Abstract
Background and Aims:
The recent COVID-19 pandemic has further increased the importance of reducing obesity and prediabetes/diabetes. We aimed to evaluate the association between adiposity and regression of prediabetes/diabetes.
Methods and results:
The San Juan Overweight Adults Longitudinal Study (SOALS) included 1,351 individuals with overweight/obesity, aged 40–65, free of major cardiovascular diseases and physician diagnosed diabetes. From the 1012 participants with baseline prediabetes/diabetes, 598 who completed the follow-up were included. Over the follow-up, 25% regressed from prediabetes to normoglycemia or from diabetes to prediabetes or normoglycemia. Poisson regression with robust standard error was used to estimate the relative risk (RRs) adjusting for major confounders. Higher neck circumference (NC) was associated with regression of prediabetes/diabetes (RR=0.45 comparing extreme tertiles; 95% CI:0.30–0.66); RR was 0.49 (95% CI:0.34–0.73) for waist circumference (WC) and 0.64 (95% CI:0.44–0.92) for BMI. Significant association were found using median cut-offs or continuous measures for weight and BMI. Greater reduction in BMI (comparing extreme tertiles) was significantly associated with regression of prediabetes/diabetes (RR=1.44; 95% CI:1.02–2.02). Continuous measures of change in adiposity (except for NC) were also associated with regression of prediabetes/diabetes for BMI and weight. Participants who reduced BMI (>5%), increased prediabetes/diabetes regression (RR=1.61; 95% CI:1.15–2.25) compared to those who did not; similarly for weight (RR=1.55; 95% CI: 1.10–2.19). Additional analysis for body fat percentage showing similar but slightly weaker results as BMI/weight, further supported our findings.
Conclusion:
Lower baseline adiposity and higher reduction in adiposity were associated with regression of prediabetes/diabetes among individuals with overweight/obesity.
Keywords: weight loss, body mass index, pre-diabetes, diabetes, regression, neck and waist circumference, obesity
Introduction
The prevalence of diabetes mellitus (diabetes) has increased rapidly over the past 2 decades[1], and it is estimated that by 2040, 642 million people will have diabetes[2]. In addition to established complications of diabetes, patients with diabetes are also at increased risk of death from coronavirus disease (COVID-19)[3]. Obesity is strongly associated with type 2 diabetes[4] and cardiovascular disease, and with regression of type 2 diabetes and prediabetes[5][6]. Baseline high neck circumference (NC)[7], is more strongly associated with the incidence of type 2 diabetes compared to BMI or waist circumference (WC)[8]. However, two meta-analyses[9,10] showed inconsistent results regarding which measures of adiposity were most pertinent for diabetes progression.
Very few studies evaluated the association between adiposity reduction and regression from prediabetes/diabetes to normoglycemia in the absence of intervention[5,11,12], assessing only one or two adiposity measures. The ADDITION-Prediabetes Cohort Study was essentially an observational study where investigators provided an information leaflet promoting the benefits of a healthy lifestyle to all participants, and evaluated changes in weight and WC over a one year follow-up period as predictors of regression of prediabetes[12]. Another population-based cohort study (KORA) found significant associations between reduction of BMI/WC and regression from prediabetes to normoglycemia[5].
Previous literature suggests that adiposity could likely reduce regression of prediabetes/diabetes, and reducing adiposity could likely increase regression of prediabetes/diabetes, but this has not been directly evaluated for a follow-up longer than one year in the absence of specific interventions. Factors associated with regression of diabetes have rarely been evaluated, and only few adiposity measures have been considered in this context. Accordingly, this observational study evaluates the associations between baseline adiposity and reduction in adiposity, and regression of prediabetes/diabetes over a three-year follow-up period, in the absence of dietary guidelines or lifestyle intervention programs, among adults with overweight/obesity.
Methods
Study Sample
The San Juan Overweight Adults Longitudinal Study (SOALS) was approved by the University of Puerto Rico Medical Sciences Campus Institutional Review Board. Informed consent was obtained from all participants. Recruitment and baseline data collection started in 2011 and the follow-up exam was completed by 2016. Participants are multiracial individuals of Hispanic ethnicity, who reported their race as White (25%), Black (14%), or mixed race (61%).
The eligibility criteria included (1) aged 40 to 65 years, (2) BMI of at least 25.0 kg/m2 (individuals with overweight/obesity). The baseline exclusion criteria were: 1) reported physician-diagnosed type 1 or type 2 diabetes, or taking either insulin or oral anti-hyperglycemic agents; 2) pregnancy; 3) physician diagnosed hypoglycemia, congenital heart murmurs, heart valve disease, congenital heart disease, endocarditis, rheumatic fever, and hemophilia or bleeding disorders; 4) active dialysis treatment; 5) having undergone procedures related to cardiovascular disease; 6) severe health conditions or psychological or physical disabilities that would interfere with participation in the study, and 7) plans on moving away during the follow-up. A total of 1,351 eligible participants completed the key components of the study at baseline.
Retention efforts included phone calls, letters, and small tokens. By the three-year follow-up (Figure 1), 255 were lost to follow-up (68 refused to continue the study, 140 could not be reached or did not complete follow-up, 6 died, and 41 moved out of Puerto Rico). Of the remaining 1,096 participants, 62 participants had incomplete outcome measures, and 32 participants had incomplete information on other key variables, resulting in 1,002 (74.2%) participants. Since the outcome of interest was regression of prediabetes/diabetes, we further excluded 404 participants who had normoglycemia as per the baseline visit assessment, and the remaining 598 participants were included in this analysis.
Figure 1.
Flow Diagram of participants of San Juan Overweight Adults Longitudinal Study (SOALS), 2011–2016
Regression of prediabetes/diabetes
We used American Diabetes Association[13] thresholds to classify glycemic status from study assessments at baseline and follow-up. Participants were classified as having diabetes if they had plasma glucose ≥ 7 mmol/l, two-hour oral glucose tolerance ≥ 11.1 mmol/l, or hemoglobin A1c (HbA1c) ≥ 48 mmol/mol[14]. For the follow-up, we also considered reported physician diagnosed diabetes during the follow-up period. People were classified as having prediabetes if they had HbA1c of 39–47 mmol/mol or as having normoglycemia if all these values were below the mentioned thresholds for prediabetes[14]. Regression was defined as reversal/remission from diabetes to prediabetes or normoglycemia, or from prediabetes to normoglycemia over the follow-up.
Anthropometric measures
Weight (0.2 kg graduation) and fat% were measured by a bioelectrical impedance analysis technology using Tanita Body Composition Analyzer-TBF-310A, and height was measured using a portable stadiometer (Seca Corporation). Circumferences were measured with Gulick tape to the nearest 0.1 cm. WC was measured at the umbilicus. NC was measured below the laryngeal prominence and perpendicular to the long axis of the neck, and the minimal circumference was recorded[15]. In the interest of clinical and public health relevance, we were primarily interested in adiposity measures that people could easily assess and monitor in their home (e.g. weight and neck circumference) to improve adiposity and health. Hence, we did not prioritize measures that require equipment like Tanita scale that is not easily available in most homes and clinics. However, we had additionally assessed body fat percentage (fat%), since it is a pertinent biological measure. All anthropometric measures were taken in duplicate according to the NHANES III procedures. When the first two measures differed at least 0.5 cm, a third measure was recorded; the average of available measures was calculated for each person at each visit. Pearson correlation coefficients between repeats for all measures were greater than 0.99, showing excellent reproducibility.
Covariates
Participants provided information on important socio-demographic and behavioral characteristics, including age, gender, years of education, smoking status (never, former, current), alcohol consumption (abstainer, former, current), physical activity, physician diagnosed diabetes and hypertension, and medication use. Physical activity was converted to metabolic equivalent of tasks (METs) in hours per week[16], and classified as meeting/not meeting WHO guidelines.
Blood pressure was measured following the gold standard Korotkoff auscultatory method after 5 minutes of rest, intervals of 1 minute between 3 measures, and then averaged. Participants were classified as hypertensive if they had physician diagnosis of hypertension, and/or if they reported taking high blood pressure medication, and/or had baseline systolic blood pressure (SBP) ≥ 140 or diastolic blood pressure (DBP) ≥ 90 mmHg; they were classified as pre-hypertensive if they were not classified as hypertensive and SBP was 120–139 and/or DBP was 80–89 mmHg[17].
Triglycerides and high-density lipoprotein cholesterol (HDL-C) levels were determined from baseline blood samples using standard procedures. Low-density lipoprotein cholesterol (LDL-C) was estimated using the Friedewald equation. Insulin resistance was estimated using Homeostatic model assessment insulin resistance index (HOMA-IR) (Fasting glucose × Fasting insulin)/405.
Statistical Analyses
Baseline characteristics of participants are described within high/low NC, WC, and BMI defined using the medians. We also described the change in glycemic status (normoglycemia, prediabetes, and diabetes) from baseline to follow up. We used a modified Poisson regression for binary outcome data (regression of prediabetes/diabetes over the three years follow up) with a log link function and robust error variance to estimate relative risks/risk ratios (RRs) and 95% confidence intervals (CIs). Poisson regression was used as it tends to provide conservative results and overcomes the convergence problem that arises when using binomial regression[18]. The logarithm of follow-up time was included as an offset variable. Since there were no standard thresholds for adiposity measures, we used gender-specific tertiles for NC, WC, weight, and BMI. We evaluated the association of baseline adiposity measures with regression of prediabetes/diabetes using the lowest tertile as the reference category; we also assessed continuous measures of adiposity.
We also evaluated the association of change in adiposity measures from baseline to follow-up (follow-up minus baseline) with regression of prediabetes/diabetes, using continuous measures and tertiles. We also evaluated more than 5% reduction and 0.1% to 5% reduction, using gain/stable adiposity measures as the reference for weight loss and other adiposity measures. We selected a threshold of 5% because for people with BMI ≥ 25 kg/m2, weight loss of at least 5% is recognized as providing meaningful health benefits, including reductions in cholesterol, blood sugar, and blood pressure[19–22].
For regression analyses, model 1 controlled for age and gender. Model 2 additionally controlled for other major risk factors for diabetes as per the literature, including education, smoking status, alcohol consumption, physical activity (MET), and hypertension status. Model 3 additionally controlled for baseline glycemic status to control extraneous variations in the outcome[23]. Other potential confounding factors based on the literature including baseline mouthwash use, statin use, family history of diabetes, biomarkers, and lipids (triglycerides, HDL-C, and LDL-C) were entered one at a time into model 2[24][17]. However, none of these confounding factors were retained as they did not change the effect estimates for the exposure by 10% or more[25]. Analyses were performed in Stata SE 13.1. The study is reported following STROBE guidelines[26].
Results
Participants with high baseline NC, WC, and BMI had significantly higher HOMA-IR, and higher frequency of hypertension and metabolic syndrome compared to people with low NC, WC, and BMI (Table 1). Additionally, NC, WC, and BMI levels were significantly different for the groups defined by low versus high baseline physical activity (METs), HOMA-IR, hypertension, metabolic syndrome, and HbA1c. Table 2 shows the baseline tertile and median values for adiposity measures for males and females separately.
Table 1.
Baseline characteristics (% or mean ± SD) of participants by high and low-neck circumference, waist circumference, and BMI defined using medians
| Characteristics | Neck circumference* | P¥ | Waist circumference** | P¥ | Body mass index*** | P¥ | |||
|---|---|---|---|---|---|---|---|---|---|
| Low (n=302) | High (n=296) | Low (n=299) | High (n=299) | Low (n=299) | High (n=299) | ||||
| Age (years) | 52.3 ± 6.5 | 51.0 ± 7.0 | < 0.05 | 52.9 ± 6.6 | 50.3 ± 6.6 | < 0.01 | 52.8 ± 6.6 | 50.5 ± 6.7 | < 0.01 |
| Female (% ) | 217 (71.9) | 213 (72.0) | 1.00 | 215 (71.9) | 215 (71.9) | 1.00 | 215 (71.9) | 215 (71.9) | 1.00 |
| Current smoker (% ) | 54 (17.9) | 38 (13.2) | 0.07 | 44 (14.7) | 49 (16.4) | < 0.05 | 47 (15.7) | 46 (15.4) | 0.12 |
| Higher than high school education (%) | 185 (61.3) | 189 (63.9) | 0.09 | 200 (66.9) | 174 (58.2) | 0.09 | 193 (64.5) | 181 (60.5) | 0.58 |
| Physical activity (METs) | 28.0 ± 53.1 | 16.1 ± 23.1 | < 0.01 | 27.2 ± 51.0 | 17.0 ± 28.2 | < 0.05 | 27.3 ± 50.8 | 16.9 ± 28.6 | < 0.05 |
| Body mass index (kg/m2) | 30.6 ± 4.1 | 36.9 ± 6.3 | < 0.01 | 29.6 ± 2.8 | 37.9 ± 5.9 | < 0.01 | 29.0 ± 1.9 | 38.5 ± 5.3 | < 0.01 |
| Alcohol grams/day (n=256) ‡ | 5.4 ± 7.5 | 5.4 ± 6.8 | 0.95 | 4.7 ± 6.7 | 6.1 ± 7.5 | 0.11 | 5.0 ± 7.0 | 5.8 ± 7.4 | 0.36 |
| Hypertension (%) | 139 (46.0) | 169 (57.1) | < 0.05 | 142 (47.5) | 166 (55.5) | < 0.05 | 138 (46.2) | 170 (56.9) | < 0.05 |
| Baseline glycemic status Prediabetes diabetes | 270 (89.4) 32 (10.6) |
256 (86.5) 40 (13.5) |
0.27 | 268 (89.6) 31 (10.4) |
258 (86.3) 41 (13.7) |
0.21 | 267 (89.3) 32 (10.7) |
259 (86.6) 40 (13.4) |
0.32 |
| HOMA-IR index | 2.5 ± 1.6 | 3.8 ± 2.2 | < 0.01 | 2.5 ± 1.6 | 3.8 ± 2.1 | < 0.01 | 2.5 ± 1.7 | 3.7 ± 2.1 | < 0.01 |
| Triglycerides (mg/dL) | 151.0 ± 86.8 | 169.9 ± 96.0 | < 0.05 | 158.2 ± 90.9 | 162.6 ± 92.9 | 0.55 | 163.3 ± 98.1 | 157.5 ± 85.2 | 0.44 |
| High density lipoprotein (mg/dL) | 49.5 ± 13.7 | 45.2 ± 11.8 | < 0.01 | 48.8 ± 12.5 | 45.9 ± 13.3 | < 0.05 | 48.8 ± 12.7 | 45.8 ± 13.1 | < 0.05 |
| Low density lipoprotein (mg/dL) | 126.2 ± 34.8 | 125.2 ± 33.9 | 0.72 | 127.8 ± 32.3 | 123.6 ± 36.2 | 0.13 | 128.3 ± 34.9 | 123.1 ± 33.6 | 0.07 |
| Metabolic syndrome (%) | 174 (57.6) | 208 (70.3) | < 0.01 | 164 (54.8) | 218 (72.9) | < 0.01 | 161 (53.8) | 221 (73.9) | < 0.01 |
| HbA1c (%) | 5.9 ± 0.6 | 6.1 ± 0.7 | < 0.01 | 5.9 ± 0.5 | 6.1 ± 0.7 | < 0.01 | 5.9 ± 0.6 | 6.1 ± 0.7 | < 0.01 |
| Fasting Glucose (mg/dl) | 100.4 ± 21.7 | 102.1 ± 24.1 | 0.38 | 99.6 ± 19.9 | 102.9 ± 25.4 | 0.08 | 101.3 ± 24.6 | 101.2 ± 21.1 | 0.95 |
| Fat (%) | 37.1 ± 7.7 | 42.2 ± 7.9 | < 0.01 | 36.5 ± 7.0 | 42.9 ± 8.0 | < 0.01 | 36.1 ± 7 | 43.3 ± 7.7 | < 0.01 |
Abbreviations: METs, Metabolic Equivalents; HOMA-IR, Homeostatic Model Assessment of Insulin Resistance; diabetes, Diabetes mellitus; SD: Standard deviation
Median cut-off:
Male 36.3 cm, Female 42.1 cm
Male 109.5 cm, Female 104.5 cm
Male 31.5 kg/m2, Female 32.4 kg/m2
Alcohol intake grams/day of current drinkers
Bold font; p-value < 0.05
P values from t-test / chi-square test
Table 2.
Tertile and median values for adiposity measures (n=598)
| Baseline Adiposity measure | Gender | Tertile 1 Range¥ (Reference) | Tertile 2 Range | Tertile 3 Range | Median Value |
|---|---|---|---|---|---|
| Neck circumference (cm) | Male Female | 35.7, 40.4 30.1, 35.2 |
40.5, 43.9 35.2, 37.5 |
44.0, 54.0 37.6, 46.9 |
42.1 36.3 |
| Waist circumference (cm) | Male Female | 86.1, 102.5 80.6, 99.4 |
102.6, 115.3 99.5, 110.3 |
115.5, 161.2 110.4, 161.2 |
109.5 104.5 |
| Weight (kg) | Male Female | 66.7, 85.1 57.1, 76.9 |
85.2, 105.0 77.0, 89.9 |
105.1, 183.0 90.0, 175.2 |
93.5 82.8 |
| BMI (kg/m2) | Male Female | 25.4, 29.1 25.0, 30.4 |
29.2, 34.4 30.5, 35.4 |
34.5, 57.3 35.5, 57.0 |
31.5 32.4 |
| Changes in Adiposity measure* | Tertile 3† Range (Reference) | Tertile 2 Range | Tertile 1 Range | Median Value | |
| Neck circumference (cm) | Male Female | +0.35, +6.15 +0.06, +4.05 |
−0.94, +0.30 −1.14, +0.05 |
−5.95, −0.95 −6.10, −1.15 |
−0.40 −0.59 |
| Waist circumference (cm) | Male Female | +2.40, +13.45 +2.48, +23.10 |
−1.85, +2.35 −2.50, +2.45 |
−18.12, −2.05 −39.95, −2.55 |
−0.42 −0.05 |
| Weight (kg) | Male Female | +2.60, +20.00 +2.30, +24.30 |
−2.05, +2.55 −1.35, +2.25 |
−21.97, −2.09 −33.20, −1.40 |
+0.40 +0.40 |
| BMI (kg/m2) | Male Female | +0.92, +6.08 +1.08, +8.34 |
−0.55, +0.88 −0.43, +1.07 |
−7.45, −0.56 −14.84, −0.44 |
+0.21 +0.31 |
Change is defined as follow-up minus baseline measures.
Positive sign (+) indicates gain/stability in adiposity measures.
Negative sign (−) indicates reduction in adiposity measures.
Tertile 1 (Lowest) is reference category.
Tertile 3 (Highest (+) positive change/Gain) is reference category.
The majority of participants (63.4% with normoglycemia, 64.8% with prediabetes, and 68.1% with diabetes) did not change glycemic status (Supplementary Table A). Three (4.2%) participants with diabetes and 135 (25.7%) participants with prediabetes at baseline regressed to normoglycemia, and 20 (27.8%) participants regressed from diabetes to prediabetes. On the other hand, 143 (35.4%) participants with normoglycemia progressed to prediabetes, five (1.2%) with normoglycemia progressed to diabetes, and 50 (9.5%) participants with prediabetes progressed to diabetes.
At baseline, 317 (53%) participants reported that they had intended to reduce their weight. Of these, 74 (31%) participants reversed their prediabetes/diabetes. Baseline adiposity measures, physical activity (METs), hypertension, HOMA-IR, HbA1c, triglycerides, and HDL-C were significantly different for the regression versus no regression groups (Supplementary Table B). Individuals with prediabetes/diabetes had higher adiposity measures, alcohol consumption, lipids (triglycerides, and LDL-C), lower physical activity (METs), and lower HDL-C compared to normoglycemic group (Supplementary Table C).
Adding more baseline covariates, including biomarkers, lipids, mouthwash use, statin use, and family history of diabetes, did not change the estimate by more than 10%, hence they were not retained in the model. No major multi-collinearity issues (tolerance index > 0.20, VIF values < 4.0) were found among independent variables[27]. Pearson correlation coefficients relating baseline adiposity measures (BMI, NC, WC, and weight) to baseline fasting glucose/HbA1c within the prediabetes (r ≤ 0.17) and diabetes (r ≤ 0.34) groups were weak.
Individuals classified in the highest tertile of NC exhibited lower regression of prediabetes/diabetes (multivariate RR=0.45; 95% CI: 0.30–0.66) compared to those in the lowest tertile (Table 3). Similar significant associations were found for the highest tertiles of WC (RR=0.49; 95% CI: 0.34–0.73), weight (RR=0.61; 95% CI: 0.42–0.88) and BMI (RR=0.64; 95% CI: 0.44–0.92) compared to the lowest tertiles. The middle tertiles were not significantly different from the lowest tertiles for any of the adiposity measures. The associations were also significant as per the median cutoffs for WC and BMI. Continuous measures of baseline NC (RR=0.88; 95% CI: 0.84–0.92) and WC scaled to 5 cm (RR=0.90; 95% CI: 0.84–0.95) were also significantly associated with regression of prediabetes/diabetes after adjusting for covariates. Continuous measures of BMI (RR=0.97; 95% CI: 0.94–1.00) and weight (RR=0.99; 95% CI: 0.98–1.00) showed borderline associations with regression of prediabetes/diabetes.
Table 3.
Multivariable adjusted risk ratios (95% CI) relating baseline adiposity measures and regression of prediabetes/diabetes over 3-year follow-up (n=598)
| Baseline Adiposity measure | RR (95% CI) | |||||||
|---|---|---|---|---|---|---|---|---|
| Tertile 1¥ (Reference) | Tertile 2¥ | P | Tertile 3¥ | P | Continuous¥ | P | ||
| Neck circumference (cm) | Model 1* | 1.00 | 0.89 (0.66, 1.21) | 0.46 | 0.45 (0.30, 0.66) | < 0.01 | 0.88 (0.83, 0.92) | < 0.01 |
| Model 2** | 1.00 | 0.91 (0.67, 1.23) | 0.53 | 0.44 (0.30, 0.66) | < 0.01 | 0.88 (0.84, 0.92) | < 0.01 | |
| Model 3† | 1.00 | 0.92 (0.68, 1.25) | 0.59 | 0.45 (0.30, 0.66) | < 0.01 | 0.88 (0.84, 0.92) | < 0.01 | |
| Waist circumference (cm) | Model 1* | 1.00 | 0.76 (0.56, 1.03) | 0.07 | 0.48 (0.33, 0.69) | < 0.01 | 0.98 (0.97, 0.99) | < 0.01 |
| Model 2** | 1.00 | 0.74 (0.54, 1.02) | 0.06 | 0.49 (0.34, 0.72) | < 0.01 | 0.98 (0.97, 0.99) | < 0.01 | |
| Model 3† | 1.00 | 0.75 (0.55, 1.02) | 0.06 | 0.49 (0.34, 0.73) | < 0.01 | 0.98 (0.97, 0.99) | < 0.01 | |
| Weight (kg) | Model 1* | 1.00 | 0.84 (0.61, 1.16) | 0.30 | 0.59 (0.41, 0.85) | < 0.01 | 0.99 (0.98, 1.00) | < 0.01 |
| Model 2** | 1.00 | 0.81 (0.60, 1.12) | 0.21 | 0.61 (0.42, 0.89) | < 0.01 | 0.99 (0.98, 1.00) | < 0.01 | |
| Model 3† | 1.00 | 0.82 (0.60, 1.13) | 0.23 | 0.61 (0.42, 0.88) | < 0.01 | 0.99 (0.98, 1.00) | < 0.01 | |
| BMI (kg/m2) | Model 1* | 1.00 | 0.82 (0.60, 1.13) | 0.23 | 0.62 (0.43, 0.89) | < 0.01 | 0.97 (0.94, 0.99) | < 0.05 |
| Model 2** | 1.00 | 0.84 (0.61, 1.15) | 0.28 | 0.64 (0.44, 0.92) | < 0.05 | 0.97 (0.94, 1.00) | < 0.05 | |
| Model 3† | 1.00 | 0.84 (0.61, 1.15) | 0.28 | 0.64 (0.44, 0.92) | < 0.05 | 0.97 (0.94, 1.00) | < 0.05 | |
Model 1 adjusted for age and gender;
Model 2 additionally adjusted for age, gender, years of education, smoking status, alcohol consumption, physical activity (MET), and hypertension status (normotensive, prehypertension, hypertension).
Model 3 additionally adjusted for glycemic status at baseline (Poisson Regression)
Bold font; p-value < 0.05
Reference category: Lowest Tertile (First Tertile)
Refer Table 2 for Tertile/continuous ranges of anthropometric measures (baseline).
Abbreviation: BMI = body mass index, P = p-value
Table 4 shows results relating the reduction in adiposity measures from baseline to follow-up with prediabetes/diabetes regression. Reduction in NC, WC, and weight was not significantly associated with regression of prediabetes/diabetes. A larger reduction in BMI (comparing extreme tertiles) was significantly associated with regression of prediabetes/diabetes (Model 3 RR=1.44; 95% CI: 1.02–2.02). Continuous measures of BMI and weight reduction were also significantly associated with regression of prediabetes/diabetes, and continuous WC showed borderline significance.
Table 4.
Multivariable adjusted risk ratios (95% CI) relating changes in adiposity measures (follow-up minus baseline) with regression of prediabetes/diabetes over 3-year follow-up (n=598)
| Changes in Adiposity measure | RR (95% CI) * | |||||||
|---|---|---|---|---|---|---|---|---|
| Tertile 3¥ (Reference) | Tertile 2¥ | P | Tertile 1¥ | P | Continuous¥ | P | ||
| Neck circumference (cm) | Model 1* | 1.00 | 0.99 (0.70, 1.40) | 0.96 | 0.95 (0.68, 1.32) | 0.74 | 1.02 (0.93, 1.11) | 0.72 |
| Model 2** | 1.00 | 1.02 (0.73, 1.43) | 0.90 | 0.98 (0.70, 1.36) | 0.88 | 1.01 (0.93, 1.10) | 0.80 | |
| Model 3† | 1.00 | 1.03 (0.73, 1.44) | 0.88 | 1.00 (0.71, 1.39) | 0.98 | 1.01 (0.93, 1.09) | 0.90 | |
| Waist circumference (cm) | Model 1* | 1.00 | 0.87 (0.60, 1.26) | 0.47 | 1.24 (0.90, 1.72) | 0.19 | 0.98 (0.97, 1.00) | 0.07 |
| Model 2** | 1.00 | 0.85 (0.59, 1.23) | 0.39 | 1.29 (0.93, 1.78) | 0.12 | 0.98 (0.96, 1.00) | 0.07 | |
| Model 3† | 1.00 | 0.86 (0.60, 1.23) | 0.40 | 1.32 (0.95, 1.82) | 0.09 | 0.98 (0.96, 1.00) | ≤ 0.05 | |
| Weight (kg) | Model 1* | 1.00 | 1.23 (0.86, 1.76) | 0.26 | 1.26 (0.89, 1.79) | 0.19 | 0.98 (0.96, 1.00) | < 0.05 |
| Model 2** | 1.00 | 1.21 (0.85, 1.72) | 0.28 | 1.26 (0.90, 1.77) | 0.18 | 0.97 (0.95, 1.00) | < 0.05 | |
| Model 3† | 1.00 | 1.22 (0.86, 1.73) | 0.26 | 1.29 (0.91, 1.81) | 0.15 | 0.97 (0.95, 0.99) | ≤ 0.01 | |
| BMI (kg/m2) | Model 1* | 1.00 | 1.15 (0.80, 1.67) | 0.45 | 1.40 (0.99, 1.97) | 0.06 | 0.93 (0.89, 0.98) | < 0.01 |
| Model 2** | 1.00 | 1.15 (0.80, 1.64) | 0.46 | 1.42 (1.01, 1.99) | < 0.05 | 0.92 (0.88, 0.97) | < 0.01 | |
| Model 3† | 1.00 | 1.16 (0.81, 1.66) | 0.41 | 1.44 (1.02, 2.02) | < 0.05 | 0.92 (0.88, 0.96) | < 0.01 | |
Model 1 adjusted for age and gender;
Model 2 additionally adjusted for age, gender, years of education, smoking status, alcohol consumption, physical activity (MET), and hypertension status normotensive, prehypertension, hypertension).
Model 3 additionally adjusted for glycemic status at baseline (Poisson Regression)
Bold font; p-value < 0.05
Reference category: Lowest Tertile (Third Tertile; Gain/stable)
Refer Table 2 for Tertile/continuous ranges of change in adiposity measures (follow-up minus baseline)
Abbreviation: BMI = body mass index, P = p-value, CI = confidence interval
Table 5 shows associations between reduction of adiposity measures (≤0.1%, >0.1% to ≤ 5%, and > 5%) and regression of prediabetes/diabetes after adjusting for confounders. NC and WC reduction (either category) were not significantly associated with regression of prediabetes/diabetes. A significant association was observed between BMI reduction (>5% versus gain/stable) and regression of prediabetes/diabetes (RR=1.61; 95% CI: 1.15–2.25) adjusting for covariates. Similarly, weight reduction (>5% versus gain/stable) was significantly associated with regression of prediabetes/diabetes (RR=1.55; 95% CI: 1.10–2.19) after adjusting for covariates. Middle categories of weight and BMI reduction (≤5% versus gain/stable) were not significantly different from the lowest categories.
Table 5.
Multivariate adjusted risk ratios (95% CI) relating reduction of adiposity measures and regression of prediabetes/diabetes over 3-year follow-up (Regression n=151, 25%)
| Reduction in Adiposity measure | Stability or Gain (≤ 0.1%) Reference) | Modest Reduction 0.1 % < > 5% | P | Large Reduction (>5%) | P | |
|---|---|---|---|---|---|---|
| Neck circumference (cm) | Total (n=598) | 209 (34.9) | 299 (50) | 90 (15.1) | ||
| No regression (n=447) | 156 (74.6) | 223 (74.6) | 68 (75.6) | |||
| Regression (n=151) | 53 (25.4) | 76 (25.4) | 22 (24.4) | |||
| Model 1* | 1.00 | 0.99 (0.73, 1.35) | 0.96 | 0.91 (0.60, 1.39) | 0.65 | |
| Model 2** | 1.00 | 1.02 (0.76, 1.38) | 0.88 | 0.92 (0.60, 1.40) | 0.69 | |
| Model 3† | 1.00 | 1.03(0.77, 1.40) | 0.82 | 0.94(0.61, 1.43) | 0.77 | |
| Waist circumference (cm) | Total (n=598) | 292 (48.8) | 194 (32.4) | 112 (18.7)) | ||
| No regression (n=447) | 223 (76.4) | 148 (76.3) | 76 (67.9) | |||
| Regression (n=151) | 69 (23.6) | 46 (23.7) | 36 (32.1) | |||
| Model 1* | 1.00 | 0.99 (0.71, 1.37) | 0.95 | 1.29 (0.92, 1.82) | 0.14 | |
| Model 2** | 1.00 | 1.02 (0.74, 1.40) | 0.93 | 1.35 (0.96, 1.90) | 0.08 | |
| Model 3† | 1.00 | 1.03(0.75, 1.42) | 0.85 | 1.38(0.98, 1.94) | 0.06 | |
| Weight (kg) | Total (n=598) | 319 (53.3) | 187 (31.3) | 92 (15.4) | ||
| No regression (n=447) | 243 (76.2) | 144 (77.0) | 60 (65.2) | |||
| Regression (n=151) | 76 (23.8) | 43 (23.0) | 32 (34.8) | |||
| Model 1* | 1.00 | 0.98 (0.70, 1.37) | 0.92 | 1.41 (1.00, 1.99) | ≤ 0.05 | |
| Model 2** | 1.00 | 0.99 (0.72, 1.38) | 0.97 | 1.46 (1.04, 2.05) | < 0.05 | |
| Model 3† | 1.00 | 1.02(0.73, 1.42) | 0.91 | 1.55(1.10, 2.19) | ≤ 0.01 | |
| BMI (kg/m2) | Total (n=598) | 342 (57.2) | 167 (27.9) | 89 (14.9) | ||
| No regression (n=447) | 261 (76.3) | 129 (77.2) | 57 (64.0) | |||
| Regression (n=151) | 81 (23.7) | 38 (22.8) | 32 (36.0) | |||
| Model 1* | 1.00 | 0.97 (0.69, 1.36) | 0.85 | 1.45 (1.04, 2.03) | < 0.05 | |
| Model 2** | 1.00 | 1.01 (0.72, 1.41) | 0.98 | 1.53 (1.09, 2.14) | ≤ 0.01 | |
| Model 3† | 1.00 | 1.03(0.73, 1.44) | 0.93 | 1.61(1.15, 2.25) | ≤ 0.01 |
Model 1 adjusted for age and gender;
Model 2 additionally adjusted for age, gender, years of education, smoking status, alcohol consumption, physical activity (MET), and hypertension status (normotensive, prehypertension, hypertension).
Model 3 additionally adjusted for glycemic status at baseline (Poisson Regression);
Bold font; p-value < 0.05
Abbreviation: BMI = body mass index, P = p-value, CI = confidence interval
We additionally evaluated the association between baseline fat% and change in fat% with regression of prediabetes/diabetes adjusting for model 3 covariates. Higher baseline fat% (comparing extreme tertiles) was associated with regression (RR=0.65; 95% CI: 0.46–0.93). Reduction of fat% comparing extreme tertiles (RR=1.38; 95% CI: 1.00–1.91) and >5% versus gain/stable (RR=1.44; 95% CI: 0.99–2.06) were also associated with regression of prediabetes/diabetes. The continuous fat% measure was of borderline significance (RR=0.98; 95% CI: 0.96–1.00).
Discussion
In this study, lower adiposity measures at baseline, as well as reduction in adiposity measures over a three-year follow-up were associated with regression of prediabetes/diabetes among adults with overweight/obesity aged 40–65 at baseline. These associations were independent of age, gender, years of education, smoking status, alcohol consumption, physical activity, and hypertension status.
Our study is novel in showing significant longitudinal associations between reduction in several anthropometric measures and regression of prediabetes/diabetes in the absence of assigned interventions. Since the results were not attributable to a particular treatment regimen, findings are more generalizable to individuals with overweight/obesity than data from highly selected populations with specific interventions in randomized controlled trials (RCTs). On the other hand, we cannot exclude possible residual confounding and limitations in causal interpretations inherent in observational studies. The effect estimates in the multivariate model were similar to those in models adjusting for age and gender, suggesting minimal confounding by other factors. Short-term intraindividual variability in fasting and 2-hour glucose levels may have led to some random misclassification of glycemic status and attenuation of the associations.
All baseline anthropometrics measures (comparing extreme tertiles, and continuous measures) were significantly associated with regression. Baseline NC showed the strongest associations both for continuous and tertile measures; this may partly be due to the scale for the continuous measure. For the baseline extreme tertiles comparisons, weight and BMI show smaller associations compared to WC and NC. The KORA cohort study reported that baseline BMI was associated with regressing from prediabetes to normoglycemia (RR = 0.96; 95% CI: 0.93–0.99 per kg/m2) and WC (RR = 0.99; 95% CI: 0.98–1.00 per cm), but did not evaluate NC, nor regression from diabetes[5]. Since most participants in our study had prediabetes and fewer participants had diabetes at baseline, the prediabetes group may have mainly driven the associations. Our study is the first to evaluate the association of NC and NC reduction with regression of prediabetes to normoglycemia
The relative associations of the adiposity measures were opposite when we evaluated adiposity changes. Compared to WC, reduction in weight or BMI (lowest tertile versus highest, continuous measures, and decline of more than 5%) showed higher associations with regression of prediabetes/diabetes. Unlike baseline NC, reduction in NC was not associated with regression of prediabetes/diabetes. The results for fat% were similar but slightly weaker than BMI and change in BMI, generally corroborating and supporting our findings.
Our findings for weight/BMI/WC change corroborate previous studies. A one year follow-up study from Japan showed that people with obesity with a reduction of more than 5% in BMI and weight had a significantly greater tendency to return from impaired to normal fasting glucose compared to people without weight reduction [28]. The ADDITION-Prediabetes cohort study which showed that weight reduction over one year (0.1% - 3% versus gain/stable) and WC reduction more than >3 cm (versus gain/stable) was associated with regression of impaired glucose regulation[12]. In the ADDITION-Prediabetes cohort study, weight reduction (>0.3% versus stable or gain) was not associated with regression of prediabetes/diabetes. The KORA study showed associations between reduction in WC (RR = 1.08; 95% CI: 1.05–1.12 per 1 cm) and BMI (RR = 1.24; 95% CI: 1.09–1.41 per 1 kg/m2) with regression of prediabetes to normoglycemia. Reduction of other adiposity measures was not evaluated in these studies.
Beyond the widely used weight, BMI and WC measures, NC reflecting upper-body fat deposition could also serve as a good measure for adiposity and may be more strongly associated with metabolic syndrome or its components than other anthropometric measures[7]. Upper body subcutaneous fat may confer risk above and beyond visceral abdominal fat. High NC may alter peripheral blood flow and lead to endothelial dysfunction[29], which may reduce insulin delivery and promote insulin resistance[30] and therefore may result in lower regression of prediabetes/diabetes. Our study showed that NC, is a novel, discrete, and pathogenic fat depot, associated with regression of prediabetes/diabetes. Systemic free fatty acid concentrations are primarily determined by upper-body subcutaneous fat, suggesting that this fat depot may play an essential role in pathogenesis[31]. Upper-body subcutaneous fat, commonly estimated from the NC[32], has been demonstrated to be the main reservoir of circulating free fatty acids in individuals with overweight/obesity[31]. Moreover, previous studies found that free fatty acid concentrations were closely related to insulin resistance and endothelial dysfunction[31]. NC is easier to measure and unaffected by last meal, fluid intake, voiding, and clothes compared to WC and may be a novel additional target for screening and interventions for diabetes regression. It is unclear why baseline NC was strongly associated but reduction of NC was not associated with regression. This may be because NC may have unique genetic basis independent of BMI[33], or because it may take longer than 3 years for NC reduction to impact pre-diabetes/diabetes regression.
Although type 2 diabetes is believed to be characterized by a progressive and irreversible loss of pancreatic insulin secretion[34], our study and the ADDITION study [12] found that prediabetes/diabetes may be reversible with weight reduction in the absence of assigned interventions. Weight reduction permits restoration of first-phase insulin secretion associated with normalization of elevated pancreatic triglyceride content. Insulin resistance and β-cell dysfunction are the major pathophysiologic factors driving type 2 diabetes[35]. In people with obesity, and decreased β-cell function has been shown to predict deterioration of glucose tolerance[36]. Evidence also suggests that β-cell dysfunction and reduction of end-differentiated β-cell phenotype can be restored by substantial weight reduction, and could be a mechanism to return to normal or near-normal blood glucose levels without glucose-lowering treatment[35].
In conclusion, this longitudinal study suggests that adiposity is strongly associated with less frequent regression of prediabetes/diabetes. NC may be a better and easier measure of adiposity and may be a novel additional target for screening and interventions for prediabetes/diabetes regression. Reversing prediabetes/diabetes by reducing adiposity challenges the assumptions that prediabetes/diabetes is uniformly irreversible and progressive. These findings highlight the possibility of reversing diabetes and prediabetes among high-risk individuals with overweight/obesity, and the findings are important to motivate people with prediabetes or newly diagnosed diabetes. People could easily assess and monitor weight, height, waist, and neck circumferences in their home to reduce adiposity and improve health. Our study provides new information and corroborates earlier findings about the importance of adiposity reduction in the regression of prediabetes/diabetes. The key take home public health message is that people with overweight/obesity and prediabetes/diabetes should focus on reducing adiposity to increase the likelihood of returning to normoglycemia.
Supplementary Material
Highlights.
25% reversed diabetes/pre-diabetes over 3 years without any assigned interventions.
Weight and BMI reduction are significantly associated with prediabetes/DM reversal.
Neck circumference may be a novel target for diabetes screening and regression.
Acknowledgments
We acknowledge Mrs. Aracelis Arroyo, Ms. Lillian Colón, Francisco J. Muñoz-Torres, Mr. Jeanpaul Fernández, Ms. Tania Ginebra, Ms. Katya Giovannetti, Mrs. Nilda González, Mr. Helson Lasanta, Dr. Sasha Martínez, Ms. Elaine Rodríguez, and Ms. Hilda Torres.
Sources of support: Research reported in this paper was supported by the National Institute of Dental and Craniofacial Research Grant R01DE020111 and the National Institute on Minority Health and Health Disparities Grant U54MD007587.
Acronyms
- SOALS
San Juan Overweight Adults Longitudinal Study
- RRs
Relative risks/risk ratios
- NC
Neck circumference
- WC
Waist circumference
- BMI
Body mass index
- COVID-19
Coronavirus disease
- CVD
Cardiovascular disease
- METs
Metabolic equivalent of tasks
- HPFS
Health Professionals Follow-Up Study
- SBP
Systolic blood pressure
- DBP
Diastolic blood pressure
- HDL-C
High-density lipoprotein cholesterol
- LDL-C
Low-density lipoprotein cholesterol
- HOMA-IR
Homeostatic model assessment-insulin resistance
- ELISA
Enzyme-linked immunosorbent assay
- CVs
Coefficients of variation
- CIs
Confidence intervals
- HbA1c
Hemoglobin A1c
- RCTs
Randomized controlled trials
- fat%
Body fat percentage
Footnotes
Data Statement
Data is available with a request to the corresponding author.
Conflict of interest
The authors declare no competing financial interests.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References:
- [1].Guariguata L, Whiting DR, Hambleton I, Beagley J, Linnenkamp U, Shaw JE. Global estimates of diabetes prevalence for 2013 and projections for 2035. Diabetes Res Clin Pract 2014;103:137–49. doi: 10.1016/j.diabres.2013.11.002. [DOI] [PubMed] [Google Scholar]
- [2].Ogurtsova K, da Rocha Fernandes JD, Huang Y, Linnenkamp U, Guariguata L, Cho NH, et al. IDF Diabetes Atlas: Global estimates for the prevalence of diabetes for 2015 and 2040. Diabetes Res Clin Pract 2017;128:40–50. doi: 10.1016/j.diabres.2017.03.024. [DOI] [PubMed] [Google Scholar]
- [3].Fang L, Karakiulakis G, Roth M. Are patients with hypertension and diabetes mellitus at increased risk for COVID-19 infection? Lancet Respir Med 2020;8:e21. doi: 10.1016/s2213-2600(20)30116-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Holvoet P. Stress in Obesity and Associated Metabolic and Cardiovascular Disorders. Sci 2012;19. doi: 10.6064/2012:205027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Kowall B, Rathmann W, Kuss O, Herder C, Roden M, Stang A, et al. Reversion from prediabetes to normoglycaemia after weight change in older persons: The KORA F4/FF4 study. Nutr Metab Cardiovasc Dis 2021;31:429–38. doi: 10.1016/j.numecd.2020.09.008. [DOI] [PubMed] [Google Scholar]
- [6].Jimenez M, Hu FB, Marino M, Li Y, Joshipura KJ. Prospective Associations Between Measures of Adiposity and Periodontal Disease. Obesity 2012;20:1718–25. doi: 10.1038/oby.2011.291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Hingorjo MR, Qureshi MA, Mehdi A. Neck circumference as a useful marker of obesity: a comparison with body mass index and waist circumference. J Pak Med Assoc 2012;62:36–40. [PubMed] [Google Scholar]
- [8].Joshipura K, Muñoz-torres F, Vergara J, Palacios C, Pérez CM. Neck Circumference May Be a Better Alternative to Standard Anthropometric Measures. J Diabetes Res 2016;2016. doi: 10.1155/2016/6058916. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Vazquez G, Duval S, Jacobs DRJ, Silventoinen K. Comparison of body mass index, waist circumference, and waist/hip ratio in predicting incident diabetes: a meta-analysis. Epidemiol Rev 2007;29:115–28. doi: 10.1093/epirev/mxm008. [DOI] [PubMed] [Google Scholar]
- [10].Kodama S, Horikawa C, Fujihara K, Heianza Y, Hirasawa R, Yachi Y, et al. Comparisons of the strength of associations with future type 2 diabetes risk among anthropometric obesity indicators, including waist-to-height ratio: a meta-analysis. Am J Epidemiol 2012;176:959–69. doi: 10.1093/aje/kws172. [DOI] [PubMed] [Google Scholar]
- [11].Kowall B, Rathmann W, Heier M, Holle R, Peters A, Thorand B, et al. Impact of weight and weight change on normalization of prediabetes and on persistence of normal glucose tolerance in an older population : the KORA S4 / F4 study. Int J Obes 2012;36:826–33. doi: 10.1038/ijo.2011.161. [DOI] [PubMed] [Google Scholar]
- [12].Bodicoat DH, Khunti K, Srinivasan BT, Mostafa S, Gray LJ, Davies MJ, et al. Incident Type 2 diabetes and the effect of early regression to normoglycaemia in a population with impaired glucose regulation. Diabet Med 2017;34:396–404. doi: 10.1111/dme.13091. [DOI] [PubMed] [Google Scholar]
- [13].American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2010;33. doi: 10.2337/dc10-S062. [DOI] [Google Scholar]
- [14].American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2010;33 Suppl 1:S62–9. doi: 10.2337/dc10-S062. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Fitch K V, Stanley TL, Looby SE, Rope AM, Grinspoon SK. Relationship between neck circumference and cardiometabolic parameters in HIV-infected and non-HIV-infected adults. Diabetes Care 2011;34:1026–31. doi: 10.2337/dc10-1983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Global Recommendations on Physical Activity for Health. Geneva: World Health Organization; 2010. [PubMed] [Google Scholar]
- [17].Morou-bermudez E. Over-the-counter mouthwash use and risk of pre-diabetes / diabetes Nitric Oxide Over-the-counter mouthwash use and risk of pre-diabetes / diabetes. Nitric Oxide 2017;71:14–20. doi: 10.1016/j.niox.2017.09.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Zou G. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004;159:702–6. doi: 10.1093/aje/kwh090. [DOI] [PubMed] [Google Scholar]
- [19].Van Gaal LF, Peiffer FW. The importance of obesity in diabetes and its treatment with sibutramine. Int J Obes 2001;25:S24–8. doi: 10.1038/sj.ijo.0801935. [DOI] [PubMed] [Google Scholar]
- [20].Blackburn G. Effect of Degree of Weight Loss on Health Benefits. Obes Res 1995;3:211s–216s. [DOI] [PubMed] [Google Scholar]
- [21].Jensen MD, Ryan DH, Apovian CM, Loria CM, Ard FJD, Millen BE, et al. AC Subcommittee on Prevention Guidelines. J Am Coll Cardiol 2013. doi: 10.1016/j.jacc.2013.11.004. [DOI] [Google Scholar]
- [22].Goldstein DJ. Beneficial health effects of modest weight loss. Int J Obes Relat Metab Disord J Int Assoc Study Obes 1992;16:397–415. [PubMed] [Google Scholar]
- [23].Goya Wannamethee S, Gerald Shaper A, Perry IJ. Smoking as a modifiable risk factor for type 2 diabetes in middle-aged men. Diabetes Care 2001;24:1590–5. doi: 10.2337/diacare.24.9.1590. [DOI] [PubMed] [Google Scholar]
- [24].Joshipura KJ, Muñoz-torres FJ, Campos M, Rivera-díaz AD, Zevallos JC. Association between within-visit systolic blood pressure variability and development of pre-diabetes and diabetes among overweight / obese individuals Association between within-visit systolic blood pressure variability and development of pre-diabetes and. J Hum Hypertens 2018. doi: 10.1038/s4137-1017-0009-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].McNamee R Confounding and confounders. Occup Environ Med 2003;60:227–34. doi: 10.1136/oem.60.3.227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Vandenbroucke JP, Von Elm E, Altman DG, Gøtzsche PC, Mulrow CD, Pocock SJ, et al. Strengthening the Reporting of Observational Studies in Epidemiology ( STROBE ): Explanation and elaboration *. Int J Surg 2014;12:1500–24. doi: 10.1016/j.ijsu.2014.07.014. [DOI] [PubMed] [Google Scholar]
- [27].O’Brien RM. A caution regarding rules of thumb for variance inflation factors. Qual Quant 2007;41:673–90. doi: 10.1007/s11135-006-9018-6. [DOI] [Google Scholar]
- [28].Shimizu S, Kawata Y, Kawakami N, Aoyama H. Effects of Changes in Obesity and Exercise on the Development of Diabetes and Return to Normal Fasting Plasma Glucose Levels at One-Year Follow-up in Middle-Aged Subjects with Impaired Fasting Glucose. Env Heal Prev Med 2001;6:127–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Hsueh WA, Quinones MJ. Role of endothelial dysfunction in insulin resistance. Am J Cardiol 2003;92:10J–17J. [DOI] [PubMed] [Google Scholar]
- [30].Chung S, Yoon I-Y, Shin Y-K, Lee CH, Kim J-W, Lee T, et al. Endothelial dysfunction and C-reactive protein in relation with the severity of obstructive sleep apnea syndrome. Sleep 2007;30:997–1001. doi: 10.1093/sleep/30.8.997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Nielsen S, Guo Z, Johnson CM, Hensrud DD, Jensen MD. Splanchnic lipolysis in human obesity. J Clin Invest 2004;113:1582–8. doi: 10.1172/JCI21047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Yoon C-Y, Park JT, Jhee JH, Kee YK, Seo C, Lee M, et al. Neck circumference predicts renal function decline in overweight women: A community-based prospective cohort study. Medicine (Baltimore) 2016;95:e4844. doi: 10.1097/MD.0000000000004844. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Liu Y, Zhang X, Lee J, Smelser D, Cade B, Chen H, et al. Genome-wide association study of neck circumference identifies sex-specific loci independent of generalized adiposity. Int J Obes 2021:1532–41. doi: 10.1038/s41366-021-00817-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].Rudenski AS, Hadden DR, Atkinson AB, Kennedy L, Matthews DR, Merrett JD, et al. Natural history of pancreatic islet B-cell function in type 2 diabetes mellitus studied over six years by homeostasis model assessment. Diabet Med 1988;5:36–41. [DOI] [PubMed] [Google Scholar]
- [35].Type Taylor R. 2 diabetes: etiology and reversibility. Diabetes Care 2013;36:1047–55. doi: 10.2337/dc12-1805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [36].Cali AMG, Man CD, Cobelli C, Dziura J, Seyal A, Shaw M, et al. Primary defects in beta-cell function further exacerbated by worsening of insulin resistance mark the development of impaired glucose tolerance in obese adolescents. Diabetes Care 2009;32:456–61. doi: 10.2337/dc08-1274. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.