Recent Advances in Genomic Studies of Gestational Duration and Preterm Birth.

GWA studies (GWAS) have been instrumental in identifying specific genetic variants and genes associated with gestational duration and PTB. 13 – 18 Similar to other complex traits or diseases, these phenotypes are affected by numerous genetic variants, each with a modest effect. 13 , 25 , 26 While genetic association studies typically examine thousands to millions of genetic variations, analyses are conducted individually for each tested variant. 25 The significantly associated genetic variants or genes, particularly the causal ones, can provide valuable information about the biology of the studied phenotype. 27 The first successful GWAS of gestational duration and PTB was published in 2017. 28 This study identified and replicated six genomic loci associated with gestational duration in over 50,000 mothers of European ancestry. Three of these loci were also associated with PTB with genome-wide significance. Parallel observations of the same associations in infant samples, albeit with smaller effect sizes, indicated the maternal origin of the identified genetic associations. The associated loci are linked to relevant biological pathways including uterine development, maternal nutrition, and vascular control. Promising results from this initial GWAS encouraged even larger GWAS aiming to identify more genetic associations and to reveal additional biological and mechanistic insights. Recently, Sole-Navais et al. 29 conducted a comprehensive genome-wide meta-analysis on gestational duration (n = 195,555) and sPTB (n = 276,218) across 18 cohorts of European ancestry. This study identified 24 independent genetic variants at 22 loci associated with gestational duration ( Table 1 ) and seven loci associated with sPTB. This research corroborated the six loci previously reported by Zhang et al. 28 Using a haplotype-based analytical method, the study differentiated between maternal and fetal genetic effects, revealing that the majority of the identified loci (15 out of 22) exerted their effects through the maternal genome. Furthermore, the study unveiled an intriguing pattern: alleles that increase gestational duration through maternal effects tend to decrease birthweight via fetal effects. This observation suggests that the maternal and fetal genomes might co-adapt 30 to achieve an optimal balance between gestational duration and birthweight, enhancing overall fitness between the mother and her baby. Another GWA meta-analysis 31 of gestational duration (n = 68,732) and sPTB (n = 98,370) included participants from Finland and summary results from 23andMe. This study identified and replicated 15 loci associated with gestational duration and four loci associated with sPTB, many overlapping with those reported by Sole-Navais et al. 29 ( Table 1 ). The functional roles of these genes highlight the intricate nature of spontaneous birth as a trait and emphasize the importance of reproductive and immune tissues. To explore the fetal genetics on the timing of birth, Liu et al. 32 conducted a large fetal genome-wide meta-analysis of gestational duration, early preterm (<34 weeks), preterm (< 37 weeks), and post-term births in 84,689 infants. This study identified a single nucleotide polymorphism (SNP) (rs7594852), located at 2q13 locus within the CKAP2L gene, which is implicated in the pro-inflammatory pathway. A joint maternal-fetal genetic association analysis using the WLM method in mother-child pairs revealed that the association at the leading SNP was driven by fetal genotype. Another fetal GWAS 33 in a Finnish population included 247 infants with sPTB (<36 weeks) and 419 term controls (38–41 weeks). The SNP, rs116461311 within the gene encoding slit guidance ligand 2 ( SLIT2 ) showed the strongest association with sPTB. The SLIT2 gene and its receptor ROBO1 were upregulated in the placenta of PTB infants which played a vital role in inflammation, decidualization, and fetal growth. 33 Post-GWAS functional studies using in vitro and in vivo models have elucidated novel molecular mechanisms through which significantly associated genetic loci influence gestational duration and risk of PTB. For example, functional studies demonstrated that a leading variant within the WNT4 gene disrupts the interaction with the estrogen receptor (ESR1), suggesting a mechanism that can impact various reproductive problems. 28 , 34 Gene expression analyses revealed that the maternal blood mRNA levels of EBF1 were reduced in mothers who delivered preterm as early as the 2 nd trimester. 35 Bioinformatic and functional investigations indicated that a risk variant proximal to AGTR2 could potentially modulate the risk of PTB by altering the expression of AGTR2 , particularly in uterine tissues. 36 Integrative analyses combining transcriptional and gene regulation datasets revealed that the HAND2 gene may regulate transcriptional programs involved in endometrium decidualization. 37 Additionally, assessments of tissue-specific expression enrichment and refined linkage disequilibrium (LD)-score regression of the top GWA genes highlighted the endometrium and other female reproductive and muscular tissues. 29 A complex trait is influenced by many genetic variants with small effects. 38 , 39 Multi-variant analysis aims to capture the cumulative effects of hundreds to thousands of genetic variations significantly associated with a disease. A multi-variant analysis approach uses polygenic scores (PGS). 40 Such scores are commonly obtained by computing the sum of risk alleles weighted by their estimated effect sizes of many genetic variants associated with a phenotype. These scores can be used to predict the genetic risks of complex diseases. 40 , 41 The recently developed PGS for gestational duration accounts for 2.2% of the variance in gestational duration. 29 While this percentage remains relatively small for clinical prediction purposes, it surpasses the variance explained by any single known environmental risk factor. Notably, the average gestational duration differs by 5 days between mothers in the lowest and highest deciles according to their PGS. This demonstrates the potential of PGS to predict gestational duration with greater precision than traditional environmental risk factors alone. Another application of PGS is to use it as a genetic instrument to investigate the causal effect of one phenotype (exposure) on another phenotype (outcome) – an approach called Mendelian randomization (MR). 42 By aggregating the cumulative genetic association of multiple variants with the exposure of interest, PGS can enhance the power of MR analysis. 43 In the framework of integrated maternal-fetal analysis, haplotype PGS can be developed based on h1, h2, and h3. These haplotype PGS can be used to dissect maternal and fetal genetic effects and to investigate the causal associations between maternal traits and pregnancy outcomes. 13 Over the past decade, many studies adopted this methodology to study the causal links between maternal phenotypes and pregnancy outcomes. An initial study leveraging this method explored the potential causal links between maternal anthropometric traits and key pregnancy outcomes, including gestational duration, birthweight, and birth length. 22 The research utilized haplotype PGS related to maternal height to distinguish between maternal and fetal genetic influences. The findings demonstrated significant associations of birth length and weight with both h1 and h3, indicating a pronounced fetal genetic impact on these growth parameters. In contrast, the duration of gestation was significantly linked to the h2 PGS, suggesting a causal effect of maternal height on gestational duration. The same group further investigated the causal associations of maternal phenotypes such as height, body mass index (BMI), blood pressure, and blood glucose level with pregnancy outcomes in 10,734 mother-child pairs. 44 The study showed that both maternal height and fetal growth are important factors in shaping the duration of gestation. The authors confirmed that taller maternal height was associated with longer gestational duration. They also demonstrated that fetal growth was influenced by both maternal and fetal genetic effects. For example, elevated maternal BMI and glucose levels were positively associated with birthweight due to maternal effects. Conversely, fetal alleles linked to heightened metabolic risk were found to negatively affect fetal growth. Another significant finding from this research was the association of rapid fetal growth with shorter gestational duration and increased maternal blood pressure, a phenomenon known as “fetal drive”. This pattern was confirmed in multiple subsequent studies. 29 , 45 – 47 Overall, these studies underscore the profound impact of both maternal and fetal genetic effects in shaping the relationship between maternal characteristics and birth outcomes, as well as the life-course associations between these birth outcomes and adult phenotypes. The findings support the fetal insulin hypothesis, 48 which emphasizes genetic determinants in linking early growth with the risk of metabolic diseases later in life. 44 , 49 The third level of analysis focuses on exploring the genetic architecture using genome-wide variants. Genetic architecture broadly refers to mapping one’s genotype to phenotype, including aspects like the proportion of phenotypic variance attributable to genetics, the distribution of allelic effects, and the genetic correlations with other phenotypes. A key method in this context is estimating heritability from genome-wide SNP data, known as SNP-based heritability. 50 This approach has been instrumental in understanding the genetic basis of complex traits and addressing the issue of ‘missing heritability’. 51 Various methods have been developed to estimate the additive genetic variance of a complex trait using genome-wide SNP data. 52 – 54 However, these methods cannot be directly applied to the analysis of pregnancy phenotypes. This is because, unlike other complex traits, pregnancy outcomes are influenced by both maternal and fetal genomes. To avoid the confounding by shared alleles between mother and child, Srivastava et al. 55 developed a haplotype-based approach (haplotype-based genome-wide complex trait analysis – H-GCTA) to estimate the genetic variance attributable to h1, h2, and h3. Their findings indicate that gestational duration is primarily genetically determined by the maternal genome whereas fetal growth measurements such as birthweight, birth length, and head circumference are mainly influenced by the fetal genome. Eaves et al. 56 introduced a contemporary method, maternal genome-wide complex trait analysis (M-GCTA), through the joint analysis of mother-child genotypes. Using this approach, they studied the genetic contribution of the maternal and fetal genomes on birth length. Later, Warrington et al. 24 estimated the SNP-heritability of birthweight. The results demonstrated the genetic variance in birthweight is mainly attributable to fetal genetics. Similar results were obtained by Qiao et al. 57 These studies, facilitated by methods such as H-GCTA and M-GCTA, revealed that both maternal and fetal genomes significantly influence pregnancy outcomes, with maternal genetics predominantly determining gestational duration and fetal genetics mainly affecting fetal growth measurements. Analysis of genome-wide variants can also shed light on the shared genetic architecture between two phenotypes. Genetic correlation estimates the proportion of genetic variance shared by two phenotypes due to common genetic factors. 58 Recently, Solé-Navais et al. 29 showed a strong genetic correlation between sex hormones with gestational duration and PTB. Specifically, testosterone and calculated bioavailable testosterone (CBAT) are negatively correlated with gestational duration, while sex hormone-binding globulin (SHBG) shows a positive correlation. These findings suggest the importance of sex hormone regulations in shaping gestational duration. In addition, the study also detected negative genetic correlations between gestational duration with preeclampsia and endometriosis suggesting shared gene pathways in these reproductive disorders. Genetic correlations might stem from either genetic pleiotropy or direct causal relationships. Indeed, Solé-Navais et al. 29 provided evidence for the genetic causality of sex hormones on birth timing.

Significant disparities exist in PTB across various racial/ethnic, socioeconomic, and geographical populations. In the United States, African-American/Black women experience PTB at a rate that is approximately 1.5 to 1.6 times higher than that of Whites. 59 , 60 Extensive research indicates that these disparities are primarily driven by social health determinants and systemic racism. 60 – 62 In addressing these disparities, genomic research in diverse populations becomes critically important. 63 The objective is not about finding genetic differences that account for varying susceptibilities to PTB. Instead, the goal is to discover population-specific risk variants and gene-environmental interactions (G × E), thereby enriching our understanding of the biological mechanisms underlying PTB. Such research can also shed light on the intricate interactions between social and environmental risk factors and genetic predispositions across different populations. 64 One of the first multi-ancestry GWAS on early sPTB examined nearly 1,000 cases of mother–infant pairs and a similar number of control pairs, including over 20% of African Americans. 65 Although several maternal and fetal candidate loci were identified in the discovery phase, none were replicated in a smaller validation cohort. Another multi-ethnic GWAS 66 investigated 1,349 extreme preterm infants (25 and 30 weeks of gestation) against 12,595 ancestry-matched controls, focusing on fetal genomic signals, and identifying two intergenic loci. The authors also attempted replication using several datasets. However, they found no evidence supporting these findings. A recent maternal GWAS in the Indian population, 67 including 521 PTB mothers and 1042 matched controls, identified 15 SNPs that reached a relaxed significance threshold (P < 2E-6). However, none of the genes implicated showed overlap with the top associated genes from much larger European studies. In a separate study, Juvinao-Quintero et al. 68 conducted a GWAS on PTB and gestational duration among 2,212 Peruvian women, uncovering several suggestive signals (P < 1E-5); again, none of them overlapped with any signals previously identified in European populations. Interestingly, this study replicated two genetic variants within the WNT4 gene, significantly associated with gestational duration in European datasets. These findings reinforce the need for larger sample sizes to ensure robust and replicable results across diverse populations. Given the importance of social and other environmental contributors to PTB risk, Hong et al. 69 investigated G × E. The authors conducted genome-wide G × E analyses of PTB in 1,733 African-American women from the Boston Birth Cohort and they found a significant interaction between a variant in the COL24A1 gene with maternal pre-pregnancy overweight/obesity on PTB risk. This interaction was replicated in African-American mothers but not replicated in mothers of European ancestry. This finding highlights the importance of taking non-genetic factors into account when conducting genetic association studies of PTB. Despite these research efforts across diverse populations, it is important to note that most of the replicated genomic findings related to gestational duration and PTB predominantly come from studies in European cohorts, 28 , 29 , 31 , 32 which often have significantly larger sample sizes. This implies that the benefits of genomic research, including understanding pathogenesis, early screening for adverse outcomes, better diagnosis, improving clinical care, and managing comorbidities during pregnancy may be eluded in those under-represented and usually high-risk populations. Moreover, differences in genetic ancestry, environment and lifestyle could further limit the transferability of genetic insights from European studies to underrepresented and high-risk populations. 70 – 72 This calls for immediate measures to address the genomic imbalance of studies in diverse populations. 64 Furthermore, genomic research in underrepresented populations is essential for mitigating future inequality in genomic medicine of pregnancy outcomes. 70 In addition, there are significant advantages of increasing diversity in genomic research, including identifying novel associations with population-enriched variants, fine-mapping causal variants, improving genetic risk prediction accuracy for all populations (particularly underrepresented populations), and understanding shared versus unique genetic and environmental risk factors. 70 , 73 One clear example of population-enriched clinically important variants is APOL1 gene variations and preeclampsia, which is identified only in populations with African ancestry. 74 , 75 Racism and social injustice have gained increased attention recently as fundamental contributors to health disparities, including adverse pregnancy outcomes such as PTB and low birthweight. 76 , 77 The impact of chronic stress associated with transgenerational racism on health requires a substantial reorientation of societal voices and leadership well beyond the healthcare system. These efforts are growing in number and impact as evidenced by prioritizing diversity, equity, and inclusion globally. Incorporating social determinants of health in electronic medical record dashboards and research databases, will guide care providers in individualizing healthcare and researchers in analyzing the social factors integrated with other covariates. This will especially be powerful as broad areas of biological factors can simultaneously be assessed and intervened. 78 , 79 Tracking and incorporating these factors into preventive strategies has considerable supportive evidence for impacting population-attributable risk for PTB. 80 Further, some of the genetic regions that are associated with PTB risk might be targets for epigenetic and post-transcriptional regulations that arise from consequences of structural racism, social injustice and adverse environmental exposures including infection. 81 To help correct the lack of diversity in genomic research and to mitigate future inequality in genomic medicine, there is an urgent need for structural, organizational, and policy changes that ensure the intentional hiring of diverse researchers by institutes, allow researchers to form genuine partnerships with communities, enhance community engagement and participation in genomic studies among minority groups, encourage the funders to set up strategic funding schemes that promote research of underrepresented populations and enact policies that create conducive environments for sustainable, diverse genomic studies of high-risk and marginalized populations. 64 , 82 – 84 Large multi-ethnic GWA studies, such as the Population Architecture through Genomics and Environment (PAGE) study, 85 have illustrated the benefits of including diverse populations in genomic research. Similarly, community-based genomic initiatives, like the All of Us Research Program ( https://allofus.nih.gov/ ), emphasize the urgent need to involve under-represented groups. Despite these efforts, there is still a critical gap in specifically addressing disparities in pregnancy health. Addressing this gap, several recent studies focusing on under-represented groups have been launched, with support from organizations like the Bill & Melinda Gates Foundation, the National Institutes of Health (NIH), the Burroughs Wellcome Fund, and the March of Dimes. These studies aim to enhance diversity in genomic research related to pregnancy outcomes, promising to yield novel genomic discoveries and benefit those most affected by adverse pregnancy outcomes. Despite its simple clinical definition, PTB is a syndrome with many causes. 86 Many pathologic processes can initiate the premature onset of labor. A broad spectrum of biological, psychosocial, behavioral, and environmental factors may influence the risk of PTB, and their effects may vary in individual cases. 87 This heterogeneity makes the prediction and management of PTB extremely challenging. Large genomic studies can potentially uncover many genetic variants statistically associated with gestation duration or PTB. However, uncovering the biological mechanisms that link these genetic differences to the final phenotypic expression remains difficult. To bridge this gap, it is crucial to also explore relevant cellular and intermediate physiological phenotypes that may have more direct connections with PTB. In addition, although identified genetic risk variants influence the baseline genetic predisposition to PTB, there may be stronger and immediate internal or environmental risk factors, when acting upon different genetic backgrounds can significantly modify an individual’s risk and ultimately trigger premature labor. Answering these questions requires an integrative approach that goes beyond genomics to also including more refined phenotyping and multiomics studies at different levels. Many prospective birth cohort studies, such as the Avon Longitudinal Study of Parents and Children (ALSPAC), 88 Born in Bradford (BiB), 89 the Generation R Study, 90 Alliance for Maternal and Newborn Health Improvement (AMANHI), 91 the Global Alliance to Prevent Prematurity and Stillbirth (GAPPS), 92 and Boston Birth Cohort, 93 have implemented comprehensive longitudinal deep phenotyping and established biobanks for a wide range of maternal and fetal biospecimens. These initiatives, when combined with cutting-edge omics technologies and innovative analytical methodologies, provide new opportunities for better prediction and understanding of PTB and related adverse pregnancy outcomes. 94 , 95

This review underscores the advances in genomic research that have begun to illuminate the intricate genetic factors that underpin gestational timing and the risk of PTB. Through genomic analyses at three different levels in mother/child pairs, studies have identified key genetic variants associated with gestational duration and PTB, revealed potential causal relationships and genetic architecture of diverse pregnancy-related phenotypes. Post-GWAS functional studies have further elucidated molecular mechanisms and biological pathways. Moreover, the review also highlighted the critical need for increasing diversity in genomic research to address disparities in PTB across different populations and to ensure that the benefits of genomic medicine extend to all, especially those most at risk. It calls for an integrative approach that combines genomics with refined phenotyping and other omics to fully capture the complexity of PTB. While significant challenges remain, the ongoing efforts to unravel the genetic basis of PTB and to incorporate comprehensive, multiomics, and socially aware research approaches represent vital steps toward better understanding, prediction, and prevention of PTB. These endeavors promise to advance our scientific understanding and pave the way for more personalized and equitable healthcare solutions for PTB.

The success of genomic research heavily depends on the precise definition and measurements of outcomes. In the context of genetic analyses of PTB, gestational duration is a widely studied outcome. However, collecting accurate data on gestational duration is challenging, especially in under-resourced and ethnically diverse communities. An early ultrasound along with fetal measurements, preferably in the 1 st trimester is the most accurate way to assess GA with an error window of 5–7 days. However, many low- and middle-income countries (LMICs) and communities with limited healthcare facilities and practitioners still depend on the date of the last menstrual period (LMP) in calculating the expected date of delivery (EDD). LMP method has low accuracy due to irregular and large variations in the length of the menstrual cycle, conception occurring up to several days after ovulation, and inability to recall the date of LMP. 4 Many developed countries and research studies combine ultrasound and LMP methods to estimate GA. This combined relatively new method of assessment referred to as “best obstetric estimate” can have better accuracy than either of the two methods. 4 , 19 The diagnosis of PTB, defined by dichotomization of gestational duration, is a common practice in clinical research to simplify statistical analyses and interpretation of results. However, such simplicity is achieved at the cost of loss in statistical power to detect true genetic associations. 20 Genetic epidemiology studies suggest a model of similar genetic and environmental contributions across the range of GAs. This supports the notion that considering gestational duration as a continuous variable could provide greater statistical power than dichotomizing preterm and term pregnancies in genetic studies. 9 Indeed, most PTB-associated genetic loci identified to date are also associated with gestational duration with higher statistical significance. Recent research further indicated the overall genetic control of birth timing appears to be shared in the term and preterm. 21 Being genetically controlled by both maternal and fetal genomes is a distinctive characteristic of human birth timing and other pregnancy-related phenotypes. Consequently, genomic studies of these phenotypes must encompass both mothers and infants. With the growing volume of genomic data from mothers and infants across many birth cohort studies, new opportunities arise to dissect the relative contributions of maternal and fetal genomes. Conventional genetic studies often consider each individual as the analytical unit, examining the association between an individual’s genotype (usually comprising two alleles) and their phenotypes. This approach has several limitations when applied to pregnancy phenotypes. Firstly, distinguishing between maternal and fetal genetic effects becomes problematic due to the sharing of alleles transmitted from mother to fetus, especially when only maternal or fetal samples are analyzed. Secondly, analyzing mothers and infants separately fails to reveal the complex phenotypic and genetic interplay between the mother and her fetus. To address these challenges, Zhang et al. 13 , 22 developed a novel approach that treats the mother-child pair (or a pregnancy) as a single analytical unit with three alleles: the maternally transmitted allele (h1), the maternally non-transmitted allele (h2), and the paternally transmitted allele (h3) ( Figure 1 ). Each of these alleles influences pregnancy outcomes differently: h1 can affect outcomes either through the mother or the fetus or both; h2 exclusively influences through the mother; and h3 acts solely through the fetus. The method is also referred to as a haplotype-based approach due to the joint estimation of haplotypic (allelic) effects of h1, h2, and h3. By considering the mother-child pair as one unit, this method enables a clear distinction between maternal and fetal genetic effects on pregnancy phenotypes. 13 This approach can also be extended to parent-offspring trios with four alleles (h1, h2, h3, and h4 – paternally non-transmitted allele). Zhang et al. 13 also proposed three different levels of genetic analyses ( Figure 2 ) – single variant analysis, multi-variant analysis, and analysis of genome-wide variants under this analytical framework. An alternative approach to estimate both maternal and fetal effects is the structural equation modelling (SEM) method proposed by Warrington et al. 23 To facilitate the genome-wide application of adjusted maternal and fetal effect estimates, they introduced a weighted linear model (WLM), which calculates unbiased estimates of maternal and fetal effects from GWA summary results obtained separately from mothers and infants. 24

Preterm birth (PTB), defined as live birth before 37 weeks of completed gestation, is the leading cause of under-five mortality and remains one of the greatest adverse public health outcomes globally. 1 Worldwide, an estimated 13.4 million neonates are born prematurely every year, and nearly 1 million of these children die each year due to complications of prematurity. 2 Despite the significant global health implications and the recognition that preventing PTB is the best way to improve health outcomes for children affected by prematurity-related complications, progress in preventing prematurity has been limited. Based on the clinical definition, PTB can be broadly grouped into medically-indicated PTB, idiopathic PTB (iPTB), and preterm premature rupture of membranes (PPROM). 3 Likewise, the World Health Organization (WHO) subcategorizes PTB based on the gestational age (GA) at birth – extreme PTB (<28 weeks), very preterm (28 to <32 weeks), and moderate or late PTB (32 to <37 weeks). 4 A substantial body of research indicates a genetic influence on the risk of PTB and, more broadly, the duration of gestation. The most direct evidence is that a history of preterm delivery in a mother is the strongest predictor of PTB in her subsequent pregnancies. 5 Numerous epidemiological studies have demonstrated the involvement of both maternal and fetal genetics, with maternal genetics often playing a more substantive role. 6 – 9 Human genome-wide genetic studies provide an unbiased way to detect associated genes and biological pathways of birth timing. This is of particular importance because each species has its own reproductive strategy and animal studies cannot provide complete information about a human pregnancy. 10 The findings from genomic research could have multiple translational values including genomic prediction of PTB, mechanistic insights, and identification of potential interventional targets that may lead to novel approaches to reduce adverse pregnancy outcomes and their long-term sequelae. In the past, various reviews have comprehensively covered different aspects of the genetics of PTB, such as genetic epidemiology, 11 candidate gene association, 12 genome-wide association (GWA), rare variants, and copy number associations along with transcriptional and epigenetic regulation of PTB. 13 – 18 In this review, we focused on recent genomic research on human birth timing in mothers and their children, aiming to distinguish between maternal and fetal genetic effects. Our discussion includes studies at three distinct levels: single-variant, multi-variant, and genome-wide variation analyses, each targeting different genetic questions. Additionally, we highlight some challenges in this field, particularly the lack of population diversity in genomic studies of pregnancy outcomes. We also explore the prospects and opportunities that lie ahead for new research directions.