Withdrawal of Medications Leads to Worsening of OGTT Parameters in Youth with Impaired Glucose Tolerance or Recently-Diagnosed Type 2 Diabetes

Abstract

Background.

The RISE Pediatric Medication Study compared strategies for preserving β-cell function, including a 9-month follow-up after treatment withdrawal to test treatment effect durability.

Objective.

Evaluate OGTT measures of glucose and β-cell response through 12 months of intervention and 9 months of medication washout.

Participants.

Youth (n=91) aged 10–19 years with BMI ≥85th percentile and impaired glucose tolerance (IGT) or recently diagnosed type 2 diabetes (T2D).

Methods.

A multi-center randomized clinical trial comparing insulin glargine for 3 months followed by metformin for 9 months (G→Met) or metformin alone (Met) for 12 months. We report within-group changes from baseline to end of medication intervention (M12), baseline to 9 months post-medication withdrawal (M21), and end of medication (M12) to M21. OGTT C-peptide index [CPI] paired with 1/fasting insulin evaluated β-cell response.

Results.

At M12, both treatments were associated with stable fasting glucose (G→Met baseline 6.0±0.1 vs. M12 5.9±0.2 mmol/L, p=0.62; Met baseline 6.1±0.2 vs. M12 6.0±0.2 mmol/L, p=0.73) and 2-hour glucose (G→Met baseline 10.2±0.4 vs. M12 9.3±0.5 mmol/L, p=0.03; Met baseline 10.2±0.4 vs. M12 10.6±0.6 mmol/L, p=0.88). Following medication withdrawal, fasting glucose worsened (G→Met M21 8.6±1.8, p=0.004; Met M21 7.8±0.7 mmol/L, p=0.003), as did 2-hour glucose (G→Met M21 13.2±1.4, p=0.002; Met M21 13.1±1.2 mmol/L, p=0.006), associated with declines in β-cell response.

Conclusions.

G→Met and Met were associated with stable glucose measures during 12 months of treatment in youth with IGT or recently diagnosed T2D. Glucose and β-cell response worsened post-medication withdrawal, suggesting treatment must be long-term or alternative treatments pursued.

Introduction

Interventions to address declining pancreatic β-cell function associated with worsening hyperglycemia in youth with impaired glucose tolerance (IGT) or recently-diagnosed type 2 diabetes (T2D) are needed. Studies in adults indicate that metformin and insulin each may be beneficial for preserving or improving β-cell function. Metformin improved β-cell function and reduced diabetes progression by 31% over 3 years in adults with IGT in the Diabetes Prevention Program [1, 2]. Short-term intensive insulin therapy improved β-cell function in adults with newly-diagnosed T2D, resulting in prolonged remission from requiring diabetes medication [3]. The RISE Pediatric Medication Study compared strategies for preservation of β-cell function in obese 10–19 year-old youth with IGT or recently diagnosed T2D: insulin glargine for 3 months titrated to target fasting glucose of 4.4–5.0 mmol/L (80–90 mg/dL) followed by metformin for 9 months (glargine followed by metformin) or metformin alone for 12 months. The primary hypothesis was that intervention with glargine followed by metformin would preserve or improve hyperglycemic clamp measures of β-cell function compared to metformin alone when measured at month 15 (M15), 3 months after medication withdrawal [4]. We previously reported that neither intervention led to improved hyperglycemic clamp measures of β-cell function and that HbA1c and BMI were not different at the end of the treatment period (month 12, M12) with no differences between treatment groups [5]. Hyperglycemic clamp measures of β-cell function, as well as HbA1c and BMI, had further worsened when measured at M15 [5].

The RISE Pediatric Medication Study also included, as pre-specified secondary outcomes, repeated OGTTs to evaluate glycemic outcomes and enterally-stimulated measures of β-cell response through 21 months post randomization (M21, 9 months following treatment withdrawal). The OGTT provides a reflection of the physiology of β-cell response under more physiologic circumstances (enterally-stimulated), in contrast to the hyperglycemic clamp which measures the response to a controlled intravenous glucose concentration. In addition to providing a clinically relevant direct assessment of glucose tolerance, OGTTs provide a physiologic stimulus of endogenous insulin and C-peptide secretion, augmented by the incretin response to enteral glucose, which is absent during a hyperglycemic clamp [6, 7]. The extended 9-month observation period following treatment withdrawal allowed for a clinically relevant window into what happens to β-cell function and glycemic control when medications are discontinued in youth with IGT or recent-onset T2D. This information is important to answer the question regarding whether temporary treatment during puberty would be effective for youth with IGT or recently diagnosed T2D, or whether long-term treatment would be needed. The objectives of the current analyses were therefore to evaluate OGTT-based 1) glucose outcomes and 2) measures of β-cell response through 12 months of intervention and 9 months of medication washout.

Methods

Study Protocol

The RISE Pediatric Medication Study was a multi-center, randomized, open-label clinical trial comparing 12-month interventions with 1) insulin glargine for 3 months followed by metformin for 9 months (glargine followed by metformin) versus 2) metformin alone for 12 months. The pediatric study, one of three protocols being performed by the RISE Consortium, enrolled youth with IGT or recently diagnosed T2D. The rationale and methods have been described previously in detail [4, 5]. The study protocol can be found at https://rise.bsc.gwu.edu/group/rise/protocols. Each center’s institutional review board (IRB) approved the protocol. Written parental or participant consent and child assent where age-appropriate were obtained prior to study procedures, consistent with the Helsinki Declaration and each center’s IRB guidelines.

Participants

A summary of participants enrolled and their rates of completion of study visits is shown in the CONSORT Diagram (Figure 1). Complete eligibility criteria for the study were previously published [5]. Briefly, 91 youth aged 10–19 years with BMI ≥85th percentile for age and sex but <50 kg/m², with IGT (60%) or recently diagnosed (<6 months duration) T2D (40%) who were negative for glutamic acid decarboxylase and IA-2 autoantibodies, Tanner stage ≥II breast development for females and testicular volume >3 mL for males were randomized. Participants were required to have fasting glucose ≥5 mmol/L (90 mg/dL), OGTT 2-hour glucose ≥7.8 mmol/L (140 mg/dL) and HbA1c ≤8.0% (64 mmol/mol) if they were drug naïve. Participants with T2D already taking metformin for less than 6 months had to have HbA1c ≤7.5% (58 mmol/mol) if on metformin for <3 months and ≤7.0% (53 mmol/mol) if on metformin for 3–6 months. Eligible participants underwent baseline evaluations followed by random 1:1 treatment assignment by study site, stratified by glycemic status.

Figure 1.

CONSORT Diagram. The glargine followed by metformin group is presented in green boxes; the metformin alone group is presented in brown boxes.

^$4 participants included in the glargine followed by metformin group at M21 decompensated between Months 15 and 21 and have imputed data for M21

*2 participants included in the metformin alone group decompensated between months 9 and 12, and 1 participant decompensated between months 15 and 21, and have imputed data for M15 and/or M21 as appropriate

†10 patients lost interest, 3 were non-compliant with run-in protocol, and 5 had poor iv access precluding studies requiring reliable iv access

Interventions

Complete details of the interventions have previously been published [4, 5]. Briefly, the glargine followed by metformin group received 3 months of insulin glargine titrated to achieve a fasting glucose of 4.4–5.0 mmol/L (80–90 mg/dL) based on daily self-monitored blood glucose. After 3 months, insulin glargine was stopped and metformin was initiated and titrated to 1000 mg twice daily (or the maximal tolerated dose) for the remainder of the 12-month intervention. The metformin alone group received metformin only, titrated to 1000 mg twice daily for 12 months.

Both groups had study medication withdrawn at 12 months and underwent study evaluations at M12 (on treatment) and at M15 and M21 (3 and 9 months after medication withdrawal, respectively) to quantify OGTT-measured β-cell function and glycemia. Anthropometrics, HbA1c testing and laboratory safety evaluations were performed every 3 months during the study. Participants were offered diabetes education and medical nutrition therapy sessions delivered via their local diabetes care and study teams according to American Diabetes Association Clinical Practice Recommendations. Throughout the entire study, participants were counseled to implement recommended levels of habitual physical activity (60 min/day), avoid sugary drinks, reduce concentrated sweets and follow recommendations from the dietary guidelines for Americans [8].

If protocol-specified HbA1c safety thresholds were exceeded, outcome measures were promptly performed if possible, and the participant was then withdrawn from the study and referred for clinical diabetes treatment [5]. Safety was monitored by an independent Data and Safety Monitoring Board.

Procedures and Calculations

A 3-hour 75-gram OGTT was performed to determine fasting and 2-hour glucose concentrations, as well as OGTT-derived β-cell response measures, at baseline, month 6 (M06), M12, M15, and M21 (Supplementary Figure S1) [4, 5, 9]. The OGTT-derived β-cell response measures included: (1) C-peptide index (CPI, nmol/mmol, ΔC-peptide_30–0/Δglucose_30–0), and (2) insulinogenic index (IGI, pmol/mmol, Δinsulin_30–0/Δglucose_30–0). Inverse fasting insulin (1/fasting insulin (pmol/L) × 10⁻³) was utilized as the estimate of whole-body insulin sensitivity in paired measures with β-cell response measures.

Laboratory assessments were performed in a central laboratory at the University of Washington [4, 5]. Glucose was measured using the glucose hexokinase method on a Roche c501 autoanalyzer (Roche Diagnostics Inc., Indianapolis, IN). C-peptide and insulin were measured by a two-site immuno-enzymometric assay performed on the TOSOH 2000 autoanalyzer (TOSOH Biosciences, Inc., South San Francisco, CA). Inter-assay CVs on quality control samples with low, medium, medium-high and high concentrations were ≤2.0% for glucose, ≤4.3% for C-peptide and ≤3.5% for insulin. HbA1c was measured on a TOSOH G8 analyzer, under Level 1 NGSP certification. The inter-assay CVs as measured on quality control samples with low and high HbA1c levels were 1.9% and 1.0%, respectively.

Statistics

The present analysis includes demographic, BMI, HbA1c and OGTT variables that were pre-specified secondary study outcomes, from baseline through the final M21 time-point (Supplementary Figure S1) [9]. For the OGTT measures of β-cell response, there were a total of three participants who had a CPI and 10 with an IGI that were negative at any time point. These negative values are may reflect temporal variability of the β-cell response as they have been reported to occur during diabetes development or may be physiologically impossible and represent measurement error, and so to be conservative, were excluded [10].

OGTT β-cell response measures paired with 1/fasting insulin at M06, M12, M15 and M21 visits, adjusted for baseline values, were compared between treatment groups. Joint models for β-cell response and 1/fasting insulin were fit simultaneously using Seemingly Unrelated Regression techniques, which provided a 2-DF chi-square test of the treatment arm difference in the joint values of β-cell response and 1/fasting insulin between treatment groups [11–13]. Due to the skewed nature of these variables, all were analyzed on a log scale and re-exponentiated for presentation.

To evaluate changes within each treatment arm over time, paired t-tests were used to compare changes in BMI, HbA1c and glucose from baseline or M12 to follow-up visits, and Hotelling’s T² method was used to simultaneously test changes in β-cell responses and 1/ fasting insulin [14]. T-tests were used to compare means across treatment groups at specific times.

Some participants met protocol-specified, pre-defined study elevations in HbA1c that required rescue diabetes therapy prior to M21 and thus had the worst β-cell function, but therefore also did not have OGTT data collected at M21. Two participants had HbA1c elevations requiring protocol-mandated study withdrawal prior to their M15 primary outcome visit and 5 additional participants had such elevations between the M15 and M21 visits. Unmeasured OGTT-derived outcome data after study withdrawal were imputed using the midpoint between zero and the lowest value among all participants at the respective visit. The final measured HbA1c and BMI were used for unmeasured M15 and/or M21 values.

Results

Participant characteristics at baseline have been previously published [5]; selected variables are shown in Table 1. Baseline anthropometric and metabolic characteristics did not differ between groups, except that the glargine followed by metformin group was older, but importantly did not differ in Tanner stage from the metformin alone group. The majority of the participants were Tanner stage V (62%). Medication and safety outcomes have been previously published [5]. Metformin adherence assessed by pill count was 88 ± 18% and did not differ by treatment group. Sixty-eight percent of participants were 80% compliant with insulin therapy at prescribed doses, as assessed by residual volume of distributed insulin.

Table 1.

Participant Characteristics by Intervention Group

	Glargine followed by Metformin (n=44)	Metformin alone (n=47)
Sex (n, % female)	27 (61%)	38 (81%)
Age (years)	14.9 ± 2.0	13.9 ± 2.1*
Tanner stage (% stage V)	3.0 (72.1%)	2.5 (53.2%)
Race/Ethnicity
White	13 (29.5%)	12 (25.5%)
Black	14 (31.8%)	9 (19.1%)
Hispanic	14 (31.8%)	20 (42.6%)
Other	3 (6.8%)	6 (12.8%)

^*p=0.03 for group difference in age at baseline

OGTT glucose, C-peptide and insulin curves at baseline, M12, M15, and M21 are shown in Figure 2 (upper panels for glargine followed by metformin, lower panels for metformin alone). OGTT glucose concentrations (red lines) were highest at M21 in both treatment groups and did not differ by treatment group. Data from all OGTT measures of C-peptide and insulin responses, and 1/fasting insulin, over time by treatment group are shown in Supplementary Figure 2 [9].

Figure 2.

OGTT glucose, C-peptide and insulin curves by treatment group at baseline (BAS, blue), M06 (purple), M12 (green), M15 (orange) and M21 (red). Data are means ± SEM.

Figure 3 presents the BMI, HbA1c, and OGTT glucose data for each treatment group across the entire study period. Figure 4 presents the trajectories of OGTT-based β-cell responses (CPI and IGI) in relation to 1/fasting insulin at each study time point. Since baseline values for these measures did not differ by group, the black box (0) represents the mean value for the whole cohort and the black line represents the baseline relationship in the whole cohort. Movement above the line represents improvement and below the line deterioration in β-cell response paired with 1/fasting insulin, whereas movement along the line represents mutually concordant changes in β-cell response and 1/fasting insulin.

Figure 3.

BMI (panel A), HbA1c (panel B), OGTT fasting glucose (panel C) and 2-hr glucose (panel D) over time, by treatment group (glargine followed by metformin in green; metformin alone in brown). Data are unadjusted means with 95% CI bars. Solid lines represent time on active treatment; dashed lines represent the period following medication withdrawal. Significant within treatment group changes from baseline are denoted with (*) and changes from M12 with (†).

Figure 4.

Relationship of OGTT-derived β-cell responses of C-peptide index (CPI) and insulinogenic index (IGI) paired with 1/fasting insulin by intervention group and study visit. Changes are shown from baseline to M06, M12, M15 and M21. The black line depicts the joint relationship between β-cell response and 1/fasting insulin at baseline for the full cohort, with the mean value indicated by the black box with a 0. The dotted lines to boxes show the trajectory of values from baseline to M06, and then to M12 (time of medication withdrawal), and then from M12 to M15 (3 months after medication withdrawal) and then to M21 (9 months after medication withdrawal) for the glargine followed by metformin group (green) and the metformin alone group (brown). The lines to points at M06, M12, M15 and M21 show the average changes between these time points. Values above the bolded baseline line indicate improved β-cell function and values below poorer β-cell function. The ellipses depict the 95% confidence bands around data for each box. Panel A) CPI paired with 1/fasting insulin. Panel B) IGI paired with 1/fasting insulin.

Change from Baseline to M12 (treatment period)

Glargine Followed by Metformin Group:

As shown in Figure 3 (green boxes), at M12, glargine followed by metformin was associated with no change in BMI (baseline 36.5±1.0 kg/m², M12 37.0±1.1 kg/m², p=0.75), no change in HbA1c (baseline 5.7±0.1% [39 mmol/mol], M12 5.8±0.1% [40 mmol/mol], p=0.75), no change in fasting glucose (baseline 6.0±0.1 mmol/L [107±2 mg/dL], M12 5.9±0.2 mmol/L [106±3 mg/dL], p=0.62), and modestly lower 2-hour glucose (baseline 10.2±0.6 mmol/L [183±7 mg/dL], M12 9.3±0.5 mmol/L [167±8 mg/dL], p=0.03). As shown in Figure 4, CPI paired with 1/fasting insulin was statistically increased from baseline (top panel, black box) to M12 (green box; p=0.002). IGI paired with 1/fasting insulin was not statistically different from baseline (Figure 4, bottom panel, black box) to M12 (green box, p=0.12).

Metformin Alone Group:

As seen in Figure 3 (brown boxes), at M12, metformin alone was associated with no change in BMI (baseline 36.9±0.9 kg/m², M12 36.5±1.0 kg/m², p=0.06), no change in HbA1c (baseline 5.7±0.1% [39 mmol/mol], M12 5.8±0.1% [40 mmol/mol], p=0.31), and no change in either fasting glucose (baseline 6.1±0.2 mmol/L [109±3 mg/dL], M12 6.0±0.2 mmol/L [108±4 mg/dL], p=0.73) or 2-hr glucose (baseline 10.2±0.4 mmol/L [184±7 mg/dL], M12 10.6±0.6 mmol/L [190±11 mg/dL], p=0.88). CPI and IGI paired with 1/fasting insulin were each statistically decreased from baseline (Figure 4, black boxes) to M12 (brown boxes; p=0.013 for CPI, p<0.001 for IGI).

There were no significant differences at M12 comparing paired β-cell response (CPI or IGI) and 1/fasting insulin between the two treatment groups.

Changes through M21 (9-months of washout)

Glargine Followed by Metformin Group:

After 9 months of medication withdrawal, BMI increased (M12 37.0±1.1 kg/m², M21 39.0±1.1 kg/m², p=0.0002), HbA1c increased (M12 5.8±0.1% [40 mmol/mol], M21 6.7±0.3% [50 mmol/mol], p=0.003), fasting glucose increased (M12 5.9±0.2 mmol/L [106±3 mg/dL], M21 8.6±1.9 mmol/L [155±17 mg/dL], p=0.004), and 2-hour glucose increased (M12 9.3±0.5 mmol/L [167±8 mg/dL], M21 13.2±1.4 mmol/L [238±25 mg/dL], p=0.002) as shown in Figure 3 (green boxes). CPI paired with 1/fasting insulin decreased from M12 to M21 (Figure 4, top panel, green boxes; p=0.006); and was significantly worse at M21 than baseline (p=0.013). While IGI paired with 1/fasting insulin was not statistically decreased from M12 to M21 (p=0.06), it was significantly lower at M21 than baseline (Figure 4, bottom panel, green boxes; p=0.007).

Metformin Alone Group:

After 9 months of medication withdrawal, BMI increased (M12 36.5±1.0 kg/m², M21 38.3±1.0 kg/m², p=0.0003), HbA1c increased (M12 5.8±0.1% [40 mmol/mol], M21 6.3±0.3% [45 mmol/mol], p=0.01), fasting glucose increased (M12 6.0±0.2 mmol/L [108±4 mg/dL], M21 7.8±0.7 mmol/L [141±13 mg/dL], p=0.003), and 2-hour glucose increased (M12 10.6±0.6 mmol/L [190±11 mg/dL], M21 13.1±1.2 mmol/L [236±21 mg/dL], p=0.006) as shown in Figure 3 (brown boxes). CPI paired with 1/fasting insulin decreased from M12 to M21 (Figure 4, top panel, brown boxes; p=0.04), but was not statistically different from baseline (p=0.19). IGI paired with 1/fasting insulin was not statistically different from M12 to M21 (p=0.15), from baseline to M21 (p=0.77), or from the baseline regression line (the confidence bands overlap the baseline regression line), possibly due to excluded negative IGI values and decreased sample size at this timepoint (Figure 4, bottom panel, brown boxes).

There were no significant differences at M21 comparing paired β-cell response (CPI or IGI) and 1/fasting insulin between the two treatment groups.

We also analyzed data from the group of participants with IGT separately to determine if results were different between participants with IGT or T2D at baseline. There were 26 participants with IGT in the glargine followed by metformin group (59%) and 28 in the metformin alone group (60%). The patterns of OGTT-measures of β-cell function were similar in the full cohort as in the cohorts defined by baseline IGT or diabetes (data not shown).

Discussion

The RISE Pediatric Medication Study compared strategies for preservation of β-cell function in obese 10–19 year-old youth with IGT or recently diagnosed T2D: insulin glargine for 3 months followed by metformin for 9 months (glargine followed by metformin), or metformin alone for 12 months. Here we describe the progression of OGTT-derived measures of glycemia and β-cell response during the entire 21-month RISE Pediatric Medication Study, reflecting 12 months on treatment and 9 months after treatment withdrawal in youth with IGT or recently diagnosed T2D. As medication withdrawal often happens in clinical practice and occurs for multiple reasons, and an increasingly frequently-asked question was whether temporary treatment in adolescence early in the disease process could have longer-lasting impacts, data during this withdrawal period are clinically relevant and important to study. Despite rising rates of obesity, IGT and T2D in youth, there are virtually no published data in the pediatric population on what happens to β-cell function after medication withdrawal in either youth with IGT or recently diagnosed T2D, making these study results both timely and novel. Ultimately, there was no signal for improving measures of glycemia or β-cell response to OGTT, nor differences between the two treatments. However, at M12, both treatments were in fact associated with stable BMI, HbA1c, and fasting and 2-hour glucose values, rather than continued decline. Despite the stability during treatment, following medication withdrawal, both groups had signficiant worsening of BMI, HbA1c, and fasting and 2-hour glucose that were associated with declines in OGTT-derived β-cell response measures. Therefore any benefits while on treatment did not appear to be sustained after treatment withdrawal.

We previously reported that hyperglycemic clamp measures of β-cell function worsened compared to baseline at M12 and M15 in both treatment arms, without differences between treatment groups [5]. The RISE study is unique in that we have data from both the OGTT and hyperglycemic clamp methods, and both add to the pediatric literature and to our understanding of the pathophysiology of progressive dysglycemia in youth with IGT and recently diagnosed T2D. While the hyperglycemic clamp measures the response to a controlled intravenous glucose concentration, the OGTT provides a reflection of the physiology of enterally-stimulated β-cell response and measures of glycemia. We prospectively planned to evaluate OGTT measures of β-cell response as secondary outcomes because of the possibility that OGTT measures, reflecting the physiology inherent to enteral glucose delivery versus intravenous glucose delivery, could differ from clamp measures [6, 7]. The OGTT not only allows for a more physiologic measurement of a hormonally integrated process that includes the incretin-driven magnification of the insulin response, it also produces clinically interpretable glucose data [15]. Moreover, an OGTT can be repeated more frequently than a hyperglycemic clamp, as the clamp measures introduce significant additional resource, blood volume and participant burden. The OGTT-derived indices of β-cell response were C-peptide and insulin responses paired with 1/fasting insulin. We recognize that 1/fasting insulin is an estimate of whole-body insulin sensitivity and not as accurate as clamp-derived measures of insulin sensitivity. However, the surrogate estimate 1/fasting insulin has been demonstrated to be strongly correlated with insulin sensitivity measured with the hyperinsulinemic-euglycemic clamp in obese youth [16]. Moreover, clamp- and OGTT-derived estimates of β-cell response relative to insulin sensitivity as estimated by 1/fasting insulin are significantly correlated with each other both in youth with IGT and in youth with controlled diabetes [17]. These data justify our choice of 1/fasting insulin to assess β-cell response relative to insulin sensitivity in intervention studies in high risk youth.

From baseline to M12 and from M12 to M21, in the glargine followed by metformin group, changes in CPI paired with 1/fasting insulin were significant whereas changes in IGI paired with 1/fasting insulin were not. Similarly, from M12 to M21, in the metformin alone group, the change in CPI paired with 1/fasting insulin was significant whereas the change in IGI paired with 1/fasting insulin was not. We do not interpret the results to be in opposition, as the two measures are showing the same pattern of change. Physiologic differences between the measures may have had an impact on the results, as insulin concentrations reflect both secretion and clearance, while C-peptide concentrations reflect secretion only. Also, the power to detect changes in IGI was lower, as more IGI values than CPI values were excluded due to negative results.

The Remission Studies Evaluating Type 2 Diabetes - Intermittent Insulin Therapy Pilot (RESET IT Pilot) was a randomized controlled trial in adults with T2D for ≤5 years (median HbA1c 6.3%) that also evaluated insulin followed by metformin to preserve beta-cell function [18]. Intensive insulin therapy for 3 weeks titrated to target fasting glucose level between 4.0 and 6.0 mmol/L (72–108 mg/dL) and 2-hour postprandial glucose level <8 mmol/L (<144 mg/dL) was utilized in all participants, followed by maintenance therapy with either metformin alone or intermittent insulin therapy every 3 months [18]. After the initial intensive insulin therapy, maintenance therapy with metformin was superior to intermittent insulin therapy for preserving β-cell function and glycemic control across 2 years of follow-up on treatment [18]. Similarly, in the youth with IGT or recently diagnosed T2D that we studied, glargine followed by metformin was associated with transient improvement or stabilization of OGTT measures of β-cell function. However, when metformin was withdrawn, deterioration in β-cell function was accompanied by worsening glycemia. Our study differed in that we utilized the more practical and clinically applicable approach of 3 months of basal insulin alone to target fasting blood glucose values of 4.4 – 5.0 mmol/L (80–90 mg/dL), and in that we also included participants with IGT. Importantly, we also withdrew metformin at M12 to evaluate whether or not changes in β-cell function would be maintained. The observed lack of any durable effect of the either treatment may be related to a failure to reduce glucose toxicity and/or continued weight gain. The lack of effect of the more aggressive therapy with glargine followed by metformin to reduce glucose toxicity or of either treatment to substantially improve insulin sensitivity was associated with progressive dysglycemia.

In the Diabetes Prevention Program, which studied adults with IGT, treatment with metformin was associated with enhanced OGTT measures of β-cell function when measured at 12 months of treatment exposure [1]. In ADOPT (A Diabetes Outcome Progression Trial), adults with T2D who were drug-naïve and randomized to treatment with metformin alone demonstrated an improvement in OGTT measures of β-cell function [19]. These measures declined slightly over the four years participants were receiving metformin, but still demonstrated a beneficial effect at the end of this more prolonged period of follow up [19]. Our results in youth who were treated with metformin alone indicate worsening OGTT measures of β-cell response at M12. These results suggest that youth are responding differently to metformin treatment compared to adults of similar glycemic status. The reason for these differences between youth and adults are incompletely understood, but may include differences in insulin sensitivity and secretion [20], developmental stage/puberty, genetics and factors yet to be adequately studied.

The Treatment of Diabetes in Adolescents and Youth (TODAY) Study treated 10–17 year-old youth with T2D of less than 2 years duration with metformin alone, metformin plus rosiglitazone, or metformin with an intensive lifestyle program [21]. Participants treated with metformin alone had a treatment failure rate of 51.7%. Of those who met criteria for treatment failure, the median time to failure was 11.8 months. TODAY youth who experienced therapeutic failure, regardless of intervention arm, had significantly decreased insulin responses to enteral glucose at baseline and over time, with no baseline differences or change over time in insulin sensitivity (1/fasting insulin) [22]. Our results reflect similar OGTT β-cell response outcomes despite the fact that the RISE participants had either IGT or more recently diagnosed T2D (duration less than 6 months prior to study entry) than in TODAY.

Our study’s strengths include the randomized study design with longitudinal, high-quality repeated measures of β-cell responses to both oral and intravenous stimuli and standardization of procedures and laboratory measures across centers. We also explicitly evaluated β-cell function both during and following treatment withdrawal to capture any persistence of treatment effects, but also to reveal possible changes in underlying β-cell biology following withdrawal of medications. The study population was diverse due to the multi-centered study design. However, the sample size did not allow for meaningful estimates of the differences in OGTT measurements according to race/ethnicity or Tanner stage. It may be considered a limitation that we did not have clamp data for the entire 21 month study. Due to resource and participant burden of completing an additional clamp study at the M21 time point, we planned for these measures to be obtained via OGTT. Another weakness is incomplete achievement of the glycemic goal during insulin treatment (although there were significant reductions in glucose on average and 50% achieved the fasting glucose target). Although hypoglycemia was rare (~0.04% of daily self-monitored blood glucose values were <70 mg/dL), some participants experienced mild hypoglycemia despite not meeting the fasting self-monitoring blood glucose targets. Others simply did not meet the glycemic target despite using a standard protocol for insulin dose adjustments and increasing insulin doses through the entire 3 months of treatment with insulin [5]. We did not lose participants due to side-effects of medications. As previously reported [5], gastrointestinal discomfort was reported in ~35% of study participants regardless of whether they were on treatment or after treatment was withdrawn per protcol. The occurrence of treatment failure and metabolic decompensation would have affected our power to detect OGTT-based differences between later time points after medications were withdrawn and cause important data to be missing. Therefore, in these cases, OGTT-derived outcome data were imputed using the midpoint between zero and the lowest value across the full cohort to retain the important impact of the participants with the worst beta-cell function at the end of the study. Finally, there were no data collected on adherence to standard lifestyle recommendations, which may have impacted study outcomes.

In summary, treatment with either glargine followed by metformin or metformin alone, was associated with stable OGTT glucose measures during 12 months of treatment in youth with IGT or recently diagnosed T2D in the RISE pediatric cohort. However, prolonged medication withdrawal in both groups was associated with progressive worsening of β-cell response, as well as worsening of OGTT measures of glycemia, HbA1c and BMI. Studies are now needed to test additional strategies for preserving favorable changes in β-cell function over time.

Supplementary Material

1

Supplementary Figure S1 The RISE Pediatric Medication Study phases and key time points.

Supplementary Figure S2. OGTT-based β-cell responses and insulin sensitivity (1/fasting insulin) over time. The box and whisker plots represent the lowest and highest quartiles (box), the median (band inside the box), and the whiskers extend to 1.5*interquartile range. Results are shown for baseline (blue), M06 (purple), M12 (green), M15 (orange), and M21 (red) by treatment assignment (glargine followed by metformin and metformin alone).

Abbreviations:

CPI

C-peptide index

IGI

insulinogenic index

IGT

impaired glucose tolerance

M06

month 6

M12

month 12

M15

month 15

M21

month 21

RISE Study

Restoring Insulin Secretion Study