To determine whether genetic risk for type 1 diabetes (T1D) differentiates the four Aβ subgroups of ketosis-prone diabetes (KPD), where A+ and A− define the presence or absence of islet autoantibodies and β+ and β− define the presence or absence of β-cell function.
We compared T1D genetic risk scores (GRS) of patients with KPD across subgroups, race/ethnicity, β-cell function, and glycemia.
Among 426 patients with KPD (54% Hispanic, 31% African American, 11% White), rank order of GRS was A+β− > A+β+ = A−β− > A−β+. GRS of A+β− KPD was lower than that of a T1D cohort, and GRS of A−β+ KPD was higher than that of a type 2 diabetes cohort. GRS was lowest among African American patients, with a similar distribution across KPD subgroups.
T1D genetic risk delineates etiologic differences among KPD subgroups. Patients with A+β− KPD have the highest and those with A−β+ KPD the lowest GRS.
Introduction
Ketosis-prone diabetes (KPD) is a heterogeneous syndrome defined by presentation with diabetic ketoacidosis (DKA) (1,2) and has been reported across a wide geographic and racial/ethnic population range (1,3,4). The Aβ classification scheme defines four subgroups of KPD based on the presence or absence of islet autoantibodies (A+ vs. A−) and β-cell functional reserve (β+ vs. β−) following the index episode of DKA (1,5) and predicts glycemic control and insulin dependence (1,5). Both β− subgroups resemble patients with type 1 diabetes (T1D) (lean, early onset, absolute insulin dependence). Both β+ subgroups resemble patients with type 2 diabetes (T2D) (with obesity or overweight, late onset, potential for insulin independence following the index DKA episode) (1,5).
Clinical and pathophysiologic features differentiate the KPD subgroups (6–10). The A− subgroups are defined by the absence of T1D-associated autoantibodies, but ∼50% of A−β− and some patients with A−β+ KPD have T-cell–mediated islet autoimmunity without circulating autoantibodies (7). Diabetes duration may affect autoantibody positivity; hence, additional markers of islet autoimmunity might improve the predictive value of the Aβ classification of KPD.
The T1D genetic risk score (GRS) accurately distinguishes autoimmune T1D from non-T1D forms of diabetes (11–14). Progression to insulin dependence in adults diagnosed with T2D is associated with the T1D GRS (15). No studies have assessed the association of T1D GRS with the various subgroups of KPD, which comprise phenotypes that range from T1D-like to T2D-like. We asked whether the T1D GRS could refine the distinctions between the KPD subgroups and define a role for islet autoimmunity in their pathogenesis.
Research Design and Methods
We studied a cohort of patients with KPD classified by Aβ criteria (1,2,16), comparing differences in T1D GRS across KPD subgroups and racial/ethnic groups and against clinical features and disease progression. The study was approved by the institutional review board of the Baylor College of Medicine. Participants were adults admitted during 1999–2017 to Ben Taub Hospital for DKA who consented to participate in a KPD registry. DKA was defined by pH <7.3, serum glucose >11 mmol/L, serum ketones >5.2 mmol/L, serum β-hydroxybutyrate >1.5 mmol/L or urine ketones >13.9 mmol/L, and/or serum bicarbonate ≤17 mmol/L. Sera and DNA were obtained 2–6 months following the index DKA episode when the patients were taking stable doses of insulin. A+ status was defined by the presence of antibodies to GAD65, IA-2 antigen, or zinc transporter 8 in serum (1). Patients with A− KPD were negative for all three autoantibodies. β+ status was defined by fasting serum C-peptide ≥1 ng/mL or peak glucagon-stimulated C-peptide ≥1.5 ng/mL (1). Patients with β− KPD had C-peptide levels below both thresholds. Race/ethnicity was defined by self-report.
Following determination of the Aβ status, attempts were made to wean patients with β+ KPD off insulin therapy (replaced by metformin) if they met specific glycemic and other criteria (1). If metabolic control deteriorated after discontinuing insulin, additional oral glucose-lowering agents were added, followed by reinstatement of insulin, if needed. Patients with A−β+ KPD maintain significant β-cell functional reserve and remain insulin independent for a median of 4 years (16).
We assayed DNA samples of 426 patients with KPD, genotyping 30 T1D single nucleotide polymorphisms (SNPs) as previously reported (11). We excluded samples where genotyping results were missing for alleles with the greatest influence on the GRS (DR3/DR4–DQ8 or HLA_DRB1_15) or more than two other SNPs or if insufficient clinical data were available to define a KPD subgroup. The T1D GRS is the sum across SNPs of the number of risk-increasing alleles at that SNP multiplied by the ln(odds ratio) for each allele divided by the number of alleles. The DR3 (DRB1*0301–DQA1*0501–DQB1*0201) and DR4-DQ8 (DRB1*04–DQA1*0301–DQB1*0302) haplotypes do not fit this log-additive model, and weights for DR3/DR4–DQ8 were assigned based on imputed haplotypes (11) (Supplementary Table 1). Comparators were GRS of childhood-onset T1D case and control subjects from the Wellcome Trust Case Control Consortium (WTCCC). Stata 17 software was used for statistical testing (17).
Results
Median (interquartile range [IQR]) age at diagnosis was 35 (24–45) years, with 0.5 (0–7.3) years duration until the index DKA episode. Median (IQR) fasting C-peptide level was 0.9 (0.25–1.8) ng/mL, with a peak glucagon-stimulated C-peptide level of 1.7 (0.5–3.5) ng/mL (Table 1). Race/ethnicity distribution was 54% Hispanic, 31% African American, 11% White, and 4% Asian. One SNP from the 30-SNP score (rs11594656) failed SNP-specific quality control, so it was not included in the score. All T1D GRS results and comparisons used the remaining 29 SNPs. Twenty-eight individuals were excluded based on missingness criteria. The range of GRS was 0.118–0.341, with a median (IQR) of 0.252 (0.225–0.275).
Patient characteristics
| KPD subgroup | |||||
|---|---|---|---|---|---|
| A+β− | A+β+ | A−β− | A−β+ | P | |
| Patients, n (%) | 89 (21) | 48 (11) | 84 (20) | 205 (48) | |
| Age at diagnosis of diabetes (years) | 23 (16–38) | 37 (31–46) | 30 (17–40) | 39 (29–50) | <0.0001 |
| Interval between diabetes diagnosis and index DKA (years) | 4.9 (0.2–11.3) | 0 (0–0.01) | 5.6 (0.3–11.5) | 0 (0–3.9) | |
| BMI (kg/m2) | 24 (22–28) | 30 (25–35) | 24 (23–27) | 31 (26–37) | <0.0001 |
| Weight category, n (%) | |||||
| Lean | 53 (60) | 13 (27) | 50 (60) | 41 (20) | |
| Overweight | 21 (24) | 9 (19) | 23 (27) | 54 (27) | |
| Obesity | 14 (16) | 26 (54) | 11 (13) | 107 (53) | |
| Race/ethnicity, n (%) | |||||
| African American | 30 (34) | 13 (27) | 26 (31) | 63 (31) | |
| Asian American | 5 (6) | 1 (2) | 4 (5) | 5 (2) | |
| White | 14 (16) | 7 (15) | 14 (17) | 13 (6) | |
| Hispanic | 40 (45) | 27 (56) | 40 (48) | 124 (60) | |
| Fasting C-peptide (ng/mL) | 0.10 (0.09–0.50) | 1.30 (0.98–1.96) | 0.24 (0.09–0.50) | 1.69 (1.10–2.41) | <0.0001 |
| Peak C-peptide (ng/mL) | 0.19 (0.09–0.50) | 2.23 (1.60–3.52) | 0.50 (0.10–1.00) | 3.12 (1.98–4.80) | <0.0001 |
| KPD subgroup | |||||
|---|---|---|---|---|---|
| A+β− | A+β+ | A−β− | A−β+ | P | |
| Patients, n (%) | 89 (21) | 48 (11) | 84 (20) | 205 (48) | |
| Age at diagnosis of diabetes (years) | 23 (16–38) | 37 (31–46) | 30 (17–40) | 39 (29–50) | <0.0001 |
| Interval between diabetes diagnosis and index DKA (years) | 4.9 (0.2–11.3) | 0 (0–0.01) | 5.6 (0.3–11.5) | 0 (0–3.9) | |
| BMI (kg/m2) | 24 (22–28) | 30 (25–35) | 24 (23–27) | 31 (26–37) | <0.0001 |
| Weight category, n (%) | |||||
| Lean | 53 (60) | 13 (27) | 50 (60) | 41 (20) | |
| Overweight | 21 (24) | 9 (19) | 23 (27) | 54 (27) | |
| Obesity | 14 (16) | 26 (54) | 11 (13) | 107 (53) | |
| Race/ethnicity, n (%) | |||||
| African American | 30 (34) | 13 (27) | 26 (31) | 63 (31) | |
| Asian American | 5 (6) | 1 (2) | 4 (5) | 5 (2) | |
| White | 14 (16) | 7 (15) | 14 (17) | 13 (6) | |
| Hispanic | 40 (45) | 27 (56) | 40 (48) | 124 (60) | |
| Fasting C-peptide (ng/mL) | 0.10 (0.09–0.50) | 1.30 (0.98–1.96) | 0.24 (0.09–0.50) | 1.69 (1.10–2.41) | <0.0001 |
| Peak C-peptide (ng/mL) | 0.19 (0.09–0.50) | 2.23 (1.60–3.52) | 0.50 (0.10–1.00) | 3.12 (1.98–4.80) | <0.0001 |
Data are median (IQR) unless otherwise indicated.
T1D GRS Is Associated With T1D-Related Etiology in KPD Subgroups
Frequencies of the four KPD subgroups were as follows: A+β−, 21%; A+β+, 11%; A−β−, 20%; and A−β+, 48%. A+β− KPD had the highest median (IQR) GRS of 0.270 (0.248–0.288) (Fig. 1), which was lower than that of children of European ancestry with T1D in the WTCCC (0.285 [0.267–0.305], P < 0.0001). The median GRS of A+β+ KPD of 0.254 (0.228–0.279) was lower than that of A+β− KPD (P = 0.04). The median GRS of A−β− KPD of 0.255 (0.225–0.270) was similar to that of A+β+ and lower than that of A+β− (P = 0.003) KPD. The median GRS of A−β+ KPD of 0.241 (0.221–0.267) was significantly lower than those of A+β− (P < 0.0001), A+β+ (P = 0.05), and A−β− (P = 0.048) KPD. However, the median A−β+ KPD GRS was higher than that of the European ancestry control cohort of individuals without diabetes (0.236 [0.212–0.258], P < 0.0003) (Fig. 1) and of a cohort of adults with T2D (0.235 [0.212–0.258], P < 0.0001) (data not shown) in the WTCCC (11,12).
Violin plot of the 29-SNP T1D GRS in patients with KPD stratified according to the Aβ classification (N = 426). A denotes autoantibody positive or negative status, and β denotes β-cell functional status (β+, fasting C-peptide ≥1.0 ng/mL or glucagon-stimulated C-peptide ≥1.5 ng/mL; β−, fasting C-peptide <1.0 ng/mL or glucagon-stimulated C-peptide <1.5 ng/mL). We compared T1D GRS in the KPD cohort with a previously described cohort of pediatric diabetes (WTCCC T1D) and control subjects (WTCCC control). The white dots represent median values, solid vertical bars represent IQRs, thin vertical bars represent the rest of the distribution, and shaded areas represent kernel density estimations of the distribution shape of the data. P values were not corrected for multiple testing.
Violin plot of the 29-SNP T1D GRS in patients with KPD stratified according to the Aβ classification (N = 426). A denotes autoantibody positive or negative status, and β denotes β-cell functional status (β+, fasting C-peptide ≥1.0 ng/mL or glucagon-stimulated C-peptide ≥1.5 ng/mL; β−, fasting C-peptide <1.0 ng/mL or glucagon-stimulated C-peptide <1.5 ng/mL). We compared T1D GRS in the KPD cohort with a previously described cohort of pediatric diabetes (WTCCC T1D) and control subjects (WTCCC control). The white dots represent median values, solid vertical bars represent IQRs, thin vertical bars represent the rest of the distribution, and shaded areas represent kernel density estimations of the distribution shape of the data. P values were not corrected for multiple testing.
T1D GRS Is Associated With T1D Biomarkers
There was a clear difference in median (IQR) GRS comparing autoantibody-positive (A+) patients with autoantibody-negative (A−) patients overall (0.267 [0.242–0.285] vs. 0.241 [0.221–0.266], P < 0.0001). Individuals with multiple autoantibodies had a higher GRS than those with a single autoantibody (multiple 0.275 [0.251–0.293] vs. single 0.265 [0.241–0.281], P < 0.0001), with a trend toward increasing genetic risk with increasing autoantibody number (Fig. 2A and Supplementary Fig. 1).
Comparison of the 29-SNP T1D GRS in patients with KPD with established biomarkers of T1D (N = 426). A: Dot plot of T1D GRS stratified by number of positive autoantibodies. B: Dot plot of T1D GRS stratified by fasting C-peptide of <1.0 ng/mL or ≥1.0 ng/mL. Red dots represent median values.
Comparison of the 29-SNP T1D GRS in patients with KPD with established biomarkers of T1D (N = 426). A: Dot plot of T1D GRS stratified by number of positive autoantibodies. B: Dot plot of T1D GRS stratified by fasting C-peptide of <1.0 ng/mL or ≥1.0 ng/mL. Red dots represent median values.
GRS was associated inversely with both baseline and glucagon-stimulated C-peptide levels (Fig. 2B and Supplementary Fig. 2). Longitudinal C-peptide measurements showed strong associations with β-cell functional status (1), with the two β+ groups having the highest levels during follow-up (Supplementary Fig. 3). GRS was not associated with follow-up C-peptide levels when the patients were stratified by KPD subgroup.
T1D GRS Is Lower Among African American Patients Compared With Hispanic Patients Within All KPD Subgroups
The GRS was lower in the African American (n = 132) group than in the Hispanic (n = 231) and White (n = 48) groups (P < 0.0001) (Supplementary Fig. 4). The GRS for African American patients was lower for each KPD subgroup compared with Hispanic patients (P < 0.0001) (Fig. 3). Within each racial/ethnic group, GRS was highest in patients with A+β− KPD and lowest in those with A−β+ KPD.
Dot plots of the 29-SNP T1D GRS in patients with KPD stratified by the Aβ classification and by self-reported ethnicity (African American, n = 132; Hispanic n = 231; White, n = 48). A denotes positive or negative autoantibody status, and β denotes β-cell functional status (β+, fasting C-peptide ≥1.0 ng/mL or glucagon-stimulated C-peptide ≥1.5 ng/mL; β−, fasting C-peptide <1.0 ng/mL or glucagon-stimulated C-peptide <1.5 ng/mL). We compared T1D GRS in the KPD cohort with a previously described cohort of pediatric diabetes patients and control subjects from the WTCCC (reference). Red dots represent median values. P values were not corrected for multiple testing.
Dot plots of the 29-SNP T1D GRS in patients with KPD stratified by the Aβ classification and by self-reported ethnicity (African American, n = 132; Hispanic n = 231; White, n = 48). A denotes positive or negative autoantibody status, and β denotes β-cell functional status (β+, fasting C-peptide ≥1.0 ng/mL or glucagon-stimulated C-peptide ≥1.5 ng/mL; β−, fasting C-peptide <1.0 ng/mL or glucagon-stimulated C-peptide <1.5 ng/mL). We compared T1D GRS in the KPD cohort with a previously described cohort of pediatric diabetes patients and control subjects from the WTCCC (reference). Red dots represent median values. P values were not corrected for multiple testing.
Conclusions
Patients with A+β− KPD had the highest and those with A−β+ KPD the lowest median T1D GRS, demonstrating that T1D genetic risk differed among KPD subgroups (1,5,6). African American patients had lower GRS than Hispanic patients, indicating that race/ethnicity is an important variable in interpreting GRS results. The distributions of GRS among the KPD subgroups were similar for all racial/ethnic groups, suggesting that the contribution of islet autoimmunity to the pathogenesis of the different subgroups is similar across races/ethnicities.
Whereas the GRS associations are consistent with an autoimmune etiology for A+β− KPD and a predominantly nonautoimmune-mediated etiology for A−β+ KPD, etiologies for KPD subgroups with intermediate GRS appear more complex, with an atypical pathophysiology of T1D-associated islet autoimmunity in A+β+ KPD (6–8) and T-cell–mediated islet autoimmunity in many patients with A−β− KPD (7). The lower GRS in patients with A+β+ compared with A+β− KPD is consistent with the higher frequency of HLA alleles protective against T1D (6) and an epitope-specific GAD65 antibody response associated with a milder clinical phenotype (9) in the former subgroup. The lower T1D GRS of patients with A−β− compared with A+β− KPD suggests that the former patients are less genetically prone to islet autoimmunity.
Our T1D GRS was developed in a European ancestry cohort and, hence, may not apply equally to all races/ethnicities (13,14,18). Most of our patients self-reported as African American or Hispanic. African American patients had a lower GRS overall, as well as within every KPD subgroup, compared with Hispanic patients, indicating differences in background genetic susceptibility to islet autoimmunity. In the future, access to large cohorts of diverse ancestries with T1D diagnosed at different ages may permit development of transancestry or ancestry-specific GRS. Better capture of diverse HLA class II alleles may facilitate the use of transancestry scores; however, this will still leave differences in the distribution of scores by ancestry (19). Onengut-Gumuscu et al. (18) highlighted that an African American–specific T1D GRS increases accuracy in defining genetic risk for islet autoimmunity among African Americans, but this approach raises the challenge that not all individuals self-identify with a single racial/ethnic group; hence, more nuanced solutions may be needed (20). Genome-wide common variant analysis could allow genetic assessment with adjustment for ancestry and assessment of newer T1D, T2D, and pathway-specific GRS to better characterize genetic contributions to KPD. Despite this limitation, we observed distribution differences among KPD groups in African Americans.
Our data show genetic differences in the propensity for islet autoimmunity among KPD subgroups. A higher T1D GRS is associated with being classified as having A+β− KPD and a lower T1D GRS with being classified as having A−β+ KPD. Future research in individuals of non-White ancestry should generate ethnic-specific cutoffs in T1D GRS to classify patients with KPD with greater etiologic precision.
This article contains supplementary material online at https://doi.org/10.2337/figshare.23638950.
D.O. and S.N.M. are joint first authors.
Article Information
Acknowledgments. The authors thank the staff of the Ben Taub General Hospital KPD clinic and all the study participants.
Funding. This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, grant R01-DK104832 and the Rutherford Chair, Baylor-St. Luke’s Medical Center/Baylor College of Medicine (to A.B.) and a Diabetes UK Harry Keen Fellowship (16/0005529) (to R.A.O.).
Duality of Interest. R.A.O. previously had UK Medical Research Council confidence-in-concept funding to develop a T1D GRS biochip with Randox Laboratories. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. D.O. and S.N.M. collected and organized the longitudinal data and biosamples. S.N.M., R.A.O., and A.B. analyzed the results and drafted the manuscript. C.S.H. performed the autoantibody assays. R.G. and N.R. performed the clinical assessments and biochemical classification and monitored the metabolic status of the patients with KPD. M.N.W. and R.A.O. performed the genetic assays and analyzed the results. R.A.O. and A.B. designed the study and researched the data. All authors reviewed the manuscript and contributed to the discussion. R.A.O. and A.B. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for accuracy of the data and the integrity of the data analysis.
