Single Islet Autoantibody at Diagnosis of Clinical Type 1 Diabetes is Associated With Older Age and Insulin Resistance

Abstract

Context

Multiple islet autoantibody positivity usually precedes clinical (stage 3) type 1 diabetes (T1D).

Objective

To test the hypothesis that individuals who develop stage 3 T1D with only a single autoantibody have unique metabolic differences.

Design

Cross-sectional analysis of participants in the T1D TrialNet study.

Setting

Autoantibody-positive relatives of individuals with stage 3 T1D.

Participants

Autoantibody-positive relatives who developed stage 3 T1D (at median age 12.4 years, range = 1.4–58.6) and had autoantibody data close to clinical diagnosis (n = 786, 47.4% male, 79.9% non-Hispanic white).

Main Outcome Measures

Logistic regression modeling was used to assess relationships between autoantibody status and demographic, clinical, and metabolic characteristics, adjusting for potential confounders and correcting for multiple comparisons.

Results

At diagnosis of stage 3 T1D, single autoantibody positivity, observed in 119 (15.1%) participants (72% GAD65, 13% microinsulin antibody assay, 11% insulinoma-associated antigen 2, 1% islet cell antibody, 3% autoantibodies to zinc transporter 8 [ZnT8]), was significantly associated with older age, higher C-peptide measures (fasting, area under the curve, 2-hour, and early response in oral glucose tolerance test), higher homeostatic model assessment of insulin resistance, and lower T1D Index60 (all P < 0.03). While with adjustment for age, 2-hour C-peptide remained statistically different, controlling for body mass index (BMI) attenuated the differences. Sex, race, ethnicity, human leukocyte antigen DR3-DQ2, and/or DR4-DQ8, BMI category, and glucose measures were not significantly associated with single autoantibody positivity.

Conclusions

Compared with multiple autoantibody positivity, single autoantibody at diagnosis of stage 3 T1D was associated with older age and insulin resistance possibly mediated by elevated BMI, suggesting heterogeneous disease pathogenesis. These differences are potentially relevant for T1D prevention and treatment.

Heterogeneity of type 1 diabetes (T1D) is increasingly recognized as a potential barrier to effective prevention, reversal, and treatment of the disease (1). In individuals with T1D, there is great variation in the aggressiveness of the autoimmune attack (2), and there is loss of beta-cell function as evidenced by the different rates of C-peptide decline following symptomatic onset and in preclinical stages of T1D (3), as well as in the time from appearance of serum islet autoantibodies until development of overt disease (4, 5). The number of positive islet autoantibodies is one of the most important determinants of risk and rate of progression to clinical (stage 3) T1D (6–8), and this has been regarded as a measure of the aggressiveness of islet autoimmunity. Positivity for 2 or more autoantibodies (ie, multiple autoantibody positivity) usually precedes clinical T1D and is broadly accepted to mark stage 1 of the disease, while stage 2 is defined by the addition of dysglycemia (9). Interestingly, 15% of the National Institutes of Health-sponsored Type 1 Diabetes TrialNet Pathway to Prevention participants who develop T1D have single autoantibody positivity at diagnosis (10). We have previously reported that individuals with single autoantibody positivity at the time of stage 3 T1D diagnosis demonstrate features suggestive of less aggressive disease, and such persons are more likely to carry a type 2 diabetes-associated TCF7L2 genetic variant (10, 11). Hence, we hypothesized that individuals with stage 3 T1D and single autoantibody positivity may have additional features that distinguish them from those who have multiple autoantibody positivity at onset. These differences may shed light on the pathogenic mechanisms contributing to the development of diabetes in individuals with relatively less aggressive islet autoimmunity, as reflected by single autoantibody positivity. Furthering our understanding of the factors that drive differences among subgroups of persons with T1D will enable the development of more effectively tailored preventive and therapeutic strategies.

Materials and Methods

Participants

Type 1 Diabetes TrialNet is a National Institutes of Health-funded, international network of centers that aim to prevent T1D and stop disease progression (12). The TrialNet Pathway to Prevention Study is an observational study that prospectively follows at-risk individuals (first- or second-degree relatives of people with T1D) for progression of islet autoimmunity and development of clinical T1D (13). We studied TrialNet Pathway to Prevention participants (autoantibody-positive relatives of individuals with T1D) who developed clinical T1D and had islet autoantibodies tested in the timeframe within 180 days prior to the clinical onset of T1D and up to 14 days after clinical onset. Nine subjects without detectable autoantibodies at or near clinical onset were excluded, resulting in a final sample size of 786 participants. Institutional Review Board approval of the study was obtained at all participating sites, and written informed consent and assent, as applicable, were obtained.

Procedures

Study protocol.

All participants were screened for islet autoantibodies to glutamic acid decarboxylase autoantibodies (GADA), insulin (microinsulin antibody assay, mIAA), and insulinoma-associated antigen 2 (IA-2A). If any of these were positive in screening, autoantibodies to zinc transporter 8 (ZnT8A) and islet cell antibodies (ICA) were also tested. TrialNet Pathway to Prevention methods for measuring islet autoantibodies have been previously described (13). Importantly, once individuals are identified as autoantibody positive, subsequent longitudinal evaluations of autoantibodies include testing for all 5 T1D-associated autoantibodies. Participants were monitored with autoantibody testing, hemoglobin A_1c, and an oral glucose tolerance test (OGTT) at 6- or 12-month intervals depending on estimated risk (14), as previously described (15).

Diagnosis of diabetes.

Clinical (stage 3) diabetes was diagnosed according to TrialNet Natural History Study of the Development of Type 1 Diabetes Protocol (TrialNet Protocol TN01), following American Diabetes Association criteria (16): fasting plasma glucose ≥7.0 mmol/L, 2-hour plasma glucose during an OGTT ≥11.1 mmol/L, a random plasma glucose ≥11.1 mmol/L with symptoms of hyperglycemia, or presence of unequivocal hyperglycemia including acute metabolic decompensation (diabetic ketoacidosis). The first 3 criteria were required to be met on 2 occasions with a strong preference that at least 1 of the 2 testing occasions includes an OGTT. Hemoglobin A_1c level ≥48 mmol/mol (6.5%) from a laboratory that is using the National Glycohemoglobin Standardization Program certified assay standardized to the Diabetes Control and Complications Trial was also accepted as a confirmatory criterion.

Human Leukocyte Antigen typing.

Human leukocyte antigen (HLA) genotyping was performed in the TrialNet HLA Laboratory at the Barbara Davis Center for Childhood Diabetes in Denver, Colorado. Prior to May 2013, DQA1/DQB1 typing was performed using the Roche Molecular Systems Linear Array typing strip, while DRB1 typing was performed using the Abbott Laboratories sequencing kit. As of May 2013, the laboratory uses Luminex microbead technology with kits from OneLambda.

Autoantibody assays.

GADA, IA-2A, mIAA, and ZnT8A were measured by radioimmunoassay in the TrialNet Core Laboratory at the Barbara Davis Center. During the 2015 Islet Autoantibody Standardization Program Workshop, sensitivities and specificities were 52% and 100%, respectively, for mIAA, 82% and 99%, respectively, for GADA, 72% and 100%, respectively, for IA-2A, and 70%, and 97% respectively, for ZnT8A (17). ICA positivity was tested by indirect immunofluorescence in the Diagnostic Referral Laboratories at the University of Florida. Metabolic measures: Participants underwent an OGTT (oral glucose dose 1.75 g/kg, maximum 75 grams) after an overnight fast. C-peptide (nmol/L) and glucose (mmol/L) measurements were performed in the fasting state followed by 30, 60, 90, and 120 minutes later. Trapezoid method was used to calculate area under the curve (AUC) for C-peptide. For each participant, we also calculated a T1D Index60, which is a composite measure of log fasting C-peptide, 60-minute glucose, and 60-minute C-peptide during OGTTs. This Index developed in autoantibody-positive individuals enrolled in the Diabetes Prevention Trial (DPT-1) and TrialNet Pathway to Prevention (18). The homeostatic model assessment of insulin resistance (HOMA-IR) was calculated as follows: fasting insulin (milliunits/L) x fasting glucose (mmol/L)/22.5 (19). The early C-peptide response was defined as the difference between the 30- and 0-minute C-peptide values. Index60 has been shown to be an earlier and better diagnostic tool than glucose alone (18), and before clinical onset of type 1 diabetes, Index60 outperforms glucose alone to identify a profile of T1D characteristics, such as younger age, lower C-peptide, and higher autoantibody number (20).

Statistical analyses

We analyzed TrialNet participants who progressed to clinical (stage 3) T1D, and compared individuals who were positive for a single islet autoantibody positive with those who were positive for multiple (ie, 2 or more) autoantibodies at the time of diagnosis. Subject characteristics were first summarized using descriptive statistics and were compared between groups of interest (eg, single vs multiple autoantibody positive at diagnosis) using Wilcoxon rank-sum tests for comparison of continuous variables or chi-square tests for comparisons of categorical variables. Associations and differences in measures between groups were also done using graphical analyses such as side-by-side boxplots. To formally evaluate the association of various factors with the frequency of being single autoantibody positive at the time of clinical diagnosis of T1D (vs multiple autoantibody positive), univariate logistic regression models were used to initially assess these relationships. In addition, we used multivariable logistic regression modeling along with regression methodologies to evaluate the combined influence of factors, when adjusting for known potential confounders, as well as to identify “optimal” models that best identify those likely to be single autoantibody positive at stage 3 T1D. Interaction effects were also evaluated in the context of the multivariable logistic regression models. Statistical significance was determined if P values were <0.05. Corrections for multiple comparisons were made within sets of analyses (eg, across 23 variable comparisons in the overall cohort) using the Benjamini–Hochberg false discovery rate adjustment. All analyses were performed using the statistical program R (version 3.5.1 for Windows). Forest plots were generated using the metafor package in R.

Results

We studied the 786 Pathway to Prevention TrialNet participants who met the aforementioned inclusion criteria. The median age at initial positive autoantibody determination was 10.3 years (range: 1.1–48.6), with 47.4% being male, 79.9% non-Hispanic white, and 86.6% first-degree relatives of a person with T1D. At the time of clinical diagnosis, the median age was 12.4 years (range: 1.4–58.6 years), 15.1% (n = 119) of the participants had a single positive autoantibody, and 84.9% (n = 667) had multiple positive autoantibodies. Of the 119 participants with single autoantibody positivity at diagnosis, 86 (72.3%) were positive for GADA, 16 for mIAA (13.4%), 13 for IA-2A (10.9%), 1 for ICA (0.8%), and 3 for ZnT8A (5.1%). Other participant characteristics are illustrated in Table 1.

Table 1.

Participant Characteristics at Diagnosis of Stage 3 Type 1 Diabetes

Characteristic	All Participants, n = 786	Single Ab+, n = 119	Multiple Ab+, n = 667	P Value (Single vs Multiple)
Age at Ab+ (years)
Median	10.3	14.7	9.9	<0.0001
Range	1.1 to 48.6	1.3 to 48.6	1.1 to 45.8
Age (years)
Median	12.4	18.1	11.7	<0.0001
Range	1.4–58.6	3.0–58.6	1.4–51.1
Sex				0.98
Female	412 (52.6%)	62 (52.1%)	350 (52.7%)
Male	371 (47.4%)	57 (47.9%)	314 (47.3%)
Missing	3	0	3
Race				0.73
White	683 (86.9%)	105 (88.2%)	578 (86.7%)
Non-white or multiracial	43 (5.5%)	7 (5.9%)	36 (5.4%)
Unknown/missing	60 (7.6%)	7 (5.9%)	53 (7.9%)
Ethnicity				0.95
Hispanic or Latino	68 (8.7%)	11 (9.2%)	57 (8.5%)
Non-Hispanic or Latino	675 (85.9%)	102 (85.7%)	573 (85.9%)
Unknown/missing	43 (5.5%)	6 (5.0%)	37 (5.5%)
HLA
DR3-DQ2 and DR4-DQ8				0.63
No	616 (83.2%)	98 (85.2%)	518 (82.9%)
Yes	124 (16.8%)	17 (14.8%)	107 (17.1%)
DR3-DQ2 or DR4-DQ8				0.36
No	129 (17.4%)	24 (20.9%)	105 (16.8%)
Yes	611 (82.6%)	91 (79.1%)	520 (83.2%)
Missing	46	4	42
Maximum number of Ab+ prior to stage 3
1	45 (5.7%)	45 (37.8%)	0 (0%)	<0.0001
2	136 (17.3%)	47 (39.5%)	89 (13.3%)
3	208 (26.5%)	22 (18.5%)	186 (27.9%)
4	239 (30.4%)	5 (4.2%)	234 (35.1%)
5	158 (20.1%)	0 (0%)	158 (23.7%)
Median	4	2	4	<0.0001
Range	1 to 5	1 to 4	2 to 5
Number of Ab+
1	119 (15.1%)	119 (100%)		n/a
2	160 (20.4%)		160 (24.0%)
3	198 (25.2%)		198 (29.7%)
4	219 (27.9%)		219 (32.8%)
5	90 (11.5%)		90 (13.5%)
Ab+ type
mIAA				<0.0001
No	439 (56.1%)	103 (86.6%)	336 (50.6%)
Yes	344 (43.9%)	16 (13.4%)	328 (49.4%)
GADA				0.019
No	153 (19.5%)	33 (27.7%)	120 (18.0%)
Yes	633 (80.5%)	86 (72.3%)	547 (82.0%)
IA-2A				<0.0001
No	250 (31.8%)	106 (89.1%)	144 (21.6%)
Yes	536 (68.2%)	13 (10.9%)	523 (78.4%)
ICA				<0.0001
No	258 (32.8%)	118 (99.2%)	140 (21.0%)
Yes	528 (67.2%)	1 (0.8%)	527 (79.0%)
ZnT8A				<0.0001
No	182 (36.4%)	57 (94.9%)	125 (28.4%)
Yes	318 (63.6%)	3 (5.1%)	315 (71.6%)
Not tested	286	59	227
BMI percentile
Median BMI percentile	70.7	76.4	70.2	0.11
BMI percentile range	0 to 99.99	4.4 to 99.93	0 to 99.99
BMI category				0.085
Normal/underweight	505	74 (64.9%)	431 (66.8%)
Overweight	122	13 (11.4%)	109 (16.9%)
Obese	132	27 (23.7%)	105 (16.3%)
Missing	27	5	22

Abbreviations: Ab+, islet autoantibody positive; BMI, body mass index; GADA, glutamic acid decarboxylase autoantibody; IA-2A, insulinoma-associated antigen 2; ICA, islet cell antibody; mIAA, microinsulin antibody assay.

All variables are at diagnosis of stage 3 type 1 diabetes except if otherwise noted. P values for comparisons of continuous variables correspond to Wilcoxon rank-sum tests, and comparisons of categorical variables correspond to chi-square or Fisher exact tests. The missing data outcome was not included in comparisons except for race. Bolded P values reflect those that were significant after a Benjamini–Hochberg false discovery rate adjustment for multiple comparisons. Glucose was expressed in mg/dL and C-peptide in ng/mL.

In univariate analyses, participants who presented with a single positive autoantibody at stage 3 T1D, as compared with multiple autoantibody-positive individuals, were older at diagnosis (median 18.1 years vs 11.7 years, P < 0.0001). Furthermore, single autoantibody-positive individuals, versus those with multiple autoantibodies, had higher C-peptide measures, including fasting (P = 0.0001), AUC (P < 0.0001), and 2-hour C-peptide values (P < 0.0001), as well as higher early C-peptide response (0–30 minutes; P = 0.007) and elevated HOMA-IR scores (P = 0.021; Table 2). Single autoantibody-positive participants had lower T1D Index60 values (P = 0.0003). Of note, fasting, AUC, and 2-hour glucose values, as well as sex, race, ethnicity, HLA, or BMI category were not statistically different between the 2 groups. These and other immunologic and metabolic characteristics are shown in Tables 1–3 (by number of positive autoantibodies), and Table 4 (by age of clinical onset).

Table 2.

Logistic Regression Models Evaluating the Association Between Single Autoantibody Positivity at Progression to Stage 3 Type 1 Diabetes With Metabolic and Immunologic Characteristics (n = 786)

	Univariate		Adjusting for Age at Stage 3 Type 1 Diabetesa
Covariate	OR	P Value	OR	P Value
Fasting glucose	1.003	0.16	1.000	0.99
Fasting C-peptide	1.38	0.0001	1.06	0.55
Mean AUC glucose	0.998	0.35	0.997	0.21
Mean AUC C-peptide	1.21	<0.0001	1.10	0.045
2-hour glucose	0.998	0.11	0.998	0.23
2-hour C-peptide	1.16	<0.0001	1.08	0.019
Early C-peptide response	1.20	0.007	1.13	0.11
Index60	0.74	0.0003	0.83	0.029
Index60 ≥2.0 (vs <2.0)	0.65	0.053	0.80	0.35
HOMA-IR	1.013	0.021	1.01	0.088
Single Ab+ at screening	20.23	<0.0001	16.97	<0.0001
mIAA positive	0.16	<0.0001	0.22	<0.0001
GADA positive	0.57	0.014	0.32	<0.0001
IA-2A positive	0.03	<0.0001	0.04	<0.0001
ICA positive	0.002	<0.0001	0.002	<0.0001

Abbreviations: Ab+, islet autoantibody positive; AUC, area under the curve; GADA, glutamic acid decarboxylase autoantibody; HOMA-IR, homeostatic model assessment of insulin resistance; IA-2A, insulinoma-associated antigen 2; ICA, islet cell antibody; mIAA, microinsulin antibody assay; OR, odds ratio.

^aAdjusted for age at screening when modeling autoantibody type at screening.All variables are at diagnosis of stage 3 type 1 diabetes except if otherwise noted. Bolded P values reflect those that were significant after a Benjamini–Hochberg false discovery rate adjustment for multiple comparisons. Glucose was expressed in mg/dL and C-peptide in ng/mL.

Table 3.

Metabolic Markers at Diagnosis of Stage 3 Type 1 Diabetes, Overall and by Number of Positive Autoantibodies

	All Subjects N = 786		Single Ab+ at T1D Diagnosis		Multiple Ab+ at T1D Diagnosis
Metabolic Measure	Median	Range	Median	Range	Median	Range	P Valuea
Fasting C-peptide	1.49	0.17–10.67	1.85	0.43–10.67	1.43	0.17–8.80	0.0002
2-hour C-peptide	4.40	0.44–28.5	5.90	1.24–27.82	4.32	0.44–28.5	<0.0001
Fasting glucose	104.5	44.5–468.5	109.25	66–337	103.5	44.5–468.5	0.03
2-hour glucose	247.5	70–688	235	129–570	251	70–688	0.085
Fasting insulin	111.0	1.0–195.0	119.75	1.0–185.5	109.2	1.0–195	0.14
2-hour insulin	24.3	0.5–938	34.15	4.7–483.6	23.45	0.5–938	<0.0001
HOMA-IR	29.2	0.14–125.2	33.72	0.16–125.23	29.03	0.14–89.03	0.06

Abbreviations: Ab, islet autoantibody; HOMA-IR, homeostatic model assessment of insulin resistance; T1D: type 1 diabetes.

^aWilcoxon rank-sum test to compare continuous markers between single and multiple autoantibody-positive subgroups. Missing values: Fasting C-peptide (n = 31), 2-hour C-peptide (n = 60), fasting glucose (n = 29), 2-hour glucose (n = 54), fasting insulin (n = 28), 2-hour insulin (n = 100), HOMA-IR (n = 29). Bolded P values reflect those that were significant after a Benjamini–Hochberg false discovery rate adjustment for multiple comparisons. Glucose was expressed in mg/dL and C-peptide in ng/mL.

Table 4.

Metabolic Markers at Diagnosis of Stage 3 Type 1 Diabetes, by Age of Onset (Children versus Adults)

	<18 Years Old at T1D Diagnosis		≥18 Years Old at T1D Diagnosis
Metabolic Marker	Median	Range	Median	Range	P Valuea
Fasting C-peptide	1.39	0.17–7.07	1.91	0.18–10.67	<0.0001
2-hour C-peptide	4.20	0.44–27.82	5.84	0.46–28.5	<0.0001
Fasting glucose	100.0	44.5–468.5	120.2	79.5–337.0	<0.0001
2-hour glucose	249	70–688	244	83–519	0.64
Fasting insulin	116	1.0–195	103.75	1.0–193	0.50
2-hour insulin	23.2	0.90–483.6	30.9	0.5–938	0.0004
HOMA-IR	28.52	0.14–125.23	32.38	0.22–83.22	0.0375

Abbreviations: HOMA-IR, homeostatic model assessment of insulin resistance; T1D: type 1 diabetes.

^aWilcoxon rank-sum test to compare continuous markers between single and multiple autoantibody-positive subgroups. Missing values: Fasting C-peptide (n = 31), 2-hour C-peptide (n = 60), fasting glucose (n = 29), 2-hour glucose (n = 54), fasting insulin (n = 28), 2-hour insulin (n = 100), HOMA-IR (n = 29). Bolded P values reflect those that were significant after a Benjamini–Hochberg false discovery rate adjustment for multiple comparisons. Glucose was expressed in mg/dL and C-peptide in ng/mL.

Given the significant difference in age between the single and multiple autoantibody-positive groups and the known influence of age on the variables under study (eg, C-peptide levels, autoantibody type), we further evaluated these relationships in multivariable models that allowed us to adjust for age as a potential confounder. With adjustment for age at diagnosis of stage 3 T1D, single autoantibody positivity was significantly associated with higher AUC (P = 0.045) and 2-hour C-peptide (P = 0.019), as well as lower Index60 (P = 0.029; Table 2). With correction for multiple comparisons, the differences for AUC C-peptide and Index60 were no longer significant. Upon age adjustment, there was also a trend for higher HOMA-IR (P = 0.088) in single autoantibody-positive participants.

Because of the association between BMI and C-peptide (21), we further adjusted for BMI. With adjustment for age and BMI percentile, the associations persisted linking single autoantibody positivity at stage 3 T1D with higher AUC (P = 0.036) and 2-hour C-peptide (P = 0.017), as well as lower Index60 (P = 0.013), but these differences were not significant after correction for multiple comparisons. We also observed that, while the correlation between age and BMI percentile at clinical onset was statistically significant and positive in both subgroups, it was stronger in single than multiple autoantibody-positive participants (respectively, r = 0.41, P < 0.0001 and r = 0.11, P = 0.007).

We observed differences between children and adults on factors associated with single or multiple autoantibody positivity at diagnosis of stage 3 T1D. In adults, but not in children, single autoantibody positivity at diagnosis of stage 3 T1D was significantly associated with older age at clinical onset (P = 0.006), higher AUC C-peptide (P = 0.005), higher 2-hour C-peptide (P = 0.001), lower AUC glucose (P = 0.005), lower 2-hour glucose (P = 0.001), and lower T1D Index60 (P = 0.0007), as well as trends for an association with higher fasting C-peptide (P = 0.031) and higher early C-peptide response (P = 0.041) (Table 5). Single autoantibody positivity was associated with higher likelihood of being obese in adults (P = 0.022) but the opposite trend (ie, for lower likelihood of overweight or obesity) in children (P = 0.015). On the other hand, children with single positivity were less likely to be GADA positive (P < 0.0001) and had higher HOMA-IR scores (P = 0.003) than those with multiple autoantibodies, but these relationships were not significant in adults.

Table 5.

Logistic Regression Models Evaluating the Association Between Single Autoantibody Positivity at Stage 3 Type 1 Diabetes With Participant Characteristics by Age at Progression

	<18 Years of Age (n = 595; 59 had single Ab+ at stage 3)		≥18 Years of Age (n = 191; 60 had single Ab+ at stage 3)
Covariate	OR	P Value	OR	P Value
Age	1.028	0.44	1.045	0.006
Age at Ab+ screening	0.995	0.88	1.037	0.022
Single Ab+ at screening	21.44	<0.0001	12.93	<0.0001
mIAA positive	0.27	<0.0001	0.06	0.006
GADA positive	0.29	<0.0001	0.72	0.57
IA-2A positive	0.05	<0.0001	0.03	<0.0001
Male (vs female)	1.28	0.37	0.90	0.74
Non-white or multiracial (vs white)	0.78	0.55	1.60	0.37
Hispanic (vs non-Hispanic)	0.94	0.90	1.51	0.46
BMI percentile	0.995	0.24	1.02	0.013
Overweight/obese (vs lean)	0.40	0.015	1.87	0.052
Overweight (vs lean)	0.17	0.017	1.36	0.48
Obese (vs lean)	0.65	0.30	2.37	0.022
Fasting glucose	1.001	0.76	1.0007	0.88
Fasting C-peptide	1.06	0.72	1.27	0.031
Mean AUC glucose	1.0008	0.75	0.987	0.005
Mean AUC C-peptide	1.11	0.15	1.16	0.005
2-hour glucose	1.0009	0.56	0.99	0.001
2-hour C-peptide	1.09	0.07	1.12	0.004
Early C-peptide response	1.12	0.29	1.23	0.041
Index60	0.94	0.61	0.67	0.0007
HOMA-IR	1.023	0.003	0.996	0.64

Abbreviations: Ab+, islet autoantibody positive; AUC, area under the curve; BMI, body mass index; GADA, glutamic acid decarboxylase autoantibody; HOMA-IR, homeostatic model assessment of insulin resistance; IA-2A, insulinoma-associated antigen 2; mIAA, microinsulin antibody assay; OR, odd ratio.

All variables are at diagnosis of stage 3 type 1 diabetes except if otherwise noted. Bolded P values reflect those that were significant after a Benjamini–Hochberg false discovery rate adjustment for multiple comparisons. Glucose was expressed in mg/dL and C-peptide in ng/mL.

In participants who had at least 1 of the T1D high-risk HLA haplotypes (ie, DR3-DQ2 and/or DR4-DQ8 [n = 611, including 472 children aged <18 years and 139 adults aged ≥18 years]), the associations between single autoantibody positivity and other factors were similar to those for the overall group, including the differences between children and adults. Specifically, the significant association between single autoantibody positivity and lower Index60 (P = 0.009), as well as trends with older age (P = 0.026), higher BMI (P = 0.033), and higher C-peptide measures (ie, fasting, AUC, 2-hour, and early response; respectively, P = 0.042, P = 0.029, P = 0.025, and P = 0.043) were observed in adults (aged ≥18 years) but not in children (aged <18 years) (Table 6). On the other hand, the association for higher HOMA-IR and single autoantibody positivity was significant in children (P = 0.003) but not adults. However, while there was no association between single positivity and GADA positivity in the DR3-DQ2/DR4-DQ8 participants, participants with neither DR3-DQ2 nor DR4-DQ8 and who were positive for a single autoantibody were less likely to be GADA positive (P < 0.0001) (Table 7). Other associations in the participants with neither DR3-DQ2 nor DR4-DQ8 were consistent with those found in the individuals with DR3-DQ2 and/or DR4-DQ8, although this subgroup was relatively small (n = 129). Because of the small sample size, we did not conduct analyses with stratification by age in this subgroup.

Table 6.

Logistic Regression Models Evaluating the Association Between Single Autoantibody Positivity at Stage 3 Type 1 Diabetes With Characteristics of Participants with ≥ 1 High-risk HLA Haplotype by Age at Progression

	All (n = 611)		Age < 18 Years (n = 472a)		Age ≥ 18 Years (n = 139b)
Covariate	OR	P Value	OR	P Value	OR	P Value
Age	1.06	<0.0001	1.05	0.21	1.04	0.026
Age ≥18 yrs (vs <18)	5.26	<0.0001
Age at Ab+ screening	1.06	<0.0001	1.03	0.53	1.03	0.062
Single Ab+ at screening	21.14	<0.0001	21.07	<0.0001	15.4	<0.0001
mIAA positive	0.09	<0.0001	0.17	<0.0001	c	c
GADA positive	0.89	0.68	0.49	0.032	0.51	0.42
IA-2A positive	0.03	<0.0001	0.05	<0.0001	0.02	<0.0001
Male (vs female)	0.94	0.78	1.03	0.93	0.91	0.79
Non-white or multiracial (vs white)	0.92	0.79	0.89	0.80	1.48	0.49
Hispanic (vs non-Hispanic)	1.37	0.40	1.12	0.84	1.71	0.36
BMI percentile	1.01	0.025	0.999	0.89	1.02	0.033
Overweight/obese (vs not)	1.23	0.39	0.56	0.14	1.86	0.094
Overweight (vs lean)	0.82	0.57	0.24	0.053	1.40	0.47
Obese (vs lean)	1.68	0.067	0.92	0.85	2.37	0.052
Fasting glucose	1.004	0.062	1.002	0.50	1.003	0.55
Fasting C-peptide	1.48	<0.0001	1.18	0.33	1.34	0.042
Mean AUC glucose	0.9996	0.85	1.003	0.29	0.9899	0.041
Mean AUC C-peptide	1.24	<0.0001	1.15	0.11	1.15	0.029
2-hour glucose	0.9985	0.31	1.002	0.26	0.992	0.013
2-hour C-peptide	1.17	<0.0001	1.10	0.13	1.11	0.025
Early C-peptide response	1.22	0.014	1.11	0.41	1.30	0.043
Index60	0.78	0.006	1.03	0.80	0.71	0.009
HOMA-IR	1.01	0.025	1.03	0.003	0.992	0.46

Abbreviations: Ab+, islet autoantibody positive; AUC, area under the curve; GADA, glutamic acid decarboxylase; HOMA-IR, homeostatic model assessment of insulin resistance; IA-2A, insulinoma-associated antigen 2; mIAA, microinsulin antibody assay; OR, odds ratio; yrs, years.

All variables are at diagnosis of stage 3 type 1 diabetes except if otherwise noted. All participants in this Table had at least one of the type 1 diabetes-associated HLA Haplotypes (DR3-DQ2 and/or DR4-DQ8).

^an = 43 had single autoantibody at diagnosis. ^bn = 48 had single autoantibody at diagnosis. ^cunstable parameter estimates as none of the participants who were 18+ years of age and mIAA-positive at diagnosis and had DR3+ and/or DR4+ were also single autoantibody positive at diagnosis. Bolded P values reflect those that were significant after a Benjamini–Hochberg false discovery rate adjustment for multiple comparisons. Glucose was expressed in mg/dL and C-peptide in ng/mL.

Table 7.

Logistic Regression Model Evaluating the Association Between Single Autoantibody Positivity at Stage 3 Type 1 Diabetes With Characteristics of Participants with neither DR3-DQ2 nor DR4-DQ8) (n = 129)

Covariate	OR	P Value
Age	1.03	0.052
Age ≥18 yrs (vs <18)	2.29	0.081
Age at Ab+ screening	1.03	0.12
Single Ab+ at screening	16.92	<0.0001
mIAA positive	0.63	0.33
GADA positive	0.14	<0.0001
IA-2A positive	0.05	<0.0001
Male (vs female)	1.80	0.20
BMI percentile	0.992	0.30
Overweight or obese (vs not)	0.57	0.31
Overweight (vs lean)	0.27	0.22
Obese (vs lean)	0.80	0.71
Fasting glucose	0.994	0.47
Fasting C-peptide	1.19	0.26
Mean AUC glucose	0.992	0.13
Mean AUC C-peptide	1.15	0.063
2-hour glucose	0.995	0.16
2-hour C-peptide	1.11	0.033
Early C-peptide response	1.14	0.33
Index60	0.63	0.023
HOMA-IR	0.997	0.82

Abbreviations: Ab+, islet autoantibody positive; AUC, area under the curve; BMI, body mass index; GADA, glutamic acid decarboxylase; HOMA-IR, homeostatic model assessment of insulin resistance; IA-2A, insulinoma-associated antigen 2; mIAA, microinsulin antibody assay; OR, odds ratio; yrs, years.

All variables are at diagnosis of stage 3 type 1 diabetes except if otherwise noted. Bolded P values reflect those that were significant after a Benjamini–Hochberg false discovery rate adjustment for multiple comparisons. Age was expressed in years, glucose in mg/dL, and C-peptide in ng/mL.

Discussion

We studied 786 participants in the TrialNet Pathway to Prevention Study who had developed stage 3 T1D, and observed that, at diagnosis, 15% of them were positive for only a single islet autoantibody. The transition from single to multiple islet autoantibodies is regarded as a critical step in the path to develop clinical T1D, and most individuals have multiple islet autoantibodies present at clinical diagnosis (9). Therefore, individuals with single autoantibody positivity at presentation of T1D are outliers, and the mechanisms that govern their progression to diabetes are currently unknown. In the present study, we have demonstrated that, at diagnosis of stage 3 T1D, single autoantibody-positive individuals, compared with those with multiple autoantibodies, are significantly older and have evidence of higher insulin resistance. The latter was suggested by higher C-peptide (fasting, AUC, and 2-hour) despite glucose levels (fasting, AUC, and 2-hour) not being different, and further supported by higher HOMA-IR (a measure of insulin resistance). The significant differences for these metabolic measures in the univariate analysis disappeared after adjustment for BMI percentile and age, suggesting that insulin resistance was likely mediated by obesity/overweight and age.

Early OGTT-stimulated C-peptide response was higher in participants with single autoantibody positivity. During progression from preclinical (ie, autoantibody positivity; stages 1 and 2) to clinical (stage 3) T1D, the early C-peptide response decreases, and the late C-peptide response increases (22–24). This finding is consistent with our previous study that showed children with single autoantibody positivity at clinical onset of T1D had higher random C-peptide levels and were less likely to present with diabetic ketoacidosis, a marker of profound insulin deficiency (11). This pattern of more preserved insulin secretion in single autoantibody-positive individuals suggests that participants in this group, although diagnosed with clinical diabetes, might be at a less advanced stage of autoimmune beta-cell destruction. Since these individuals have developed diabetes (despite their more preserved beta-cell function), other factors, such as insulin resistance, may have contributed to the progression and onset of their disease. This hypothesis is supported by previous studies in autoantibody-positive relatives showing that elevated insulin resistance, particularly if accompanied by impaired insulin secretion (25, 26), accelerates progression to clinical diabetes. Thus, in individuals with just mild islet autoimmunity, as reflected by single autoantibody positivity, the development of diabetes may be the result of a combination of insulin insufficiency and insulin resistance (27, 28). On the other hand, in individuals with multiple autoantibody positivity, the profound insulin deficiency resulting from aggressive islet autoimmunity may be sufficient to cause diabetes without additional predisposing metabolic factors.

Individuals with single autoantibody positivity had lower a T1D Index60 than those with multiple autoantibodies. This was expected, as Index60 measures C-peptide in relation to glucose during an OGTT, signals insulin deficiency, and predicts progression to T1D (18, 20, 29). A low T1D Index60 in individuals with single autoantibody-positive stage 3 T1D further supports that their diabetes has elements less characteristic of classical T1D, such as older age at onset, insulin resistance, and fewer abnormalities in insulin secretion. These findings are consistent with previous literature, which suggests that an elevated Index60 is associated with classical T1D (18, 20). Furthermore, stratified analysis by HLA demonstrated that in the absence of T1D HLA genotypes, single autoantibody positivity was associated with a low Index60. Therefore, individuals who develop clinical T1D despite absence of high-risk HLA genotypes and presence of only mild islet autoimmunity (as reflected by single autoantibody positivity), also have low a T1D Index60 but elevated insulin resistance. Taken together, these observations indicate that the pathogenesis leading to diabetes in these individuals may involve classical T1D mechanisms combined with other pathways, such as insulin resistance, suggesting an involvement of type 2 diabetes-like pathways (27). Consistent with this concept, a recent study by Hippich et al suggested that, in relatives with low-risk genetic markers, it is additional factors that increase their risk of T1D compared with that in the general population (30). On the other hand, among participants with high-risk HLA genotypes, single autoantibody positivity was associated with older age at clinical diagnosis. This observation may indicate that, despite high genetic risk, single autoantibody individuals have a protective mechanism resulting in slower progression of islet autoimmunity and slower decline in beta-cell function. The slower deterioration of beta-cell function could lead them to be older than multiple autoantibody individuals by the time clinical diabetes occurs.

BMI (percentile or category) was not associated with single autoantibody positivity in the overall cohort. This likely occurred because single autoantibody status was positively associated with BMI among adults and negatively associated with BMI in children. Among children, single autoantibody positivity, as compared with multiple, was associated with a trend toward lower BMI percentile at clinical onset despite having higher insulin resistance as reflected by higher HOMA-IR. This finding is consistent with our earlier findings in other pediatric cohorts (11, 31) and may suggest that factors other than obesity, such as puberty, may be driving insulin resistance in children. However, Buryk et al found that, although adiposity was not associated with the number of islet autoantibodies, it was associated with higher T-cell responses to diabetes antigens (32). Of note, previous reports indicate that higher BMI in the first 2 years of life was associated with islet autoantibody positivity in pediatric relatives of persons with T1D (33), and we have observed that progression from single to multiple islet autoantibody positivity is accelerated by elevated BMI (Ferrara et al, unpublished). Libman et al reported a secular increase in the prevalence of overweight children with newly diagnosed T1D (34). Further studies are required to clarify these relationships.

This study’s strengths include the uniqueness of the cohort of autoantibody-positive individuals followed longitudinally, availability of multiple metabolic measures, and a large sample size. One of the limitations of this study is that, being a secondary analysis of an established cohort, there are data such as Tanner stage and waist circumference that were not available to us. Also, the observational design of the study precluded us from drawing conclusions on causality. Finally, we cannot rule out the possibility that the observed differences in insulin resistance are related to 1 or more unmeasured variables, such as exposure to hyperglycemic excursions in the weeks prior to diagnosis and measurement of the biomarkers reported here.

In summary, our data suggest that individuals who progress to stage 3 T1D with single autoantibody positivity are older and have higher insulin resistance than those who present with multiple autoantibody positivity. These findings support the concept that insulin resistance and relative insulin deficiency (28) may accelerate clinical presentation and, thus, strategies aimed at preventing T1D in these individuals should include agents targeting this pathogenic mechanism. In addition, our observations highlight the need for further studies to understand the heterogeneity of T1D.

Glossary

Abbreviations

AUC

area under the curve

BMI

body mass index

GADA

glutamic acid decarboxylase autoantibody

HLA

human leukocyte antigen

HOMA-IR

homeostatic model assessment of insulin resistance

ICA

islet cell antibod

mIAA

microinsulin antibody assay

OGTT

oral glucose tolerance test

T1D

type diabetes

ZnT8A

zinc transporter 8 autoantibody

Financial Support: We acknowledge the support of the Type 1 Diabetes TrialNet Study Group, which identified study participants and provided samples and follow-up data for this study. The Type 1 Diabetes TrialNet Study Group is a clinical trials network funded by the National Institutes of Health (NIH) through the National Institute of Diabetes and Digestive and Kidney Diseases, the National Institute of Allergy and Infectious Diseases (NIAID), and the Eunice Kennedy Shriver National Institute of Child Health and Human Development, through cooperative agreements U01 DK061010, U01 DK061034, U01 DK061042, U01 DK061058, U01 DK085465, U01 DK085453, U01 DK085461, U01 DK085466, U01 DK085499, U01 DK085504, U01 DK085509, U01 DK103180, U01 DK103153, U01 DK085476, U01 DK103266, U01 DK103282, U01 DK106984, U01 DK106994, U01 DK107013, U01 DK107014, UC4 DK097835, UC4 DK106993, and Juvenile Diabetes Foundation (JDRF).

Author Contributions: M.J.R. designed the study, interpreted the data and wrote the first draft of the manuscript. J.S., I.L., J.M., M.T., M.A. and D.B. contributed to study design and data interpretation, and reviewed/edited the manuscript. S.G. contributed to study design, analyzed the data, contributed to data interpretation, and reviewed/edited the manuscript. All authors are members of the Type 1 Diabetes TrialNet Study Group. M.J. Redondo is the guarantor of this article and takes full responsibility for the work as a whole, including the study design, access to data, and the decision to submit and publish the manuscript.

Type 1 Diabetes TrialNet Study Group participants: Steering Committee: C.J. Greenbaum (Benaroya Research Institute), M.A. Atkinson (University of Florida), D.A. Baidal (University of Miami), M. Battaglia (San Raffaele University), D. Becker (University of Pittsburgh), P. Bingley (University of Bristol), E. Bosi (San Raffaele University), J. Buckner (Benaroya Research Institute), M. Clements (The Children’s Mercy Hospital), P.G. Colman (Walter & Eliza Hall Institute of Medical Research), L. DiMeglio (Indiana University), C. Evans-Molina (Indiana University), S.E. Gitelman (University of California, San Francisco), R. Goland (Columbia University), P. Gottlieb (Barbara Davis Center for Childhood Diabetes), K. Herold (Yale University), M. Knip (University of Helsinki), J.P. Krischer (University of South Florida), A. Lernmark (Skane University Hospital), W. Moore (The Children’s Mercy Hospital), A. Moran (University of Minnesota), A. Muir (Emory Children’s Center), J. Palmer (University of Washington), M. Peakman (King’s College), L. Philipson (University of Chicago), P. Raskin (University of Texas Southwestern), M. Redondo (Baylor College of Medicine), H. Rodriguez (University of South Florida Diabetes and Endocrinology Center), W. Russell (Vanderbilt Eskind Diabetes Clinic), D.A. Schatz (University of Florida), J.M. Sosenko (University of Miami), L. Spain (National Institute of Diabetes and Digestive and Kidney Diseases [NIDDK]), J. Wentworth (Walter & Eliza Hall Institute of Medical Research), D. Wherrett (University of Toronto), D.M. Wilson (Stanford University), W. Winter (University of Florida), A. Ziegler (Technical University Munich). Past Members: M. Anderson (University of California, San Francisco), P. Antinozzi (Wake Forest University), R. Insel (Juvenile Diabetes Research Foundation [JDRF]), T. Kay (St. Vincent’s Institute of Medical Research), J.B. Marks (University of Miami), A. Pugliese (University of Miami), B. Roep (Leiden University Medical Center), J.S. Skyler (University of Miami), J. Toppari (Hospital District of Southwest Finland). Executive Committee: C.J. Greenbaum (Benaroya Research Institute), J.P. Krischer (University of South Florida), E. Leschek (National Institute of Diabetes and Digestive and Kidney Diseases [NIDDK]), L. Spain (National Institute of Diabetes and Digestive and Kidney Diseases [NIDDK]). Past Members: K. Bourcier (National Institute of Allergy and Infectious Diseases [NIAID]), R. Insel (Juvenile Diabetes Research Foundation [JDRF]), J. Ridge (National Institute of Allergy and Infectious Disease [NIAID]), J.S. Skyler (University of Miami). Chair’s Office: C.J. Greenbaum (Benaroya Research Institute), L. Rafkin (University of Miami), J.M. Sosenko (University of Miami). Past Members: J.S. Skyler (University of Miami), I. Santiago (University of Miami). TrialNet Coordinating Center (University of South Florida): J.P. Krischer, B. Bundy, M. Abbondondolo, T. Adams, I. Asif, J. Bjellquist, M. Boonstra, C. Burroughs, M. Cleves, D. Cuthbertson, M. DeSalvatore, C. Eberhard, S. Fiske, J. Ford, J. Garmeson, S. Geyer, B. Hays, C. Henderson, M. Henry, K. Heyman, B. Hsiao, C. Karges, B.-G. Koziol, L. Lane, S. Liu, J. Lloyd, K. Maddox, J. Malloy, J. Martin, C. McNeill, M. Moore, S. Muller, T. Nguyen, J. Nunez, R. O’Donnell, M. Parker, MJ Pereyra, A. Roberts, K. Sadler, C. Sullivan, R. Tamura, E. Walker-Veras, M.V. Warnock, K. Wood, R. Wood, P. Xu, V. Yanek, K. Young. Past Members: D. Amado, A. Kinderman, A. Leinbach, J. Miller, N. Reed, T. Stavros. National Institute of Diabetes and Digestive and Kidney Diseases [NIDDK]: E. Leschek, L. Spain. Data Safety and Monitoring Board: E. Blumberg (University of Pennsylvania), S. Aas (Georgetown University), G. Beck (Cleveland Clinic Foundation), R. Gubitosi-Klug (Case Western Reserve University), L. Laffel (Joslin Diabetes Center), R. Vigersky (Medtronic), D. Wallace (Research Triangle Institute). Past Members: D. Brillon (Cornell University), R. Veatch (Georgetown University). Infectious Disease Safety Committee: B. Loechelt (Children’s National Medical Center), L. Baden (Brigham and Women’s Hospital), P. Gottlieb (Barbara Davis Center for Childhood Diabetes), M. Green (University of Pittsburgh), E. Leschek (National Institute of Diabetes and Digestive and Kidney Diseases [NIDDK]), A. Weinberg (University of Colorado). Laboratory Directors: S. Marcovina (University of Washington), J.P. Palmer (University of Washington), J. Tischfield (Rutgers University), A. Weinberg (University of Colorado), W. Winter (University of Florida), L. Yu (Barbara Davis Center for Childhood Diabetes). TrialNet Clinical Network Hub (Benaroya Research Institute): A. Shultz, E. Batts, A. Pagryzinski, M. Ramey, M. Tobin. Past Members: K. Fitzpatrick, R. Guerra, M. Romasco, C. Webb. Active Personnel at Clinical Centers Participating in the TN01 Protocol: Barbara Davis Center for Childhood Diabetes, Aurora, Colorado: A.K. Steck, J. Albery, B. Bradfield, L. Chesshir, P.A. Gottlieb, A.W. Michels, M.J. Rewers, K. Simmons, M. VanDyke. Baylor College of Medicine, Houston, Texas: M.J. Redondo, M. Pietropaolo, S. Pietropaolo, M. Tosur. Benaroya Research Institute, Seattle, Washington: C.J. Greenbaum, J.H. Buckner, W. Hao, S. Lord, M. McCulloch-Olson, M. Ramey, E. Sachter, J. Snavely, M. Tobin, C. Tordillos, D. VanBuecken, N. Wickstrom. The Children’s Mercy Hospital, Kansas City, Missouri: W. Moore, M. Beidelschies, D. Brenson-Hughes, J. Boyd, J. Cernich, M. Clements, J. Dolan, A. Elrod, E. Haith, K. Halpin, J. James, T. Luetjen, C. McClain, R. McDonough, S. Mitchell, F. Al Muhaisen, S. Orlich, E. Paprocki, B. Seuferling, J. Sexton, J. Slover. Columbia University, New York, New York: R. Goland, A. Alvarez, M. Bogun, R. Gandica, S. Pollak, K. Williams. Emory Children’s Center, Atlanta, Georgia: A. Muir, A. Antich, K. Cossen, E. Felner, M. Jenkins, L. Panagiotakopoulos, B. Powell-Lee, W. Sanchez. The Hospital for Sick Children, Toronto, Ontario: D.K. Wherrett, D. Dias, L. Eisel, R. Kovalakovska, B. Perro, M.J. Ricci. Indiana University, Indianapolis, Indiana: L.A. DiMeglio, C. Evans-Molina, E. Grubbs, M. Hildinger, H.M. Ismail, M. Legge, A. Newnum, J. Sanchez, E.K. Sims, M. Spall, M. Tebbe, S. Woerner. San Raffaele Hospital, Milan, Italy: E. Bosi, M. Battaglia, E. Bianconi, A. Bolla, R. Bonfanti, A. Caretto, G. Frontino, P. Grogan, A. Laurenzi, C. Molinari, M. Pastore, A. Petrelli. Skane University Hospital, Scania, Sweden: Å. Lernmark, L. Ahlkvist, H. Borg, B. Jónsdóttir, A. Katsarou, J.Å. Kördel, H. Larsson, M. Lundgren, U. Ulvenhag. Stanford University, Stanford, California: D.M. Wilson, N. Arrizon-Ruiz, T. Aye, L. Bachrach, T. Esrey, L. Nally. Technical University Munich, Munich, Germany: A. Ziegler, P. Achenbach, M. Bunk, M. Herbst, J. Hirte, A. Hofelich, V. Hoffmann, M. Kroetz, C. Ramminger, F. Reinmüller, K. Warncke, T. Welzhofer. University of Bristol, Bristol, England: C. Anderson, P. Bingley, T. Hughes, V. King, Y. Liu, S. Loud, B. Thorne, B. Wilson. University of California, San Francisco, California: S.E. Gitelman, F. Abdulhussein, M.S. Anderson, G. Auerback, J. Buchanan, H. Chesser, A. Erkin-Cakmak, C.T. Ferrara, A. Huang, K. Ko, J. Lee, R. Long, S. Sanda, C. Schulmeister, S.K. Selling, C. Torok, R. Wesch, J. Wong. University of Chicago, Chicago, Illinois: L. Philipson, D. Deplewski, G. Gannon, S. Greeley, L. Jones, H. Kolluri, M. Lemelman, C. Miles, M. Miller, R. Naylor, K. O’Sullivan, M. Pusinelli, S. Tucker, C. Yu. University of Florida, Gainesville, Florida: D.A. Schatz, A. Abraham, J. Adams, A. Albanese-O’Neill, M.A. Atkinson, B. Bruggeman, T.M. Brusko, M.J. Claire-Salzler, K. Dayton, M.J. Haller, P. Hiers, J. Hosford, L.M. Jacobsen, J. Marks, H. Rohrs, J. Silverstein, P. Towe, W.E. Winter, C. Zimmerman. University of Helsinki, Helsinki, Finland: M. Knip, M. Hirvasniemi, K. Koski, K. Luopajärvi, H. Siljander, R. Veijola. University of Miami, Miami, Florida: D.A. Baidal, C. Blaschke, D. Matheson, J. Sanchez, N. Sanders-Branca, J.S. Skyler, J.M. Sosenko. University of Minnesota, Minneapolis, Minnesota: A.M. Moran, M.D. Bellin, J. Leschyshyn, J. McVean, B.M. Nathan, B. Nelson, B. Pappenfus, J. Ruedy, A. Street, D. Weingartner. University of Pittsburgh, Pittsburgh, Pennsylvania: D.J. Becker, K. DeLallo, D. Groscost, M.B. Klein, I.M. Libman, K. Riley, F. Toledo. University of Texas Southwestern, Dallas, Texas: P. Raskin, L. Boyles. University of South Florida Diabetes and Endocrinology Center, Tampa, Florida: H. Rodriguez, C. Alverex, S. Bollepalli, J. Brown, R. Brownstein, A. Castro, E. Eyth, M. Feldman, R. Franks, D. Gomez, L. Guerra, C. Jacovino, V. Jorgensen, B. Moran, H. Navarte, A. Oxner, D. Shulman. Vanderbilt Eskind Diabetes Clinic, Nashville, Tennessee: W.E. Russell, F. Brendle, A. Brown, B. Hammel, J. Leshko, D.J. Moore, K. Rainer, T. Smith, J.W. Thomas, G. Williams, S. Wright. Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria: P. Colman, Marika Bjorasen, Spiros Fourlanos, L.C. Harrison, F. Healy, L. Redl, J.M. Wentworth. Yale University, New Haven, Connecticut: K.C. Herold, L. Feldman, R. Sherwin, W.V. Tamborlane, S.A. Weinzimer.

Additional Information

Disclosure statement: The authors have no relevant conflict of interest to disclose. The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the NIH or the JDRF.

Data Availability: All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.

Prior Presentation: Parts of these data were presented at the American Diabetes Association 78^th Scientific Sessions in June 2018.