The Genetic Risk of Gestational Diabetes in South Asian women

South Asian women are at increased risk of developing gestational diabetes (GDM). Few studies have investigated the genetic contributions to GDM risk. We investigated the association of a type 2 diabetes (T2D) polygenic risk score (PRS) , on its own , and with GDM risk factors, on GDM- related traits using data from two birth cohorts in which South Asian women were enrolled during pregnancy. 837 and 4,372 pregnant South Asian women from the Sou Th Asian Bi Rth Cohor T (START) and Born in Bradford (BiB) cohort studies underwent a 75-gram glucose tolerance test. PRSs were derived using GWAS results from an independent multi-ethnic study (~18% South Asians). Associations with fasting plasma glucose (FPG); 2h post-load glucose (2hG); area under the curve glucose ; and GDM were tested using linear and logistic regression s. The population attributable fraction (PAF) of the PRS was calculated. Every 1 SD increase in the PRS was associated with a 0.085 mmol/L increase in FPG ([95%CI=0.07-0.10], P= 2.85 × 10 -20); 0.2 1 mmol/L increase in 2hG ([95%CI=0.16-0.26], P=5.49 × 10 -16); and a 45% increase in the risk of GDM ([95%CI=32-60%], P=2.27 × 10 -14), independent of parental history of diabetes and other GDM risk factors. PRS tertile 3 accounted for 12.5% of the population’s GDM. No consistent interactions of the PRS with BMI, or diet quality were observed. A T2D PRS is strongly associated with multiple GDM-related traits in women of South Asian descent, and accounts for a substantial proportion of the PAF of GDM. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 4

1 Gestational diabetes mellitus (GDM) is defined as hyperglycaemia first d iagnosed during 2 pregnancy. This abnormal increase in blood glucose levels is associated with an increased risk of 3 adverse health outcomes for both mother and their fetus/child during pregnancy , and later in 4 life.(Farrar et al., 2016) It is estimated that 1% to >30% of live births are affected by GDM 5 worldwide. This prevalence has been shown to vary widely depending on the participants ethnicity, 6 countries/regions, and on the diagnostic criteria used .(Archambault, Arel, & Filion, 2014; 7 McIntyre et al., 2019) South Asian women (whose ancestry derives from the Indian subcontinent) 8 have a 2-fold increased odds of developing GDM, compared to white European women.(Anand et 9 al., 2016; Cosson et al., 2014; Farrar et al., 2015; McIntyre et al., 2019) The reasons for this 10 disproportionate risk have not been fully characterized. 11 Gestational diabete s is a complex disorder influenced by multiple genetic and environmental 12 factors such as maternal age , ethnicity, obesity, poor diet quality, and family history of 13 diabetes.(Anand et al., 2017; Hedderson, Darbinian, Quesenberry, & Ferrara, 2011; Solomon et 14 al., 1997) Most genetic and environmental GDM risk factors are shared with type 2 diabetes 15 ,(Sattar & Greer, 2002; Zhang & Ning, 2011) another condition which is thought to be very closely 16 related to GDM. For example , women with GDM have a higher probability of having at least one 17 parent with type 2 diabetes, compared to those with normal gestational glycemia.(Jang, Min, Lee, 18 Cho, & Metzger, 1998) Furthermore, women with a GDM history have a 10-fold higher risk of 19 subsequently being diagnosed with type 2 diabetes compared to those without a history of 20 GDM.(Vounzoulaki et al., 2020) In terms of genetic architecture, both candidate gene and genome-21 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 5 wide association studies (GWASs) demonstrated a considerable overlap between GDM and type 22 2 diabetes.(Hayes et al., 2013; Kwak et al., 2012; Pervjakova et al., 2021) Finally, type 2 diabetes 23 polygenic risk scores (PRSs) have also be en associated with GDM risk.(Lamri et al., 2020; 24 Pervjakova et al., 2021) 25 It has been demonstrated that environmental exposures such as diet and/or physical activity may 26 modulate the effect of type 2 diabetes loci (such as TCF7L2, PPARG and CDKAL1) on the risk of 27 type 2 diabetes .(Dietrich et al., 201 9) Nevertheless, only a handful of studies have investigated 28 genetic×environmental interactions on GDM,(Chen et al., 2019; Grotenfelt et al., 2016; Popova et 29 al., 2017) and to date, no study has tested the interaction between a genome-wide PRS with other 30 GDM risk factors, on the risk of GDM. 31 The aims of this investigation were to: i) test the association of a type 2 diabetes PRS, generated 32 from an external multi-ethnic GWAS (~18% South Asians), with GDM and related traits (fasting 33 plasma glucose (FPG), 2h post-load glucose (2hG) and area under the curve glucose (A UCg) 34 levels) in pregnant South Asian women from the SouTh Asian biRth cohorT (START) and the 35 Born in Bradford (BiB) studies; ii) To estimate the population attributable fraction (PAF) of the 36 PRS on GDM, and iii) To determine whether the effect of the PRS is modulated by other GDM 37 risk factors including age, BMI, diet quality, birth country, education, and parity. 38

39 Study design and participants 40 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 6 START is a prospective cohort study designed to evaluate the environmental and genetic 41 determinants of cardio-metabolic traits among South Asian women and their offspring living in 42 Canada.(Anand et al., 2013) In brief, 1,012 South Asian pregnant women , aged between 18 and 43 40 years old, were recruited during the ir second trimester of pregnancy from the Peel Region 44 (Ontario, Canada) through physician referrals between 2011 and 2015 . All START participants 45 provided informed consent, and the study was approved by local ethic s committees ( Hamilton 46 Integrated Research Ethics B oard [ID:10-640], William Osler Health System [ID:11-0001], and 47 Trillium Health Partners [RCC:11-018, ID:492]). 48 BiB is a prospective, longitudinal family cohort study designed to investigate the causes of illness, 49 and develop interventions to improve health in a deprived multi -ethnic population in Bradford, 50 England, UK .(Wright et al., 2013) Between 2007 and 2011, 1 2,453 women of various ethnic 51 backgrounds (~46% South Asian origin) were recruited between their 24 th and 28 th week of 52 pregnancy. Detailed information on socio -economic characteristics, ethnicity , family history, 53 environmental and physical risk factors has been collected.(Farrar et al., 2015; Wright et al., 2013) 54 Ethical approval for all aspects of the research was granted by Bradford Research Ethic s 55 Committee [Ref 07/H1302/112]. 56 Measurements and questionnaires: 57 START: A detailed description of the maternal measurements has been published 58 previously.(Anand et al., 2017) Briefly, weight and height were measured using standard 59 procedures, and information about pre-pregnancy weight, family and personal medical history was 60 collected using questionnaires . Parental history of diabetes was derived from baseline 61 questionnaires and categorized as neither parents had a history of diabetes or either one or both 62 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 7 parents did. Birth country, number of years spent in Canada, and education-related variables were 63 self-reported. Participants’ highest level of education was coded as a five-category ordinal variable 64 as: 1) less than high school; 2 ) high school completed; 3) Diploma or certificate from trade, 65 technical or vocational school; 4) Bachelor's or undergraduate degree, or teacher's college; 5) 66 Master's, Doctorate or professional degree. A binary “born in South Asia” variable was categorized 67 as participants born in South Asia ( India, Pakistan, Sri Lanka or Bangladesh versus participants 68 were born in any other country. A validated ethnic-specific food frequency questionnaire (FFQ) 69 was used to collect dietary information.(Kelemen et al., 2003) To calculate diet quality, 1 point 70 was given for consuming ≥ the study population median of : 1) green leafy vegetables, 2) raw 71 vegetables, 3) other cooked vegetables, and 4) fruits , and less than the population median of 5) 72 fried foods/fast food/ or snacks, and 6) meat/poultry. The continuous scores (ranging from 0 to 6 73 points) was categorized into a binary variable: low diet quality vs. medium or high quality] before 74 analysis.(Anand et al., 2017) 75 BiB: Maternal height was measured during the recruitment visit (24-28th weeks of pregnancy) 76 using standard procedures . In the absence of pre-pregnancy weight data, weight from the first 77 antenatal clinic visit ( average 12 weeks of pregnancy ) was used to calculate BMI. Ethnicity of 78 participants and years spent in UK were self-reported at recruitment through an interview 79 administered questionnaire; missing ethnicity data was backfill ed from primary care data when 80 available. The South Asian ethnicity of all participants included in this analysis was validated 81 using genetic data. Parental history of diabetes and “born in South Asia” variables were derived 82 from the baseline questionnaire data and coded as in START. Since only a very small proportion 83 of BiB ’s participants completed an FFQ that included information about fruits and vegetables 84 intake, the diet quality score could not be derived in BiB. Data regarding the participant’s highest 85 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 8 educational qualification was equalized (using UK standards) and recoded into the following 86 categories: 1) less than 5 General Certificate of Secondary Edu cation (GCSE) equivalent; 2) 5 87 GCSE equivalent; 3) A-level equivalent; 4) higher than A-level. Data for unclassifiable foreign 88 degrees was considered as missing. 89 Outcomes: Study participants without prior type 2 diabetes were invited to undertake a 75-gram 90 oral glucose tolerance test (OGTT ) in both START and BiB , and FPG, and 2h G levels were 91 measured (1h post-load glucose was measured in START only) . AUCg was calculated using the 92 FPG and 2hG glucose levels in BiB, and u sing the FPG, 1h post-load glucose, and 2hG levels in 93 START.(Anand et al., 2017) Given the difference in the number of data points included in the 94 calculation of AUC between the two studies and the skewness of the distributions, values were 95 log-transformed, winsorized and standardized in each study before analysis . Gestational diabetes 96 status of women without pre-existing type 2 diabetes was primarily defined based on OGTT results 97 in both studies using the International Association of Diabetes and Pregn ancy Study Group 98 (IADPSG) GDM criteria (FPG ≥5.1 mmol/L or higher, or a 1hG ≥10.8 or a 2hG ≥ 8.5 mmol/L or 99 higher)(International Association of et al., 2010). Our secondary outcome was GDM using BiB’s 100 South Asian specific definition (FPG of 5.2 mmol/L or higher, or a 2hG of 7.2 mmol/L or higher) 101 (Farrar et al., 2015), which will be referred to as the South Asian-specific definition hereafter. Self-102 reported GDM status or data from the birth chart was used to determine GDM’s status if OGTT 103 measures were unavailable (N=65 and 31 in START and BiB respectively). Women with pre -104 existing diabetes at baseline were not included in this analysis. Pre-pregnancy diabetes status was 105 determined using maternal self-reported data (about diabetes diagnosis, diabetes medication and/or 106 insulin intake prior to pregnancy) in START. In BiB, information on pre-pregnancy diabetes was 107 backfilled from electronic medical records. 108 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 9 In order to keep a single pregnancy (and a single GDM status) per mother in BiB, only pregnancies 109 with no missing data for GDM were included. For mothers with available data at multiple 110 pregnancies at this stage, pregnancies with no missing data across all covariates (age, BMI, family 111 history, birth country, parity, education level) were prioritized. Next, only pregnancies with the 112 least amount of missing data across all covariates were kept. The following two additional filtering 113 approaches were then applied for mothers with multiple pregnancies remaining: i) if GDM was 114 not diagnosed at any of the pregnancies, phenotype data at the latest available time point was kept 115 (ie. keep older GDM controls). ii) if GDM was diagnosed during any of the pregnancies included 116 in the study, the earliest time point where GDM was diagnosed was kept (ie. keep younger GDM 117 cases). 118 DNA extraction, Genotyping, Imputation and Filtering: 119 START: DNA was extracted and genotyped for 867 mothers using the Illumina Human 120 CoreExome-24 and Infinium CoreExome-24 arrays (Illumina, San-Diego, CA, USA). 837 samples 121 passed standard quality control procedures.(Anderson et al., 2010) Genotypes were phased and 122 imputed using SHAPEIT v2.12 (Delaneau, Marchini, Genomes Project, & Genomes Project, 123 2014), and IMPUTE v2.3.2(Howie, Donnelly, & Marchini, 2009) respectively using the 1000 124 Genomes (phase 3) data as a reference panel.(Consortium et al., 2015) Variants with an info score 125 <0.7 were removed from analysis. In total, 837 START participants with both genotypes and 126 available GDM status, FPG, 1h- and/or 2hG levels were included in the analysis (Figure S1). 127 Born in Bradford: DNA was extracted and genotyped for 16,267 and 3,663 BiB participants using 128 the Illumina HumanCoreExome (12v1.0, 12v1.1 or 24v1.0) and InfiniumGlobal Screening Array 129 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 10 (24v2.0) arrays respectively (Illumina, San-Diego, CA, USA). 4,372 South Asian mothers passed 130 genotyping quality controls, had GDM status, FPG, and/or 2hG levels available, and were included 131 in our analysis (Figure S1). 132 Deriving the PRS : Given the absence of publicly available South Asian -specific T2D or GDM 133 GWAS data at the time of the analysis, weights were derived from the DIAGRAM’s 2014 multi-134 ethnic T2D GWAS meta-analysis, which included over 18% of South Asians (~63% European and 135 19% other ethnic backgrounds).(Mahajan et al., 2014) A grid search approach was used to identify 136 the optimal parameters ( 17 P -values tested , ranging from 5x10-8 to 1 with 0.1 increase ; 4 137 heritability values tested: 0.023; 0.06; 0.08; 0.12). START and BiB genotypes were pooled. 70% 138 of the samples’ data were used for training and 30% for validation (random sampling stratified by 139 study) in order to minimize the impact of population stratification . The PRS was derived using 140 LDpred2.(Prive, Arbel, & Vilhjalmsson, 2020) The best PRS (i.e. that maximized the AUC) was 141 characterized by a P value ≤ 0.0014 and an h2=0.08 (NSNVs=6,492). The PRS was standardized 142 (mean=0, standard deviation=1) in both studies before analysis. 143 Principal component analysis of genetic data 144 A principal component analysis (PCA) was performed using the PC -Air function from the 145 GENESIS R package (v2.20.0).(Conomos, Miller, & Thornton, 2015) Kinship matrices (required 146 to derive PCs with PC-Air) were derived using KING (v2.2.5).(Manichaikul et al., 2010) 147 Statistical Analysis 148 The statistical analysis was conducted using R (v3.6.3).(R core Team, 2016) Linear regression 149 models were used to test the association between the PRS and FPG, 2hG and AUCg. PRS and 150 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 11 GDM associations were tested using logistic regression. Both univariate and multivariate models 151 were constructed with adjustment for GDM risk factors (age, BMI, parity, birth in South Asia (yes 152 vs. no), education level, and diet quality (in START only) and the first 5 PCs (in order to minimize 153 the effect of population stratification). Interactions between the PRS and each risk factors was also 154 tested. The estimated population attributable fractions (PAFs) and their corresponding standard 155 errors were calculated using the AF R package (v.0.1.5). To this end, continuous variables were 156 recoded into categorical variables: age was divided in two categories [(29-31, 32-43) vs. 19-28]; 157 BMI was stratified into a two categories variable using South Asian obesity cutoff points suggested 158 by Gray et al. (Gray et al., 2011) (<23 vs. ≥23); The PRS was divided into two categories (tertiles 159 1+2 vs tertile 3); Parity was divided in to two categories (primiparity vs. 1 pregnancy or more); 160 Education level variables were divided in two categories ( completed high school or lower vs. 161 higher degree, diploma or certificate in START; and A-level equivalent or lower vs. higher than 162 A-level in BiB). 163

164 The proportion of women classified with GDM using the IADPSG criteria was 25% and 11.2% in 165 START and BiB respectively, which was lower than the proportion using the South Asian-specific 166 definition of 36.2% and 22.9% respectively. Notably the proportion of women with GDM was 167 higher in START compared to BiB irrespective of the classification method used. 168 The proportion of women of Indian origin in START and BiB was 71.8% and 5.1%, while the 169 proportion of Pakistani women was 23.4% and 94.3% respectively. The proportion of participants 170 born in the Indian sub -continent was higher in START (88 .6%) than in BiB (5 5.6%), and the 171 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 12 average number of years spent in Canada or the UK among these participants was lower in START 172 compared to BiB (6.6 vs. 9.7 years respectively). The proportions of primiparous women (40.9% 173 vs 31.7%) and women with 1 prior pregnancy (42.4% vs 26.9%) were higher in START than in 174 BiB. Conversely, participants with 2 or more prior pregnancies were more frequent in BiB than 175 START (41.4% vs. 16.6% respectively). The proportion of vegetarian participants was higher in 176 START than in BiB (36.4% vs 1.3%). Finally, the proportion of participants with a post-secondary 177 degree/diploma or higher was greater in START than BiB (84.0% vs 29.0%). 178 Table 1 shows the baseline characteristics of the South Asian women from the START and BiB 179 stratified by GDM case vs non GDM (IADPSG criteria). As expected, women with GDM had a 180 higher mean fasting, 2hG and AUCg levels than non -GDM participants. Participants with GDM 181 were older, had a higher BMI, and were more likely to report a family history of diabetes compared 182 to women without GDM, in both studies. The overall diet quality was lower in participants with 183 GDM compared to non-GDM participants in START (data not available in B iB). Of note, the 184 average difference in BMI between GDM cases and controls was higher in BiB than in START 185 (2.8 and 1.8 respectively) (Table 1). 186 The standardized PRS ranged between -3.23 and 3.12 in START as compared to -3.51 and 4.16 in 187 BiB. The full list of genetic variants included in the PRS as well as their characteristics are shown 188 in Table S1 . Women with GDM had a higher mean PRS compared to women without GDM. 189 Similarly, women with GDM were more likely to have PRS categorized in tertile 2 or 3, compared 190 to tertile 1 (Table 1). 191 Genetic risk and GDM related traits in univariate models 192 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 13 The continuous PRS was associated with FPG, 2hG, and AUCg in START and BiB in univariate 193 models. Every 1 SD increase in the PRS was associated with a 0.09 mmol/L increase in FPG 194 [95%CI=0.07-0.10], 0.23 mmol/L increase in 2hG [95%CI=0.18-0.28], and a 0.17 unit increase in 195 AUCg z-scores [0.14 - 0.20] in the meta-analysed results (Table S2). 196 The PRS was also associated with the risk of GDM IADPSG in univariate models whereby a 1 SD 197 increase in PRS was associated with a 47% increase in risk of GDM after meta -analysis 198 [95%CI=35-60%]. A similar association is observed using the South Asian-specific definition of 199 GDM, with moderate between-study heterogeneity observed (Table S2). 200 Overall, the risk of GDM IADPSG increased progressively comparing tertile 2 of the PRS to tertile 1, 201 and tertile 3 to tertile 1 (43% and 230% respectively, Table S2). Higher PRS categories were also 202 associated with higher FPG, 2hG and AUCg levels (Table S2). 203 Multivariable models of GDM risk factors and GDM related traits. 204 The continuous PRS was strongly and independently associated with FPG, 2hG and AUCg levels 205 in a multivariable model adjusted for age, BMI, parity, parental history of diabetes, region of birth 206 (South Asia vs. other), education level, and diet quality (available in START only), and the first 5 207 PCs (Table 2). For example, every 1 SD increase in the PRS was associated with a 0.08 mmol/L 208 increase in FPG, and 0.21 mmol/L increase in 2hG levels (Table 2). The continuous PRS was also 209 associated with a higher risk of GDM in a model with similar adjustments whereby every 1 SD 210 increase in the PRS was associated with a 45% increase in the risk of GDM IADPSG (Table 2 ). 211 Association results for GDM using the South Asian-specific criteria are shown in Table S3 212 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 14 When testing tertiles of PRS with similar covariates, our results show that p articipants in the 2 nd 213 and 3 rd PRS tertile s have a 37% and 19% increase in the risk of GDM IADPSG compared to 214 participants in tertile 1 respectively (Table S4). Higher PRS tertiles w ere also associated with 215 higher FPG, 2hG and AUCg levels ( Table S4). The effect sizes associated with tertiles 2 were 216 higher in START than BiB across multiple GDM related traits (2hG, AUCg and GDM, Table S4). 217 Population attributable fraction 218 In a model adjusted for maternal age, BMI, education, birth in South Asia (yes/no), parental history 219 of diabetes, and diet quality (in START only), the PRS tertile 3 accounted for 12.5% of the 220 population’s total GDM IADPSG cases overall, and was higher in START than in BiB (Table 3). The 221 combined effect of P RS and parental history of diabetes o n GDM accounted for ~21.7% of the 222 population’s GDM cases in the two studies combined (Table 3). 223 Interactions between the PRS and GDM risk factors on GDM 224 No consistent interactions were observed between the PRS and maternal age; parity; or education 225 level modulating FPG, 2hG, AUCg, or GDM IADPSG in START or BiB (Table 4 and Table S5). 226 Some nominally significant interactions modulating the continuous trait of FPG were observed in 227 START which were not confirmed in BiB. These included the PRS with BMI and the PRS×birth 228 in South Asia (yes/no) interactions (Pinteraction=0.01 and 0.04 respectively), yet non-significant in 229 BiB (Pinteraction=0.05 and 0. 07 respectively), with different effect sizes between the two studies 230 (Table S6), resulting in non-significant meta -analysis of these effects (Pinteraction=0.45 and 0.2 6 231 respectively). Lastly, a PRS×diet quality interaction on FPG was detected in START 232 (Pinteraction=0.002, Table 4) whereby the effect of the PRS appeared to be stronger in participants 233 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 15 with a low diet quality (Beta=0.17 [95%CI=0.10-0.24]) than in participants with a medium or high 234 diet quality (Beta=0.05 [95%CI=0.00-0.09]) (Table S6 and Figure S2). The overall diet quality 235 score was not available BiB, hence this interaction could not be tested for replication. 236

237 We demonstrate that a type 2 diabetes polygenic risk score, based on an independent and multi-238 ethnic GWA S meta -analysis (with 18% South Asian participants), is strongly associated with 239 GDM and related glucose traits among South Asian pregnant women settled in Canada or the UK. 240 This association is independent of other known GDM risk factors, including maternal age, BMI, 241 parental history of diabetes, and maternal birth country. The PRS highest tertile accounted for 13.8 242 % of the PAF of GDM. Consistent with a recent trans-ethnicity GWAS of GDM, and these results 243 support the hypothesis that GDM and type 2 diabetes are part of the same underlying 244 pathology.(Pervjakova et al., 2021) 245 Family history of type 2 diabetes is often used as a surrogate marker of the genetic risk of type 2 246 diabetes. Our results show that the addition of the PRS to the multivariate models does not nullify 247 the impact of parental history on GDM and vice versa. This suggests that the PR S and family 248 history of diabetes both partially convey independent information. This partial independence could 249 be explained by the fact that the PRS does not entirely capture the genetic association signals with 250 GDM. On the other hand, family history reflects not only genetic similarity, but also shared non-251 genetic lifestyle factors. 252 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 16 Overall, we found no robust evidence for modulation of the PRS ’s effect on GDM-related traits 253 by other GDM risk factors. While marginal PRS×BMI and PRS×South Asia born interactions on 254 FPG were observed, these did not replicate, both in terms of statistical significance and effect sizes, 255 suggesting that these environmental and genetic factors may act independently. Furthermore, these 256 interactions would not pass multiple testing corrections if applied. A potentially stronger PRS×diet 257 quality interaction modulating FPG was observed in START. However, since it was not possible 258 to replicate this interaction in BiB, f uture investigations are required in order to validate this 259 observation. If confirmed, this interaction may help identify a subpopulation who will benefit the 260 most from a targeted diet intervention for the prevention of GDM. 261 The overall clinical implications of our findings should be carefully considered. At present, the 262 use of laboratory-derived genetic information in the clinical setting remains expensive and is not 263 implemented for complex diseases like GDM. Given the enhanced predictive value of our genome-264 wide PRS, future evaluations of whether the knowledge of one’s genetic risk improves adherence 265 to lifestyle recommendations and contributes to reducing gestational hyperglycemia would be of 266 great interest. 267 Our study has been considerably strengthened by the use of a PRS optimized for a large population 268 of South Asians from two independent cohorts , as well as by the fact that GDM status was 269 determined using objective OGTT measures . Nevertheless, there are some limitations to our 270 analysis that should be considered: i) The weights attributed to the genetic variants included in the 271 PRS are derived from a type 2 diabetes study. Overall, evidence points to a strong correlation 272 between top variants from type 2 diabetes and GDM GWASs. However, variants at some common 273 loci (eg. MTNR1B) might have significantly different effect si ze depending on the phenotype 274 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 17 studied.(Pervjakova et al., 2021) In addition, variants in at least one locus ( HKDC1) have been 275 strongly associated to GDM but not type 2 diabetes.(Pervjakova et al., 2021) More GDM-specific 276 loci, or loci with a different magnitude of effect between GDM and type 2 diabetes might be 277 identified from future, larger studies. These observations suggest that future PRSs based on a GDM 278 GWAS may have a higher predictive value a nd more power to detect gene×environment 279 interactions. ii) Second, some differences in measurements exist between START and BiB studies, 280 including the tim ing of weight measurements, and the number of data points included in the 281 calculation of AUCg. However, since data was standardized in both studies, we do not expect that 282 AUCg measurements differences had a major impact on the results. iii) Finally, the comparison of 283 genetic data between START and BiB revealed the existence of gene tic heterogeneity , both 284 between and within the samples of these two cohorts (Figure S3). It is our assumption that these 285 differences can be explained by the difference of sample size (START being smaller than BiB), as 286 well as by historical differences in migration patterns from South Asia to Canada and the UK. For 287 example, most START participants were first generation migrants from India, whereas the 288 majority of South Asians in BiB are descendants of Pakistani migrants who settled in the UK for 289 several generations. In order to account for this genetic heterogeneity, we derived a new type 2 290 diabetes PRS that combined samples from the two studies. This PRS should be more generalizable 291 to other South Asian studies. Another measure implemented to reduce the effe ct of population 292 stratification was the adjustment for the PC axes in our analysis. Given the absence of 293 heterogeneity in our FPG, 2hG, or GDM IADPSG PC adjusted models, we consider that population 294 stratification effects have been accounted for. 295 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 18

A type 2 diabetes derived PRS is strongly associated with the risk of GDM in pregnant 296 women of South Asian descent, independent of parental history of diabetes , and other GDM risk 297 factors. 298 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 19

Research projects in START and Born in Bradford are only possible because of the enthusiasm and commitment of the parents and children involved in these two studies. We are grateful to all the participants, teachers, school staff, health professionals and researchers and other contributors who have made these studies happen. Funding Studies: The South Asian Birth Cohort ( START) study data were collected as part of a program funded by the Indian Council of Medical Research in Canada and by the Canadian Institutes of Health Research (grant INC-109205), and the Heart and Stroke Foundation (grant NA7283) with founding principal investigators: Sonia S Anand, Anil Vasudevan, Milan Gupta, Katherine Morrison, Anura Kurpad, Koon K Teo, and Krishnamachari Srinivasan. The Born in Bradford ( BiB) The Born in Bradford cohort is funded by the National Institute for Health Research Collaboration for Applied Health Research and Care (NIHR CLAHRC) and the Programme Grants for Applied Research funding scheme (RP -PG-0407-10044). The study also receives funding from the Wellc ome Trust (WT101597MA), a joint grant from the UK Medical Research Council (MRC) and Economic and Social Science Research Council (ESRC) (MR/N024397/1) and the British Heart Foundation (CS/16/4/32482). DNA extraction was funded by the UK Medical Research C ouncil via the Integrative Epidemiology Unit (MRC IEU; MC_UU_12013/5) and genotyping via the MRC IEU and a National Institute of Health Research Senior Investigator Award to D.A.L. (NF-0616-10102). Authors: Research associate (A.L.) and g raduate student ( J.L.) costs were covered by two Canadian Institutes of Health Research Grants [Project grant number: 298104, Foundation Scheme . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 20 grant number: FDN -143255, Study grant numbers: INC 109205, NA 7283] awarded to S .S.A; D.A.L’s contribution to this study is supported by the Bristol NIHR Biomedical Research Centre, the UK Medical Research Council ( MC_UU_00011/6) and the British Heart Foundation (CH/F/20/90003). S.S.A. is supported by a Tier 1 Canada Research Chain in Ethnic Diversity and Cardiovascular Disease, and a Heart and Stroke Foundation/Michael G. DeGroote Chair in Population Health Research at McMaster University. Competing/Duality of interests D.A.L has received support from Medtronic Ltd and Roche Diagnostics for research unrelated to that presented here. No financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work. Other than those declared by D.A.L above, no authors have any conflict of interest. Author contributions A.L. derived the PRS, performed the association and interaction analysis, drafted and revised the manuscript. J.L. performed the association tests and PAR analysis in START, drafted and revised the manuscript. K.S. reviewed and supervised the statistical analysis and commented on the paper. D.D. is the study coordinator for the START birth cohort and provided comments on the manuscript. B.K. directed BiB data acquisition, p rovided comments on statistical analysis and reviewed the manuscript, R.J.D. provided comments on statistical analysis and reviewed manuscript. G.P. performed the genotyping laboratory analysis and provided comments on the manuscript. D.A.L. provided comments on the analysis and reviewed the manuscript. J.R. the BiB . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 21 study’s principal investigator, he p rovided comments on statistical analysis and reviewed the manuscript. S.S.A. (guarantor) oversaw the development, analysis, and writing of this manuscript. Data and code availability Data from START is not publicly available, since the study is bound by consent which indicates the data will not be used by an outside group. Requests for collaboration or replicatio n will be considered for research purposes only (no commerc ial use allowed, as per the study’s informed consent). Requests should be addressed to the study’s principal investigator (Sonia Anand, [email protected]) via a form which will be provided upon request by emailing [email protected]. The request will be evaluated by PIs and co -investigators, and projects deemed of scientific interest will be further evaluated/validated by local REB chair. Born in Bradford data are available for research purposes only by sending an expression of interest form downloadable from https://borninbradford.nhs.uk/wp- content/uploads/BiB_EoI_v3.1_10.05.21.doct to [email protected] . The proposal will be reviewed by BiB’s executive team. If the request is approved, the requester will be asked to sign a Data Sharing Contract and a Data Sharing Agreement. Full details on how to access data and forms can be found her e https://borninbradford.nhs.uk/research/how-to-access-data/. The code used to analyze the data is available at https://github.com/AmelLamri/Paper_T2dPrsGdm_StartBiB. All Sharable processed versions of the datasets used in the manuscript are made available as supplementary material or at https://github.com/AmelLamri/Paper_T2dPrsGdm_StartBiB. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 22

Anand, S. S., Gupta, M., Teo, K. K., Schulze, K. M., Desai, D., Abdalla, N., . . . South Asian Birth Cohort - Canada, I. (2017). Causes and consequences of gestational diabetes in South Asians living in Canada: results from a prospective cohort study. CMAJ Open, 5(3), E604- E611. doi:10.9778/cmajo.20170027 Anand, S. S., Gupta, M. K., Schulze, K. M., Desai, D., Abdalla, N., Wahi, G., . . . Teo, K. K. (2016). What accounts for ethn ic differences in newborn skinfold thickness comparing South Asians and White Caucasians? Findings from the START and FAMILY Birth Cohorts. Int J Obes (Lond), 40(2), 239-244. doi:10.1038/ijo.2015.171 Anand, S. S., Vasudevan, A., Gupta, M., Morrison, K., Kurpad, A., Teo, K. K., . . . Investigators, S. C. S. (2013). Rationale and design of South Asian Birth Cohort (START): a Canada - India collaborative study. BMC Public Health, 13, 79. doi:10.1186/1471-2458-13-79 Anderson, C. A., Pettersson, F. H., Clarke, G. M., Cardon, L. R., Morris, A. P., & Zondervan, K. T. (2010). Data quality control in genetic case-control association studies. Nat Protoc, 5(9), 1564-1573. doi:10.1038/nprot.2010.116 Archambault, C., Arel, R., & Filion, K. B. (2014). Gestational diabetes and risk of cardiovascular disease: a scoping review. Open Med, 8(1), e1-9. Chen, Y., Liu, H., Wang, L., Zhou, T., Liang, Z., Li, W., . . . Qi, L. (2019). Lifestyle intervention modifies the effect of the MC4R genotype on changes in insulin resistance among women with prior gestational diabetes: Tianjin Gestational Diabetes Mellitus Prevention Program. Am J Clin Nutr. doi:10.1093/ajcn/nqz121 Conomos, M. P., Miller, M. B., & Thornton, T. A. (2015). Robust inference of population structure for ancestry prediction and correction of stratification in the presence of relatedness. Genet Epidemiol, 39(4), 276-293. doi:10.1002/gepi.21896 Consortium, T. G. P., Auton, A., Brooks, L. D., Durbin, R. M., Garrison, E. P., Kang, H. M., . . . Abecasis, G. R. (2015). A global reference for human genetic variation. Nature, 526(7571), 68-74. doi:10.1038/nature15393 Cosson, E., Cussac-Pillegand, C., Benbara, A., Pharisien, I., Jaber, Y., Banu, I., . . . Carbillon, L. (2014). The diagnostic and prognostic performance of a selecti ve screening strategy for gestational diabetes mellitus according to ethnicity in Europe. J Clin Endocrinol Metab, 99(3), 996-1005. doi:10.1210/jc.2013-3383 Delaneau, O., Marchini, J., Genomes Project, C., & Genomes Project, C. (2014). Integrating sequence and array data to create an improved 1000 Genomes Project haplotype reference panel. Nat Commun, 5, 3934. doi:10.1038/ncomms4934 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 23 Dietrich, S., Jacobs, S., Zheng, J. S., Meidtner, K., Schwingshackl, L., & Schulze, M. B. (2019). Gene-lifestyle interaction o n risk of type 2 diabetes: A systematic review. Obes Rev . doi:10.1111/obr.12921 Farrar, D., Fairley, L., Santorelli, G., Tuffnell, D., Sheldon, T. A., Wright, J., . . . Lawlor, D. A. (2015). Association between hyperglycaemia and adverse perinatal outcomes in south Asian and white British women: analysis of data from the Born in Bradford cohort. Lancet Diabetes Endocrinol, 3(10), 795-804. doi:10.1016/S2213-8587(15)00255-7 Farrar, D., Simmonds, M., Bryant, M., Sheldon, T. A., Tuffnell, D., Golder, S., . . . Lawlor, D. A. (2016). Hyperglycaemia and risk of adverse perinatal outcomes: systematic review and meta-analysis. BMJ, 354, i4694. doi:10.1136/bmj.i4694 Gray, L. J., Yates, T., Davies, M. J., Brady, E., Webb, D. R., Sattar, N., & Khunti, K. (2011). Defining obesity cut -off points for migrant South Asians. PLoS One, 6 (10), e26464. doi:10.1371/journal.pone.0026464 Grotenfelt, N. E., Wasenius, N. S., Rono, K., Laivuori, H., Stach -Lempinen, B., Orho-Melander, M., . . . Eriksson, J. G. (2016). Interaction between rs10830963 polymorphism in MTNR1B and lifestyle intervention on occurrence of gestational diabetes. Diabetologia, 59(8), 1655- 1658. doi:10.1007/s00125-016-3989-1 Hayes, M. G., Urbanek, M., Hivert, M. F., Armstrong, L. L., Morrison, J., Guo, C., . . . Group, H. S. C. R. (2013). Identification of HKDC1 and BACE2 as genes influencing glycemic traits during pregnancy through genome-wide association studies. Diabetes, 62(9), 3282-3291. doi:10.2337/db12-1692 Hedderson, M. M., Darbinian, J. A., Quesenberry, C. P ., & Ferrara, A. (2011). Pregravid cardiometabolic risk profile and risk for gestational diabetes mellitus. Am J Obstet Gynecol, 205(1), 55 e51-57. doi:10.1016/j.ajog.2011.03.037 Howie, B. N., Donnelly, P., & Marchini, J. (2009). A flexible and accurate ge notype imputation

for the next generation of genome -wide association studies. PLoS Genet, 5 (6), e1000529. doi:10.1371/journal.pgen.1000529 International Association of, D., Pregnancy Study Groups Consensus, P., Metzger, B. E., Gabbe, S. G., Persson, B., Buchanan, T. A., . . . Schmidt, M. I. (2010). International association of diabetes and pregnancy study groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care, 33(3), 676-682. doi:10.2337/dc09-1848 Jang, H. C., Min, H. K., Lee, H. K., Cho, N. H., & Metzger, B. E. (1998). Short stature in Korean women: a contribution to the multifactorial predisposition to gestational diabetes mellitus. Diabetologia, 41(7), 778-783. doi:10.1007/s001250050987 Kelemen, L. E., Anand, S. S., Vuksan, V., Yi, Q., Teo, K. K., Devanesen, S., . . . Investigators, S. (2003). Development and evaluation of cultural food frequency questionnaires for South Asians, Chinese, and Europeans in North America. J Am Diet Assoc, 103 (9), 1178-1184. doi:10.1053/jada.2003.50578 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 24 Kwak, S. H., Kim, S. H., Cho, Y. M., Go, M. J., Cho, Y. S., Choi, S. H., . . . Park, K. S. (2012). A genome-wide association study of gestational diabetes mellitus in Korean women. Diabetes, 61(2), 531-541. doi:10.2337/db11-1034 Lamri, A., Mao, S., Desai, D., Gupta, M., Pare, G., & Anand, S. S. (2020). Fine-tuning of Genome- Wide Polygenic Risk Scores and Prediction of Gestational Diabetes in South Asian Women. Sci Rep, 10(1), 8941. doi:10.1038/s41598-020-65360-y Mahajan, A., Go, M. J., Zhang, W., Below, J. E., Gaulton, K. J., Ferreira, T., . . . Type 2 Diabetes Genetic Exploration by Nex-generation sequencing in muylti-Ethnic Samples, C. (2014). Genome-wide trans-ancestry meta-analysis provides insight into the genetic archite cture of type 2 diabetes susceptibility. Nat Genet, 46(3), 234-244. doi:10.1038/ng.2897 Manichaikul, A., Mychaleckyj, J. C., Rich, S. S., Daly, K., Sale, M., & Chen, W. M. (2010). Robust relationship inference in genome -wide association studies. Bioinformatics, 26(22), 2867- 2873. doi:10.1093/bioinformatics/btq559 McIntyre, H. D., Catalano, P., Zhang, C., Desoye, G., Mathiesen, E. R., & Damm, P. (2019). Gestational diabetes mellitus. Nat Rev Dis Primers, 5 (1), 47. doi:10.1038/s41572 -019- 0098-8 Pervjakova, N., Moen, G. -H., Borges, M. -C., Ferreira, T., Cook, J. P., Allard, C., . . . Mägi, R. (2021). Trans -ancestry genome -wide association study of gestational diabetes mellitus highlights genetic links with type 2 diabetes. medRxiv. doi:10.1101/2021.10.11.21264235 Popova, P. V., Klyushina, A. A., Vasilyeva, L. B., Tkachuk, A. S., Bolotko, Y. A., Gerasimov, A. S., . . . Grineva, E. N. (2017). Effect of gene -lifestyle interaction on gestational diabetes risk. Oncotarget, 8(67), 112024-112035. doi:10.18632/oncotarget.22999 Prive, F., Arbel, J., & Vilhjalmsson, B. J. (2020). LDpred2: better, faster, stronger. Bioinformatics. doi:10.1093/bioinformatics/btaa1029 R core Team. (2016). R: A language and environment for statistical computing (Version 3.3). Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.R- project.org/ Sattar, N., & Greer, I. A. (2002). Pregnancy complications and maternal cardiovascular risk: opportunities for intervention and scre ening? BMJ, 325 (7356), 157 -160. doi:10.1136/bmj.325.7356.157 Solomon, C. G., Willett, W. C., Carey, V. J., Rich -Edwards, J., Hunter, D. J., Colditz, G. A., . . . Manson, J. E. (1997). A prospective study of pregravid determinants of gestational diabetes mellitus. JAMA, 278(13), 1078-1083. Vounzoulaki, E., Khunti, K., Abner, S. C., Tan, B. K., Davies, M. J., & Gillies, C. L. (2020). Progression to type 2 diabetes in women with a known history of gestational diabetes: systematic review and meta-analysis. BMJ, 369, m1361. doi:10.1136/bmj.m1361 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 25 Wright, J., Small, N., Raynor, P., Tuffnell, D., Bhopal, R., Cameron, N., . . . Born in Bradford Scientific Collaborators, G. (2013). Cohort Profile: the Born in Bradford multi -ethnic family cohort study. Int J Epidemiol, 42(4), 978-991. doi:10.1093/ije/dys112 Zhang, C., & Ning, Y. (2011). Effect of dietary and lifestyle factors on the risk of gestational diabetes: review of epidemiologic evidence. Am J Clin Nutr, 94 (6 Suppl), 1975S-1979S. doi:10.3945/ajcn.110.001032 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 26 URLs 1000 Genomes study https://www.internationalgenome.org/ Born in Bradford Study https://borninbradford.nhs.uk/ DIAGRAM consortium https://www.diagram-consortium.org/ GCTA: a tool for Genome-wide Complex Trait Analysis https://cnsgenomics.com/software/gcta KING: Relationship Inference https://people.virginia.edu/~wc9c/KING/manual.html PC-AiR: Principal Components Analysis in Related Samples https://rdrr.io/bioc/GENESIS/man/pcair.html PLINK https://www.cog-genomics.org/plink/ R Project https://cran.r-project.org/ . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 27 TABLES . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 28 Table 1: Characteristics of START and BiB study participants included in the analysis. START BiB No GDM GDM P-value No GDM GDM P-value N (%) 759 (75) 253 (25) - 3809 (88.8) 481 (11.2) - Age, years 29.8 (3.8) 31.6 (4) 5.55 × 10-10 27.7 (5) 30.5 (5.4) 1.40 × 10-22 Height, cm 162.5 (6.27) 161.13 (6.01) 0.002 159.9 (5.69) 158.3 (5.66) 6.19 × 10-08 Weight, kg a 61.7 (11.7) 65.6 (12.9) 2.00 × 10-05 64.8 (14.1) 71.1 (15.1) 2.22 × 10-16 BMI, kg/m2 b 23.4 (4.3) 25.3 (4.9) 4.93 × 10-08 25.4 (5.2) 28.4 (5.8) 5.89 × 10-23 Parity, n (%) 0 328 (44.3%) 78 (31.1%) 0.001 1189 (32.2%) 129 (27.4%) 5.03 × 10-07 1 299 (40.4%) 122 (48.6%) 1026 (27.8%) 94 (20%) 2 or more 114 (15.4%) 51 (20.3%) 1473 (39.9%) 248 (52.7%) Post-secondary education, n (%) 641 (84.6%) 208 (82.2%) 0.43 952 (29.4%) 110 (26.3%) 0.22 Country of origin/ancestry, n (%) India 567 (74.7%) 160 (63.2%) 0.001 175 (5.2%) 19 (4.3%) 0.37 c Pakistan 163 (21.5%) 74 (29.2%) 3198 (94.2%) 425 (95.5%) Other 29 (3.8%) 19 (7.5%) 23 (0.7%) 1 (0.2%) Born in South Asia, n (%) 671 (88.5%) 225 (88.9%) 0.95 1836 (54.2%) 291 (65.7%) 6.50 × 10-06 Years in recruitment country (Canada/UK) d 6.4 (5.8) 7.4 (5.8) 0.02 9.3 (9) 12.1 (9.4) 3.88 × 10-06 Parental history of diabetes, n (%) 282 (37.3%) 142 (56.1%) 2.25 × 10-07 891 (27.4%) 170 (38.9%) 8.88 × 10-07 Vegetarians, n (%) 266 (37%) 84 (34.6%) 0.54 12 (1.3%) 1 (1.1%) > 0.99 c Low diet quality, n (%) 180 (24) 88 (35.1) 8.00 × 10-04 - - - Polygenic risk score (z-scores) -0.11 (1) 0.347 (0.93) 1.51 × 10-08 -0.04 (0.99) 0.32 (1.04) 4.98 × 10-12 Polygenic risk score Tertile 1 240 (37.7%) 39 (19.4%) 2.74 × 10-06 1309 (34.4%) 117 (24.3%) 7.60 × 10-10 Tertile 2 206 (32.4%) 73 (36.3%) 1291 (33.9%) 142 (29.5%) Tertile 3 190 (29.9%) 89 (44.3%) 1209 (31.7%) 222 (46.2%) Fasting plasma glucose, mmol/L 4.27 (0.32) 5.02 (0.83) 5.51 × 10-32 4.53 (0.41) 5.34 (1.14) 3.18 × 10-43 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 29 1h post-load glucose, mmol/L 7.31 (1.38) 10.26 (2.02) 6.04 × 10-57 - - - 2h post-load glucose, mmol/L 5.96 (1.16) 8.47 (2.16) 1.53 × 10-42 5.49 (1.02) 9.14 (1.97) 1.57 × 10-155 Area under curve glucose, mmol.h e 12.43 (1.83) 17.02 (2.89) 2.27 × 10-63 10.02 (1.21) 14.48 (2.77) 3.82 × 10-133 Characteristics of participant s with available PRS and GDM IADPSG, FPG, 1h, 2h post -load glucose levels or AUC glucose data. Presented data are means (Standard Deviation) unless otherwise indicated. P-Values are calculated from Chi-squared test for categorical variables and independent t-test for continuous variables. a pre-pregnancy values in START vs. weight at antenatal clinic (average 12 completed weeks of pregnancy) in BiB. b Derived using height measured at initial visit (in both studies) and pre-pregnancy weights (START) or antenatal clinic weights (BiB).c approximation may be incorrect due to small counts. d Canada for START samples and UK for BiB. e derived using fasting, 1h and 2h post-load measurements in START vs. fasting and 2h post-load measurements in BiB. Abbreviations: AUC, area under the curve glucose; BiB, Born in Bradford; BMI, body mass index; GDM, gestational diabetes mellitus; IADPSG, International Association of Diabetes and Pregnancy Study Groups; START, south Asian birth cohort; T2D, type 2 diabetes; UK, United Kingdom; vs. versus. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 30 Table 2: Association between GDM risk factors and GDM related traits: results from multivariate models in START and BiB cohorts. START BiB Meta-analysis Dependent Variable Independent Variables Beta/OR [95%CI] P-value Beta/OR [95%CI] P-value Beta (SE) / OR [95%CI] P-value I2 QE P- value Fasting glucose PRS (per 1SD increase) 0.083 [0.043 - 0.123] 6.00 × 10-05 0.085 [0.065 - 0.105] 1.67 × 10-16 0.085 [0.067 - 0.103] 2.85 × 10-20 0 0.92 Age (year) 0.021 [0.01 - 0.032] 2.00 × 10-04 0.014 [0.009 - 0.019] 2.49 × 10-08 0.015 [0.011 - 0.02] 4.19 × 10-11 22 0.26 BMI (kg/m2) 0.024 [0.014 - 0.033] 5.51 × 10-07 0.032 [0.028 - 0.036] 6.53 × 10-53 0.031 [0.027 - 0.034] 7.99 × 10-60 63 0.1 Born in South Asia (Yes/No) 0.037 [-0.088 - 0.162] 0.56 0.08 [0.039 - 0.122] 2.00 × 10-04 0.076 [0.037 - 0.115] 2.00 × 10-04 0 0.52 Parental history of T2D (Yes/No) 0.04 [-0.043 - 0.123] 0.34 0.066 [0.02 - 0.111] 0.005 0.06 [0.02 - 0.1] 0.003 0 0.6 Parity -0.046 [-0.102 - 0.01] 0.11 -0.015 [-0.033 - 0.004] 0.13 -0.018 [-0.036 - 0] 0.05 9 0.29 Education level (per level) -0.031 [-0.068 - 0.006] 0.1 -0.016 [-0.035 - 0.002] 0.09 -0.019 [-0.036 - -0.002] 0.02 0 0.49 Low diet quality (Yes/No) 0.102 [0.01 - 0.193] 0.03 - - - - - - 2h postload glucose PRS (per 1SD increase) 0.189 [0.068 - 0.311] 0.002 0.211 [0.156 - 0.266] 7.87 × 10-14 0.207 [0.157 - 0.257] 5.49 × 10-16 0 0.75 Age (year) 0.127 [0.093 - 0.161] 8.34 × 10-13 0.068 [0.055 - 0.082] 8.82 × 10-23 0.076 [0.064 - 0.089] 1.49 × 10-32 90 0.002 BMI (kg/m2) 0.047 [0.019 - 0.074] 0.001 0.064 [0.053 - 0.075] 1.43 × 10-29 0.062 [0.051 - 0.072] 3.44 × 10-32 23 0.25 Born in South Asia (Yes/No) 0.298 [-0.08 - 0.675] 0.12 0.308 [0.195 - 0.422] 1.14 × 10-07 0.308 [0.199 - 0.416] 3.09 × 10-08 0 0.96 Parental history of T2D (Yes/No) 0.361 [0.109 - 0.613] 0.005 0.242 [0.117 - 0.366] 1.00 × 10-04 0.265 [0.154 - 0.377] 3.23 × 10-06 0 0.4 Parity -0.279 [-0.45 - -0.109] 0.001 -0.095 [-0.146 - -0.043] 3.00 × 10-04 -0.11 [-0.16 - -0.061] 1.00 × 10-05 76 0.04 Education level (per level) -0.063 [-0.176 - 0.051] 0.28 -0.073 [-0.124 - -0.022] 0.005 -0.071 [-0.118 - -0.025] 0.002 0 0.87 Low diet quality (Yes/No) 0.365 [0.086 - 0.644] 0.01 - - - - - - AUC glucose PRS (per 1SD increase) 0.165 [0.099 - 0.231] 1.08 × 10-06 0.152 [0.119 - 0.185] 2.50 × 10-19 0.155 [0.125 - 0.184] 7.74 × 10-25 0 0.74 Age (per year) 0.068 [0.05 - 0.087] 1.16 × 10-12 0.043 [0.035 - 0.051] 1.01 × 10-24 0.047 [0.039 - 0.054] 3.34 × 10-35 84 0.01 BMI (kg/m2) 0.047 [0.032 - 0.062] 2.12 × 10-09 0.047 [0.041 - 0.054] 8.89 × 10-44 0.047 [0.041 - 0.053] 4.27 × 10-53 0 0.94 Born in South Asia (Yes/No) 0.081 [-0.123 - 0.285] 0.44 0.201 [0.133 - 0.269] 8.15 × 10-09 0.189 [0.124 - 0.253] 1.01 × 10-08 17 0.27 Parental history of T2D (Yes/No) 0.122 [-0.015 - 0.258] 0.08 0.138 [0.063 - 0.213] 3.00 × 10-04 0.134 [0.069 - 0.2] 6.00 × 10-05 0 0.83 Parity -0.122 [-0.214 - -0.029] 0.01 -0.057 [-0.088 - -0.026] 3.00 × 10-04 -0.063 [-0.093 - -0.034] 2.00 × 10-05 41 0.19 Education level (per level) -0.045 [-0.106 - 0.016] 0.15 -0.045 [-0.075 - -0.014] 0.004 -0.045 [-0.072 - -0.017] 0.001 0 0.99 Low diet quality (Yes/No) 0.215 [0.064 - 0.366] 0.005 - - - - - - PRS (per 1SD increase) 1.56 [1.3 - 1.88] 2.97 × 10-06 1.42 [1.27 - 1.59] 1.09 × 10-09 1.45 [1.32 - 1.6] 2.27 × 10-14 0 0.4 Age (year) 1.13 [1.07 - 1.19] 2.50 × 10-06 1.1 [1.07 - 1.13] 1.07 × 10-13 1.11 [1.08 - 1.13] 1.98 × 10-18 0 0.4 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 31 GDM (IADPSG criteria) BMI (kg/m2) 1.08 [1.04 - 1.12] 1.00 × 10-04 1.08 [1.06 - 1.11] 1.01 × 10-14 1.08 [1.06 - 1.1] 6.25 × 10-18 0 0.87 Born in South Asia (Yes/No) 1.35 [0.78 - 2.43] 0.3 1.72 [1.35 - 2.19] 1.00 × 10-05 1.65 [1.33 - 2.06] 8.37 × 10-06 0 0.44 Parental history of T2D (Yes/No) 1.67 [1.17 - 2.38] 0.005 1.53 [1.21 - 1.94] 5.00 × 10-04 1.57 [1.29 - 1.92] 7.06 × 10-06 0 0.69 Parity 0.86 [0.68 - 1.09] 0.23 0.87 [0.79 - 0.95] 0.003 0.87 [0.79 - 0.95] 0.001 0 0.99 Education level (per level) 0.89 [0.76 - 1.05] 0.18 0.9 [0.81 - 0.99] 0.04 0.9 [0.82 - 0.98] 0.01 0 0.96 Low diet quality (Yes/No) 1.68 [1.14 - 2.47] 0.008 - - - - - - Models were additionally adjusted for the first 5 PCs of each study. Abbreviations: BiB, Born in Bradford; BMI, Body mass index; CI, Confidence interval; GDM, Gestational diabetes mellitus; IADPSG, International Association of Diabetes and Pregnancy Study Groups; OR, Odds ratio; SA, South Asia; SD, Standard deviation; QE P, P-value from the test for (residual) heterogeneity; START, South Asian birth cohort; T2D, Type 2 diabetes. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 32 Table 3: Population attributable fractions of GDM risk factors in mothers from the START and Born in Bradford studies (multivariable models). START BiB Meta-analysis Independent Variable AF [95%CI] P-value AF [95%CI] P-value AF [95%CI] P-value I2 QE P- value Age (29-31 vs. 32yr vs. <29 years) 31.2 [17.1 - 45.3] 1.00 × 10-05 20.2 [14.8 - 25.7] 4.72 × 10-13 21.7 [16.6 - 26.8] 9.19 × 10-17 50 0.16 Body mass index (≥ 23 vs. < 23) 21.8 [8.7 - 34.9] 0.001 33.8 [25.4 - 42.2] 2.47 × 10-15 30.3 [23.3 - 37.4] 3.59 × 10-17 56 0.13 Born in SA (Yes vs. No) 13.5 [-17.2 - 44.3] 0.39 19.3 [12.6 - 26] 1.47 × 10-08 19 [12.5 - 25.6] 1.07 × 10-08 0 0.72 Education (Post-secondary vs. less) -18.2 [-46.8 - 10.5] 0.21 -0.8 [-4.6 - 3.1] 0.7 -1.1 [-4.9 - 2.7] 0.58 28 0.24 Parental history of T2D (Yes vs. No) 15.1 [4.4 - 25.7] 0.005 8.3 [4.1 - 12.5] 1.00 × 10-04 9.2 [5.3 - 13.1] 3.54 × 10-06 26 0.24 PRS (Tertile 3 vs. 1+2) 13.8 [4.9 - 22.6] 0.002 12.2 [7.8 - 16.6] 5.14 × 10-08 12.5 [8.6 - 16.5] 4.47 × 10-10 0 0.76 Low Diet Quality (Yes vs. No) 8.9 [1.5 - 16.4] 0.02 - - - - - - Sum PAF of PRS (T3) and parental history of diabetes 28.9 20.5 21.7 GDM status derived using IADPSG criteria. Multivariate models included age, BMI, region of birth (South Asia vs other), education, parental history of diabetes, parity, principal components 1 to 5, and diet quality (START only) when applicable. Abbreviations: BiB, Born in Bradford; BMI, Body mass index, CI, Confidence interval; GDM, Gestational diabetes mellitus ; IADPSG, International Association of Diabetes and Pregnancy Study Groups; PAF, Population attributable fraction; PRS, Polygenic risk score; QE P, P-value from the test for (residual) heterogeneity; START, South Asian birth cohort. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 33 Table 4: Interaction effects between GDM risk factors and T2D PRS in START and BiB. START BiB Meta-analysis Dependent Variable Interaction term Beta/OR [95%CI] a Pinteraction Beta/OR [95%CI] a Pinteraction Beta/OR [95%CI] a Pinteraction I2 QE P- value Fasting glucose PRS x Age -0.006 [-0.016 - 0.003] 0.2 0.004 [0 - 0.008] 0.07 0.002 [-0.001 - 0.006] 0.23 72 0.06 PRS x BMI -0.01 [-0.019 - -0.002] 0.01 0.004 [0 - 0.008] 0.05 0.001 [-0.002 - 0.005] 0.42 89 0.002 PRS x Born in South Asia -0.137 [-0.268 - -0.006] 0.04 0.037 [-0.003 - 0.078] 0.07 0.022 [-0.016 - 0.061] 0.26 84 0.01 PRS x Parental history of T2D -0.059 [-0.139 - 0.022] 0.15 0.016 [-0.028 - 0.061] 0.48 -0.001 [-0.04 - 0.037] 0.94 61 0.11 PRS x Parity -0.014 [-0.062 - 0.034] 0.56 0.004 [-0.01 - 0.018] 0.6 0.002 [-0.011 - 0.016] 0.73 0 0.48 PRS x Education level -0.014 [-0.05 - 0.022] 0.45 -0.005 [-0.022 - 0.013] 0.6 -0.006 [-0.022 - 0.009] 0.42 0 0.65 PRS x Low diet quality 0.141 [0.053 - 0.228] 0.002 - - - - - - 2h post- load glucose PRS x Age 0 [-0.03 - 0.03] 0.98 0.01 [-0.001 - 0.021] 0.07 0.009 [-0.001 - 0.019] 0.08 0 0.54 PRS x BMI -0.022 [-0.047 - 0.003] 0.09 0 [-0.01 - 0.011] 0.94 -0.003 [-0.012 - 0.007] 0.56 61 0.11 PRS x Born in South Asia -0.191 [-0.586 - 0.205] 0.34 0.072 [-0.039 - 0.182] 0.2 0.053 [-0.054 - 0.159] 0.33 36 0.21 PRS x Parental history of T2D -0.092 [-0.335 - 0.151] 0.46 0.055 [-0.066 - 0.177] 0.37 0.026 [-0.083 - 0.135] 0.64 11 0.29 PRS x Parity -0.039 [-0.184 - 0.107] 0.6 0.009 [-0.03 - 0.047] 0.66 0.005 [-0.032 - 0.043] 0.77 0 0.54 PRS x Education level 0.037 [-0.072 - 0.146] 0.51 0.008 [-0.039 - 0.056] 0.73 0.013 [-0.031 - 0.056] 0.56 0 0.64 PRS x Low diet quality 0.068 [-0.199 - 0.335] 0.62 - - - - - - AUC glucose PRS x Age -0.007 [-0.023 - 0.009] 0.41 0.004 [-0.002 - 0.011] 0.19 0.003 [-0.003 - 0.009] 0.36 37 0.21 PRS x BMI -0.014 [-0.027 - 0] 0.05 0.002 [-0.004 - 0.008] 0.52 -0.001 [-0.006 - 0.005] 0.82 77 0.04 PRS x Born in South Asia -0.126 [-0.34 - 0.088] 0.25 0.015 [-0.051 - 0.081] 0.65 0.003 [-0.06 - 0.066] 0.93 35 0.22 PRS x Parental history of T2D -0.027 [-0.158 - 0.105] 0.69 0.025 [-0.048 - 0.098] 0.49 0.013 [-0.051 - 0.077] 0.68 0 0.5 PRS x Parity -0.057 [-0.135 - 0.022] 0.16 0.006 [-0.017 - 0.029] 0.6 0.001 [-0.021 - 0.023] 0.91 56 0.13 PRS x Education level 0.007 [-0.052 - 0.066] 0.82 -0.008 [-0.036 - 0.021] 0.6 -0.005 [-0.03 - 0.021] 0.71 0 0.67 PRS x Low diet quality 0.07 [-0.074 - 0.214] 0.34 - - - - - - PRS x Age 0.99 [0.94-1.03] 0.59 0.99 [0.96-1.01] 0.17 0.99 [0.97-1] 0.14 0 0.92 . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint 34 GDM (IADPSG criteria) PRS x BMI 0.97 [0.94-1.01] 0.15 0.98 [0.96-1] 0.03 0.98 [0.96-0.99] 0.01 0 0.76 PRS x Born in South Asia 0.65 [0.33-1.23] 0.2 1.04 [0.82-1.31] 0.76 0.98 [0.79-1.23] 0.89 41 0.19 PRS x Parental history of T2D 0.72 [0.5-1.04] 0.08 1.07 [0.85-1.35] 0.59 0.95 [0.78-1.16] 0.63 67 0.08 PRS x Parity 0.88 [0.71-1.09] 0.23 0.99 [0.92-1.07] 0.86 0.98 [0.91-1.05] 0.57 15 0.28 PRS x Education level 1.01 [0.85-1.19] 0.93 0.93 [0.85-1.02] 0.12 0.95 [0.87-1.03] 0.2 0 0.4 PRS x Low diet quality 1.26 [0.85-1.89] 0.26 - - - - - -

from models adjusted for age, BMI, education level, birth region (South Asia vs. other), parity, parental history of diabetes, and genetic PC axes 1 to 5. aValues are Beta for continuous dependent variables (fasting 2h, and AUC glucose), and OR for binary variable (i.e. GDM). Abbreviations: AUC, area under the curve; BiB, Born in Bradford; BMI, body mass index; CI, confidence interval; GDM, gestational diabete s mellitus; IADPSG, International Association of Diabetes and Pregnancy Study Groups ; OR, odds ratio; PRS, polygenic risk score; QE P, P-value from the test for (residual) heterogeneity; START, South Asian birth cohort. . CC-BY 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 16, 2022. ; https://doi.org/10.1101/2022.07.13.22277599doi: medRxiv preprint