebsen, Mahns, Smuda, Miller, Beer, Kolbe and Hearing2015). Variants in MFSD12, OCA2 and MC1R regions affect skin and eye color variation in Asian populations (Adhikari et al., Reference Adhikari, Mendoza-Revilla, Sohail, Fuentes-Guajardo, Lampert, Chacón-Duque, Hurtado, Villegas, Granja, Acuña-Alonzo, Jaramillo, Arias, Lozano, Everardo, Gómez-Valdés, Villamil-Ramírez, Silva de Cerqueira, Hunemeier, Ramallo and Ruiz-Linares2019; Eaton et al., Reference Eaton, Edwards, Krithika, Cook, Norton and Parra2015; Yamaguchi et al., Reference Yamaguchi, Watanabe, Kawaguchi, Sato, Naka, Shindo, Moromizato, Aoki, Ishida and Kimura2012). In Europeans, genes associated with skin colour variation include ASIP, SLC45A2, IRF4, HERC2/OCA2, EIF2S2, GSS and MC1R (Liu et al., Reference Liu, Visser, Duffy, Hysi, Jacobs, Lao, Zhong, Walsh, Chaitanya, Wollstein, Zhu, Montgomery, Henders, Mangino, Glass, Bataille, Sturm, Rivadeneira, Hofman and Kayser2015), one of the key genes involved in the regulation of mammalian skin, eye and hair color (Pavan & Sturm, Reference Pavan and Sturm2019). Of note, genetic variations in many pigmentation genes, including MC1R, have been associated with other visible traits, especially skin aging, various types of skin cancer, freckling and the appearance of moles (melanocytic nevi; Duffy et al., Reference Duffy, Zhu, Li, Sanna, Iles, Jacobs, Evans, Yazar, Beesley, Law, Kraft, Visconti, Taylor, Liu, Wright, Henders, Bowdler, Glass, Ikram and Martin2018; Jacobs et al., Reference Jacobs, Hamer, Gunn, Deelen, Lall, van Heemst, Uh, Hofman, Uitterlinden, Griffiths, Beekman, Slagboom, Kayser, Liu and Nijsten2015; Landi et al., Reference Landi, Bishop, MacGregor, Machiela, Stratigos, Ghiorzo, Brossard, Calista, Choi, Fargnoli, Zhang, Rodolfo, Trower, Menin, Martinez, Hadjisavvas, Song, Stefanaki, Scolyer and Law2020; Law et al., Reference Law, Medland, Zhu, Yazar, Viñuela, Wallace, Shekar, Duffy, Bataille, Glass, Spector, Wood, Gordon, Barbour, Henders, Hewitt, Montgomery, Sturm, Mackey and Macgregor2017; Maccioni et al., Reference Maccioni, Rachakonda, Scherer, Bermejo, Planelles, Requena, Hemminki, Nagore and Kumar2013), implying shared genetic liability amongst pigmentation phenotypes.
Polygenic risk score (PRS) analysis is a useful statistical approach for predicting disease susceptibility and detecting associations both within and across traits. To date, a number of studies have explored shared genetic pathways associated with pigmentation phenotypes. Loci containing important pigmentary genes have been associated with various skin cancers (Roberts et al., Reference Roberts, Asgari and Toland2019), with, for example, genetic variants in the MC1R gene having a major effect on melanoma susceptibility (Cust et al., Reference Cust, Drummond, Kanetsky, Goldstein, Barrett, MacGregor, Law, Iles, Bui, Hopper, Brossard, Demenais, Taylor, Hoggart, Brown, Landi, Newton-Bishop, Mann, Bishop and Bishop2018). Additional loci associated with skin color, mole count, telomere length and cellular senescence (skin aging) also harbor genetic variants implicated in melanoma, and contribute to the prediction of melanoma risk when included in a PRS (Cust et al., Reference Cust, Drummond, Kanetsky, Goldstein, Barrett, MacGregor, Law, Iles, Bui, Hopper, Brossard, Demenais, Taylor, Hoggart, Brown, Landi, Newton-Bishop, Mann, Bishop and Bishop2018). While these reports support the notion that pigmentation traits share an underlying genetic etiology, genetic relationships between pigmentation-related traits and PLN have not been investigated, and the contribution of MC1R to these relationships is unknown. To address these questions, here we present our exploration of the genetics underlying PLN. We performed the first genomewide association study (GWAS), to our knowledge, for PLN, estimated PLN prevalence and heritability, and assessed its relationships with other pigmentation phenotypes, specifically skin color and mole count.
Materials and Methods
PLN Study Population and Phenotyping
We used phenotype and single nucleotide polymorphism (SNP) genotype data collected within the Nausea and Vomiting during Pregnancy (NVP; Colodro-Conde et al., Reference Colodro-Conde, Cross, Lind, Painter, Gunst, Jern, Johansson, Lund Maegbaek, Munk-Olsen, Nyholt, Ordoñana, Paternoster, Sánchez-Romera, Wright and Medland2017) and Women’s Health and Lifestyle (WHLS) studies (Painter et al., Reference Painter, Grasby, Actkins, Blostein, Straub, Madrid-Valero, Laisk, Lu, Karczewski, Westergaard, Steinthorsdottir, Thorleifsson, Lu, Bragantini, Corfield, Morosoli, Olsen, Pandeya, Brøns and Medland2025) conducted between 2013 and 2016 at the QIMR Berghofer Medical Research Institute (QIMR Berghofer), Brisbane, Australia. The detailed study protocol is included in the Supplementary Methods. Both projects were approved by the QIMR Berghofer Human Research Ethics Committee (under project numbers P1515 and P199), and all participants provided informed consent to participate in these studies. Women who self-reported PLN in any of their pregnancies (28%, n = 275) were considered as cases and women who had been pregnant but did not experience PLN (72%, n = 701) were included as controls.
Genotyping and Imputation
Participants recruited within the NVP study were genotyped using the Illumina Global Screening Array. Participants from the WHLS study had previously been genotyped using the Illumina 317K, 370K, 610K, Core Exome or Omni-arrays (Medland et al., Reference Medland, Nyholt, Painter, McEvoy, McRae, Zhu, Gordon, Ferreira, Wright, Henders, Campbell, Duffy, Hansell, Macgregor, Slutske, Heath, Montgomery and Martin2009). Stringent quality control procedures on the genotype data were conducted before GWAS analysis using PLINK 1.9b4 (Chang et al., Reference Chang, Chow, Tellier, Vattikuti, Purcell and Lee2015). Individuals and -SNPs- with genotype call rates < 95% were excluded from the analysis. We also excluded SNPs with a minor allele frequency (MAF) < 1% or significant (p value <1 × 10-6) deviation from Hardy-Weinberg equilibrium. Heterozygosity checks were conducted and only individuals within ±3 standard deviations (SD) of the sample mean heterozygosity rate were included in the GWAS analysis. Population stratification was examined using principal components of allele frequencies, with all individuals being within ±6 SD of the mean (Medland et al., Reference Medland, Nyholt, Painter, McEvoy, McRae, Zhu, Gordon, Ferreira, Wright, Henders, Campbell, Duffy, Hansell, Macgregor, Slutske, Heath, Montgomery and Martin2009).
Genotype imputation was carried out on the Michigan Imputation Server (Das et al., Reference Das, Forer, Schönherr, Sidore, Locke, Kwong, Vrieze, Chew, Levy, McGue, Schlessinger, Stambolian, Loh, Iacono, Swaroop, Scott, Cucca, Kronenberg, Boehnke, Abecasis and Fuchsberger2016) using the Haplotype Reference Consortium (HRC) reference panel (Version r1.1 2016) (McCarthy et al., Reference McCarthy, Das, Kretzschmar, Delaneau, Wood, Teumer, Kang, Fuchsberger, Danecek, Sharp, Luo, Sidore, Kwong, Timpson, Koskinen, Vrieze, Scott, Zhang and Durbin2016). Post-imputation quality control included the exclusion of variants with an MAF < 1% or low imputation quality (R 2 < 0.6) from the imputed dataset. In total, 7,802,598 autosomal and X chromosome SNPs were included in the PLN GWAS analysis.
Single Nucleotide Polymorphism (SNP) Heritability
The proportion of phenotypic variance in PLN explained by all SNPs was estimated using the GREML (genomewide complex trait analysis with genome-based restricted maximum likelihood) method in the GCTA genetic analysis software package (Yang et al., Reference Yang, Lee, Goddard and Visscher2011).
Genomewide Association Analysis of PLN
To assess the association between common genetic variants and PLN, a genome-wide association analysis including 275 cases (women self-reporting PLN) and 701 controls (women who did not self-report PLN) was performed using Scalable and Accurate Implementation of Generalized mixed model (SAIGE) (version 0.41) with saddlepoint approximation (Zhou et al., Reference Zhou, Nielsen, Fritsche, Dey, Gabrielsen, Wolford, LeFaive, VandeHaar, Gagliano, Gifford, Bastarache, Wei, Denny, Lin, Hveem, Kang, Abecasis, Willer and Lee2018). The first four ancestry-informative principle components were included as covariates. A linear mixed model was used to account for cryptic relatedness and an unbalanced case-control ratio, and the leave-out-one-chromosome (LOCO) option implemented to avoid contamination for correlated markers (Zhou et al., Reference Zhou, Nielsen, Fritsche, Dey, Gabrielsen, Wolford, LeFaive, VandeHaar, Gagliano, Gifford, Bastarache, Wei, Denny, Lin, Hveem, Kang, Abecasis, Willer and Lee2018). An association p-value threshold of ≤ 1 × 10-6 was considered to be suggestively genomewide significant. The GWAS lambda (λ = 0.998) indicates no inflation in our association results (refer to quantile-quantile plot in Supplementary Figure S2).
Polygenic Risk Score Analyses
Polygenic risk score analysis (PRS) is a powerful tool that is used to explore shared genetic etiology between phenotypes (Dudbridge, Reference Dudbridge2013; Wray et al., Reference Wray, Lin, Austin, McGrath, Hickie, Murray and Visscher2021). As PLN is a pigmentation phenotype, we tested whether a PRS for skin color and/or a PRS for mole count could predict phenotypic variance in PLN (see Supplementary Methods). First, including only individuals of European ancestry, we conducted a GWAS for skin colour (very fair to brown; Supplementary Methods) in the UK Biobank cohort (N = 427,893; Bycroft et al., Reference Bycroft, Freeman, Petkova, Band, Elliott, Sharp, Motyer, Vukcevic, Delaneau, O’Connell, Cortes, Welsh, Young, Effingham, McVean, Leslie, Allen, Donnelly and Marchini2018) using BOLT-LMM v2.3 (Loh et al., Reference Loh, Tucker, Bulik-Sullivan, Vilhjálmsson, Finucane, Salem, Chasman, Ridker, Neale, Berger, Patterson and Price2015). We also performed a GWAS for mole count (melanocytic nevi; self-reports of none, few, some and many) in the QSkin Sun and Health Study (QSkin) cohort (N = 16,134; Olsen et al., Reference Olsen, Green, Neale, Webb, Cicero, Jackman, O’Brien, Perry, Ranieri and Whiteman2012) using PLINK 1.90b4.1 (Chang et al., Reference Chang, Chow, Tellier, Vattikuti, Purcell and Lee2015). Further details for both GWAS are included in the Supplementary Methods.
We then used these GWAS results (risk alleles and their corresponding GWAS effect sizes) to compute skin color and mole count PRSs in the PLN sample (see Supplementary Methods). Briefly, using the ‘clumping and thresholding’ method, PRSs were calculated for five p-value thresholds (p values ≤ 1 × 10-08, 1 × 10-06, 1 × 10-04, 1 × 10-02 and 1) using the PLINK1.9 ‘score’ function (Chang et al., Reference Chang, Chow, Tellier, Vattikuti, Purcell and Lee2015). Due to the major effect of the MC1R gene locus on both skin color and mole count, two PRSs were computed for each trait, one including and one excluding SNPs located within 1 Mb of the MC1R gene. Associations between each PRS and PLN, and the variance explained by the SNPs included in each PRS, were estimated using the GCTA ‘REML’ method (Yang et al., Reference Yang, Lee, Goddard and Visscher2011). Associations were considered significant at p values ≤ 2.5 × 10-3 (Bonferroni correction for 20 PRS-PLN association tests).
Results
PLN Heritability
We calculated the SNP-based heritability of PLN to be 35% (h2 SNP = 0.35, SE = 0.2), assuming a sample prevalence of 28% and including relatives in the estimation. Although this point estimate is not significant, most likely due to low power from the small sample size, it does suggest the presence of a genetic liability to PLN.
PLN GWAS
Our PLN GWAS including 275 cases and 701 controls detected no associations surpassing the genome-wide significant threshold (i.e., p values ≤ 5 × 10-8), likely due to the small sample size and consequent low power of this study (see Material and Methods). However, SNPs at four genomic loci were suggestively associated with PLN (p values of lead SNPs ≤ 1 × 10-6) (Table 1, Supplementary Figure S1): rs72693263 on chromosome 14 downstream of the fibronectin leucine rich transmembrane protein 2 (FLRT2) gene (p = 1.1 × 10-7), rs78371540 on chromosome 1 in intron 1 of olfactomedin 3 (OLFM3; p = 5.5 × 10-7), rs26331 on chromosome 5 downstream of semaphorin 6A (SEMA6A; p = 6.6 × 10-7) and rs1263154 on chromosome 2 downstream of uridine phosphorylase 2 (UPP2; p = 9.0 × 10-7).
Table 1. List of independent single nucleotide polymorphisms nominally associated (p ≤ 1 × 10-6) with pregnancy-related linea nigra
Note: SNP ID, single nucleotide polymorphism (SNP) dbSNP153 rs number; CHR:BP, hg19 positional information; Effect allele freq, allele frequencies in the HRC imputation reference panel; BETA, effect sizes; SE, standard error of the effect sizes; N, sample size; p values for the association between SNPs and pregnancy-related linea nigra.
Polygenic Risk Score Analyses
We performed GWAS for skin color and for mole count in two additional cohorts (see Supplementary Methods) to enable the calculation of two polygenic risk scores for each of these pigmentation traits, with one score per trait including SNPs at the skin pigmentation-associated MC1R gene region on chromosome 16 and one score excluding SNPs at this location. The full skin color PRS was significantly associated with PLN at multiple thresholds, accounting for 3.0% of the variance in developing PLN when only genomewide significant (p GWAS ≤ 1 × 10-08) skin colour variants were included in the PRS (p PRS = 3.8 × 10-8; Figure 1; Supplementary Table S1). These associations were attenuated but remained significant when the MC1R gene region was excluded from the PRS, explaining 1.8% of the variance in PLN at the p GWAS ≤ 1 × 10-08 threshold (p PRS = 1.7 × 10-05). These results indicate the association between genetically predicted skin color and PLN is due to variants of both small and large effect on skin pigmentation in individuals of European ancestry.
Figure 1. Association between mole count and skin color polygenic risk scores (PRS) and pregnancy-related linea nigra (PLN). PRS analyses were conducted using mole count- and skin colour-associated SNPs at five levels of significance (1 × 10-8, 1 × 10-6, 1x10-4, 1 × 10-2 and 1). Results are expressed as % of the variance explained in developing PLN, with the p values of two-sided association tests reported above each variance estimate.
The mole count PRS was most significantly associated with PLN at the p GWAS ≤ 1 × 10-6 threshold, explaining 1.2% of the variance in developing PLN (p PRS = 5.7 × 10-4; Figure 1). The variance explained slightly decreased, to 1.0%, when the MC1R region was excluded (p PRS = 7.0 × 10-4; Supplementary Table S2). Overall, our PRS results may imply a shared genetic liability for PLN and both skin color and mole count.
Discussion
This study marks a significant step forward in our understanding of the genetics of PLN and its relationship with other pigmentary phenotypes. Assuming a sample prevalence estimate of 28%, our heritability analysis showed evidence of genetic contribution in the development of PLN, with 35% of the PLN liability explained by all common SNPs; to our knowledge this has not been reported previously. Even though this point estimate is not significant, it does suggest that PLN is a heritable trait, with a considerable proportion of the variance in the occurrence of PLN among women explained by genetic factors. Previous twin studies have revealed high heritability for other pigmentation traits, including hair color (73−99%) and mole count (70%) (Lin et al., Reference Lin, Mbarek, Willemsen, Dolan, Fedko, Abdellaoui, de Geus, Boomsma and Hottenga2015). Recent GWAS have also supported a strong genetic influence on pigmentation, with SNP-based estimates of up to 40% of the variance in hair color and 53% of the variance in eye-color explained by common SNPs (Hysi et al., Reference Hysi, Valdes, Liu, Furlotte, Evans, Bataille, Visconti, Hemani, McMahon, Ring, Smith, Duffy, Zhu, Gordon, Medland, Lin, Willemsen, Jan Hottenga, Vuckovic and Spector2018; Morgan et al., Reference Morgan, Pairo-Castineira, Rawlik, Canela-Xandri, Rees, Sims, Tenesa and Jackson2018; Simcoe et al., Reference Simcoe, Valdes, Liu, Furlotte, Evans, Hemani, Ring, Smith, Duffy, Zhu, Gordon, Medland, Vuckovic, Girotto, Sala, Catamo, Concas, Brumat, Gasparini and Hysi2021). That SNP-based heritability estimates are typically lower than twin-based heritability estimates has been discussed previously, and may reflect influences from factors such as epistasis and/or gene-environment interactions, or the presence of low frequency rare variants with large effects and structural variations such as copy number variants, inversions and translocations that are not considered in GWAS studies (Manolio et al., Reference Manolio, Collins, Cox, Goldstein, Hindorff, Hunter, McCarthy, Ramos, Cardon, Chakravarti, Cho, Guttmacher, Kong, Kruglyak, Mardis, Rotimi, Slatkin, Valle and Visscher2009).
Data on PLN prevalence are limited, and a large variation among prevalence estimates has been reported. One cross-sectional study examining pregnancy-related physiological changes in a Brazilian cohort reported linea nigra in 54% of women (Fernandes & do Amaral, Reference Fernandes and do Amaral2015). Other studies of cutaneous changes in pregnancy observed PLN in 87% of women of Indian ancestry (Panicker et al., Reference Panicker, Riyaz and Balachandran2017) and 92% of Nigerians (George et al., Reference George, Shittu, Enwerem, Wachtel and Kuti2005). The prevalence of PLN in our European ancestry sample is lower (28%) compared to previous reports, and could be explained by an observation of lower PLN occurrence in fair-skinned compared to darker-skinned populations (Barnawi et al., Reference Barnawi, Barnawi and Alamri2021), with a decrease in skin pigmentation in populations at increasing distance from the Equator attributed as an adaptation to lower UV radiation levels (Jablonski & Chaplin, Reference Jablonski and Chaplin2010), although age group disparities within the cohorts included here cannot be ruled out. Results from a previous study suggest that the heritability of some cutaneous traits (e.g., mole count, color, and morphology) may increase with older age (Lee et al., Reference Lee, Duffy, McClenahan, Lee, McEniery, Burke, Jagirdar, Martin, Sturm, Soyer and Schaider2016). Future studies using larger samples sizes and including individuals of varying ages may provide more accurate estimates of the variance in PLN that can be attributed to genetic factors.
Despite the lack of genomewide significant associations, likely due to the small sample size and consequent low power of this PLN GWAS, four genetic variants were suggestively associated with PLN. Of these SNPs, rs78371540 on chromosome 1 is located within intron 1 of the OFLM3 gene. A different SNP at this gene locus (rs10874518 in intron 4) has been shown to be associated with freckles in a forensic context (Kukla-Bartoszek et al., Reference Kukla-Bartoszek, Pośpiech, Woźniak, Boroń, Karłowska-Pik, Teisseyre, Zubańska, Bronikowska, Grzybowski, Płoski, Spólnicka and Branicki2019), although rs10874518 is in very low linkage disequilibrium (r 2 = .003) with rs78371540. The three other associated SNPs are near genes that have not been previously associated with pigmentation or cutaneous phenotypes. UPP2 plays an important role in the metabolism of pyrimidine analogues (in nucleotide metabolism) and catalyses the reversible conversion of uridine to uracil and ribose-1-phosphate (Johansson, Reference Johansson2003; Roosild et al., Reference Roosild, Castronovo, Villoso, Ziemba and Pizzorno2011). Previous GWAS have linked genetic variants in UPP2 to congenital left-sided heart lesions (Agopian et al., Reference Agopian, Goldmuntz, Hakonarson, Sewda, Taylor and Mitchell2017), unipolar depression (Biernacka et al., Reference Biernacka, Sangkuhl, Jenkins, Whaley, Barman, Batzler, Altman, Arolt, Brockmöller, Chen, Domschke, Hall-Flavin, Hong, Illi, Ji, Kampman, Kinoshita, Leinonen and Weinshilboum2015), severe gingival inflammation (Zhang et al., Reference Zhang, Divaris, Moss, Yu, Barros, Marchesan, Morelli, Agler, Kim, Wu, North, Beck and Offenbacher2016) and sudden cardiac arrest in patients with coronary artery disease (Aouizerat et al., Reference Aouizerat, Vittinghoff, Musone, Pawlikowska, Kwok, Olgin and Tseng2011). SEMA6A participates in cell-cell signaling and governs cell migration in the developing central nervous system. A SNP in SEMA6A (rs9327007) has been shown to be associated with sleepiness during menstruation in Japanese women (Hirata et al., Reference Hirata, Koga, Johnson, Morino, Nakazono, Kamitsuji, Akita, Kawajiri, Kami, Hoshi, Tada, Ishikawa, Hine, Kobayashi, Kurume, Fujii, Kamatani and Osuga2018). Lastly, FLRT2 participates in cell-cell adhesion and axon guidance, and genetic variations within/near this gene are associated with age at menarche in African American women (Demerath et al., Reference Demerath, Liu, Franceschini, Chen, Palmer, Smith, Chen, Ambrosone, Arnold, Bandera, Berenson, Bernstein, Britton, Cappola, Carlson, Chanock, Chen, Chen and Haiman2013). Although our dataset was underpowered to detect genomewide significant associations with PLN, our results suggest a polygenic contribution to PLN and provide motivation for future research. We anticipate that extending GWAS analyses to larger samples may increase our understanding of the variation in PLN and enable identification of associated genetic variants that could be used for functional follow-up analyses.
Human pigmentary phenotypes, such as skin and hair color, degree of freckling and mole count, are widely acknowledged to be associated with coding variation in the MC1R gene (Smit et al., Reference Smit, Collazo-Roman, Vadaparampil, Valavanis, Del Rio, Soto, Flores, Dutil and Kanetsky2020; Sturm et al., Reference Sturm, Duffy, Box, Newton, Shepherd, Chen, Marks, Leonard and Martin2003). Given the influence of MC1R variants on skin pigmentation, we further explored relationships of pigmentation-associated traits and PLN including and excluding variants in the MC1R locus from the PRS analysis. Including the MC1R locus, PRSs for skin color and mole count significantly predicted PLN, accounting for 1.2% and 3% of the PLN variance. The associations of the skin color and mole count PRSs with PLN were both attenuated but remained significant when the MC1R region was excluded from the calculations, explaining 1% and 1.8% of the variance in PLN. This is of interest as it indicates that the association between genetically predicted skin color, mole count, and PLN are due to variants of both small and large effect on skin pigmentation in individuals of European ancestry.
While PLN was best predicted by the skin color PRS, our finding of significant genetic relationships between mole count and PLN is of interest as it suggests a shared genetic liability for PLN and both skin color and mole count. To the best of our knowledge, there are no published studies that have used PRSs to explore shared genetic liability among pigmentation phenotypes and future interrogations may bring to light new relationships between pigmentary traits. For example, in the psychiatric genetics field, PRSs have been extensively used to explore shared genetic risk across disorders, linking schizophrenia to bipolar disorder, depression, anxiety, substance-use disorder, and to immune disorders (Stringer et al., Reference Stringer, Kahn, de Witte, Ophoff and Derks2014; Zheutlin et al., Reference Zheutlin, Dennis, Karlsson Linnér, Moscati, Restrepo, Straub, Ruderfer, Castro, Chen, Ge, Huckins, Charney, Kirchner, Stahl, Chabris, Davis and Smoller2019; Musliner et al., Reference Musliner, Mortensen, McGrath, Suppli, Hougaard, Bybjerg-Grauholm, Bækvad-Hansen, Andreassen, Pedersen, Pedersen, Mors, Nordentoft, Børglum, Werge and Agerbo2019). Investigating PLN in diverse populations of other ancestries is also of special interest, in order to unravel the contribution of genetic variation in major pigmentation loci to the variability of PLN between populations.
In summary, this study supports a heritable component to PLN and provides the first examination of the genetic architecture of this pregnancy-related pigmentation trait. Although our sample size is currently limited, we provide evidence that PLN is a complex heritable trait that is influenced by both small and large gene effects. Our results suggest significant genetic relationships between PLN and skin color and susceptibility to melanocytic nevi. The extent to which these traits are influenced by common or unique biological mechanisms remains to be discovered. Our findings also provide motivation for further data collection and future studies using larger and ethnically diverse populations that would expand our understanding of common and rare variations influencing PLN, and clarify their complex interactions with other cutaneous and pigmentation phenotypes.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/thg.2025.10026.
Data availability statement
Pregnancy-related linea nigra GWAS summary statistics will be available for download from a QIMR Berghofer-hosted website upon publication.
Acknowledgments
QIMR Berghofer thanks the participants of the Nausea and Vomiting during Pregnancy Study (NVP), the Women’s Health and Lifestyle Study (WHLS) and the QSkin study for their participation. We also thank Simone Cross and Richard Parker for their assistance in the collection of these data. All data collections were approved by the Human Research Ethics Committee of the QIMR Berghofer Medical Research Institute, and followed the national regulations regarding data protection. The Helsinki Declaration, as well as applicable institutional and governmental regulations concerning the ethical use of human volunteers, were followed during all the phases of this research.
Author contributions
The study was designed by SB and SEM, and cohort collections and data acquisition led by CMO and DCW (Qskin), and PAL, JNP and SEM (NVP and WHLS). SB and MS performed the GWAS analyses, SB and JNP performed the PRS analyses, JNP prepared Table 1 and Figure 1, and MHL, BLM, LCC and PAL provided critical comments on the analysis methods and results. SB wrote the initial manuscript, and all authors contributed to the overall interpretation of the data, drafting and critical review of the manuscript, and have approved the submitted version and agreed to be personally accountable for the work.
Funding
QIMR Berghofer data collections have been funded by the Australian Research Council [grant numbers A79600334, A79906588, A79801419, DP0212016, DP0343921] and the National Health and Medical Research Council (NHMRC) Project Grant [grant numbers 241944, 389875, 552485, 552471, 1031119, 1049894, 1084325]. The Nausea and Vomiting during Pregnancy and Women’s Health and Lifestyle studies were funded by an NHMRC Project Grant APP1084325. The QSkin Study was supported by an NHMRC Clinical Trials and Cohort Grant [APP1185416]. The QSkin Genetics Study was supported by an NHMRC Research Grant [APP1063061]. Sarah Medland was supported by NHMRC Investigator grant APP1172917. David Whiteman was supported by an NHMRC Research Fellowship APP2026567.
Competing interests
The authors state no conflict of interest.
Open access
