Genome-wide discovery for diabetes-dependent triglycerides-associated loci

Margaret Sunitha Selvaraj; Kaavya Paruchuri; Sara Haidermota; Rachel Bernardo; Stephen S Rich; Gina M Peloso; Pradeep Natarajan

doi:10.1371/journal.pone.0275934

. 2022 Oct 21;17(10):e0275934. doi: 10.1371/journal.pone.0275934

Genome-wide discovery for diabetes-dependent triglycerides-associated loci

Margaret Sunitha Selvaraj ^1,^2,³, Kaavya Paruchuri ^1,², Sara Haidermota ^1,², Rachel Bernardo ^1,², Stephen S Rich ⁴, Gina M Peloso ^5,^‡,^*, Pradeep Natarajan ^1,^2,^3,^‡,^*

Editor: Arnar Palsson⁶

PMCID: PMC9586367 PMID: 36269708

Abstract

Purpose

We aimed to discover loci associated with triglyceride (TG) levels in the context of type 2 diabetes (T2D). We conducted a genome-wide association study (GWAS) in 424,120 genotyped participants of the UK Biobank (UKB) with T2D status and TG levels.

Methods

We stratified the cohort based on T2D status and conducted association analyses of TG levels for genetic variants with minor allele count (MAC) at least 20 in each stratum. Effect differences of genetic variants by T2D status were determined by Cochran’s Q-test and we validated the significantly associated variants in the Mass General Brigham Biobank (MGBB).

Results

Among 21,176 T2D and 402,944 non-T2D samples from UKB, stratified GWAS identified 19 and 315 genomic risk loci significantly associated with TG levels, respectively. Only chr6p21.32 exhibited genome-wide significant heterogeneity (I² = 98.4%; p_{heterogeneity} = 2.1x10^-15), with log(TG) effect estimates of -0.066 (95%CI: -0.082, -0.050) and 0.002 (95%CI: -0.002, 0.006) for T2D and non-T2D, respectively. The lead variant rs9274619:A (allele frequency 0.095) is located 2Kb upstream of the HLA-DQB1 gene, between HLA-DQB1 and HLA-DQA2 genes. We replicated this finding among 25,137 participants (6,951 T2D cases) of MGBB (p_{heterogeneity} = 9.5x10^-3). Phenome-wide interaction association analyses showed that the lead variant was strongly associated with a concomitant diagnosis of type 1 diabetes (T1D) as well as diabetes-associated complications.

Conclusion

In conclusion, we identified an intergenic variant near HLA-DQB1/DQA2 significantly associates with decreased triglycerides only among those with T2D and highlights an immune overlap with T1D.

Introduction

Diabetes, largely due to type 2 diabetes (T2D), was estimated to afflict 9.3% of population in 2019 and projected to increase to 10.9% by 2045 [1]. Despite ongoing scientific advances [2], T2D remains a leading cause of morbidity and mortality in the US and increasingly worldwide [3, 4]. Novel approaches to discover the factors influencing T2D-related metabolic alterations may yield new insights toward the prevention of T2D-related complications.

Plasma lipid, particularly triglycerides (TG), alterations represent early metabolic changes linked to insulin resistance. Hypertriglyceridemia is often observed among individuals at risk for T2D and is more severe among individuals with poorly controlled T2D [5]. Enhanced hepatic secretion of TG rich lipoproteins due to insulin resistance and delayed clearance involving lipoprotein lipase-mediated lipolysis may further exacerbate hypertriglyceridemia [6]. Hypertriglyceridemia is an independent predictor of cardiovascular disease in T2D [7, 8], as well as a predictor of T2D itself [9]. Characterizing the genetic determinants of TG concentrations specific to those with T2D may yield new insights into diabetes pathogenesis and complications.

Here, we tested the hypothesis that there are genetic variants associated with TG levels specific to T2D using GWAS and heterogeneity analysis in 424,120 participants of the UKB. Further, we assessed the role of the identified lead variant for multiple diabetes-related phenotypes.

Results

Baseline characteristics

The overall study schematic is depicted in S1 Fig Among 424,120 and 25,137 included samples, 21,176 (5.0%) and 6,951 (27.7%) of samples had T2D in UKB and MGBB, respectively. Overall UKB was composed of participants with a mean age (standard deviation [SD]) of 56.6 (8.1) years, 195,966 (46.2%) male, and 356,023 (83.9%) White British self-reported race. MGBB participants were mean 62.1 (16.2) years, 11,579 (46.1%) male, and 21,172 (84.2%) White British self-reported race. As expected, individuals with T2D versus non-T2D had greater median TG concentrations in both cohorts (Table 1).

Table 1. Baseline characteristics for discovery and replication cohorts.

Distribution of samples across the T2D strata in discovery (UKB) and replication (MGBB) cohorts are provided. Number of samples by gender, ancestry and lipid lowering medications are documented. Lipid measurements for the four main lipid class is tabulated based on T2D strata.

Cohorts	UK Biobank		Mass General Brigham Biobank
Strata	T2D	non-T2D	T2D	non-T2D
Number of samples (%)	21176(4.99)	402944(95.01)	6951(27.65)	18185(72.34)
Age (years) mean (SD)	60.22(6.87)	56.36(8.11)	67.79(13.28)	59.93(16.73)
Female samples (%)	8107(38.28)	220047(54.60)	3283(47.23)	10274(56.40)
European samples (%)	16636(78.56)	339387(84.23)	5457(78.51)	15715(86.42)
Lipid lowering medication prescription (%)	13433(63.44)	56045(13.91)	2197(31.61)	2625(14.43)
TG concentration (mg/dL) median [IQR]	171.83[125.09]	129.58[95.66]	125.00[93.00]	97.00[70.00]
HDL-C concentration (mg/dL) mean (SD)	46.04(12.44)	56.51(14.71)	49.78(17.82)	58.88(19.80)
LDL-C concentration (mg/dL) mean (SD)	111.59(34.17)	138.98(33.04)	90.74(35.88)	104.08(35.89)
TC concentration (mg/dL) mean (SD)	183.09(45.98)	222.06(43.27)	168.91(44.12)	185.39(42.83)

Open in a new tab

HLD-C–High-Density Lipoprotein Cholesterol; IQR–Inter quartile range; LDL-C–Low-Density Lipoprotein Cholesterol; MGBB–Mass General Brigham Biobank; SD–Standard Deviation; T2D –Type 2 Diabetes; TG–Triglycerides; TC–Total Cholesterol; UKB–UK Biobank.

T2D-stratified GWAS of TG identified an associated locus on chromosome 6

We performed GWAS on normalized natural log TG stratified by T2D status in the discovery cohort. Among the 402,944 non-T2D samples, 315 significant loci were identified. Among the 21,176 T2D samples, 19 significant loci were identified (S2 Fig and S1 Table). We then assessed for differential TG effects for 67M variants by T2D status using Cochran’s Q-test for heterogeneity. We identified 478 variants which were genome-wide significant, all at chr6p21.32 (lead variant: rs9274619:G>A; I² = 98.4%; p_{heterogeneity} = 2x10^-15) (Fig 1). The most significantly heterogenous variant was an intergenic variant near the HLA-DQB1/DQA2 genes (S3 Fig), where the minor allele (frequency 0.095) decreases natural log TG among those with T2D but yields no difference among those with non-T2D (T2D group: beta = -0.066, p-value = 3.9x10^-15; non-T2D group: beta = 0.002, p-value = 0.21; p_interaction = 1.9x10^-11). We observed that the difference test (Z_Diff) identified this top significant lead variant (rs9274619:A; Z_diff = -7.935) in the HLA locus as well. The mean raw TG measurements were significantly different between the reference and alternative genotypes in T2D samples (rs9274619(G/A or A/G): mean_diff = 5.45 mg/dl; p-value = 1.26x10^-02; rs9274619(A/A): mean_diff = 25.6 mg/dl; p-value = 3.78x10^-03). To evaluate the independence of the lead variant, we clumped the 478 genome-wide significant variants and performed conditional analyses. After clumping, we retained 8 individual signals, and rs3957148 was in strong LD with the lead variant. However, conditional analyses with all these eight clumped variants for rs9274619:A interaction with T2D status did not abrogate the lead variant’s signal (S2 Table). Since BMI is associated with both TG and T2D, we used scaled BMI as an additional covariate in testing the association of the identified lead variant with TG. We observed evidence for persistent albeit attenuated interaction between rs9274619:A and T2D after adjusting for BMI (p_interaction = 2.9x10^-7). We observed a slight reduction in effects in the interaction model with T2D (beta_interaction = -0.046), which shows that BMI has a confounding effect.

Fig 1 — Manhattan plot for heterogeneity p-values comparing T2D and non-T2D groups. Only one locus achieved genome-wide significance and the corresponding variants are colored in green. Lambda GC values from GWAS stratified by T2D status is shown in the figure. Red line: Genome-wide significance (p-value = 5x10^-8), Blue line: Suggestive significance (p-value = 1x10^-5). GWAS–Genome wide association studies; T2D –Type 2 Diabetes.

Individuals with T2D with lower TG concentrations were enriched for rs9274619:A (Fig 2A). Among individuals with normal TG (i.e., <150 mg/dL), rs9274619:A was associated with T2D by 1.23-fold (95% CI 1.15,1.29; p-value 2.8x10^-11). However, among individuals with TG > 450 mg/dl, rs9274619:A was not associated with T2D (OR 0.96, 95% CI 0.76–1.18; p-value 0.67). We replicated the findings in an independent cohort of 25,137 participants (6,951 T2D cases) of MGBB (p_{heterogeneity} = 9.5x10^-3) (Fig 2B). Additionally, we evaluated the association of the lead variant interacting with T2D status with other lipids in discovery cohort (S3 Table). We observed a significant interaction between rs9274619:A and T2D on HDL-C (p_interaction = 2.9x10^-8) with higher concentrations among those with T2D, and nominally greater reductions in LDL-C among those with T2D (p_interaction = 6.0x10^-4).

We next bioinformatically prioritized the putative causal gene responsible for the T2D-dependent TG genetic association observed. Using T2D GWAS summary statistics PoPS prioritized the HLA-DQB1 gene to be one of the top 20 genes along with other known TG genes such as APOE, LPL and APOB. However, HLA-DQB1 was not prioritized in the non-T2D GWAS (S4 Table). Intersecting rs9274619:A with GTEx eQTL data for five different tissues (Methods) shows that the variant is an eQTL for multiple HLA genes including HLA-DQB1 but more significantly for HLA-DQA2 and HLA-DRB6 (S5 Table). We curated all the eQTLs of the three HLA genes from GTEx database, T2D GWAS and correlated the Z-scores. eQTLs of all three HLA genes had a similar degree of correlations, but with opposite directions (S4 Fig). We further interrogated pQTL and mQTL data. rs9274619:A is a pQTL for HLA-DQA2 (beta = 0.31; p-value = 6.2x10^-14) but it is an mQTL for multiple CpG regions at genome-wide significance. We identified 133 cis-associations and mapped the CpGs to Illumina HumanMethylation450 BeadChip (Illumina Inc., San Diego, USA) identification numbers (GEO data: GPL13534) to obtain the corresponding genes. Multiple HLA genes and other genes in chromosome 6 were identified (S6 Table). Gene prioritization using PoPS and QTL curation identified multiple HLA-genes (S5 Fig).

rs9274619:A tags *HLA-DQB1**0302

Since the significant locus was at the HLA region, we correlated the rs9274619:A with 362 imputed HLA genotypes from 11 classes in the UKB. DQB1 and DQA1 were the most strongly correlated with rs9274619:A. Furthermore, DQB1_302 and DQA1_301 were most strongly correlated with rs9274619:A (DQB1_302: r = 0.95, p_correlation<3.83x10^-313; DQA1_301: r = 0.62, p_correlation<3.83x10^-313) (S6A Fig). We subsequently tested the interaction of all 362 HLA genotypes with T2D status on log(TG) as outcome. From this focused assessment of 362 HLA genotypes (S7 Table), 7 passed Bonferroni corrected significance (0.05/362 = 1x10^-4) (S6B Fig). Consistent with our discovery and correlation analyses, only DQB1_302 had a genome-wide significant interaction (p_interaction = 1.05x10^-9). Although rs9274619:A is associated with increased expression of HLA-DQA2 gene in eQTL and pQTL analysis, alleles from these HLA types were not previously imputed in UKB. Since rs9274619:A located between both HLA-DQB1 and HLA-DQA2, the variant could be a potential quantitative trait loci to both the genes.

Since the HLA locus is a strong predictor of type 1 diabetes (T1D), we used polygenic risk scores (PRS) from 67 variants previously associated with T1D as reported by Oram et al. [10], to exclude potential T1D cases in sensitivity analysis. First, we used the 67 SNPs from both HLA and non-HLA loci to create a genome-wide PRS for all the UKB samples. The top one percentile of the samples based on PRS were classified as T1D (N = 4242). Next, from the whole UKB samples, we removed any sample which identified as T1D by either ICD codes or PRS scores. With the remaining 18460 T2D cases, we observe a nominal interaction of the lead SNP with T2D (beta_interaction = -0.026; p_interaction = 1.16x10^-02).

Phenome-wide interaction analyses implicates multiple diabetes-related complications

We assessed the interactions between the rs9274619:A and T2D with 1567 disease conditions as outcomes (combination of incidence and prevalence) adjusted for all the covariates (age, age², sex, race, PC1-10). Using a Bonferroni correction (0.05/1567 = 3.19x10^-5), 45 disease phenotypes exhibited significant interactions. The strongest interaction was for the concomitant diagnosis of type 1 diabetes (T1D) among those with T2D (p_interaction<1.72x10^-274) (S8 Table). We applied logistic regression models stratified by T2D status on the 45 significant phenotypes, while adjusting for all covariates as mentioned above (S9 Table). Multiple diabetes-related microvascular and macrovascular complications including hypoglycemia, retinopathy, polyneuropathy, angiopathy, atherosclerosis and osteomyelitis were significantly associated, with the T2D-specific TG-lowering rs9274619:A allele leading to increased risks (Fig 3). Interestingly, a recent investigation on Anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitides (AAV) by Dahlqvist et al. have identified rs9274619 as a lead variant for myeloperoxidase (MPO)-ANCA association [11], showing its importance in immune related vascular diseases. However, this allele was associated with reduced odds for obesity and related phenotypes.

Fig 3 — Interaction between the rs9274619:A and T2D with multiple disease conditions as outcomes (combination of incidence and prevalence) was modeled while adjusting for all covariates (sex, age, age², race, PC1-10). Bonferroni corrected 45 significant phenotypes were stratified by T2D status and analyzed using logistic regression. The T2D effect estimates for rs9274619:A and the interaction p-value are documented, the disease conditions are ordered based on T2D beta. From the 45 disease conditions tested, highly correlated Type1 diabetic conditions were removed while plotting the figure. PC–Principal Components; PheWAS–Phenome Wide Association Studies; T2D –Type 2 Diabetes.

Since HLA-DQB1 is a GWAS locus for an overlap between T1D and T2D previously referred to as latent autoimmune diabetes in adults (LADA) [12], as well as T1D itself [13], we further explored the relationships between the rs9274619:A, T1D, T2D, and their respective interactions on TGs (S10 Table). The TG-lowering rs9274619:A was strongly associated with T1D after adjusting for T2D (p-value = 8.6x10^-113), but not significantly associated with T2D status after adjusting for T1D (p-value = 0.29). However, when assessing for interactions on TGs, there was still a significant interaction with T2D independent of T1D (beta_interaction = -0.055; p_interaction = 5.28x10^-9) and more strongly with T1D independently of T2D (beta_interaction = -0.252; p_interaction = 5.53x10^-49). Furthermore, we removed T1D samples and the interaction of rs9274619:A with T2D on TG was nominally significant (beta_interaction = -0.022; p_interaction = 2.9x10^-2). We additionally checked the main effects of rs9274619:A for presumed LADA samples (N = 2580) in our dataset and observed a significant association (p-value = 8.37x10-41), showing that overlapping T1D and T2D samples could potentially contribute to the observed association of this locus.

Cousminer et al. reported four loci to be significantly associated with LADA [14], therefore we assessed T2D/TG interactions for these lead variants (S11 Table). None of the variants tested had a genome-wide significant interaction. The variant rs9273368 from HLA-DQB1 (p_interaction = 2.58x10^-5) was genome-wide significant in our heterogeneity analysis (p_{heterogeneity} = 1.8x10^-8) and was in moderate LD with rs9274619:A (R² = 0.3). The four LADA loci were examined for interactions for additional diabetes-related phenotypes as noted in S12 Table.

Metabolic characterizations of interactions with T2D

We secondarily explored the relationship between the rs9274619:A, interacting with T2D, and relationships with other metabolic features in UKB. Both the main effects and interaction models were adjusted for all the covariates (age, age², sex, race, PC1-10). Outcomes assessed included waist/hip ratio (WHR), body mass index (BMI), macronutrients from 24-hour dietary recall surveys, and 60 plasma biomarkers (S13 and S14 Tables). Several features were associated with the rs9274619:A itself, including increased eosinophil and neutrophil counts as well as hemoglobin A1c and C-reactive protein concentrations. We identified several outcomes that demonstrated differential association by T2D status in interaction testing (alpha 0.05/72 = 6.94x10^-4). The TG-lowering rs9274619:A allele was associated with greater concentrations for T2D vs non-T2D for hemoglobin A1c, sex hormone binding globulin, HDL-C, glucose, and apolipoprotein-A1 concentrations. However, the TG-lowering rs9274619:A allele was associated with reduced concentrations for T2D vs non-T2D for urate, BMI, WHR, and reticulocyte count (S13 and S14 Tables). Since HbA1c is a strong predictor of T2D, HbA1c is the most significant biomarker interacting with the lead SNP (S14 Table). When also using HbA1c > = 48 mmol/mol to define T2D samples as described by Young et al. [15], 15,180 samples were identified, with an additional 5212 samples to the previous T2D group. The interaction model of HbA1c-based only T2D with the lead SNP showed higher significance with TG (beta_interaction = -0.159; p_interaction = 1.26x10^-54). Next, we added the 5212 samples to our existing ICD10-based T2D groups and conducted a sensitivity analysis for the interaction. The addition of samples showed consistent significant interaction (beta_interaction = -0.102; p_interaction = 1.03x10^-35).

We further assessed lipid metabolomic data comprising 102,575 UKB samples (5200 T2D cases) and 249 metabolites for interactions. For each normalized metabolomic phenotype we analyzed the main effect of rs9274619:A and the interaction of rs9274619:A with T2D, with both models adjusted for all aforementioned covariates. Using a Bonferroni correction (0.05/249 = 2.01x10^-4), we identified 6 metabolomic features associated with the rs9274619:A and only 1 interaction with T2D (S15 and S16 Tables). With respect to the interaction detected, we observed that the average diameter for LDL particles among rs9274619:A carriers was greater for T2D versus non-T2D.

Discussion

Independent GWAS studies have identified multiple loci strongly associated with TG and T2D separately [16, 17], and we now observe a variant tagging HLA-DQB1/DQA2 associated with TGs only among those with T2D and not among those without T2D. Despite being associated with reduced TGs specifically in T2D, the lead variant is associated with greater diabetes-related complications and an overlap with T1D. Previous studies have shown TG as an independent risk factor for cardiovascular diseases in T2D patients, and a predictor of T2D itself. Through a genome-wide stratified analysis, we now show that the HLA locus may triangulate some of these relationships by specifically influencing T2D-associated triglycerides.

Our study permits several conclusions regarding T2D pathogenesis. First, our study highlights heterogeneous metabolism among individuals with adult-onset diabetes. Hyperinsulinemia contributes to reduced hydrolysis and clearance of TG rich lipoproteins and thus persistence of these atherogenic lipoproteins toward heightened macrovascular risk [18]. TGs have more moderate associations with microvascular complications among diabetics [19]. Indeed, glycemic control is a more potent risk factor for microvascular complications. Here, we find that an immune-related locus (i.e., HLA-DQB1/DQA2 genotype of major histocompatibility class II) linked to reduced TGs interestingly associates with greater microvascular versus macrovascular risk among individuals with T2D. Lipid homeostasis plays a key role in immune cells, where lipids are key constituents of major histocompatibility complex molecules and other cell membrane microdomains [20]. These observations highlight complementary roles of immune dysfunction and hyperinsulinemia in adult-onset diabetes pathogenesis.

Second, the distinct lipid pattern observed by HLA-DQB1*0302 genotype may reflect etiologically distinct subgroups of adult-onset diabetes. The HLA-DQB1*0302 has long been recognized as a very strong risk factor for T1D and tags the potent T1D DR4 risk haplotype [21–24]. Approximately 2–12% of adults diagnosed with type 2 diabetes have glutamic acid decarboxylase autoantibodies (GADA), thereby leading to the proposed term of latent autoimmune diabetes in adults (LADA) [25–27]. Such individuals are often classified as T2D because they typically do not initially require insulin. Indeed, a recent GWAS of LADA showed that HLA-DQB1 was the most significant locus [14]. Recent data-driven approaches to cluster diabetes have grouped diabetes with GADA, traditionally classified as T1D or LADA, as severe autoimmune diabetes (SAID) [28]. Consistent with separate T1D and LADA analyses, the HLA-DQB1 locus is significantly associated with SAID unlike with other diabetes subgroups [29]. Thus, the relatively reduced TG concentrations among adults classified as having T2D and the HLA-DQB1 risk allele may reflect the lack of hypertriglyceridemia typically observed with more typical hyperinsulinemic T2D.

Third, TG concentrations among individuals diagnosed with T2D may help identify individuals with features more consistent with T1D. GADA testing only in adult-onset diabetics with normal TGs would optimize diagnostic yield. Furthermore, with increasingly available genotyping through expanding research testing and widely used direct-to-consumer approaches, HLA genotypes may further improve efficiency of testing. While large-scale randomized controlled trials for LADA are lacking, expert consensus recommend personalized management approaches deviating from conventional T2D and surveillance management [30]. Such approaches include biomarker-based surveillance of residual beta cell function and to determine insulin initiation.

A few limitations of our study deserve mention. First, causal gene prioritization through multiple methods did not converge on a single gene. Whether our observations reflect coordinated regulation merits further study. Based on the current results we were not able to elucidate the exact mechanism of TG lowering by HLA rs9274619:A in the context of T2D. Second, GADA or other islet autoantibodies and C-peptide levels are not present, despite using ICD codes and T1D polygenic risk score in sensitivity analyses, we cannot rule out the possibility of T1D-like features in some of the classified T2D cases. Third, UKB and MGBB are predominantly White and this may limit generalizability to other genetic backgrounds. Finally, Cochran’s Q test is under powered to detect differences in heterogeneity of effect sizes and we may have missed other loci with differences in effects. However, we also performed a difference test and found similar results to using the Cochran’s Q test.

In conclusion, we observed that rs9274619:A linked to HLA-DQB1/DQA2 is associated with reduced TGs only among adults with diabetes. Presence of this allele reflects an autoinflammatory subgroup of adult-onset diabetes most consistent with T1D, without characteristic hyperinsulinemia and thus relatively reduced TG concentrations. Among individuals classified as T2D, these individuals have greater risks for diabetes-associated complications.

Materials and methods

Study participants

We used the UK Biobank (UKB), which is a prospective population-based cohort composed of approximately 500,000 samples with rich phenotypic and genotypic information, as the discovery cohort [31]. UKB includes volunteer residents of the UK aged 40 to 69 years recruited during 2006–2010. The phenotypic information includes details on lifestyle, medical history, food habits, weight, height, body measurements, scans, blood routines, and electronic medical record (EMR) coded data. Out of 488,377 total individuals, we removed unconsented individuals and samples with >10% genotypic missingness thereby retaining 424,120 individuals for whom the TGs were also available.

We used data from the Mass General Brigham Biobank (MGBB), which comprises volunteer patients of the large Mass General Brigham Healthcare system in Massachusetts with greater than 105,000 participants, as replication [32]. In total 36,424 randomly selected individuals were genotyped using three versions of the Multi-Ethnic Genotyping Array (MEGA) Single-Nucleotide Polymorphism (SNP) array (Multiethnic Exome Global (Meg), Human multi-ethnic array (Mega), Expanded multi-ethnic genotyping array (Megex)). Out of 36,424 individuals, we retained 25,137 samples for whom T2D status and TG measurements were available for the current study.

Ethics

All UK Biobank participants gave written, informed consent per the UKB primary protocol. Secondary use of these data was approved by the Massachusetts General Hospital Institutional Review Board (protocol 2021P002228) and was facilitated through UKB application 7089. All MGB Biobank participants provided written, informed consent per the MGBB primary protocol. Secondary use of these data was approved by the Massachusetts General Hospital Institutional Review Board (protocol 2020P000904). All participants consent to broad use of their samples and data for research, no minor participants were included in this study.

Phenotypes

In the UKB, we defined T2D based on self-reported status (data field 20002) and ICD10 codes E11:0–9 (data fields 41202, 41204, 40001, and 40002). The first instance of TG measurement (data field 30870) was defined as the primary lipid phenotype of interests. We also included other lipid levels as secondary outcomes: total cholesterol (TC) (data field 30690), low-density lipoprotein cholesterol (LDL-C) (data field 30780), and high-density lipoprotein cholesterol (HDL-C) (data field 30760). TG measurements were converted to mg/dL by multiplying mmol/L values by 88.57 and natural log transformed. TC, LDL-C and HDL-C values in mmol/L were converted to mg/dL by multiplying 38.67. When lipid-lowering medications were prescribed, TC measurements were divided by 0.8 and LDL-C by 0.7, as previously done [17]. All four lipid measurements were further inverse rank normalized to the residuals scaled by the standard deviation, where the model was adjusted for covariates (sex, age, age², PC1-10).

We curated multiple diseases for UKB samples into phecodes for PheWAS analysis. The PheWAS R package (version PheWAS_0.99.5–4) was used to map ICD codes to phecodes based on the phecode map 1.2 and 1.2b1 from https://phewascatalog.org/ [33]. Codes that failed to map were excluded, which were relatively few and often procedural. Mapped codes were defined as multiple disease conditions and specified as incident or prevalent based on the time of sample collection. Next, we obtained the secondary phenotypes, which included waist circumference (data field 48), hip circumference (data field 49), body mass index (BMI) (data field 23104) and 24-hour diet recall (data field 110001) for downstream analysis. Additionally, we included blood biochemistry (category id 18518) and count (category id 9081) measures. These phenotypes were normalized to a mean 0 and standard deviation 1 for analysis. We obtained the NMR metabolic biomarkers generated by Nightingale Health (Helsinki, Finland) from the first tranche of 249 metabolic biomarkers in 118,032 UKB participants [34]. We included 102,528 samples that intersected with the discovery cohort and each of the metabolites were inverse rank normalized and regressed against the covariates (age, age², sex, race, PC1-10). The residuals were used as our phenotypes in analyses.

In MGBB, electronic health record (EHR) data were used to define incident and prevalent cases based on enrollment date and ICD-9/ICD-10 codes on clinical phenotype definitions from phecode groups [35, 36], where samples with phecode 250.2X were defined as T2D in our study. Similarly, lipid test results, medication information, demographic status of genotyped samples were curated from EHR records. LDL-C was measured directly or calculated using Friedewald equation when TG were <400 mg/dL, all lipid measurements were in mg/dL units. The lipid measurements closest to sequencing date was curated. A sample was defined as on statin medication, if statin treatment was prescribed within the last one year of the sequencing date. We performed phenotype harmonization and normalization for the validation data as described above.

Genotypes

Genetic data from 488,377 UKB samples were assayed using two similar genotyping arrays from Affymetrix (Santa Clara, CA): i) Applied Biosystems UK BiLEVE Axiom Array ii) Applied Biosystems UK Biobank Axiom Array. 49,950 participants with 807,411 markers were genotyped at using the Applied Biosystems UK BiLEVE Axiom Array and 438,427 participants with 825,927 markers were genotyped using the closely related Applied Biosystems UK Biobank Axiom Array. Both arrays shared 95% of marker content and the UK Biobank Axiom array was chosen to capture genome-wide genetic variation (single nucleotide polymorphism (SNPs) and short insertions and deletions (indels)) [31]. The imputation from the UKB array-derived genotypes was performed using merged UK10K and 1000 Genomes phase 3 reference panels [37] and was combined to the Haplotype Reference Consortium (HRC) [38] panel using IMPUTE4 program (https://jmarchini.org/software/) as implemented in IMPUTE2 [39]. We obtained the UKB imputed human leukocyte antigen (HLA) genotypes (data field 22182) composed of classical allelic variation of 11 HLA types (A, B, C, DRB5, DRB4, DRB3, DRB1, DQB1, DQA1, DPB1, DPA1). HLA imputation from allele pairs was performed using HLA*IMP:02 in the UKB, as previously described [31]. Genotypic data in MGBB cohort was generated using three different arrays (Multiethnic Exome Global [MEG], Human multi-ethnic array [MEGA], Expanded multi-ethnic genotyping array [MEGEX]) from Illumina (San Diego, CA).

Statistical analysis

We performed genome wide association analysis (GWAS) stratified based on T2D status. We first performed quality control (QC) of the full UKB dataset regardless of T2D status by applying additional filters, including minor allele frequency (MAF) < 1%, Hardy-Weinberg equilibrium p-value not exceeding 1x10^-15 and genotype missingness > 10% to filter variants, and sample-level genotype missingness > 10%. The QC-passed dataset was used to create NULL model with sex, age, age², genotype array, race and PC1-10 as covariates. We employed REGENIE with leave-one-out-cross-validation (LOOCV) [40] approach adjusted for covariates stated above to perform GWAS on UKB imputed data with minor allele count (MAC) 20 in both T2D and non-T2D samples, independently. We annotated the genomic risk-loci from GWAS summary statistics using FUMA [41].

We tested for differences in effect estimates per genotype by T2D status using Cochran’s Q-test for heterogeneity in the METAL package [42]. For the genome-wide heterogeneity assessment, we used the conventional alpha threshold of 5x10^-8 to assign statistical significance accounting for multiple-hypothesis testing. We validated the outcomes from the Cochran’s Q-test using the Difference Test [43]. We clumped the genome significant variants using plink [44] (p1: 5e-08; r2:0.5; kb:250). Significant lead variant in the discovery dataset were replicated at an alpha threshold of 0.05 accounting for the single SNP assessed.

The significant and replicated variant (lead variant) was pursued for further downstream analysis. Given its genomic location, we correlated the UKB imputed classical HLA genotypes with the significant lead variant using corplot R-package (method-pearson; version-0.90). We applied sample and variant quality measures to the initial data (—geno 0.05;—mind 0.05;—hwe 1E-06) and calculated PRS using plink [44]. We performed regression-based interaction analyses using the model where adiposity-related, diet-related and other blood biomarker phenotypes were separately analyzed with the lead variant along with T2D status. The regression analysis (main and interaction model) was carried out in R, adjusting for sex, age, age², genotype array, race, and PC1-10 as covariates. Bonferroni corrected alpha threshold of 0.05/number of tests was considered statistically significant for these analyses. Fisher’s exact test was performed to test the significance of sample proportions among group of samples.

We implemented the Polygenic Priority Score (PoPS) enrichment method [45] for gene prioritization with the GWAS summary statistics. PoPS integrates multiple public bulk and single-cell expression datasets, protein-protein interaction and pathway databases to implement enrichment analysis using MAGMA [46] based gene association scores to identify top list of genes functionally linked to the phenotype of interest. We complementarily performed quantitative trait locus (QTLs) interrogations using multiple publicly available datasets. We downloaded expression quantitative trait locus (eQTLs) data from GTEx (v8_eQTL_all_associations) database (https://gtexportal.org/home/datasets) and curated significant hits (p-value < 5x10^-8) for the lead SNP from 5 different tissues relevant to diabetes, lipids, and inflammation (i.e., Liver, Adipose Subcutaneous, Adipose Visceral Omentum, Whole Blood, and Pancreas). We utilized protein quantitative trait loci (pQTL) data in blood from the INTERVAL study [47] and GoDMC database [48] for methylation quantitative trait loci (mQTL) in blood to curate pQTLs and mQTLs related to the lead variant (p-value < 5x10^-8).

Supporting information

S1 Fig. Overall study schematic.

We carried out stratified GWAS on UKB discovery cohort based on T2 status, using Regenie LOOCV models adjusting for age, age², sex, race, and PC1-10. We implemented heterogeneity analysis to identify loci that was differentially associated between the two strata. Out of the 67M variants analyzed, only one locus achieved genome-wide significance. We replicated the significant locus using MGBB, an independent cohort, and further analyzed the lead variant using various secondary analysis in the discovery cohort. GWAS–Genome wide association; LOOCV—leave-one-out-cross-validation; MGBB–Mass General Brigham Biobank; T2D –Type 2 Diabetes; UKB–UK Biobank.

(TIF)

Click here for additional data file.^{(240.1KB, tif)}

S2 Fig. Manhattan (MH) plots for T2D stratified GWAS.

A) MH plots for T2D GWAS B) MH plots for non-T2D GWAS. Genes near to the most significant lead variant in each loci are documented, full list of lead SNPs are tabulated in S1 Table. Red line: Genome significance (p-value = 5x10^-8), Blue line: Suggestive significance (p-value = 1x10^-5). GWAS–Genome wide association studies; MH–Manhattan; T2D –Type 2 Diabetes.

(ZIP)

Click here for additional data file.^{(471.5KB, zip)}

S3 Fig. HLA-DQB1/DQA2 lead variant locus in T2D and non-T2D strata.

A) Locus zoom plot for HLA-DQB1/DQA2 loci in T2D strata. B) Locus zoom plot for HLA-DQB1/DQA2 loci in non-T2D strata. X-axis defines the genomic position where variants +/-500 kb on either side of rs9274619—chr6:32635954:G:A (grc37) is mapped on the genome. The variants are colored based on the r² with the lead variant and the genes are mapped based on their genomic position. Y-axis is the -log10(p-values) from the respective strata and the scale of y-axis is different between the two plots. HLA–Human Leukocyte Antigen; T2D –Type 2 Diabetes.

(ZIP)

Click here for additional data file.^{(678.2KB, zip)}

S4 Fig. Correlation between GTEx and GWAS Z-scores of eQTLs from HLA-DQB1, HLA-DQA2 and HLA-DRB6.

Z-scores were calculated from T2D GWAS and GTEx (version 8) summary statistics for all the eQTLs for the three genes. Pearson correlation coefficient was calculated, and scatter plots were generated for eQTL data from five different tissues. Most of the TG lowering variants increases the expression of HLA-DQA2/HLA-DRB6, whereas decreases the expression of HLA-DQB1.

(ZIP)

Click here for additional data file.^{(345KB, zip)}

S5 Fig. Gene prioritization using PoPS and QTL mining.

PoPS method was used to prioritize genes using GWAS summary statistics from both T2D and non-T2D stratum. Multiple HLA genes were prioritized, where HLA-DQB1 topped the list. eQTL, pQTL and mQTL curation of rs9274619:A from various public repositories mapped the lead variant to multiple HLA-genes, where HLA-DQA2 was identified by all three QTL searches.

(TIF)

Click here for additional data file.^{(341.3KB, tif)}

S6 Fig. Imputed HLA alleles and its correlation with variant of interest—rs9274619:A.

A) Correlation between rs9274619:A and HLA alleles in DQB1 and DQA1 class. DQB1-302 is the most strongly correlated allele. B) Forest plot showing the different alleles that passed the Bonferroni correction on interacting with T2D with log(TG) as outcome, the model was adjusted age, age², sex, race, PC1-10 and rs9274619:A. DQB1-302 allele is the only allele with significant interaction with T2D and highly correlated to rs9274619:A (mapped in red). HLA–Human Leukocyte Antigen; T2D –Type 2 Diabetes; TG–Triglycerides; VOI–Variant of interest.

(ZIP)

Click here for additional data file.^{(1.2MB, zip)}

S1 Table. Genomic risk loci identified from GWAS stratified by T2D status in UKB.

19 and 315 risk loci were identified by FUMA on T2D and non-T2D GWAS respectively. Significant loci and lead variants identified from FUMA analysis for each risk loci are documented. GWAS–Genome wide association studies; UKB–UK Biobank; T2D –Type 2 Diabetes.

(XLSX)

Click here for additional data file.^{(82.1KB, xlsx)}

S2 Table. List of variants identified after clumping the 478 SNPs in the chromosome 6 loci.

The r2 of the clumped variant with the lead SNP and their respective summary data are documented. Each clumped SNP was additionally used to carry out conational analysis with the lead SNP interaction model and the summary statistics are documented.

(XLSX)

Click here for additional data file.^{(10.4KB, xlsx)}

S3 Table. Summary statistics of lead SNP with lipids.

Summary statistics from linear regression model where rs9274619:A interacting with T2D is documented for log(TG) as outcome for the 3 lipids including LDL-C, HDL-C and TC. HLA-DQB1/DQA2 is enriched only in T2D and not in non-T2D. HLD-C–High-Density Lipoprotein Cholesterol; LDL-C–Low-Density Lipoprotein Cholesterol; T2D –Type 2 Diabetes; TG–Triglycerides; TC–Total Cholesterol.

(XLSX)

Click here for additional data file.^{(9.6KB, xlsx)}

S4 Table. Gene prioritization scores.

PoPS enrichment scores for the top100 genes from T2D and nonT2D GWAS summary statistics. The Genes are ordered based on PoPS scores within each strata. T2D –Type 2 Diabetes.

(XLSX)

Click here for additional data file.^{(16.8KB, xlsx)}

S5 Table. Summary statistics for curated eQTLs.

eQTL–gene pairs for rs9274619:A was curated from GTEx data and genome significant hits were obtained (p-value = 5x10^-8) from 5 different tissues. Each associated gene is tabulated and ordered based on GTEx p-value. eQTL–Expression Quantitative Trait Locus.

(XLSX)

Click here for additional data file.^{(11KB, xlsx)}

S6 Table. Summary statistics for curated mQTLs.

mQTL–CpG pairs for rs9274619:A and the corresponding genes that regulate the CpG regions was curated from GoDMC database where the genome significant hits were obtained (p-value = 5x10^-8) from the cis-associations. The data is ordered based on p-value and the corresponding CpG island regions were curated from Illumina HumanMethylation450 BeadChip resource. mQTL–Methylation Quantitative Trait Locus.

(XLSX)

Click here for additional data file.^{(19.1KB, xlsx)}

S7 Table. Summary statistics for curated HLA-alleles.

Summary statistics for linear regression HLA interaction model with T2D status adjusted for age, age^2, sex, race, PC1-10 and rs9274619:A, where the outcome was normalized log(TG). The correlation coefficient of each HLA allele to rs9274619:A is documented. The HLA alleles are ordered based on their significance, out of the total 362 HLA alleles 331 alleles had interaction summary statistics. PC–Principal Components; T2D –Type 2 Diabetes; TG–Triglycerides.

(XLSX)

Click here for additional data file.^{(35KB, xlsx)}

S8 Table. Summary statistics from PheWAS.

Summary statistics for logistic regression model for PheWAS where disease conditions used as outcomes, adjusted for age, age^2, sex, race and PC1-10. The effects from the rs9274619:A interacting with T2D status is documented and the disease conditions are ordered based on significance. PC–Principal Components; PheWAS–Phenome Wide Association Studies; T2D –Type 2 Diabetes.

(XLSX)

Click here for additional data file.^{(114.5KB, xlsx)}

S9 Table. Summary statistics from significant disease associations.

The 45 significant disease conditions identified using interaction model were analyzed using logistic regression, adjusted for age, age^2, sex, race and PC1-10 and stratified by T2D status. The summary statistics from T2D and non-T2D models are tabulated, where the phenotypes are ordered based on T2D beta. PC–Principal Components; T2D –Type 2 Diabetes.

(XLSX)

Click here for additional data file.^{(15KB, xlsx)}

S10 Table. Summary statistics from T1D and LADA linear models.

Multiple linear main and interaction models used to validate the influence of T1D and LADA samples. LADA–Latent Autoimmune Diabetes in Adults; T1D –Type 1 Diabetes.

(XLSX)

Click here for additional data file.^{(10KB, xlsx)}

S11 Table. Summary statistics from LADA interaction models.

Interaction model between the four LADA loci and T2D with TG as outcome. LADA–Latent Autoimmune Diabetes in Adults; T2D –Type 2 Diabetes; TG–Triglycerides.

(XLSX)

Click here for additional data file.^{(9.9KB, xlsx)}

S12 Table. Significant disease phenotypes with LADA loci.

Interaction p-values between the four LADA loci and T2D with T1D related and hyperlipidemia related phenotypes as outcomes. The top 5 and bottom 5 phenotypes significantly associated with rs9274619:A were selected (Fig 3). All four loci have significant interaction with diabetes related diseases, but not with obesity related phenotypes. LADA–Latent Autoimmune Diabetes in Adults; T1D –Type 1 Diabetes; T2D –Type 2 Diabetes.

(XLSX)

Click here for additional data file.^{(10.4KB, xlsx)}

S13 Table. Summary statistics from linear models with biomarkers.

Summary statistics for linear regression main model with fat, diet and biomarkers phenotypes as outcomes, adjusted for age, age^2, sex, race and PC1-10. The phenotypes are orders based on p-value. PC–Principal Components.

(XLSX)

Click here for additional data file.^{(14.6KB, xlsx)}

S14 Table. Summary statistics from interaction models with biomarkers.

Summary statistics for linear interaction model with fat, diet and biomarkers phenotypes as outcomes, adjusted for age, age^2, sex, race and PC1-10. The phenotypes are orders based on p-value. PC–Principal Components.

(XLSX)

Click here for additional data file.^{(14.8KB, xlsx)}

S15 Table. Summary statistics from linear models with metabolites.

Summary statistics for linear regression main model with metabolomic phenotype as outcomes, adjusted for age, age^2, sex, race and PC1-10. The effects from rs9274619:A are documented and the metabolomes are ordered based on significance. PC–Principal Components.

(XLSX)

Click here for additional data file.^{(25KB, xlsx)}

S16 Table. Summary statistics from interaction models with metabolites.

Summary statistics for linear regression interaction model with metabolomic phenotype as outcomes, adjusted for age, age^2, sex, race and PC1-10. The effects from rs9274619:A interacting with T2D status is documented and the metabolomes are ordered based on significance. PC–Principal Components; T2D –Type 2 Diabetes.

(XLSX)

Click here for additional data file.^{(25KB, xlsx)}

Acknowledgments

We thank all the participants from UKB and MGBB.

Data Availability

All relevant data are within the paper and its Supporting Information files.

Funding Statement

P.N. is supported by grants from the National Institutes of Health (R01HL142711, R01HL148050, R01HL151283, R01HL127564, R01HL148565, R01HL151152, R01DK125782), Fondation Leducq (TNE-18CVD04), and Massachusetts General Hospital (Fireman Chair). GMP is supported by NIH grants R01HL127564 and R01HL142711. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript

References

1.Saeedi P. et al., “Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: Results from the International Diabetes Federation Diabetes Atlas, 9th edition,” Diabetes Res. Clin. Pract., vol. 157, p. 107843, Nov. 2019, doi: 10.1016/j.diabres.2019.107843 [DOI] [PubMed] [Google Scholar]
2.Gillies C. L. et al., “Pharmacological and lifestyle interventions to prevent or delay type 2 diabetes in people with impaired glucose tolerance: systematic review and meta-analysis,” BMJ, vol. 334, no. 7588, p. 299, Feb. 2007, doi: 10.1136/bmj.39063.689375.55 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Tancredi M. et al., “Excess Mortality among Persons with Type 2 Diabetes,” N. Engl. J. Med., vol. 373, no. 18, pp. 1720–1732, Oct. 2015, doi: 10.1056/NEJMoa1504347 [DOI] [PubMed] [Google Scholar]
4.Wright A. K. et al., “Life Expectancy and Cause-Specific Mortality in Type 2 Diabetes: A Population-Based Cohort Study Quantifying Relationships in Ethnic Subgroups,” Diabetes Care, vol. 40, no. 3, pp. 338–345, Mar. 2017, doi: 10.2337/dc16-1616 [DOI] [PubMed] [Google Scholar]
5.Low S. et al., “The role of triglyceride glucose index in development of Type 2 diabetes mellitus,” Diabetes Res. Clin. Pract., vol. 143, pp. 43–49, Sep. 2018, doi: 10.1016/j.diabres.2018.06.006 [DOI] [PubMed] [Google Scholar]
6.Pang J., Chan D. C., and Watts G. F., “Origin and therapy for hypertriglyceridaemia in type 2 diabetes,” World J. Diabetes, vol. 5, no. 2, pp. 165–175, Apr. 2014, doi: 10.4239/wjd.v5.i2.165 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Feher M., Greener M., and Munro N., “Persistent hypertriglyceridemia in statin-treated patients with type 2 diabetes mellitus,” Diabetes Metab. Syndr. Obes. Targets Ther., vol. 6, pp. 11–15, 2013, doi: 10.2147/DMSO.S35053 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Leiter L. A. et al., “Persistent lipid abnormalities in statin-treated patients with diabetes mellitus in Europe and Canada: results of the Dyslipidaemia International Study,” Diabet. Med. J. Br. Diabet. Assoc., vol. 28, no. 11, pp. 1343–1351, Nov. 2011, doi: 10.1111/j.1464-5491.2011.03360.x [DOI] [PubMed] [Google Scholar]
9.Zhao J. et al., “Triglyceride is an independent predictor of type 2 diabetes among middle-aged and older adults: a prospective study with 8-year follow-ups in two cohorts,” J. Transl. Med., vol. 17, no. 1, p. 403, Dec. 2019, doi: 10.1186/s12967-019-02156-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Oram R. A. et al., “Utility of Diabetes Type–Specific Genetic Risk Scores for the Classification of Diabetes Type Among Multiethnic Youth,” Diabetes Care, vol. 45, no. 5, pp. 1124–1131, May 2022, doi: 10.2337/dc20-2872 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Dahlqvist J. et al., “Identification and functional characterization of a novel susceptibility locus for small vessel vasculitis with MPO-ANCA,” Rheumatology, vol. 61, no. 8, pp. 3461–3470, Aug. 2022, doi: 10.1093/rheumatology/keab912 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Basile K. J., Guy V. C., Schwartz S., and Grant S. F. A., “Overlap of genetic susceptibility to type 1 diabetes, type 2 diabetes, and latent autoimmune diabetes in adults,” Curr. Diab. Rep., vol. 14, no. 11, p. 550, 2014, doi: 10.1007/s11892-014-0550-9 [DOI] [PubMed] [Google Scholar]
13.Noble J. A. and Valdes A. M., “Genetics of the HLA region in the prediction of type 1 diabetes,” Curr. Diab. Rep., vol. 11, no. 6, pp. 533–542, Dec. 2011, doi: 10.1007/s11892-011-0223-x [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Cousminer D. L. et al., “First Genome-Wide Association Study of Latent Autoimmune Diabetes in Adults Reveals Novel Insights Linking Immune and Metabolic Diabetes,” Diabetes Care, vol. 41, no. 11, pp. 2396–2403, Nov. 2018, doi: 10.2337/dc18-1032 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Young K. G., McDonald T. J., and Shields B. M., “Glycated haemoglobin measurements from UK Biobank are different to those in linked primary care records: implications for combining biochemistry data from research studies and routine clinical care,” Int. J. Epidemiol., vol. 51, no. 3, pp. 1022–1024, Jun. 2022, doi: 10.1093/ije/dyab265 [DOI] [PubMed] [Google Scholar]
16.Scott R. A. et al., “An Expanded Genome-Wide Association Study of Type 2 Diabetes in Europeans,” Diabetes, vol. 66, no. 11, pp. 2888–2902, Nov. 2017, doi: 10.2337/db16-1253 [DOI] [PMC free article] [PubMed] [Google Scholar]
17.NHLBI TOPMed Lipids Working Group et al., “Deep-coverage whole genome sequences and blood lipids among 16,324 individuals,” Nat. Commun., vol. 9, no. 1, p. 3391, Dec. 2018, doi: 10.1038/s41467-018-05747-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Miller M. et al., “Triglycerides and Cardiovascular Disease: A Scientific Statement From the American Heart Association,” Circulation, vol. 123, no. 20, pp. 2292–2333, May 2011, doi: 10.1161/CIR.0b013e3182160726 [DOI] [PubMed] [Google Scholar]
19.Sacks F. M. et al., “Association Between Plasma Triglycerides and High-Density Lipoprotein Cholesterol and Microvascular Kidney Disease and Retinopathy in Type 2 Diabetes Mellitus: A Global Case–Control Study in 13 Countries,” Circulation, vol. 129, no. 9, pp. 999–1008, Mar. 2014, doi: 10.1161/CIRCULATIONAHA.113.002529 [DOI] [PubMed] [Google Scholar]
20.Hiltbold E. M., Poloso N. J., and Roche P. A., “MHC Class II-Peptide Complexes and APC Lipid Rafts Accumulate at the Immunological Synapse,” J. Immunol., vol. 170, no. 3, pp. 1329–1338, Feb. 2003, doi: 10.4049/jimmunol.170.3.1329 [DOI] [PubMed] [Google Scholar]
21.Erlich H. et al., “HLA DR-DQ Haplotypes and Genotypes and Type 1 Diabetes Risk: Analysis of the Type 1 Diabetes Genetics Consortium Families,” Diabetes, vol. 57, no. 4, pp. 1084–1092, Apr. 2008, doi: 10.2337/db07-1331 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Noble J. A. et al., “HLA Class I and Genetic Susceptibility to Type 1 Diabetes: Results From the Type 1 Diabetes Genetics Consortium,” Diabetes, vol. 59, no. 11, pp. 2972–2979, Nov. 2010, doi: 10.2337/db10-0699 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Noble J. A. and Erlich H. A., “Genetics of Type 1 Diabetes,” Cold Spring Harb. Perspect. Med., vol. 2, no. 1, pp. a007732–a007732, Jan. 2012, doi: 10.1101/cshperspect.a007732 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Jacobi T. et al., “HLA Class II Allele Analyses Implicate Common Genetic Components in Type 1 and Non–Insulin-Treated Type 2 Diabetes,” J. Clin. Endocrinol. Metab., vol. 105, no. 3, pp. e245–e254, Mar. 2020, doi: 10.1210/clinem/dgaa027 [DOI] [PubMed] [Google Scholar]
25.Buzzetti R. et al., “High Titer of Autoantibodies to GAD Identifies a Specific Phenotype of Adult-Onset Autoimmune Diabetes,” Diabetes Care, vol. 30, no. 4, pp. 932–938, Apr. 2007, doi: 10.2337/dc06-1696 [DOI] [PubMed] [Google Scholar]
26.Radtke M. A., Midthjell K., Nilsen T. I. L., and Grill V., “Heterogeneity of Patients With Latent Autoimmune Diabetes in Adults: Linkage to Autoimmunity Is Apparent Only in Those With Perceived Need for Insulin Treatment: Results from the Nord-Trondelag Health (HUNT) study,” Diabetes Care, vol. 32, no. 2, pp. 245–250, Feb. 2009, doi: 10.2337/dc08-1468 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Tuomi T., Santoro N., Caprio S., Cai M., Weng J., and Groop L., “The many faces of diabetes: a disease with increasing heterogeneity,” The Lancet, vol. 383, no. 9922, pp. 1084–1094, Mar. 2014, doi: 10.1016/S0140-6736(13)62219-9 [DOI] [PubMed] [Google Scholar]
28.Ahlqvist E. et al., “Novel subgroups of adult-onset diabetes and their association with outcomes: a data-driven cluster analysis of six variables,” Lancet Diabetes Endocrinol., vol. 6, no. 5, pp. 361–369, May 2018, doi: 10.1016/S2213-8587(18)30051-2 [DOI] [PubMed] [Google Scholar]
29.Mansour Aly D. et al., “Genome-wide association analyses highlight etiological differences underlying newly defined subtypes of diabetes,” Nat. Genet., vol. 53, no. 11, pp. 1534–1542, Nov. 2021, doi: 10.1038/s41588-021-00948-2 [DOI] [PubMed] [Google Scholar]
30.Buzzetti R. et al., “Management of Latent Autoimmune Diabetes in Adults: A Consensus Statement From an International Expert Panel,” Diabetes, vol. 69, no. 10, pp. 2037–2047, Oct. 2020, doi: 10.2337/dbi20-0017 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Bycroft C. et al., “The UK Biobank resource with deep phenotyping and genomic data,” Nature, vol. 562, no. 7726, pp. 203–209, Oct. 2018, doi: 10.1038/s41586-018-0579-z [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Smoller J. et al., “An eMERGE Clinical Center at Partners Personalized Medicine,” J. Pers. Med., vol. 6, no. 1, p. 5, Jan. 2016, doi: 10.3390/jpm6010005 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Denny J. C. et al., “PheWAS: demonstrating the feasibility of a phenome-wide scan to discover gene-disease associations,” Bioinformatics, vol. 26, no. 9, pp. 1205–1210, May 2010, doi: 10.1093/bioinformatics/btq126 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Julkunen H., Cichońska A., Slagboom P. E., Würtz P., and Nightingale Health UK Biobank Initiative, “Metabolic biomarker profiling for identification of susceptibility to severe pneumonia and COVID-19 in the general population,” eLife, vol. 10, p. e63033, May 2021, doi: 10.7554/eLife.63033 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Wei W.-Q. et al., “Evaluating phecodes, clinical classification software, and ICD-9-CM codes for phenome-wide association studies in the electronic health record,” PloS One, vol. 12, no. 7, p. e0175508, 2017, doi: 10.1371/journal.pone.0175508 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Wu P. et al., “Mapping ICD-10 and ICD-10-CM Codes to Phecodes: Workflow Development and Initial Evaluation,” JMIR Med. Inform., vol. 7, no. 4, p. e14325, Nov. 2019, doi: 10.2196/14325 [DOI] [PMC free article] [PubMed] [Google Scholar]
37.UK10K Consortium et al., “Improved imputation of low-frequency and rare variants using the UK10K haplotype reference panel,” Nat. Commun., vol. 6, no. 1, p. 8111, Nov. 2015, doi: 10.1038/ncomms9111 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.the Haplotype Reference Consortium, “A reference panel of 64,976 haplotypes for genotype imputation,” Nat. Genet., vol. 48, no. 10, pp. 1279–1283, Oct. 2016, doi: 10.1038/ng.3643 [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Howie B., Fuchsberger C., Stephens M., Marchini J., and Abecasis G. R., “Fast and accurate genotype imputation in genome-wide association studies through pre-phasing,” Nat. Genet., vol. 44, no. 8, pp. 955–959, Aug. 2012, doi: 10.1038/ng.2354 [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Mbatchou J. et al., “Computationally efficient whole-genome regression for quantitative and binary traits,” Nat. Genet., vol. 53, no. 7, pp. 1097–1103, Jul. 2021, doi: 10.1038/s41588-021-00870-7 [DOI] [PubMed] [Google Scholar]
41.Watanabe K., Taskesen E., van Bochoven A., and Posthuma D., “Functional mapping and annotation of genetic associations with FUMA,” Nat. Commun., vol. 8, no. 1, p. 1826, Dec. 2017, doi: 10.1038/s41467-017-01261-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Willer C. J., Li Y., and Abecasis G. R., “METAL: fast and efficient meta-analysis of genomewide association scans,” Bioinformatics, vol. 26, no. 17, pp. 2190–2191, Sep. 2010, doi: 10.1093/bioinformatics/btq340 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Winkler T. W. et al., “Approaches to detect genetic effects that differ between two strata in genome-wide meta-analyses: Recommendations based on a systematic evaluation,” PLOS ONE, vol. 12, no. 7, p. e0181038, Jul. 2017, doi: 10.1371/journal.pone.0181038 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Purcell S. et al., “PLINK: A Tool Set for Whole-Genome Association and Population-Based Linkage Analyses,” Am. J. Hum. Genet., vol. 81, no. 3, pp. 559–575, Sep. 2007, doi: 10.1086/519795 [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Weeks E. M. et al., “Leveraging polygenic enrichments of gene features to predict genes underlying complex traits and diseases,” Genetic and Genomic Medicine, preprint, Sep. 2020. doi: 10.1101/2020.09.08.20190561 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.de Leeuw C. A., Mooij J. M., Heskes T., and Posthuma D., “MAGMA: Generalized Gene-Set Analysis of GWAS Data,” PLOS Comput. Biol., vol. 11, no. 4, p. e1004219, Apr. 2015, doi: 10.1371/journal.pcbi.1004219 [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Sun B. B. et al., “Genomic atlas of the human plasma proteome,” Nature, vol. 558, no. 7708, pp. 73–79, Jun. 2018, doi: 10.1038/s41586-018-0175-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Min J. L. et al., “Genomic and phenotypic insights from an atlas of genetic effects on DNA methylation,” Nat. Genet., vol. 53, no. 9, pp. 1311–1321, Sep. 2021, doi: 10.1038/s41588-021-00923-x [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0275934.r001

Decision Letter 0

Arnar Palsson

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

21 Jul 2022

PONE-D-22-11074Genome-wide Discovery for Diabetes-Dependent Triglycerides-Associated LociPLOS ONE

Dear Dr. Natarajan,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The manuscript is based on very impressive datasets, and integrates both genetic associations and expression data. The reviewers were generally positive and notice the strengths of both dataset and analyses approach. The DQB1*03:02 SNP for T2D and triglycerides is particularly interesting.

Their main concerns were the following.

1. Could the association be with DQB1*03:02 SNP be caused by type 1 diabetes patients being intermixed with the T2D patients? (rev 2 and 3). Can this be addressed by analysing certain subsets of the data?

2. Elaborate on the relationship of SNP, TG(endophenotype) and T2D and T1D. Rev 2. “from previous studies, seems TG was an independent risk factor of cardiovascular diseases in T2D patients, and a predictor of T2D itself. The authors may discuss the difference of the association in T2D and non-T2D subjects.”

3. Is the lead SNP independent or not? Rev 3. “The authors need to provide further analysis that the present high risk SNP is not in linkage disequilibrium with other genetic factors.” This can be done with conditional analyses. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4572002/

4. It would be interesting, though not required by journal to address the question. Is there a “… relationship between the HLA-DQB1*03:02 SNP and TG levels in the type 1 diabetes subjects in the UKBB “?

5. Explain in the manuscript how other researchers can gain access to the data, note the PLOS one data policy.

Please submit your revised manuscript by Sep 02 2022 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Arnar Palsson, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following financial disclosure:

Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

If this statement is not correct you must amend it as needed.

Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf.

3. Thank you for stating the following in the Acknowledgments Section of your manuscript:

We thank all the participants from UKB and MGBB. P.N. is supported by grants from the National Institutes of Health (R01HL142711, R01HL148050, R01HL151283, R01HL127564, R01HL148565, R01HL151152, R01DK125782), Fondation Leducq (TNE-18CVD04), and Massachusetts General Hospital (Fireman Chair). GMP is supported by NIH grants R01HL127564 and R01HL142711.

However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

4. Thank you for stating the following in the Competing Interests section:

P.N. reports grants from Amgen, Apple, AstraZeneca, Boston Scientific, and Novartis, personal fees from Apple, AstraZeneca, Blackstone Life Sciences, Foresite Labs, Genentech / Roche, Novartis, and TenSixteen Bio, equity in geneXwell, and TenSixteen Bio, co-founder of TenSixteen Bio, and spousal employment at Vertex, all unrelated to the present work.

We note that you received funding from a commercial source: Amgen, Apple, AstraZeneca, Boston Scientific, and Novartis, personal fees from Apple, AstraZeneca, Blackstone Life Sciences, Foresite Labs, Genentech / Roche, Novartis, and TenSixteen Bio, equity in geneXwell, and TenSixteen Bio, co-founder of TenSixteen Bio, and spousal employment at Vertex

Please provide an amended Competing Interests Statement that explicitly states this commercial funder, along with any other relevant declarations relating to employment, consultancy, patents, products in development, marketed products, etc.

Within this Competing Interests Statement, please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials by including the following statement: "This does not alter our adherence to PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests). If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your amended Competing Interests Statement within your cover letter. We will change the online submission form on your behalf.

5. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

6. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

Additional Editor Comments:

Their main concerns were the following.

5. Explain in the manuscript how other researchers can gain access to the data, note the PLOS one data policy.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Dear Author,

I strongly recommend this publication for the following perspectives. First, the large cohort sample sizes for both the cases and controls increases the statistical power for detection of variants with small effect sizes, a vital limitation in analysis of complex genetic traits, were the genetic effect is due to multiple variants with small effect sizes. Second, adjustment for BMI as a potential confounder is of high relevance giving the precision of the statistical model. Third, the implementation of Cochran statistical method in METAL, is highly recommended in conduction meta-analysis of GWAS.

Fourth, linking the GWAS results to the gene expression in different tissues associated with lipid metabolism strongly augment and support the finding. Fifth, even though identification of genetic association of T2D with the well-known T1D HLA locus adds to the complexity of the phenotype yet it could be of valuable information reflecting the protective role of immune system in this category of T2D patients. Overall, great research and work, well written, it adds value for understanding the heterogeneity of T2D supporting the fact that there are underlying subtypes and identifying the role of immune system in T2D not only T1D and diabetic complications.

Reviewer #2: Selvaraj et al. performed genome-wide association and interaction studies among UKB and MGBB subjects, they found that the HLA-DQB1 locus was associated with lower triglyceride levels in T2D patients, but not in non-T2D subjects.

Several issues need to be addressed before acceptance for publication: 1) The affection status of T2D in UKB subjects was based on self-reported data, the incidence of T2D (5.0%) in UKB participants seems lower than general population, given the mean age of 56.6 years. Should fasting glucose and HbA1c be considered for the diagnosis of T2D? 2) The "concomitant diagnosis of T1D among those with T2D" is confusing. It is highly unlikely that a subject had both T1D and T2D, however, it could be an issue since affection statuses were self-reported. Indeed, the HLA-DQB1 locus was only associated with T1D, not with T2D, in previous GWASs in large cohorts. Since no information of LADA diagnosis was available in UKB, testing LADA loci in UKB (the number of LADA patients was limited) provided little information. 3) It is interesting to know that the HLA-DQB1 polymorphisms were associated with higher HbA1c and glucose levels, higher risks of diabetic marco- and microvascular complications, but lower TG, urate, and body weight phenotypes in T2D subjects. From previous studies, seems TG was an independent risk factor of cardiovascular diseases in T2D patients, and a predictor of T2D itself. The authors may discuss the difference of the association in T2D and non-T2D subjects.

Reviewer #3: 1. The study suffers from the habit of classifying the bulk of diabetes patients above 20-25 years of age as type 2 diabetes. The diagnosis is diabetes but the classification is type 2 diabetes based on loose clinical criteria. The strong association between low levels of TG and the A variant in the HLA-DQ B1*03:02 is a distinct example that the authors should discuss.

2. The authors should also discuss the lack of information on GADA or other islet autoantibodies, c-peptide levels as well as HbA1c and the weakness that most of the diabetes classification is self-reported.

3. Genetic Risk Scores (GRS) for autoimmune type 1 diabetes has been developed using the UKBB (Oram and others) and these data should be run in parallel with the present analysis. It may resolve the many issues of re-classifying the patients with autoimmune type 1 diabetes rather than maintaining that they are type 2 diabetes patients.

4. The region around the reported DQB1*03:02 SNP should be further analysed to provide a heat-map of the region to indicate that the particular SNP reported has the highest risk, or p-value, compared to neighbouring SNPs. The authors need to provide further analysis that the present high risk SNP is not in linkage disequilibrium with other genetic factors.

5. There are type 1 diabetes patients (usually more clearly defined when self-reported) in the UKBB. What is the risk of the present DQB1*03:02 SNP for type 1 diabetes? Is there data to indicate that this particular SNP marked low level TG also in type 1 diabetes patients?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Dina Mansour Aly

Reviewer #2: No

Reviewer #3: Yes: Åke Lernmark

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2022 Oct 21;17(10):e0275934. doi: 10.1371/journal.pone.0275934.r002

Author response to Decision Letter 0

2 Sep 2022

EDITORS COMMENT:

Their main concerns were the following.

Author Response:

Thank you for this comment. We observed that rs9274619:A was more significantly associated with T1D (p-value 1.29x10-124) compared to T2D (p-value 9.69x10-14) as previously shown in Supplementary Table 10. We agree that sample intermixing could be one of the possible reasons. The concomitant diagnosis of T2D along with T1D is an active area of research with overlap often termed as latent autoimmune diabetes in adults (LADA), which could also be a potential explanation for the association at this locus. We have discussed this in the manuscript from lines 178-196.

To add clarity of the LADA specific SNP interaction we have added a sentence discussing the overlapping samples from both T1D and T2D forming the LADA group in the results section.

Text in the manuscript:

We additionally checked the effect of rs9274619:A for LADA samples (N=2580) in our dataset and observed significant association (p-value=8.37x10-41), showing that overlapping T1D and T2D samples could potentially drive the association of this loci.

Additionally, we subsetted samples based on T1D polygenic risk score (PRS) from Oram et al as suggested by Reviewer 3. Briefly, we used 67 SNPs to calculate a T1D PRS in UKB. Based on 99th percentile of the PRS, we identified and removed potential T1D samples (N= 4242) in UKB. Next, we removed residual T1D samples identified based on ICD codes as well. Finally, we show that the interaction model of the lead SNP with T2D is still nominally significant for TG (beta= -0.0261; p-value= 0.01). Through this analysis we show that even after the removal of potential T1D samples based on validated 67 SNPs of T1D GRS (Oram et al), the signal from the lead SNP rs9274619:A is nominally significant. While we cannot rule out the possibility of residual T1D case mixing, these results remain consistent with an interaction for T2D. We have added this analysis to the results section.

Text in the manuscript:

Since the HLA locus is a strong predictor of type 1 diabetes (T1D), we used polygenic risk scores (PRS) from 67 variants previously associated with T1D as reported by Oram et al(10), to exclude potential T1D cases in sensitivity analysis. First, we used the 67 SNPs from both HLA and non-HLA loci to create a genome-wide PRS for all the UKB samples. The top one percentile of the samples based on PRS were classified as T1D (N=4242). Next, from the whole UKB samples, we removed any sample which identified as T1D by either ICD codes or PRS scores. With the remaining 18460 T2D cases, we observe a nominal interaction of the lead SNP with T2D (betainteraction=-0.026; pinteraction=1.16x10-02).

Author Response:

Thank you for this comment. It is true that TG is an independent risk factor for cardiovascular diseases and a predictor of T2D. In this study, we performed a stratified analysis to capture the heterogeneity by genetic variants by T2D status. Though the independent GWAS of T2D and non-T2D groups identified significant loci, from our analysis we were able to show that genome-wide significant heterogeneity was present at the chromosome 6 locus. This may have bearing on pathways that may specifically influence T2D-associated cardiovascular disease with less relevance outside the context of T2D. The chromosome 6 HLA locus is a strong locus for T1D odds but given the heterogeneity in the associations between T2D and non-T2D, we show that this locus has the strongest signal to differentiate T2D to non-T2D. Our analysis could identify hidden signals between the stratified groups of samples and could pave way to identify new loci. We have added a sentence in the discussion section of the manuscript addressing the question.

Text in the manuscript:

Previous studies have shown TG as an independent risk factor for cardiovascular diseases in T2D patients, and a predictor of T2D itself. Through a genome-wide stratified analysis, we now show that the HLA locus may triangulate some of these relationships by specifically influencing T2D-associated triglycerides.

Author Response:

Thank you for this comment and suggestion. We have investigated the locus in detail showing that the association from lead SNP (rs9274619:A) is independent of other variants in the region. To demonstrate this, we performed variant clumping with all 478 genome-wide significant SNPs in chr6p21.32 loci and identified 8 independent signals. We tabulated the R2 for these SNPs (using 1000G reference panel) with the lead SNP analyzed in this study and provided summary statistics for these clumped SNPs as Supplementary Table 2. Next, we performed conditional analyses of these clumped SNPs with the lead SNP for the interaction of T2D status with triglycerides. From this conditional analysis, we observed that none of the clumped SNPs abrogated the signal from the lead SNP.

We have added a results section to the manuscript describing this additional analysis.

Text in the manuscript:

To evaluate the independence of the lead variant, we clumped the 478 genome-wide significant variants and performed conditional analyses. After clumping, we retained 8 individual signals, and rs3957148 was in strong LD with the lead variant. However, conditional analyses with all these eight clumped variants for rs9274619:A interaction with T2D status did not abrogate the lead variant’s signal (Supplementary Table 2).

Author Response:

Thank you for this comment. Yes, there is a relationship between the lead SNP from this study and TG in T1D samples. In summary, the TG-lowering rs9274619:A was strongly associated with T1D after adjusting for T2D (p-value=8.6x10-113), but not significantly associated with T2D status after adjusting for T1D (p-value=0.29). However, when assessing for interactions on TGs, there was still a significant interaction with T2D independent of T1D (betainteraction=-0.055; pinteraction=5.28x10-9) and more strongly with T1D independently of T2D (betainteraction=-0.252; pinteraction=5.53x10-49). We included this in the manuscript text (lines 199-205) and tabulated the findings in the Supplementary Table 10.

5. Explain in the manuscript how other researchers can gain access to the data, note the PLOS one data policy.

Author Response:

Thank you for this comment. We have included a section describing the data availability for UK Biobank and MGB Biobank.

Text in the manuscript:

Data availability:

Data cannot be shared publicly by the authors because of information governance restrictions around health data. The UK Biobank data can however be downloaded following a project approval process. Researchers wishing to access the data can apply directly to the UK Biobank https://www.ukbiobank.ac.uk/enable-your-research/apply-for-access and the process involves registering on the access management system, submitting a research study protocol and paying a fee directly to the UK Biobank. The Mass General Brigham Biobank (MGBB) individual-level data are available from https://personalizedmedicine.partners.org/Biobank/Default.aspx, where the data is available through institutional review board (IRB) approval, therefore not publicly available. Summary statistics from the current study will be shared through the Cardiovascular Disease Knowledge Portal (CVDKP) https://cvd.hugeamp.org/.

REVIEWERS COMMENTS:

Reviewer #1: Dear Author,

Author Response:

We thank the reviewer for the positive comments.

Several issues need to be addressed before acceptance for publication:

1) The affection status of T2D in UKB subjects was based on self-reported data, the incidence of T2D (5.0%) in UKB participants seems lower than general population, given the mean age of 56.6 years. Should fasting glucose and HbA1c be considered for the diagnosis of T2D?

Author Response:

Thank you for this comment. UK Biobank is well recognized to have a healthy volunteer bias with generally lower prevalences of chronic conditions relative to the UK population at large (Fry A et al. Am J Epidemiol 2017). This contrasts with our healthcare associated biobank, MGB Biobank (prevalence 27.65%). Reassuringly, in both contexts, the findings hold. Consistent with prior studies, we have focused on clinician recognized and self reported T2D. To improve the T2D diagnosis sensitivity, we now consider HgA1c (i.e., >= 48 mmol/mol) for the case definition; unfortunately, fasting glucose is not available. Indeed in secondary analysis, we associated blood biomarkers including random glucose and HbA1c with the lead SNP and observe significant associations (Supplementary Tables 13 &14).

We investigated using HbA1c for the diagnosis of T2D status. As an additional analysis, we stratified the cohort based on HbA1c values (UKB data field: 30750), where HbA1c-based T2D had HbA1c >=48 mmol/mol. Using this criterion, we were able to identify 15,180 HbA1c-based T2D samples. The interaction model of the lead SNP with HbA1c-augmented T2D showed higher significance on TG (betainteraction= -0.159; pinteraction= 1.26x10-54). Through this classification, we were able to add 5212 samples more to the existing T2D stratified group based on self-reporting and ICD codes. We finally included the ~5K samples to the ~21K T2D group and tested the interaction modeled with the lead SNP to show a stronger significance (betainteraction= -0.102; pinteraction= 1.03x10-35). We included these results in the revised manuscript.

Text in the manuscript:

Since HbA1c is a strong predictor of T2D, HbA1c is the most significant biomarker interacting with the lead SNP (Supplementary Table 14). When also using HbA1c >= 48 mmol/mol to define T2D samples as described by Young et al (14), 15,180 samples were identified, with an additional 5212 samples to the previous T2D group. The interaction model of HbA1c-based only T2D with the lead SNP showed higher significance with TG (betainteraction= -0.159; pinteraction= 1.26x10-54). Next, we added the 5212 samples to our existing ICD10-based T2D groups and conducted a sensitivity analysis for the interaction. The addition of samples showed consistent significant interaction (betainteraction= -0.102; pinteraction= 1.03x10-35).

2) The "concomitant diagnosis of T1D among those with T2D" is confusing. It is highly unlikely that a subject had both T1D and T2D, however, it could be an issue since affection statuses were self-reported. Indeed, the HLA-DQB1 locus was only associated with T1D, not with T2D, in previous GWASs in large cohorts. Since no information of LADA diagnosis was available in UKB, testing LADA loci in UKB (the number of LADA patients was limited) provided little information.

Author Response:

Thank you for this comment. We agree that overlapping samples between T1D and T2D are not well-defined and this also occurs clinically. In the current paper, we tried to dissect the association of this locus by comparing T1D, T2D, and presumed LADA (overlap of T1D and T2D). As shown in Supplementary Table 10, the lead variant has strongest association with T1D, followed by T2D. The number of samples with LADA is limited but the lead variant’s association is still genome-wide significant. We have discussed this in the manuscript lines 178-196 and added a sentence to emphasize the effect of overlapping samples.

Text in the manuscript:

We additionally checked the main effects of rs9274619:A for presumed LADA samples (N=2580) in our dataset and observed a significant association (p-value=8.37x10-41), showing that overlapping T1D and T2D samples could potentially contribute to the observed association of this locus.

3) It is interesting to know that the HLA-DQB1 polymorphisms were associated with higher HbA1c and glucose levels, higher risks of diabetic marco- and microvascular complications, but lower TG, urate, and body weight phenotypes in T2D subjects. From previous studies, seems TG was an independent risk factor of cardiovascular diseases in T2D patients, and a predictor of T2D itself. The authors may discuss the difference of the association in T2D and non-T2D subjects.

Author Response:

Thank you for this comment. We added a sentence in the discussion explaining the rationale for the stratified GWAS. Please check Editors comment #2 for the response.

Reviewer #3:

1. The study suffers from the habit of classifying the bulk of diabetes patients above 20-25 years of age as type 2 diabetes. The diagnosis is diabetes but the classification is type 2 diabetes based on loose clinical criteria. The strong association between low levels of TG and the A variant in the HLA-DQ B1*03:02 is a distinct example that the authors should discuss.

Author Response:

Thank you for this comment. We would like to point out that we defined T2D based on self-reported status and based on ICD10 codes. We recognize that clinicians may erroneously classify T2D for all adult-onset diabetes because it comprises the overwhelming majority. We have addressed a similar question, please check reviewer #2 comment #1 for the additional analysis that we conducted. In brief, we consider diagnostic overlaps for T1D and T2D and also consider undiagnosed T1D by a T1D PRS.

Author Response:

Thank you for this comment. We agree that the classification of T2D status is complicated where multiple factors play a very important role and particularly the potential lack of LADA or T1D recognition. HgA1c is indeed available and we now include this in new analyses. We have added a sentence in the limitation section. Please also refer to the comments above about further sensitivity analyses now included.

Text in the manuscript:

Second, GADA or other islet autoantibodies and C-peptide levels are not present, despite using ICD codes and T1D polygenic risk score in sensitivity analyses, we cannot rule out the possibility of T1D-like features in some of the classified T2D cases.

Author Response:

Thank you for this comment and the valuable suggestion. We have integrated the PRS from the 67 SNPs reported by Oram et al to prioritize undiagnosed T1D samples. We show that the lead SNP in nominally significant in its interaction to T2D even after the removal of T1D samples based on ICD codes and T1D PRS. Please see our response in Editors comment #1.

Author Response:

Thank you for this comment and the valuable suggestion. We have addressed this comment with additional conditional analyses. Please see our response in Editors comment #3.

Author Response:

Thank you for this comment. We have conducted this analysis. Please see our response in Editors comment #4.

Attachment

Submitted filename: Response_to_Reviewers.docx

Click here for additional data file.^{(36KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0275934.r003

Decision Letter 1

Arnar Palsson

12 Sep 2022

PONE-D-22-11074R1Genome-wide Discovery for Diabetes-Dependent Triglycerides-Associated LociPLOS ONE

Dear Dr. Natarajan,

few minor things need your attention, see reviewers comments.

Please submit your revised manuscript by Oct 27 2022 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

We look forward to receiving your revised manuscript.

Kind regards,

Arnar Palsson, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Additional Editor Comments:

Thanks for responding so well to the reviewers suggestions, this is very nearly ready.

One reviewer suggest minor touch ups to the manuscript. Should be relatively quickly completed.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #3: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #1: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #1: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Reviewer #1: This study highlights the necessity of understanding the genetics of T2D and underlying pathways.

T2D is heterogeneous complex genetic trait, in order to proceed for personalized medical care for each patient, tailored medical care based on genetics, genetic risk scores offers valuable tool for patient T2D characterization and drug _repurposing.

Reviewer #3: The authors have answered my queries. It would strengthen the observed association to check if TG levels are related to the rs9274619 genotypes A/A, A/G and G/G. As rs9274619 is located between the HLA-DQB1 and HLA-DQA2 loci, the authors may want to refer this potentially quantitative trait loci as linked to HLA-DQB1/HLA-DQA2 rather than only HLA-DQB1.

LD to either gene may be about the same.

The authors should consider to add this recent investigation of small vessel vasculitis, an immune related disease also related to rs9274619: Dahlqvist et al. Identification and functional characterization of a novel susceptibility locus for small vessel vasculitis with MPO-ANCA. Rheumatology (Oxford). 2022 Aug 3;61(8):3461-3470.

Finally, HLA genotypes seemed to influence levels of TG in children at increased HLA genetic risk and positive for islet autoantibodies but prior to clinical diagnosis. TG levels were lower in the not-yet diabetes affected subjects and may indeed reflect the rs9274619 variant.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Dina Mansour Aly

Reviewer #3: No

**********

PLoS One. 2022 Oct 21;17(10):e0275934. doi: 10.1371/journal.pone.0275934.r004

Author response to Decision Letter 1

23 Sep 2022

Editors Comment:

Authors response:

Thank you for the comment. We have corrected one reference and removed one reference.

Reference no 14:

Young KG, McDonald TJ, Shields BM. Glycated haemoglobin measurements from UK Biobank are different to those in linked primary care records: implications for combining biochemistry data from research studies and routine clinical care. Int J Epidemiol. 2022 Jun 13;51(3):1022–4.

Reference no 27:

Removed

Reviewers' comments:

Reviewer #1: This study highlights the necessity of understanding the genetics of T2D and underlying pathways.

Authors response:

Thank you for the positive comment.

LD to either gene may be about the same.

Authors response:

Thank you for the suggestion, we have added the below sentences to address the comments. Throughout the manuscript we have made changes to sentences referring the lead variants proximity to HLA-DQB1/DQA2. Additionally, we agree that our results showing the main effect of rs9274619 on TGs is consistent with the prior literature and here we show its dependency on diabetes status.

Line number 114: The mean raw TG measurements were significantly different between the reference and alternative genotypes in T2D samples (rs9274619(G/A or A/G) : meandiff=5.45 mg/dl; p-value=1.26x10-02; rs9274619(A/A) : meandiff=25.6 mg/dl; p-value=3.78x10-03).

Line number 165: Since rs9274619:A located between both HLA-DQB1 and HLA-DQA2, the variant could be a potential quantitative trait loci to both the genes.

Line number 193: Interestingly, a recent investigation on Anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitides (AAV) by Dahlqvist et al have identified rs9274619 as a lead variant for myeloperoxidase (MPO)-ANCA association (11), showing its importance in immune-related vascular diseases.

Figures:

Authors response:

Thank you for the information. We have uploaded the figures through this tool.

Attachment

Submitted filename: Response_to_Reviewers_AfterRevision.docx

Click here for additional data file.^{(23KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0275934.r005

Decision Letter 2

Arnar Palsson

26 Sep 2022

Genome-wide Discovery for Diabetes-Dependent Triglycerides-Associated Loci

PONE-D-22-11074R2

Dear Dr. Natarajan,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Arnar Palsson, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

PLoS One. doi: 10.1371/journal.pone.0275934.r006

Acceptance letter

Arnar Palsson

13 Oct 2022

PONE-D-22-11074R2

Genome-wide Discovery for Diabetes-Dependent Triglycerides-Associated Loci

Dear Dr. Natarajan:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Arnar Palsson

Academic Editor

PLOS ONE

Associated Data

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

Supplementary Materials

S1 Fig. Overall study schematic.

(TIF)

Click here for additional data file.^{(240.1KB, tif)}

S2 Fig. Manhattan (MH) plots for T2D stratified GWAS.

(ZIP)

Click here for additional data file.^{(471.5KB, zip)}

S3 Fig. HLA-DQB1/DQA2 lead variant locus in T2D and non-T2D strata.

(ZIP)

Click here for additional data file.^{(678.2KB, zip)}

S4 Fig. Correlation between GTEx and GWAS Z-scores of eQTLs from HLA-DQB1, HLA-DQA2 and HLA-DRB6.

(ZIP)

Click here for additional data file.^{(345KB, zip)}

S5 Fig. Gene prioritization using PoPS and QTL mining.

(TIF)

Click here for additional data file.^{(341.3KB, tif)}

S6 Fig. Imputed HLA alleles and its correlation with variant of interest—rs9274619:A.

(ZIP)

Click here for additional data file.^{(1.2MB, zip)}

S1 Table. Genomic risk loci identified from GWAS stratified by T2D status in UKB.

(XLSX)

Click here for additional data file.^{(82.1KB, xlsx)}

S2 Table. List of variants identified after clumping the 478 SNPs in the chromosome 6 loci.

(XLSX)

Click here for additional data file.^{(10.4KB, xlsx)}

S3 Table. Summary statistics of lead SNP with lipids.

(XLSX)

Click here for additional data file.^{(9.6KB, xlsx)}

S4 Table. Gene prioritization scores.

PoPS enrichment scores for the top100 genes from T2D and nonT2D GWAS summary statistics. The Genes are ordered based on PoPS scores within each strata. T2D –Type 2 Diabetes.

(XLSX)

Click here for additional data file.^{(16.8KB, xlsx)}

S5 Table. Summary statistics for curated eQTLs.

(XLSX)

Click here for additional data file.^{(11KB, xlsx)}

S6 Table. Summary statistics for curated mQTLs.

(XLSX)

Click here for additional data file.^{(19.1KB, xlsx)}

S7 Table. Summary statistics for curated HLA-alleles.

(XLSX)

Click here for additional data file.^{(35KB, xlsx)}

S8 Table. Summary statistics from PheWAS.

(XLSX)

Click here for additional data file.^{(114.5KB, xlsx)}

S9 Table. Summary statistics from significant disease associations.

(XLSX)

Click here for additional data file.^{(15KB, xlsx)}

S10 Table. Summary statistics from T1D and LADA linear models.

Multiple linear main and interaction models used to validate the influence of T1D and LADA samples. LADA–Latent Autoimmune Diabetes in Adults; T1D –Type 1 Diabetes.

(XLSX)

Click here for additional data file.^{(10KB, xlsx)}

S11 Table. Summary statistics from LADA interaction models.

Interaction model between the four LADA loci and T2D with TG as outcome. LADA–Latent Autoimmune Diabetes in Adults; T2D –Type 2 Diabetes; TG–Triglycerides.

(XLSX)

Click here for additional data file.^{(9.9KB, xlsx)}

S12 Table. Significant disease phenotypes with LADA loci.

(XLSX)

Click here for additional data file.^{(10.4KB, xlsx)}

S13 Table. Summary statistics from linear models with biomarkers.

(XLSX)

Click here for additional data file.^{(14.6KB, xlsx)}

S14 Table. Summary statistics from interaction models with biomarkers.

(XLSX)

Click here for additional data file.^{(14.8KB, xlsx)}

S15 Table. Summary statistics from linear models with metabolites.

(XLSX)

Click here for additional data file.^{(25KB, xlsx)}

S16 Table. Summary statistics from interaction models with metabolites.

(XLSX)

Click here for additional data file.^{(25KB, xlsx)}

Attachment

Submitted filename: Response_to_Reviewers.docx

Click here for additional data file.^{(36KB, docx)}

Attachment

Submitted filename: Response_to_Reviewers_AfterRevision.docx

Click here for additional data file.^{(23KB, docx)}

Data Availability Statement

All relevant data are within the paper and its Supporting Information files.

[pone.0275934.ref001] 1.Saeedi P. et al., “Global and regional diabetes prevalence estimates for 2019 and projections for 2030 and 2045: Results from the International Diabetes Federation Diabetes Atlas, 9th edition,” Diabetes Res. Clin. Pract., vol. 157, p. 107843, Nov. 2019, doi: 10.1016/j.diabres.2019.107843 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref002] 2.Gillies C. L. et al., “Pharmacological and lifestyle interventions to prevent or delay type 2 diabetes in people with impaired glucose tolerance: systematic review and meta-analysis,” BMJ, vol. 334, no. 7588, p. 299, Feb. 2007, doi: 10.1136/bmj.39063.689375.55 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref003] 3.Tancredi M. et al., “Excess Mortality among Persons with Type 2 Diabetes,” N. Engl. J. Med., vol. 373, no. 18, pp. 1720–1732, Oct. 2015, doi: 10.1056/NEJMoa1504347 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref004] 4.Wright A. K. et al., “Life Expectancy and Cause-Specific Mortality in Type 2 Diabetes: A Population-Based Cohort Study Quantifying Relationships in Ethnic Subgroups,” Diabetes Care, vol. 40, no. 3, pp. 338–345, Mar. 2017, doi: 10.2337/dc16-1616 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref005] 5.Low S. et al., “The role of triglyceride glucose index in development of Type 2 diabetes mellitus,” Diabetes Res. Clin. Pract., vol. 143, pp. 43–49, Sep. 2018, doi: 10.1016/j.diabres.2018.06.006 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref006] 6.Pang J., Chan D. C., and Watts G. F., “Origin and therapy for hypertriglyceridaemia in type 2 diabetes,” World J. Diabetes, vol. 5, no. 2, pp. 165–175, Apr. 2014, doi: 10.4239/wjd.v5.i2.165 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref007] 7.Feher M., Greener M., and Munro N., “Persistent hypertriglyceridemia in statin-treated patients with type 2 diabetes mellitus,” Diabetes Metab. Syndr. Obes. Targets Ther., vol. 6, pp. 11–15, 2013, doi: 10.2147/DMSO.S35053 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref008] 8.Leiter L. A. et al., “Persistent lipid abnormalities in statin-treated patients with diabetes mellitus in Europe and Canada: results of the Dyslipidaemia International Study,” Diabet. Med. J. Br. Diabet. Assoc., vol. 28, no. 11, pp. 1343–1351, Nov. 2011, doi: 10.1111/j.1464-5491.2011.03360.x [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref009] 9.Zhao J. et al., “Triglyceride is an independent predictor of type 2 diabetes among middle-aged and older adults: a prospective study with 8-year follow-ups in two cohorts,” J. Transl. Med., vol. 17, no. 1, p. 403, Dec. 2019, doi: 10.1186/s12967-019-02156-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref010] 10.Oram R. A. et al., “Utility of Diabetes Type–Specific Genetic Risk Scores for the Classification of Diabetes Type Among Multiethnic Youth,” Diabetes Care, vol. 45, no. 5, pp. 1124–1131, May 2022, doi: 10.2337/dc20-2872 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref011] 11.Dahlqvist J. et al., “Identification and functional characterization of a novel susceptibility locus for small vessel vasculitis with MPO-ANCA,” Rheumatology, vol. 61, no. 8, pp. 3461–3470, Aug. 2022, doi: 10.1093/rheumatology/keab912 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref012] 12.Basile K. J., Guy V. C., Schwartz S., and Grant S. F. A., “Overlap of genetic susceptibility to type 1 diabetes, type 2 diabetes, and latent autoimmune diabetes in adults,” Curr. Diab. Rep., vol. 14, no. 11, p. 550, 2014, doi: 10.1007/s11892-014-0550-9 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref013] 13.Noble J. A. and Valdes A. M., “Genetics of the HLA region in the prediction of type 1 diabetes,” Curr. Diab. Rep., vol. 11, no. 6, pp. 533–542, Dec. 2011, doi: 10.1007/s11892-011-0223-x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref014] 14.Cousminer D. L. et al., “First Genome-Wide Association Study of Latent Autoimmune Diabetes in Adults Reveals Novel Insights Linking Immune and Metabolic Diabetes,” Diabetes Care, vol. 41, no. 11, pp. 2396–2403, Nov. 2018, doi: 10.2337/dc18-1032 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref015] 15.Young K. G., McDonald T. J., and Shields B. M., “Glycated haemoglobin measurements from UK Biobank are different to those in linked primary care records: implications for combining biochemistry data from research studies and routine clinical care,” Int. J. Epidemiol., vol. 51, no. 3, pp. 1022–1024, Jun. 2022, doi: 10.1093/ije/dyab265 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref016] 16.Scott R. A. et al., “An Expanded Genome-Wide Association Study of Type 2 Diabetes in Europeans,” Diabetes, vol. 66, no. 11, pp. 2888–2902, Nov. 2017, doi: 10.2337/db16-1253 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref017] 17.NHLBI TOPMed Lipids Working Group et al., “Deep-coverage whole genome sequences and blood lipids among 16,324 individuals,” Nat. Commun., vol. 9, no. 1, p. 3391, Dec. 2018, doi: 10.1038/s41467-018-05747-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref018] 18.Miller M. et al., “Triglycerides and Cardiovascular Disease: A Scientific Statement From the American Heart Association,” Circulation, vol. 123, no. 20, pp. 2292–2333, May 2011, doi: 10.1161/CIR.0b013e3182160726 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref019] 19.Sacks F. M. et al., “Association Between Plasma Triglycerides and High-Density Lipoprotein Cholesterol and Microvascular Kidney Disease and Retinopathy in Type 2 Diabetes Mellitus: A Global Case–Control Study in 13 Countries,” Circulation, vol. 129, no. 9, pp. 999–1008, Mar. 2014, doi: 10.1161/CIRCULATIONAHA.113.002529 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref020] 20.Hiltbold E. M., Poloso N. J., and Roche P. A., “MHC Class II-Peptide Complexes and APC Lipid Rafts Accumulate at the Immunological Synapse,” J. Immunol., vol. 170, no. 3, pp. 1329–1338, Feb. 2003, doi: 10.4049/jimmunol.170.3.1329 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref021] 21.Erlich H. et al., “HLA DR-DQ Haplotypes and Genotypes and Type 1 Diabetes Risk: Analysis of the Type 1 Diabetes Genetics Consortium Families,” Diabetes, vol. 57, no. 4, pp. 1084–1092, Apr. 2008, doi: 10.2337/db07-1331 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref022] 22.Noble J. A. et al., “HLA Class I and Genetic Susceptibility to Type 1 Diabetes: Results From the Type 1 Diabetes Genetics Consortium,” Diabetes, vol. 59, no. 11, pp. 2972–2979, Nov. 2010, doi: 10.2337/db10-0699 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref023] 23.Noble J. A. and Erlich H. A., “Genetics of Type 1 Diabetes,” Cold Spring Harb. Perspect. Med., vol. 2, no. 1, pp. a007732–a007732, Jan. 2012, doi: 10.1101/cshperspect.a007732 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref024] 24.Jacobi T. et al., “HLA Class II Allele Analyses Implicate Common Genetic Components in Type 1 and Non–Insulin-Treated Type 2 Diabetes,” J. Clin. Endocrinol. Metab., vol. 105, no. 3, pp. e245–e254, Mar. 2020, doi: 10.1210/clinem/dgaa027 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref025] 25.Buzzetti R. et al., “High Titer of Autoantibodies to GAD Identifies a Specific Phenotype of Adult-Onset Autoimmune Diabetes,” Diabetes Care, vol. 30, no. 4, pp. 932–938, Apr. 2007, doi: 10.2337/dc06-1696 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref026] 26.Radtke M. A., Midthjell K., Nilsen T. I. L., and Grill V., “Heterogeneity of Patients With Latent Autoimmune Diabetes in Adults: Linkage to Autoimmunity Is Apparent Only in Those With Perceived Need for Insulin Treatment: Results from the Nord-Trondelag Health (HUNT) study,” Diabetes Care, vol. 32, no. 2, pp. 245–250, Feb. 2009, doi: 10.2337/dc08-1468 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref027] 27.Tuomi T., Santoro N., Caprio S., Cai M., Weng J., and Groop L., “The many faces of diabetes: a disease with increasing heterogeneity,” The Lancet, vol. 383, no. 9922, pp. 1084–1094, Mar. 2014, doi: 10.1016/S0140-6736(13)62219-9 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref028] 28.Ahlqvist E. et al., “Novel subgroups of adult-onset diabetes and their association with outcomes: a data-driven cluster analysis of six variables,” Lancet Diabetes Endocrinol., vol. 6, no. 5, pp. 361–369, May 2018, doi: 10.1016/S2213-8587(18)30051-2 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref029] 29.Mansour Aly D. et al., “Genome-wide association analyses highlight etiological differences underlying newly defined subtypes of diabetes,” Nat. Genet., vol. 53, no. 11, pp. 1534–1542, Nov. 2021, doi: 10.1038/s41588-021-00948-2 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref030] 30.Buzzetti R. et al., “Management of Latent Autoimmune Diabetes in Adults: A Consensus Statement From an International Expert Panel,” Diabetes, vol. 69, no. 10, pp. 2037–2047, Oct. 2020, doi: 10.2337/dbi20-0017 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref031] 31.Bycroft C. et al., “The UK Biobank resource with deep phenotyping and genomic data,” Nature, vol. 562, no. 7726, pp. 203–209, Oct. 2018, doi: 10.1038/s41586-018-0579-z [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref032] 32.Smoller J. et al., “An eMERGE Clinical Center at Partners Personalized Medicine,” J. Pers. Med., vol. 6, no. 1, p. 5, Jan. 2016, doi: 10.3390/jpm6010005 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref033] 33.Denny J. C. et al., “PheWAS: demonstrating the feasibility of a phenome-wide scan to discover gene-disease associations,” Bioinformatics, vol. 26, no. 9, pp. 1205–1210, May 2010, doi: 10.1093/bioinformatics/btq126 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref034] 34.Julkunen H., Cichońska A., Slagboom P. E., Würtz P., and Nightingale Health UK Biobank Initiative, “Metabolic biomarker profiling for identification of susceptibility to severe pneumonia and COVID-19 in the general population,” eLife, vol. 10, p. e63033, May 2021, doi: 10.7554/eLife.63033 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref035] 35.Wei W.-Q. et al., “Evaluating phecodes, clinical classification software, and ICD-9-CM codes for phenome-wide association studies in the electronic health record,” PloS One, vol. 12, no. 7, p. e0175508, 2017, doi: 10.1371/journal.pone.0175508 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref036] 36.Wu P. et al., “Mapping ICD-10 and ICD-10-CM Codes to Phecodes: Workflow Development and Initial Evaluation,” JMIR Med. Inform., vol. 7, no. 4, p. e14325, Nov. 2019, doi: 10.2196/14325 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref037] 37.UK10K Consortium et al., “Improved imputation of low-frequency and rare variants using the UK10K haplotype reference panel,” Nat. Commun., vol. 6, no. 1, p. 8111, Nov. 2015, doi: 10.1038/ncomms9111 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref038] 38.the Haplotype Reference Consortium, “A reference panel of 64,976 haplotypes for genotype imputation,” Nat. Genet., vol. 48, no. 10, pp. 1279–1283, Oct. 2016, doi: 10.1038/ng.3643 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref039] 39.Howie B., Fuchsberger C., Stephens M., Marchini J., and Abecasis G. R., “Fast and accurate genotype imputation in genome-wide association studies through pre-phasing,” Nat. Genet., vol. 44, no. 8, pp. 955–959, Aug. 2012, doi: 10.1038/ng.2354 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref040] 40.Mbatchou J. et al., “Computationally efficient whole-genome regression for quantitative and binary traits,” Nat. Genet., vol. 53, no. 7, pp. 1097–1103, Jul. 2021, doi: 10.1038/s41588-021-00870-7 [DOI] [PubMed] [Google Scholar]

[pone.0275934.ref041] 41.Watanabe K., Taskesen E., van Bochoven A., and Posthuma D., “Functional mapping and annotation of genetic associations with FUMA,” Nat. Commun., vol. 8, no. 1, p. 1826, Dec. 2017, doi: 10.1038/s41467-017-01261-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref042] 42.Willer C. J., Li Y., and Abecasis G. R., “METAL: fast and efficient meta-analysis of genomewide association scans,” Bioinformatics, vol. 26, no. 17, pp. 2190–2191, Sep. 2010, doi: 10.1093/bioinformatics/btq340 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref043] 43.Winkler T. W. et al., “Approaches to detect genetic effects that differ between two strata in genome-wide meta-analyses: Recommendations based on a systematic evaluation,” PLOS ONE, vol. 12, no. 7, p. e0181038, Jul. 2017, doi: 10.1371/journal.pone.0181038 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref044] 44.Purcell S. et al., “PLINK: A Tool Set for Whole-Genome Association and Population-Based Linkage Analyses,” Am. J. Hum. Genet., vol. 81, no. 3, pp. 559–575, Sep. 2007, doi: 10.1086/519795 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref045] 45.Weeks E. M. et al., “Leveraging polygenic enrichments of gene features to predict genes underlying complex traits and diseases,” Genetic and Genomic Medicine, preprint, Sep. 2020. doi: 10.1101/2020.09.08.20190561 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref046] 46.de Leeuw C. A., Mooij J. M., Heskes T., and Posthuma D., “MAGMA: Generalized Gene-Set Analysis of GWAS Data,” PLOS Comput. Biol., vol. 11, no. 4, p. e1004219, Apr. 2015, doi: 10.1371/journal.pcbi.1004219 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref047] 47.Sun B. B. et al., “Genomic atlas of the human plasma proteome,” Nature, vol. 558, no. 7708, pp. 73–79, Jun. 2018, doi: 10.1038/s41586-018-0175-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0275934.ref048] 48.Min J. L. et al., “Genomic and phenotypic insights from an atlas of genetic effects on DNA methylation,” Nat. Genet., vol. 53, no. 9, pp. 1311–1321, Sep. 2021, doi: 10.1038/s41588-021-00923-x [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Genome-wide discovery for diabetes-dependent triglycerides-associated loci

Margaret Sunitha Selvaraj

Kaavya Paruchuri

Sara Haidermota

Rachel Bernardo

Stephen S Rich

Gina M Peloso

Pradeep Natarajan

Roles

Abstract

Purpose

Methods

Results

Conclusion

Introduction

Results

Baseline characteristics

Table 1. Baseline characteristics for discovery and replication cohorts.

T2D-stratified GWAS of TG identified an associated locus on chromosome 6

Fig 1. Genome-wide heterogeneity between T2D strata.

Fig 2. HLA-DQB1/DQA2 rs9274619:A significantly interacts with T2D on triglycerides.

rs9274619:A tags HLA-DQB1*0302

Phenome-wide interaction analyses implicates multiple diabetes-related complications

Fig 3. Phenome wide association (PheWAS) of disease conditions stratified by T2D status.

Metabolic characterizations of interactions with T2D

Discussion

Materials and methods

Study participants

Ethics

Phenotypes

Genotypes

Statistical analysis

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Arnar Palsson

Roles

Transfer Alert

Author response to Decision Letter 0

Decision Letter 1

Arnar Palsson

Roles

Author response to Decision Letter 1

Decision Letter 2

Arnar Palsson

Roles

Acceptance letter

Arnar Palsson

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

rs9274619:A tags *HLA-DQB1**0302