Early Impairment of Insulin Sensitivity, β-Cell Responsiveness, and Insulin Clearance in Youth with Stage 1 Type 1 Diabetes

Abstract

Context

Clinical onset of type 1 diabetes (Stage 3 T1D) is preceded by a presymptomatic phase characterized by multiple islet autoantibodies with normal glucose tolerance (Stage 1 T1D).

Objective

The aim was to explore the metabolic phenotypes of β-cell function and insulin sensitivity and clearance in normoglycemic youth with Stage 1 T1D and compare them with healthy nonrelated peers during a 3-hour oral glucose tolerance test (OGTT).

Methods

Twenty-eight lean youth, 14 with ≥2 islet autoantibodies (cases) and 14 healthy controls underwent a 3-hour 9-point OGTT with measurement of glucose, C-peptide, and insulin. The oral minimal model was used to quantitate β-cell responsiveness (${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$) and insulin sensitivity (SI), allowing assessment of β-cell function by the disposition index ($\mathrm{D}\mathrm{I}={\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}\times \mathrm{S}\mathrm{I}$). Fasting insulin clearance (CL₀) was calculated as the ratio between the fasting insulin secretion rate (ISR) and plasma insulin levels (ISR₀/I₀), while postload clearance (CL₁₈₀) was estimated by the ratio of AUC of ISR over the plasma insulin AUC for the 3-hour OGTT (ISR_AUC/I_AUC). Participants with impaired fasting glucose, impaired glucose tolerance, or any OGTT glucose concentration ≥200 mg/dL were excluded.

Results

Cases (10.5 years [8, 15]) exhibited reduced DI (P < .001) due to a simultaneous reduction in both ${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$ (P < 0.001) and SI (P = .008) compared with controls (11.5 years [10.4, 14.9]). CL₀ and CL₁₈₀ were lower in cases than in controls (P = .005 and P = .019).

Conclusion

Presymptomatic Stage 1 T1D in youth is associated with reduced insulin sensitivity and lower β-cell responsiveness, and the presence of blunted insulin clearance.

Clinically symptomatic type 1 diabetes (Stage 3 T1D) is preceded by a prolonged presymptomatic phase characterized by progressive loss of functional β-cell mass after the onset of islet autoimmunity (1). During this presymptomatic phase, disease can be identified by measuring islet autoantibodies, thus defining a unique window of opportunity for diabetes prevention trials.

Early stages of T1D are defined as the presence of at least 2 islet autoantibodies and normal glucose tolerance (Stage 1) or dysglycemia (Stage 2 T1D) by the oral glucose tolerance test (OGTT). Individuals with a single diabetes autoantibody, who have an approximately 15% 10-year risk of progression to symptomatic (Stage 3) diabetes, are not considered to have diabetes. In contrast, those with 2 or more autoantibodies are considered to have early diabetes because of a 70% 10-year risk of progression to Stage 3 T1D, which increases to nearly 100% risk over a lifetime (2).

Early defects in insulin secretion have been described in Stage 1 T1D using intravenous glucose tolerance tests (3) or serial C-peptide measurements during the OGTT, suggesting that even though oral glucose tolerance is normal, β-cell damage begins early in the process (4-7). T1D has been thought of as primarily a disease of declining β-cell function (8), and little attention has been paid to the role that insulin sensitivity and clearance might play in the onset or progression of T1D.

We hypothesized that β-cell function measurements in the absence of quantitation of insulin sensitivity provide a limited picture of T1D early stages, which is necessary to optimize preventive strategies based on the functional impairment that underpins progression to overt diabetes (9, 10). The current study aimed to simultaneously estimate β-cell responsiveness, insulin sensitivity, and insulin clearance in youth with Stage 1 T1D compared with healthy peers, matched for age, sex, and body weight, by use of a 3-hour OGTT.

Materials and Methods

We conducted a multisite, prospective (cases), and retrospective (controls) analysis in a cohort of normoglycemic youth with at least 2 positive islet autoantibodies (Stage 1 T1D) and a matched control cohort of healthy nonrelated peers. Participants were enrolled at 3 clinical sites, with cases from the University of Minnesota (UMN) and Indiana University (IU), and controls from Yale University (Yale).

Cases

The cases from UMN and IU were relatives of individuals with T1D who were screened for diabetes autoantibodies as part of the Type 1 Diabetes TrialNet Pathway to Prevention study. TrialNet screens around 15 000 relatives of persons with T1D each year, and about 5% are found to be positive for 1 or more diabetes autoantibodies. Participants are first tested for anti-insulin (mIAA), antiglutamic acid decarboxylase (GAD65A), and protein tyrosine phosphatase (IA-2A) autoantibodies, and, if these are negative, for anti-islet cell (ICA) and anti-zinc transporter 8 antibodies (11). Those with 2 or more autoantibodies are invited to have a monitoring OGTT to differentiate Stage 1 from Stage 2 T1D.

Subjects <18 years of age who were scheduled for TrialNet OGTT monitoring were invited to participate in the current protocol. Those who agreed underwent a 3-hour, 9-point OGTT (see below). Participants from UMN and IU would ordinarily have had a 2-hour OGTT performed as part of TrialNet monitoring, and parents and participants enrolled in the current study consented/assented to extend the OGTT to 3 hours and to allow their data to be used for these analyses. A physical evaluation was performed prior to the OGTT that included weight and height measurements. The study protocol was approved by the Human Subjects Committee at each site.

Consent to participate in the study and to the use of de-identified registry-based data was obtained at the time of the visit. After obtaining the OGTT results, subjects were included in the current analysis if they had normal fasting and 2-hour glucose tolerance, and no glucose level ≥200 mg/dL during the OGTT (Stage 1 T1D). The TrialNet Pathway to Prevention study was registered at clinicaltrials.gov NCT00097292.

During the study period, 20 youth underwent TrialNet OGTT screening at UMN, and 11 of them agreed to the expanded 3-hour testing for the current protocol, while at IU 7 of 41 TrialNet OGTT participants agreed to the 3-hour test. The only documented reason for subject refusal was the inconvenience of the length of the expanded study. Excluded participants did not differ from the analyzed cohort with respect to age, fasting, or 2-hour glucose levels (data not shown).

Twelve months after the baseline study OGTT, a follow-up OGTT was planned. This was not possible due to research interruption by the SARS-COV2 pandemic, so all cases were followed-up by telephone to assess for the presence of diabetes symptoms or the known development of clinical Stage 3 T1D.

Controls

Matched healthy controls were retrospectively selected from the Yale Pediatric Diabetes Obesity Clinic repository. This repository includes control individuals who are not obese. Each participant from the Yale cohort underwent a 3-hour OGTT and was screened for islet autoantibodies as part of other studies. Healthy controls agreed, at the time of the OGTT, to permit use of their data for future studies. The matching procedure of cases and controls was conducted by grouping the cases and controls by age, sex, and body mass index (BMI) categories, then a nested case–control dataset was generated by the use of sequential run cycles with progressive narrowed categories by the automated STATA13 software (StataCorp 2013; Stata Statistical Software: Release 13; College Station, TX: StataCorp LP).

Islet Autoantibody Assay

The radiobinding assay for insulin autoantibodies (IAA) (12), glutamic acid decarboxylase antibodies (GADA) (13), protein tyrosine phosphatase IA-2 antibodies (IA-2A) (13), and Zinc transporter 8 antibodies (anti ZnT8A) (14) from samples collected at UMN and IU were all performed at the Barbara Davis Center for Diabetes, University of Colorado, using previously published assay methods. In the 2020 IASP (Islet Autoantibody Standardization Program) workshop, sensitivities and specificities were 62% and 99% for IAA, 78% and 99% for GADA, 72% and 100% for IA-2A, and 74% and 100% for ZnT8A. Islet autoantibodies from the Yale control cohort were performed at the Yale Center for Clinical Investigation core lab facility, which uses the same methodology described above.

Oral Glucose Tolerance Test

Prior to the OGTT, all subjects were instructed to follow a weight maintenance diet consisting of at least 250 g of carbohydrates per day for 7 days before the study and to avoid strenuous physical activity. Flavored dextrose in a dose of 1.75 g per kilogram of body weight (up to a maximum of 75 g) was given orally with time 0 set as immediately before the onset of beverage consumption. Blood samples were obtained at –10, 0, 10, 20, 30 minutes, and then every 30 minutes for 180 minutes (for a total of 9 time points) for the measurement of plasma glucose, insulin, and C-peptide. The average of the –10 minutes and baseline measurements was considered the “baseline” time point. Plasma insulin was measured by radioimmunoassay (Linco, St. Charles, MO) that had <1% cross-reactivity with C-peptide and proinsulin. Plasma C-peptide levels were determined with an assay from Diagnostic Product (Tosoh Bioscience, Los Angeles, CA).

In accordance with the American Diabetes Association classification (15), impaired glucose tolerance (IGT) was defined as a 2-hour plasma glucose level between 140 and 199 mg/dL, impaired fasting glucose as fasting glucose values 100 to 125 mg/dL, and diabetes as a fasting glucose level ≥126 mg/dL or a 2-hour plasma glucose level ≥200 mg/dL, or a HbA_1c ≥6.5% (48 mmol/mol). Dysglycemia, an exclusion criterion for this study, was defined as the presence of impaired fasting glucose, IGT, or any OGTT glucose concentration ≥200 mg/dL (15, 16).

OGTT-derived Measures of β-Cell Function, Insulin Sensitivity, and Clearance

The oral minimal model method (17) was used to estimate insulin sensitivity (S_I), β-cell responsiveness (${\phi }_{total}$) and β-cell function with respect to the prevailing insulin sensitivity (disposition index, DI). This model has been successfully validated in adults, children and adolescents against model-independent measurements using multiple-tracer meal protocols and euglycemic and hyperglycemic clamps (18, 19).

Insulin sensitivity (S_I) was estimated from the 3-hour OGTT plasma glucose and insulin concentrations using the oral glucose minimal model, which measures the overall effect of insulin on stimulating glucose disposal and inhibiting glucose production. A lower number represents increased insulin resistance. β-Cell responsiveness (${\phi }_{total}$) was quantified from 3-hour OGTT C-peptide and glucose concentrations using the oral C-peptide minimal model. The model assumes that glucose-stimulated insulin secretion is made up of 2 components: a dynamic component, representing the secretion of readily releasable insulin stimulated by the rate of increase in glucose concentration (${\phi }_d$), and a static component, which measures new insulin provision in response to a given increment in glucose above basal concentrations (${\phi }_s$) after a constant delay time. From ${\phi }_d$ and ${\phi }_s$, a total responsiveness index (${\phi }_{total}$) can be derived, which measures the ability of the β-cell to respond to a glucose stimulus (17, 20, 21). A higher number suggests better insulin secretion in response to glucose. The DI, which assesses the appropriateness of β-cell function relative to the level of insulin sensitivity, was calculated by multiplying ${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$ by S_I.

Insulin sensitivity was further explored by the use of the hepatic insulin resistance index (HIRI), computed to identify the hepatic component based on the OGTT plasma insulin and glucose measures. HIRI was calculated as the product of the area under the curve (AUC) of plasma glucose and plasma insulin during the first 30 minutes of the OGTT (glucose AUC_0-30 × insulin AUC_0-30). The HIRI has been shown to be correlated with hepatic insulin resistance (r = 0.64, P < .0001) estimated as the product of basal endogenous glucose production (measured with labelled glucose) and fasting plasma insulin (22). Additionally, compared with other estimates of insulin sensitivity, such as the homeostasis model assessment index, that simply rely on fasting glucose and insulin (23), HIRI accounts for the early (30 minutes) glucose and insulin response to an oral load (22).

Insulin clearance

Fasting insulin clearance (CL₀) was calculated as the ratio between the fasting insulin secretion rate (ISR) and plasma insulin levels (ISR₀/I₀), while postload clearance (CL₁₈₀) was estimated by the ratio of the AUC of ISR over the plasma insulin AUC for the 3-hour OGTT (ISR_AUC/I_AUC), as previously reported (24-27). The insulin secretion rate (ISR) was estimated by C-peptide deconvolution (20).

Statistical Analysis

The primary outcomes of the study were comparisons of β-cell function (DI) and CL₁₈₀ between cases (Stage 1 T1D) and controls (healthy nonrelated peers). Using data from a previous adolescent cohort (19), we estimated that with 12 participants per group we would have 99% power at the 2-sided α-level of 0.05 to detect a standardized difference of at least 10% for the DI, using a 2-sided Mann–Whitney test. Non-normally distributed variables were analyzed by the Kruskal–Wallis test, and categorical variables were compared using the chi-squared test. Data were summarized in tables using median (25th percentile, 75th percentile) for continuous variables and count (percent, %) for categorical variables. AUCs for glucose, C-peptide, and insulin during the 3-hour OGTT were computed by using the trapezoidal rule and compared between the 2 groups.

Results are shown as mean ± SD or median (25th, 75th centile) as appropriate. Analyses were performed using STATA.13 software (StataCorp 2013; Stata Statistical Software: Release 13; College Station, TX; StataCorp LP) and Prism 8.0 (GraphPad Software, San Diego, CA). The oral minimal model (OMM) was numerically identified by nonlinear least squares, as implemented in SAAM II v.2.3 (The Epsilon Group 2012-2013).

Results

Participant Characteristics

Twenty-eight participants were included for the final analysis (14 cases and 14 controls). Only 1 of these cases was diagnosed with diabetes during 12 months of follow-up. Baseline characteristics of cases and controls are displayed in Table 1. The 2 groups were matched with respect to age, BMI, and sex, as per study design. Out of the 14 cases, 3 were positive for 2 antibodies and 11 had 3 or more autoantibodies.

Table 1.

Participants’ characteristics, controls vs autoantibody-positive cases (AAbs+)

	Cases (n = 14)	Controls (n = 14)	P
Age (y)	10.5 (8.0, 15.0)	11.5 (10.4, 14.9)	.151
BMI (kg/m2)	19.5 (16.2, 24.6)	19.6 (18.4, 20.4)	.981
z-score BMI	0.75 (–0.07, 1.25)	0.80 (–0.07, 0.90)	.549
Sex (F) n (%)	7 (50)	6 (43)	.705
Fasting glucose (mg/dL)	91 (88, 100)	88 (84, 89.5)	.095
1-hour glucose (mg/dL)	149 (123, 183)	129.5 (103, 132)	.033
2-hour glucose (mg/dL)	119 (86, 133)	102.5 (85, 110)	.174
Fasting insulin (µU/mL)	7 (4.6, 15.6)	14 (10.5, 17.0)	.164
Fasting C-peptide (pmol/L)	503.0 (288, 857.5)	682.5 (508,790)	.349

Data are expressed as median (25th, 75th centile) or n (%). Values with P < .05 are presented in bold font.

Metabolic Data

Glycemic profiles during the 3-hour OGTT are displayed in Fig. 1A. Baseline and 2-hour glucose levels did not differ between the 2 groups, and thus all were considered to have normal glucose tolerance (Fig. 1A). However, the mid-OGTT glucose excursion for cases was modestly but significantly higher than for controls during the 3-hour test (AUC_glucose 22 119 ± 1267 mg/dL × minutes vs 19 357 ± 1046 mg/dL × minutes for cases and controls, respectively; P < .001), with greater 1-hour glucose in cases than controls (P = .033) (Table 1), although no mid-OGTT values reached the dysglycemia threshold of ≥200 mg/dL.

Figure 1.

Glucose (A), Insulin (B), and C-peptide (C) during oral glucose tolerance test (OGTT) by time in controls (blue) and cases (red). Data are expressed as mean ± SD.

Fasting baseline insulin and C-peptide levels were similar in cases and controls (P = .164 and P = .349, respectively) (Table 1). Though the trajectory of insulin during the 3-hour OGTT was similar for cases and controls (AUC_insulin 10 128 ± 8280 and 12 590 ± 4177, respectively, P = .330), C-peptide excursion was less in cases (AUC_c-peptide 308 426 ± 153 903 vs 441 204.9 ± 131 940.5 pmol × L/minute, for cases and controls, P = 0.021), supporting the presence of reduced β-cell responsiveness to a glucose load in cases (Fig. 1B and 1C).

β -Cell Function and Insulin Sensitivity

β-Cell function as quantified by the DI was significantly reduced in the cases as compared with controls (98.3 [79.5, 131.7] vs 568 [202.0, 857.3] × 10^–12 dL/kg/minute² per pmol/L, P < .001), as shown in Fig. 2A. This was due to a reduction of both β-cell responsiveness (${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}=36.2$ [29.3, 44.3] vs 86.4 [59.6, 100] × 10^–4/minute for cases and controls, P < .001, Fig. 2B) and insulin sensitivity (3.2 [2.2, 3.9] vs 8.1 [3.2, 10.1] × 10^–4 dL/kg/minute per µU/mL for cases and controls, P = .008, Fig. 2C). Lower β-cell responsiveness (${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$) resulted from a reduction of ${\phi }_s$ in cases compared with controls (P = .035), while ${\phi }_d$ was similar between the 2 groups (P = .066) (Fig. 2D and 2E). Consistent with increased whole body insulin resistance in cases, hepatic insulin resistance (HIRI) was also higher in cases than controls (3.2 [1.9, 6.2] vs 1.5 [1.3, 2.4] mg/dL/min × μU/mL/minute, P = .005). The reduction of insulin sensitivity was paralleled by a lower insulin clearance at baseline (3.6 [2.3, 4.9] vs 5.6[3.5, 11.9] pmol/L per μU/mL, P = .005) (Fig. 2F) and during the 3-hour OGTT (2.8 [1.7, 3.7] vs 6.0 [2.8, 9.2] pmol/L per μU/mL × minutes² for cases vs controls, P = .019). (Fig. 2G)

Figure 2.

Disposition index (DI) (A), ${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$ (B) and insulin sensitivity (SI) (C), ${\phi }_{\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{c}}$ (D) and ${\phi }_{\mathrm{d}\mathrm{y}\mathrm{n}\mathrm{a}\mathrm{m}\mathrm{i}\mathrm{c}}$ (E), baseline (F) and 180-min (G) insulin clearance for controls (blue) and cases (red). CL_0, baseline clearance; CL₁₈₀, 180-minute clearance; ISR_AUC/I_AUC, area under the curve (AUC) for insulin secretion rate over AUC of insulin over 180 minutes; ISR₀/I₀ , baseline insulin secretion rate over baseline insulin. Data are expressed as median (25th, 75th centile).

The relationship between β-cell responsiveness and insulin sensitivity (β-cell function, DI) is displayed in Fig. 3 for cases and controls: cases exhibited a left-downward shift of β-cell function compared with controls, with simultaneous reduction of both the ${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$ and SI (Fig. 3). No obvious difference was noted between subjects with 2 or 3 vs those with 4 or 5 autoantibodies, although the numbers were too small for statistical analysis.

Figure 3.

β-cell responsiveness (${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$) and insulin sensitivity (SI) for controls (blue circles) and cases with 2 or 3 antibodies (empty red triangle) or 4 or 5 antibodies (filled red triangle). The red and blue dashed fit-lines represent the relationship between ${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$ and SI for cases and controls, respectively. Data are expressed as naturally log-transformed values.

Discussion

The oral minimal model is a robust measure that has been used to quantify β-cell responsiveness in the context of prevailing SI using data from the 3-hour OGTT (19, 28, 29). The model has been previously validated in healthy adolescents (19) and adults (28, 29), and reflects postabsorption C-peptide, insulin, and glucose kinetics (30, 31). Herein, we describe the first use of the oral minimal model to compare youth with Stage 1 T1D (at least 2 islet autoantibodies in the absence of dysglycemia), to antibody-negative healthy peers. We found that preclinical diabetes in youth was associated with a lower DI. This was due to simultaneous reduction of insulin sensitivity and β-cell responsiveness (insulin secretion in response to glucose). While reduced β-cell responsiveness has previously been described in this population (7), the observation of an insulin resistant state in Stage 1 T1D is novel, as is our finding of lower baseline and postload insulin clearance in cases compared with controls. We also noted higher HIRI, a model-independent index of hepatic insulin resistance, suggesting that both hepatic and peripheral insulin resistance contributed to the reduced insulin sensitivity seen in youth with islet autoimmunity (22, 32).

The finding of impaired β-cell responsiveness (${\phi }_{\mathrm{t}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l}}$) is consistent with the lower C-peptide response to oral glucose and mixed meal tests previously described up to 5 years before the onset of clinical T1D (4). Autoantibody-positive children and adolescents have also been shown to exhibit reduced first phase insulin secretion during intravenous glucose tolerance tests (3). Using the oral minimal model, we were able to assess both early, readily-releasable insulin (${\phi }_d$), and the static component of insulin secretion which occurs later (${\phi }_s$) and reflects insulin granule translocation and maturation relative to the prevailing glucose concentration. We found impairment of the static component of β-cell responsiveness in Stage 1 T1D. In contrast to (${\phi }_s$), we did not find a difference in the dynamic component (${\phi }_d$) of β-cell responsiveness. It is possible that we would have found changes in this early phase in insulin secretion with a larger population sample (17, 33).

While the cases examined did not reach the threshold for dysglycemia as it is standardly defined (1), they did show higher glucose concentrations at 60 minutes after the oral glucose load than the control group, as well as greater glucose excursion (AUC) during the 3-hour test. At the same time, insulin excursion was similar between the 2 groups, which in the presence of higher glucose levels supports both increased insulin resistance in those with Stage 1 T1D (similar insulin levels did not have similar glucose-lowering impact) and β-cell dysfunction (absence of the compensatory increase in insulin secretion that would have been necessary to completely normalize glucose levels) (Fig. 1A and 1B). Higher 1-hour glucose has been previously associated with β-cell dysfunction in youth (26, 40) and might represent an early marker for T1D that is may be potentially overlooked if 1-hour glucose sampling is not performed during the standard OGTT (41, 42).

Insulin resistance was previously studied in a clinically similar group of autoantibody-positive children (7). That study characterized β-cell responsiveness and insulin resistance at baseline and then, after a period of 2.2 years, assessed whether these baseline factors predicted “progressors” (those who progressed to Stage 3 diabetes) and “nonprogressors” (those who did not). β-Cell glucose responsiveness was worse at baseline in progressors compared to nonprogressors, demonstrating that those destined to advance to clinical diabetes within about 2 years already had a greater defect in insulin secretion. In contrast, there was no difference in insulin sensitivity at baseline between the groups (7). Our methodology differed significantly from that study in that our autoantibody-positive children were compared with a healthy control population rather than with each other, allowing us to describe for the first time that abnormal increased insulin resistance is a feature of Stage 1 T1D, accompanying the decline of β-cell responsiveness. We speculate that the combined message of these 2 studies is that early autoantibody positivity is associated with abnormal insulin resistance, but insulin resistance does not appear to worsen during the progression from Stage 1 to Stage 3 diabetes.

Several factors could be postulated to explain the development of insulin resistance at this early stage of diabetes. The most likely causes may be islet inflammation and/or subclinical hyperglycemia. Systemic inflammation is well known to induce severe insulin resistance. The autoimmune attack on pancreatic β-cells that characterizes T1D is associated with smoldering, local islet inflammation (34), which could be the root of the insulin resistance that was measured in the current study. Modest hyperglycemia related to impaired insulin secretion, as displayed in our cohort, might also create an insulin resistant state. Hyperglycemia per se, not associated with diabetes, has been previously shown to reduce whole body and muscle glucose disposal in adults with T1D (35) as well as in healthy adults (36), reducing insulin sensitivity. Thus, mild hyperglycemia might play a dual role in the pathology of early diabetes, reducing both insulin sensitivity and β-cell responsiveness, the latter occurring via glucose toxicity.

Other factors known to be associated with insulin resistance do not appear to be relevant in the current cohort. Increased adiposity results in insulin resistance, and longitudinal analysis of the TrialNet cohort has shown that a higher BMI in childhood is associated with an increased risk of developing T1D (37). While there was a slightly wider range of BMI z-scores in cases (0.75 [–0.07, –1.25]) compared with controls (0.80 [–0.007, –0.90]) this was not statistically significant. Because obesity is associated with insulin resistance, we did a separate SI analysis where we excluded the individual with the BMI z-score of 1.25 and found that this did not change our finding of insulin resistance in the cases. Insulin resistance is a component of normal puberty, peaking at Tanner stage 3 in both sexes (38), which usually corresponds to an age of about 13 in boys and 12 in girls. While we did not assess Tanner stage in the current study, the average age and the age ranges were similar between groups (cases: 10.5 years with a range of 8-15 years, and controls: 11.5 years with a range of 10-15 years). Finally, some individuals might have a genetic background predisposing to insulin resistance (39, 40), but testing whether the 2 groups showed differences in gene variants associated with insulin resistance was beyond the scope of this study. Whatever the etiology, the finding of insulin resistance in Stage 1 T1D suggests that this may represent an early therapeutic target.

Lower fasting and postabsorptive insulin clearance was found in cases compared with normal controls in our cohort. This could have resulted from reduced liver insulin extraction as a compensatory mechanism for primary β-cell failure. Lower insulin clearance may also be related to the insulin-resistant state, as described in obese youth (24, 41-43), to avoid hyperglycemia. The insulin clearance measure adopted in this study tends to underestimate insulin clearance when insulin secretion increases, as described by the saturation curve during the early insulin response to an oral load (24, 25, 27). Thus, it is notable that we observed a lower insulin clearance in spite of the reduced secretion in Stage 1 T1D.

The major strengths of this study include (1) the use of a robust model-based assessment of the DI in this population, which provides a comprehensive metabolic phenotype with respect to insulin secretion and sensitivity; (2) the inclusion of healthy controls not related to the cases, which provides the first evidence of differential β-cell function and insulin sensitivity in individuals with early islet autoimmunity compared with their healthy peers; (3) evidence for lower insulin sensitivity and clearance in autoantibody-positive individuals compared with healthy peers in the absence of standardly defined dysglycemia; and (4) the method adopted to quantify insulin clearance was validated against euglycemic clamp data in adolescents (26, 44) and provides an estimate of endogenous insulin clearance under physiological conditions.

The major weakness of the study is the lack of longitudinal assessment. The SARS-COV2 pandemic limited our ability to perform 1-year follow-up OGTTs, and instead we relied on subject report. The diagnosis of Stage 3 T1D is not subtle, and thus we feel confident that we did not miss any cases of progression to clinical diabetes. Although we noted hepatic insulin resistance, we cannot dissect the role of liver vs peripheral insulin resistance in determining the observed metabolic phenotype in the absence of a tracer study. Insulin resistance is impacted by pubertal stage; although cases and controls were matched for age, sex, and gender, a limitation of this study is the lack of Tanner stage assessment. An additional potential source of bias is the inclusion of 3 different clinical sites, though the anthropometric characteristics of participants as well as the site procedures for the OGTT were uniform.

In summary, the new observation of lower insulin sensitivity and reduced insulin clearance in the presence of early islet autoimmunity holds the potential to shift the paradigm of an insulin secretion–centered model of T1D development to one that also includes a role for insulin resistance. These findings, in addition to the well-known presence of reduced insulin secretion, may inform novel investigational paths aimed at targeting both insulin secretion and sensitivity for future diabetes prevention strategies.

Glossary

Abbreviations

AUC

are under the curve

BMI

body mass index

CL

clearance

DI

disposition index

HIRI

hepatic insulin resistance index

IGT

impaired glucose tolerance

ISR

insulin secretion rate

OGTT

oral glucose tolerance test

SI

insulin sensitivity

T1D

type 1 diabetes

Additional Information

Disclosures. The authors have no conflict of interest to disclose regarding the present manuscript.

Data Availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.