Trajectories of changes in glucose tolerance in a multiethnic cohort of obese youths: An observational prospective analysis

Abstract

Background.

Type 2 diabetes (T2D) is preceded by a pre-diabetic stage of impairment of glucose tolerance (IGT) that affects 10–23% of obese youth and is expected to double over the next decade. The natural history of IGT and the determinants of beta cell dynamic response have never been longitudinally investigated in youth. Herein, we evaluated the clinical and metabolic determinants of longitudinal glucose tolerance changes and beta cell function in a multiethnic cohort of obese youth.

Methods.

We prospectively followed a multi-ethnic cohort of 526 obese youths (non-Hispanic-White (NHW) 206; non-Hispanic-Black (NHB) 163; Hispanic 157), aged 12•7 years (10•6–14•2), with baseline normal (n=364) or impaired glucose tolerance (n=162). All subjects underwent a 3hr-oral-glucose-tolerance-test at baseline and after 2-years to estimate insulin secretion (oral disposition index, oDI) in the context of body insulin sensitivity. Participants underwent dietary evaluation every 5–6 months in the absence of structured lifestyle or pharmacological intervention.

Results.

In the presence of IGT, 65% reverted to normal glucose tolerance (NGT), 27% persisted as IGT and 8% progressed to T2D. The reversion to NGT was featured by a ~4-fold increase in oDI (p<0•001) that persisted after returning to NGT, as compared to subjects who remained NGT across the study. In contrast, a decrease in insulin-secretion was observed in subjects who remained IGT (p<0•001) or progressed to T2D (p<0•001). NHW ethnicity conferred an odds for reversion of 5 times higher than NHB (p=0•001) with a ~2 times greater yearly increase of oDI (p=0.045).

Conclusions.

The condition of IGT is highly reversible in obese youth, with ethnicity acting as the main clinical modifier of the dynamic beta cell response to pre-diabetic hyperglycemia and thus determining the reversibility of IGT or its persistence. Therapeutic interventions for IGT should target the specific mechanisms underpinning glucose tolerance changes in high risk-groups.

Introduction

The growing incidence of type 2 diabetes (T2D) in youth has propelled this condition to become a new metabolic disease in pediatrics.(1, 2) Recently, the SEARCH for Diabetes in Youth study indicated an alarming increase in T2D incidence from 2002 to 2012 in adolescents aged 10–19 years, particularly among Non-Hispanic Black (NHB) and Hispanic ethnic groups.(1–7) Notably, NHB obese adolescents with T2D have been found to be less responsive to the current treatment options for diabetes than Non-Hispanic Whites (NHW) and Hispanics.(8–10)

T2D is preceded by a pre-diabetic stage known as impaired glucose tolerance (IGT)(11) with a prevalence in the US pediatric population of 10–23%.(2) While the transition from prediabetes to T2D in adults is usually a gradual phenomenon that occurs over 5–10 years,(12) this deterioration in youth seems to be accelerated.(13, 14) However, the clinical predictors of the reversibility of prediabetes in adolescents and the dynamic changes in beta cell function and insulin sensitivity, underpinning shifts in glucose tolerance in obese youth, have never been assessed longitudinally.(8, 15, 16)

To gain insights into the natural history of glucose tolerance changes in obese youths and to understand the physiologic basis of ethnic disparities in susceptibility to the disease, we conducted an observational prospective study in a large multi-ethnic cohort of obese adolescents with either normal (NGT) or impaired glucose tolerance (IGT), assessing their clinical and metabolic hallmarks over time, in the absence of structured therapeutic interventions.

Research Design and Methods

We designed an observational two-points prospective study in obese youth, recruited from the large multi-ethnic cohort of adolescents at the Yale Pediatric Obesity Clinic from 2010 to 2016 (Yale Pathophysiology of Youth Onset prediabetes/T2D(PYOD Study), NCT01967849).(17) The inclusion criteria for the study were BMI ≥85^th percentile for age and sex, and age between 8–21 years at the screening visit. Subjects using medications affecting glucose metabolism, diagnosed with syndromic obesity or participating in other clinical trials were excluded from the study. Additionally, participants who tested positive for at least one of the autoantibodies associated with type 1 diabetes(18) were excluded.

Ethnicity was assessed by a self-reporting questionnaire, with NHB, NHW, Hispanic or others as the major categories.(9) Pubertal stage and anthropometric measures were recorded at each clinical encounter. The first and the last available oral glucose tolerance test (OGTT) for each subject were used for the analysis.

As part of the standard of care of the Yale Pediatric Obesity clinic, all the participants underwent a dietary evaluation every 5–6 month. Our first line intervention consisted of discontinuing sugar-laden drinks (juice, soda, lemonade, ice team) along with advices aimed to decrease screen time and promote physical activity. During the following visits, we advices all the participants to switch high-sugar to low-sugar cereal, to eliminate the sugar-sweetened beverages and encourage the consumption of home-made fresh lunch at school instead of hot meals, preferring low-fat protein sandwiches, vegetables, and fruit, and switched whole milk to 1% or fat-free milk. We do not start the children on a diet, but work together on making and achieving sustainable lifestyle changes as their long-term goal.(19)

The study protocol was approved by the Human Investigations Committee of the Yale School of Medicine. Participants provided assent and parents provided written informed consent to participate in the study.

Oral glucose tolerance test.

Prior to the OGTT, all subjects followed a weight-maintenance diet consisting of at least 250 g of carbohydrates per day for seven days before the study and were instructed to avoid strenuous physical activity. Subjects were studied in the Yale Center for Clinical Investigation (YCCI) at 8 a.m. after a 12-hour overnight fast. After the local application of a topical anesthetic cream containing (Emla, Astra Zeneca, Wilmington, Del.), one antecubital intravenous catheter was inserted for blood sampling, and its patency was maintained by slow infusion of normal saline. Each subject then rested watching a videotape throughout the test. Two baseline samples were then obtained for measurements of plasma glucose, insulin, and c-peptide. Thereafter, flavored glucose in a dose of 1.75 g per kilogram of body weight (up to a maximum of 75 g) was given orally, and blood samples were obtained every 30 minutes for 180 minutes for the measurement of plasma insulin and and c-peptide. Glucose was measured at the bedside using the YSI2700-STAT-Analyzer (Yellow Springs Instruments, Yellow Springs, OH). Plasma insulin was measured by radioimmunoassay (Linco, St. Charles, MO) that has <1% cross-reactivity with c-peptide and proinsulin. Plasma c-peptide levels were determined by ELISA using ALPCO-Immunoassays (Salem, NH) with a 3.87% intra-assay variability.

In accordance with the American Diabetes Association,(20) IGT was defined as a 2-hr plasma glucose level between 140 and 199 mg/dl; T2D as a fasting glucose level of 126 mg per deciliter or higher or a two-hour plasma glucose level of more than 200 mg/dl confirmed in two OGTTs. The diagnosis of T2D was confirmed by a second OGTT.(20, 21)

OGTT-derived measures of insulin secretion and sensitivity.

The oral disposition index (oDI) was adopted as a surrogate measure of β-cell function in the context of insulin sensitivity and calculated as the product of insulin sensitivity index (1/fasting insulin [1/I_F, ISI]) and β−cell function (c-peptide index [CPI] as the ratio of incremental c-peptide and glucose over the first 30 minutes [ΔCpep₃₀/ΔG₃₀]) as previously described.(9) The oDI was proved to strongly correlate with clamp-derived disposition index (DI), with a predictive power for clamp-derived DI comparable to the 2-h glucose. (22)

The hepatic insulin clearance (HIC) was determined as a ratio of the incremental areas under the curve (AUC) of OGTT (AUC_{C-peptide 0–180 min}/AUC_{insulin 0–180 min}), as previously described (23), adopting a surrogate measure proved to be strongly correlated with metabolic insulin clearance determined by hyperinsulinemic-euglycemic clamp.

Study outcomes.

The primary outcome was the “reversion” from IGT to NGT at follow-up. Secondary outcomes included: adverse “progression” of glucose tolerance to IGT among those with NGT at baseline, percentage change from baseline in insulin secretion (oDI), defined as percentage ratio of difference in oDI between last and first OGTT and the baseline oDI.

Statistical analysis.

Using a prior study, a sample of at least 60 NHW, 40 NHB, and 50 H participants with IGT would have 80% power at alpha 0•025 (Bonferroni correction for two primary comparisons: NHW vs. either NHB or H) to detect a minimum difference of 28–29% in the percentage of IGT to NGT reversion between the ethnic group comparisons of interest (primary outcome), if the reversion rate among NHW was in the range of 50–80%.(4)

All the analyses were stratified by baseline NGT/IGT grouping variable. Kruskal-Wallis test, followed by post-hoc pair-wise Mann-Whitney test, were used to compare continuous variables, and categorical variables were compared using the Chi-square test. Data were summarized in tables using median (25^th percentile, 75^th percentile) for continuous variables and count (percent, %) for categorical variables.

Adjusted analyses of the effect of ethnicity on the binary outcomes were performed using multivariable logistic regression modeling. The following adjustment variables were selected a priori based on the published literature(24): sex, age at baseline (years), baseline BMI and its change from baseline to the follow-up point, family history of T2D, pubertal stage at baseline (Tanner stage), fasting and 2-hours glucose at baseline, and follow-up time (years).(4, 24, 25) Statistical significance for the effect of ethnicity and the adjustment variables was established with alpha=0•05. Results were summarized as Odds Ratios (ORs) and 95% Confidence Intervals (95%CI). Multivariable linear regression was used to identify predictors of change in the beta cell function, using the same predictors used for the multivariable logistic regression, plus the baseline disposition index (oDI). Results were summarized as differences in the means (standard errors) of levels of predictors.

Analyses were performed using STATA.13 software (StataCorp. 2013. Stata Statistical Software: Release 13. College Station, TX: StataCorp LP). For a detailed statistical analysis, see Supplemental appendix.

Results.

Baseline characteristics.

(Figure 1) We screened 1620 subjects for study eligibility and analyzed data from 526 participants who completed at least two OGTTs over a period of 2•9 years [2•74–3•09]. At enrollment, 69% (n=364) of the eligible subjects had NGT and 31% (n=162) had IGT. Duration of follow-up was similar among the ethnic groups in both strata of initial glucose impairment, as well as the number of follow-up visits attended. (Table 1) Twenty-three out of the 1620 screened subjects were excluded because treated with metformin, as consequence of a previous diagnosis of IGT. One hundreds-fifty-six out of 682 eligible participants did not return for the follow-up OGTT. Their baseline characteristics did not differ from the follow-up cohort with respect to the anthropometric and metabolic features. (Table S1)

Figure 1.

Flow diagram of participant inclusion and classification by glucose tolerance at baseline and ethnic background. (NHW, non-Hispanic White; NHB, non-Hispanic Black; H, Hispanic; NGT, normal glucose tolerance; IGT, impaired glucose tolerance; T2D, type 2 diabetes). Data are expressed as numbers (percentage). NHW has been adopted as the reference group for paired comparison.

Table 1.

Baseline clinical and metabolic characteristics by baseline glucose tolerance status and ethnic background.

	NGT (n=364)					IGT (n=162)
Ethnicity (n, %)	NHW 140 (38·46)	NHB 121 (33·24)	P (NHB vs NHW)	H 103 (28·36)	P (H vs NHW)	NHW 66 (40·74)	NHB 42 (25·93)	P (NHB vs NHW)	H 54 (33·33)	P (H vs NHW)
Sex (male)	51 (36·43)	36 (29·75)	0·134	52 (50·48)	0·036	16 (24·2)	10 (23·8)	0·481	22 (40·7)	0·025
Family History T2D	34 (24·28)	44 (36·36)	0·016	29 (28·15)	0·257	15 (22·7)	13 (30·9)	0·188	25 (46·2)	0·032
Tanner Stage (n, %) I–II III–IV	39 (28) 101 (72)	46 (38) 75 (62)	0·385	33 (32) 70 (68)	0·097	19 (29) 47 (71)	13 (31) 29 (69)	0·227	19 (35) 34 (65)	0·444
Age (y)	13·05 (10·5, 14·65)	12·00 (10·5, 13·9)	0·065	11·9 (10, 14·2)	0·035	13·75 (11·6–15·3)	12·85 (11·1–14·8)	0·085	12·3 (10·7, 13·7)	0·070
Follow-up (y)	2·25 (1·7, 3·8)	2·6 (1·6, 4·1)	0·239	2·7 (1·8, 4·1)	0·078	1·9 (1·2–3·1)	2·05 (1·1–3·3)	0·403	2·35 (1·20, 3·70)	0·095
Follow-up visits (n)	6·4 (3·6, 9·5)	6.5 (4·0, 8·2)	0·667	6·8 (3·8, 10·2)	0·938	5·8 (2·4, 7·8)	5.3 (3·6, 7·8)	0·271	5·6 (3·1, 9·6)	0.619
BMI (kg/m2)	32·84 (27·53, 36·96)	33·40 (30·11, 39·10)	0·025	33·20 (27·71, 36·95)	0·427	33·42 (29·15, 38·80)	36·83 (33·69, 40·82)	0·004	33·18 (27·58, 37·72)	0·368
HbA1c % [mmol/mol]	5·3 (5·1–5·5) 34 (32–37)	5·5 (5·3–5·7) 37 (34–39)	<0·001	5·3 (5·2–5·5) 34 (33–37)	0·154	5·4 (5·2–5·6) 36 (33–38)	5·7 (5·5–6·1) 39 (37–43)	<0·001	5·6 (5·4–5·8) 38 (36–40)	<0·001
Fasting glucose (mg/dl)	91 (86, 96)	91 (86·75, 96·50)	0·452	91(87, 95)	0·356	92·50 (87, 98)	95·25 (89, 102)	0·122	99·5 (93, 106)	<0·001
2h-glucose (mg/dl)	116 (106, 125)	115 (103·5, 127)	0·416	115 (102, 125)	0·453	149·50 (145, 160)	151 (145, 163)	0·227	146·5 (142, 158)	0·048
Fasting Insulin (μU/mL)	29 (20, 38·75)	29 (24, 40·5)	0·080	30·5(19·5, 43)	0·218	34·25 (24, 48)	46 (29, 59·5)	0·007	39·75 (28·99, 52·99)	0·122
Fasting c-peptide (pmol/L)	1050 (820, 1390)	960 (681·25, 1173·75)	0·008	1150 (910, 1465)	0·118	1310 (920, 1610)	1215 (1060, 1473)	0·483	1300 (995, 1655)	0·261
Disposition index (oDI)	1·72 (1·24, 2·67)	1·88 (1·33, 2,62)	0·241	1·73 (1·13, 2·87)	0·473	1·12 (0·85, 1·65)	1·00 (0·71, 1·45)	0·217	1·17 (0·85, 1·76)	0·296
C-peptide index (CPI)	52·17 (36·75, 64·62)	54·17 (39·47, 79·52)	0·053	56·82 (38,49, 76·35)	0·151	37·21 (27·65, 53·68)	49·10 (35·64, 65·92)	0·015	46·82 (33·84, 63·80)	0·040
ISI (1/Fasting Insulin)	0·03 (0·02, 0·08)	0·03 (0·02, 0·04)	0·080	0·03 (002, 0·05)	0·218	0·03 (0·02, 0·04)	0·02 (0·02, 0·03)	0·007	0·02 (0·02, 0·03)	0·123
Hepatic Insulin Clearance (AUC C-PEP0–180/INS0–180)	3·24 (2·53, 4·36)	2·51 (2·01, 3·28)	<0·001	2·97 (2·17, 4·18)	0·095	2·80 (2·06, 3·68)	2·01 (1·28, 2·81)	0·002	2·62 (1·66, 3·11)	0·287

The distribution of the three ethnic groups (NHW, NHB and Hispanic) did not differ between the NGT and IGT (p=0•214) at baseline. In the NGT group, NHB as compared with NHW, had significantly higher proportion with family history of T2D, higher BMI and HbA_1C, and significantly lower fasting c-peptide and hepatic insulin clearance (all p<0•05). In spite of comparable fasting and 2h-glucose, NHB exhibited a higher HbA_1c in both NGT and IGT group than NHW. (Table 1) Hispanic obese adolescents in the NGT group, as compared to NHW, were more likely to be male (p=0•036) and younger (p=0•035) as compared with their NHW counterparts. In the IGT group, NHB had significantly higher BMI, HbA_1C, fasting insulin and c-peptide index, and subsequently significantly lower ISI and hepatic insulin clearance, as compared with NHW youth (all p<0•05). Hispanic participants with IGT at baseline were more likely than NHW to be male, to report family history of T2D, to have higher HbA1c, fasting glucose and c-peptide index (all p<0•05). (Table 1)

Dynamics of glucose tolerance over time.

At follow-up, only 14% of those with NGT at baseline progressed to IGT (Progressor, NGT➔IGT), while 65% of those with IGT at baseline reverted to NGT (Reverter, IGT➔NGT), 27% remained IGT (Non-Progressor, IGT➔IGT) and 8% progressed to T2D (Progressor, IGT➔T2D) (Figure 1)

Demographic and baseline anthropometric characteristics of subjects who maintained NGT did not significantly differ from those who progressed to IGT at follow up, including the duration of follow-up (all p=NS). Participants who reverted from IGT to NGT were similar to those who remained IGT at follow up with respect to gender, family history of T2D, age, Tanner stage and BMI at baseline, and duration of follow up (all p=NS). The small group (n=13) who progressed from IGT to T2D had the longest duration of follow-up (p<0•001), and more advanced pubertal development (p<0•001) as compared with subjects who maintained IGT. (Table S2)

Determinants of glucose tolerance change over time.

After adjusting for family history for T2D, gender, pubertal stage, BMI and its change over time, age and follow-up duration, fasting and 2h-glucose at baseline, NHW ethnic background conferred an odds of reverting from IGT to NGT five times greater than NHB [5•06 (1•86, 13•76) p=0•001]. Ethnicity was not a statistically significant predictor of glycemic tolerance at follow up among youth with NGT at baseline. The 2h-glucose was the only other variable significantly associated with probability of progressing from NGT to IGT and with probability of reverting from IGT to NGT (p 0•009 and 0•047 respectively). (Table S3)

Dynamic changes in beta cell function.

The progression from NGT to IGT was featured by a decrease in oDI, from 1•57[1•27, 2•17] to 0•86[0•67, 1•32] (p=0•025) (Figure 2, panel A). Conversely, reverters (IGT➔NGT) revealed a dynamic hyper-responsiveness to impaired glucose tolerance featured by a 4-fold increase in oDI (from 0•94[0•68, 1•35] to 3•90 [2•58, 6•08], p <0•001) as compared with the decline in the insulin secretion among those with persistent IGT (1•73 [1•34, 2•64] to 1•31 [1•01, 1•85] p=0•020) or progressed to T2D (from 1•66 [1•12, 2•29] to 0•20 [0•12, 0•58] p=0•002). (Figure 2, panel A)

Figure 2.

(A) Baseline and follow up disposition index (oDI) by glucose tolerance change. (B) Rate of Change per year in insulin-derived indices by glucose tolerance change, and (C) by the ethnic background. oDI, disposition index; CPI, c-peptide index (pmol/L per mg/dL); ISI, insulin sensitivity index. (x10ml/μU)

Notably, the baseline oDI in participants who reverted to NGT was significantly lower (p<0•001) than the group that remained IGT at follow up (IGT➔IGT). This was consistent with the former also having lower C-peptide index (CPI) (p=0•001) as compared to the latter group, in the absence of significant differences in the baseline insulin sensitivity index (p=0•181). (Figure 2, panel B and Table S2)

The yearly rate of change of oDI did not differ between the group who remained NGT and the group who progressed to IGT (Figure 2, panel B). Reverters also exhibited a median yearly increase in oDI of 35%, whereas participants who maintained IGT did not show significant yearly change in oDI (−6.7% per year). The small number of progressors to T2D experienced a significant decline in insulin secretion (−11% per year). (Figure 2, panel B) The beta-cell hyper-responsiveness persisted even after the observed reversion from IGT to NGT, as the follow-up oDI among the reverters remained significantly higher than the one measured in subjects with persistent NGT (1•59[1•12, 2•23] vs 3•90[2•58, 6•08], p 0•001). (Figure 2, panel A)

The return to NGT did not determinate a reversal of oDI to values comparable to the ones measured in participants who did not experience transient impairment of glucose tolerance. Indeed, the oDI of participants who reverted from IGT to NGT was higher than the one observed in those who remained NGT at follow-up (3•90 [2•58, 6•08] vs. 1•59[1•12, 2•23], p<0•001), while it dropped in subjects progressing from NGT to IGT (follow-up oDI 0•86[0•67, 1•32]) or persisting IGT/progressing to T2D (1•31[1•01, 1•85] and 0•20[0•12,0•58] respectively) when compared to the group that persisted with NGT across the entire study (p<0•01). (Figure 2, panel A)

Determinants of beta cell function changes over time.

Ethnicity was the sole significant and independent predictor of the oDI levels at follow-up among subjects with baseline IGT, but not among those with NGT. Compared with NHB participants with IGT, follow-up oDI index among their NHW counterparts was higher by an average of 2•32(0•30, 4•60) points (p=0•045), after controlling for other covariates. (Table S4)

The effect of ethnicity on the change in oDI in the two strata of glucose tolerance was also confirmed when analyzing the percentage change in oDI from baseline among the three ethnic groups. NHW with IGT at baseline exhibited a significantly higher median yearly increase in oDI (189%) as compared with NHB (83%, p=0•011) and Hispanic peers (135%, p 0•274), while no significant differences were observed among the three ethnic groups with baseline NGT. (Figure 2, panel C)

Discussion

We provided the first insight into the natural history of the dynamics in glucose tolerance from a large prospective multiethnic cohort of obese youth. The novel findings of this study include identifying the highly transient nature of IGT in this population and ethnicity being the major predictor of the beta cell response to pre-diabetic hyperglycemia.

The highly transient nature of IGT, as 65% of our pediatric cohort reverted from IGT to NGT, is in contrast with ~25% of adults with IGT reverting to NGT in the Diabetes Prevention Program (DPP) and in the ACT-NOW cohorts (25, 26) describing, for the first time in pediatrics, the impact of transient dysglycemia on beta cell function, and the clinical determinant of IGT.

Herein, we demonstrated that NHB ethnic background determines the persistence of the IGT phenotype, with only 50% of subjects reverting to normal glucose tolerance as compared to 77% of NHW group, when the other clinical and individual variables were held constant. This finding does not negate the roles of other known risk factors for T2D, like BMI and its change, 2-hours glucose, or the pubertal stage, as their effects on the progression to T2D have been largely described by our group and others in the past.(4, 27–29) However, demonstrating the pivotal role that ethnicity plays in the pathogenesis of youth onset impaired glucose tolerance, we support the concept that the genetic background, of which ethnicity is a reliable proxy, (30) is the major determinant of change in glucose tolerance in this age group, and, potentially, of the response to the therapeutic options. Indeed, genetic factors affecting both the individual insulin sensitivity (31, 32) and the beta cell response in the context of the whole body insulin sensitivity could explain the dysfunctional response to impaired glucose tolerance observed in the NHB group. (33)

In support of our findings, the TODAY study proved how the rates of failure in the three treatments arms (metformin plus rosiglitazone, metformin alone, and metformin plus lifestyle) was influenced by the ethnic background of participants, with NHB showing higher rate of treatment failure than NHW and Hispanic.(8, 9) Subsequently, the same cohort showed how treatment-associated changes in biochemical markers, like adiponectin, were associated with ethnicity, further supporting its role as a major determinant of response-to-treatment.(9) The mechanisms underpinning the influence of ethnicity on diabetes progression are still unclear. Genome wide association studies have proved that NHBs carry a greater number of risk allele for T2D than NHW, supporting the idea of ethnicity as a proxy for specific risk genes. (30, 34) Additionally, NHB background could act as an amplifier of the risk for T2D progression associated with specific single nucleotide polymorphism like the ones described in the TCF7L2 gene. NHB confers an almost two- fold increased risk for T2D when compared with NHW youths in the presence of the same risk allele. (35) This demonstrates the role of ethnicity as an effect modifier of the genetic response to dysglycemia.

NHB background has been reported to be associated with decreased hepatic insulin clearance, with consequent relative hyperinsulinemia, suggesting a potential mechanism of the derailed response to impaired glucose tolerance in this group.(36, 37) Adopting a surrogate measure of insulin clearance, previously validated against euglycemic-clamp, (23) we confirmed lower insulin clearance in NHB and Hispanic groups, as compared with NHW, in the presence of normal and impaired glucose tolerance. NHB ethnicity has been associated to 0.4% higher HbA_1c than NHW for a given mean glucose concentration, (38) as confirmed by the baseline HbA_1c of our cohort. Even though this finding does not affect the measured outcome, it suggests that NHB background may increase the future risk for diabetes complications influencing tissues’ glycosylation.

The effect of ethnicity on the longitudinal change in glucose tolerance in our cohort, in the presence of prediabetes, was characterized by its influence on the dynamic change in insulin secretion. Indeed, compared with NHB obese youth who reverted from IGT to NGT, their NHW counterparts exhibited higher levels in insulin secretion at follow-up, as measured by the oDI. Even if the increase in insulin secretion, could be a natural physiological response to hyperglycemia, this mechanism has never been described in a pediatric longitudinal cohort. Remarkably, once normal glucose tolerance is restored, the reverter group persists in having an almost double oDI when compared to subjects who have never experienced transient hyperglycemia. This feature is a hallmark of the effect of transient hyperglycemia on beta cell function in this population.

These findings suggest that natural history of youth-onset prediabetes can be featured either by a progressive loss of insulin secretive response (drop in oDI) with consequent persistence of the IGT phenotype or progression to T2D, or by an hypersecretive status that is associated with regression to NGT (increase in oDI). The individual genetic background could be responsible for the dynamic response to the ambient insulin sensitivity and ultimately of the glucose tolerance change.

The impairment of glucose tolerance in our young cohort occurs in the framework of the pubertal transition: (39)^, this phase is physiologically featured by a transient doubling of insulin response.(28)^, The effect of overweight, and as its consequence, of reduced insulin sensitivity, during this age, deeply affects the risk for T2D, and explains the peculiar response to transient hyperglycemia of our large pediatric cohort, and the high vulnerability of this age group to develop a disrupted response to transient hyperglycemia.

Accounting for ethnicity as a proxy of genetic background,(33) our findings are suggestive for a prevailing role of individual genetic factors, over other modifiable variables, in driving glucose changes in youth onset prediabetes. Ethnic differences in T2D prevalence (1) as well in the success of therapeutic interventions (8) have been partially attributed to social and economic barriers that prevent certain ethnic groups to be engaged in the pharmaceutical and lifestyle interventions.

A limitation of our study is represented by the lack of baseline and prospective data on nutritional intake for this cohort, that could have influenced the individual risk for glucose tolerance change over time, as well as detailed data on the compliance to lifestyle modifications suggested at the time of clinical follow-up appointments. Assuming the change in BMI as a reliable proxy of adherence, the absence of a significant change in the BMI, as well as its insignificant effect on the odds for reverting to NGT, support the predominant role of ethnic background, as proxy of specific genetic features, in determining the natural history of prediabetes over the other factors. Such consideration should suggest tailored interventions in high-risk groups and explain the higher rate of failure of previous therapeutic treatments in certain ethnic groups.(24)

One of the limitations of our study is that the classification of glucose tolerance relied on a single OGTT. This may not reliably establish glucose tolerance status, given the well-known poor reproducibility of the OGTT in both youth and adults.(40) However, we took steps to mitigate this limitation in our study. Prior to the OGTT, we implemented a number of measures, such as dietary standardization, child-friendly environment, avoidance of strenuous physical activity, and bed-side analysis of glucose. We also obtained additional indices derived from the OGTT, such as the oDI, to corroborate our classification of glucose tolerance status.(41) This is also supported by our finding that participant with transient IGT exhibited higher oDI values at follow up than those who maintained NGT.

Another potential limitation is that transient impairment of glucose tolerance could occur in our participants, but go undetected. However, since our study subjects came from an on-going longitudinal cohort study with a standardized follow up, this limitation is unlikely. Despite a longer follow-up would have allowed a more comprehensive assessment of the transient IGT, as participants with IGT at the end of the study could have potentially turned back to NGT later on, our study was conducted during peripubertal transition that is expected to trigger the youth-onset dysglycemia, representing the ideal timeframe for describing the natural history of youth-onset prediabetes. (42) Even though we reported a limited number of participants progressing to T2D (n=13, 8% of the IGT cohort), the ethnic distribution of this small group reflects the higher prevalence of T2D in NHB adults reported by the SEARCH study.(1) However, our findings prompt the urgent need for large multiethnic studies aimed to assess the natural history of prediabetes in youths and its determinants.

The strength of our study is the active prospective surveillance of outcomes in a cohort of youths with similar features of obesity and dysglycemia. In particular, our measurement were obtained using standardized methods for measures of insulin sensitivity and insulin secretion.

In conclusion, our study describes the transient nature of IGT in obese youth, and the role of ethnicity as the main predictor of the dynamic response of beta cells to pre-diabetic hyperglycemia. The ethnic differences in the beta-cell response to hyperglycemia characterize the natural history of glucose tolerance as related to T2D with youth onset. This suggests that treatment for impaired glucose tolerance in obese youth should target high risk populations and the specific mechanisms underpinning changes in glucose tolerance in these groups.

Supplementary Material

1

Supplemental Table S1. Baseline clinical and metabolic characteristics of the cohort that completed the follow-up OGTT (Study Cohort) and the group lost at follow-up assessment (Lost Follow-up). NGT, normal glucose tolerance; IGT, impaired glucose tolerance; NHW, non-Hispanic White; NHB, non-Hispanic Black; H, Hispanic. Data are expressed as mean±SD or number(percentage).

Supplemental Table S2. Baseline clinical and metabolic characteristics according to changes in glucose tolerance status. NGT, normal glucose tolerance; IGT, impaired glucose tolerance, T2D, type 2 diabetes; NHW, non-Hispanic White; NHB, non-Hispanic Black; H, Hispanic. Data are expressed as median(25th, 75th percentile) or number(percentage). NGT>NGT and IGT>IGT are the reference group for paired comparison in presence of baseline NGT or IGT respectively.

Supplemental Table S3. Left panel. Multivariable logistic regression model for progression in glucose tolerance from NGT to IGT. Right panel. Multivariable logistic regression model for transient IGT phenotype. NHB is the reference group. (95%CI, 95% Confidence Interval; NHW, non-Hispanic White; NHB, non-Hispanic Black; H, Hispanic; NGT, normal glucose tolerance; IGT, impaired glucose tolerance)

Supplemental Table S4. Multivariable linear regression model for the yearly rate of change in disposition index (DI) among participants with baseline NGT (left side) and IGT (right side)· NHW, non-Hispanic White; H, Hispanic· NGT, normal glucose tolerance; IGT, impaired glucose tolerance; T2D, type 2 diabetes· Data are expressed as the estimated adjusted differences in mean levels of DI (β [95%CI]).

Research in context

Evidence before this study.

We searched Pubmed and clinicaltrials.gov for clinical trials and observational studies conducted from January 2000 and April 2018 using the following keywords: “Obesity”, “Impaired glucose tolerance”, “Type 2 Diabetes”, “Youth”, “Epidemiology”, “prediabetes”, “ethnicity”. The SEARCH for Diabetes in Youth study indicated a steep increase in T2D incidence in adolescents, particularly among Non-Hispanic Black (NHB) and Hispanic ethnic groups. NHB obese adolescents with T2D have been found to be less responsive to the current treatment options for diabetes than Non-Hispanic Whites (NHW) and Hispanics in the TODAY study. T2D is preceded by a prediabetic impairment of glucose tolerance that is reversible in 25% of adults, as described in in the Diabetes Prevention Program (DPP) and in the ACT-NOW cohorts. However, the clinical predictors of the reversibility of prediabetes in youths and the underpinning mechanisms in obese youth, have never been assessed longitudinally.

Added value of this study.

Herein, we described the natural history of prediabetes in obese youth, by the mean of OGTT derived measures of insulin secretion (oDI) along with the clinical determinants of changes in glucose tolerance. We demonstrated, for the first time, that an increase in insulin secretion is associated with reversion to NGT, and a persistent hyper-responsiveness of beta cells function in subjects with transient IGT, as compared to those who never experienced the prediabetic hyperglycemia. The ethnic background was the main predictor for reverting to NGT, when compared to the other clinical and anthropometric variables like BMI, age, pubertal stage.

Implications of all the available evidence.

Impaired glucose tolerance is a highly transient condition in obese youth, with 65% reverting to NGT. However, the rate of reversion is markedly lower in the NHB group. The failure to increase beta cell response in the presence of IGT has a pivotal role in determining the risk of persistence in the prediabetes stage, and NHB ethnic background is associated with a reduced insulin response to prediabetic hyperglycemia. This finding points out a major role for the genetic background associated with NHB ethnicity and the requirement for targeted intervention in the group who holds the higher risk of persistent IGT and consequently of T2D.