Leptin and Soluble Leptin Receptor Levels in Plasma and Risk of Type 2 Diabetes in U.S. Women: A Prospective Study

The NHS cohort was established in 1976 with 121,700 female registered nurses aged 30–55 years who were residing in 11 U.S. states and had completed a mailed questionnaire on their medical history and lifestyle characteristics. During 1989–1990, upon request, 32,826 women provided blood samples, the majority (97%) of which were received within 26 h of blood draw. Immediately upon arrival, whole blood samples were centrifuged and aliquoted into cryotubes as plasma, buffy coat, and erythrocytes, which were then stored in the vapor phase of liquid nitrogen freezers at a temperature ≤ −130°C.

Among the participants who provided blood samples and were free of diabetes, cardiovascular disease, and cancer at blood draw, we prospectively identified and confirmed 1,054 type 2 diabetes case subjects from the date of blood draw through June 2004. Using risk-set sampling, one or two control subjects were randomly selected for each case subject from the rest of the population who remained free of diabetes and matched to case subjects by age at blood draw (±1 year), date of blood draw (±3 months), fasting status (fasting for ≥8 h or not fasting), and race. After excluding eight control subjects with missing leptin or sOB-R data, 1,054 incident type 2 diabetes case subjects and 1,254 control subjects were available for the current analysis.

The study protocol was approved by the institutional review board of the Brigham and Women's Hospital and the Human Subjects Committee Review Board of Harvard School of Public Health.

In the NHS, women who reported a diagnosis of type 2 diabetes in the biennial follow-up questionnaires were sent supplementary questionnaires inquiring about symptoms, diagnostic tests, and treatment for the purpose of confirmation. The self-report of diagnosis of type 2 diabetes has been proven to be accurate in a validation study (21). Of a random sample of 62 diabetic nurses initially confirmed by the supplementary questionnaire, 61 (98%) were confirmed by medical records reviewed by an endocrinologist blinded to the supplementary questionnaire information (21). In the current study, all case subjects were confirmed using the supplementary questionnaire. For case subjects diagnosed before 1998, we used the following criteria from the National Diabetes Data Group for the confirmation of type 2 diabetes diagnosis: 1) an elevated glucose concentration (fasting plasma glucose ≥7.8 mmol/l, random plasma glucose ≥11.1 mmol/l, or plasma glucose ≥11.1 mmol/l after an oral glucose load) and at least one symptom (excessive thirst, polyuria, weight loss, or hunger) related to diabetes; 2) no symptoms but elevated glucose concentrations on two occasions; and 3) treatment with insulin or oral hypoglycemic medication. For case subjects of type 2 diabetes identified after 1998, the cutoff point used for fasting plasma glucose concentrations was lowered to 7.0 mmol/l according to the American Diabetes Association criteria.

Each case-control pair or triplet was shipped in the same batch and analyzed in the same run. Within each batch, samples were assayed by the same technicians in a random sequence under identical conditions.

Total leptin was measured by a radioimmunoassay (Millipore, Billerica, MA), the sensitivity of which is 0.5 ng/ml. sOB-R was measured by ELISA technique (R&D Systems, Minneapolis, MN) with a sensitivity of 0.06 ng/ml. Laboratory control samples (n = 20) were run along with the case-control samples. Based on the measurements of these control samples, the average intra-assay coefficient of variation was 7.7% for the leptin assay and 7.3% for the sOB-R assay. In addition to leptin and sOB-R, total and high-molecular-weight (HMW) adiponectin, resistin, C-reactive protein, tumor necrosis factor-α receptor 2, interleukin-6 (IL-6), interleukin-18 (IL-18), fasting insulin, and C-peptide were also measured for all or some of the case and control subjects. The assays for these biomarkers have been described elsewhere (22,–24).

In baseline and/or subsequent biennial follow-up questionnaires, information on major lifestyle risk factors for chronic diseases, such as body weight, cigarette smoking, physical activity, family history of diabetes, menopausal status, and postmenopausal hormone use, was collected and updated in the NHS cohort. BMI as weight in kilograms divided by the square of height in meters (kg/m²) was calculated to assess overall adiposity. We asked the participants to measure their waist circumference (at umbilicus) and hip circumference (the largest circumference) in 1986. From 1980, diet was assessed using a validated semiquantitative food frequency questionnaire every 2–4 years. Nutrient intake was calculated based on responses to the food frequency questionnaire and the nutrient content of foods was derived from the Harvard Food Composition Database.

We evaluated the correlation of leptin and sOB-R with anthropometric measurements and plasma concentrations of inflammatory markers and adipokines among control subjects. Spearman partial correlation coefficients were calculated, adjusted for age at blood draw (years), date of blood draw, fasting status (8 h versus <8 h since last meal), race, BMI (kg/m²), smoking status (never smoked, former smoker, or current smoker), postmenopausal status (yes/no), hormone use (never used, former user, current user), family history of diabetes (yes/no), and physical activity and alcohol intake (both in quintiles). Because we observed a strong correlation between leptin biomarkers and BMI (Table 2), we calculated the BMI-adjusted residuals of the biomarkers by regressing each biomarker on BMI at blood draw using linear regressions and used these residuals in the subsequent analysis. These residuals are statistically independent of BMI and, therefore, are subject to minimal residual confounding by BMI. This approach has been described in detail elsewhere (25).

We categorized the study population into quintiles according to the distribution of the BMI-adjusted levels of biomarkers among control subjects. We used both conditional and unconditional logistic regressions to examine the associations of interest, and because the findings were similar, we presented the results only from unconditional logistic regressions. In the multivariate analysis, we controlled for the aforementioned lifestyle covariates, as well as intake of alcohol, cereal fiber, heme iron, trans fat, magnesium, coffee, and red meat (all in quintiles). All lifestyle covariates were derived from the questionnaire administered in 1990 or the nearest year prior to 1990. Because we have multiple assessments of dietary factors, we calculated the average of nutrient intake in 1980, 1984, 1986, and 1990 to minimize measurement errors and better represent long-term diet (26). The chosen lifestyle and dietary covariates were either matching factors that should be adjusted when using unconditional regression to derive ORs in a nested case-control study (27) or established diabetes risk factors that were also correlated with leptin or sOB-R levels. P values for linear trend were calculated by entering an ordinal score based on the median value in each quintile of leptin biomarkers into the multivariate models. When we examined potential interactions between leptin biomarkers and lifestyle/dietary risk factors, we used tertiles of the biomarkers to preserve statistical power and to gain more stable estimates of the OR. We constructed interaction terms between biomarker tertiles and the interacting factors of interest and used likelihood ratio tests to assess the significance of these interaction terms. Likelihood ratio tests are calculated as the difference of −2 log likelihood in models with and without interaction terms and follow the χ² distribution with the degree of freedom equal to the number of parameters for the interaction terms. In addition, we used restricted cubic spline regressions with three knots (28) to examine possible nonlinear relationships between leptin biomarkers and risk of type 2 diabetes. Tests for nonlinearity used the likelihood ratio test, comparing the model with only the linear term to the model with the linear and the cubic spline terms.

All P values were two-sided, and 95% CIs were calculated for ORs. Data were analyzed with the Statistical Analysis Systems software package, version 9.1 (SAS Institute, Cary, NC).

Table 1 shows the baseline characteristics of study population. As expected, type 2 diabetes case subjects had a higher BMI, lower physical activity levels, higher probability of having family history of diabetes, and less healthy dietary intake at baseline than control subjects. Case subjects also had significantly higher leptin levels but lower sOB-R levels than control subjects. The distribution of other inflammatory biomarkers and adipokines in case and control subjects was consistent with our previous findings (22,–24).

In control subjects, after adjustment for multiple covariates, plasma leptin levels were strongly correlated with BMI; the partial Spearman correlation coefficient (r) was 0.72 (Table 2). In contrast, sOB-R was significantly inversely correlated with BMI (r = −0.43). To better control for confounding by BMI, we used BMI-adjusted residuals of these biomarkers in the subsequent analyses. BMI-adjusted leptin and sOB-R were not correlated with waist circumference or waist-to-hip ratio. Significantly positive correlations were found between leptin and C-reactive protein (r = 0.25) and between sOB-R and HMW adiponectin (r = 0.26). In a subset of the control subjects with insulin and C-peptide data, we found no significant correlations with insulin but significant correlations for leptin (r = 0.18) and sOB-R (r = −0.18) with C-peptide.

After multivariate adjustment for established and potential lifestyle and dietary risk factors for type 2 diabetes, BMI-adjusted leptin levels were not significantly associated with diabetes risk (Table 3). In contrast, BMI-adjusted sOB-R levels were significantly associated with a lower risk of type 2 diabetes. Compared with women in the lowest quintile of sOB-R, women in the highest quintile had an OR of 0.39 (95% CI: 0.28–0.54; P for trend < 0.0001). This association was attenuated but remained statistically significant after further adjustment for HMW adiponectin; the OR (95% CI) was 0.67 (0.47–0.95; P for trend = 0.005). Further adjustment for other biomarkers, including inflammatory markers and other adipokines, did not change the point estimates of these associations materially, although the 95% CIs were much wider and P values less significant because only half of the study population had all of these biomarkers available (data not shown). Lastly, when we further adjusted for fasting insulin in a subset of the study participants (39% of total participants) with fasting insulin data, the associations for sOB-R levels were slightly attenuated, but a statistically significant trend persisted (data not shown).

We found no significant deviation from a linear relationship between sOB-R and risk of type 2 diabetes in spline regression models; the P value for nonlinearity is 0.35 (Fig. 1).

Joint associations for sOB-R and leptin, as well as HMW adiponectin, are shown in Fig. 2. We did not find a significant interaction between sOB-R and leptin (P for interaction = 0.09) or HMW adiponectin (P for interaction = 0.30). Within each tertile of leptin levels, sOB-R levels were consistently associated with a lower risk of type 2 diabetes. In our previous study, HMW adiponectin was significantly associated with a lower risk of type 2 diabetes (24). In the current analysis, sOB-R was associated with a lower risk within each HMW adiponectin level and vice versa. Women who were in the highest tertiles of both sOB-R and HMW adiponectin had the lowest odds of developing type 2 diabetes; the OR (95% CI) was 0.13 (0.09–0.20) in comparison with women in the lowest tertiles of these two markers.

We subsequently examined potential interactions of leptin biomarkers with other risk factors in relation to type 2 diabetes risk (Table 4). Although we did not find significant interactions between leptin biomarkers and risk factors such as age, fasting status, and physical activity, the association for leptin with diabetes risk was significantly modified by BMI (P for interaction = 0.001). Leptin was significantly associated with a lower risk of type 2 diabetes in participants with a BMI ≥30 kg/m². In contrast, in lean participants (BMI <25 kg/m²), leptin levels were associated with a significantly increased risk of type 2 diabetes. Additional adjustment for waist circumference or waist-to-hip ratio did not change these associations (data not shown).

In this nested case-control study conducted in middle-aged U.S. women, we found that high sOB-R levels were strongly associated with a lower risk of type 2 diabetes independent of baseline leptin or adiposity levels. On the other hand, BMI-adjusted leptin levels were not significantly associated with risk of type 2 diabetes.

Our study provided the first piece of evidence for an inverse association between sOB-R levels and risk of developing type 2 diabetes independent of plasma leptin levels. Leptin generates its central and peripheral effects by binding to its receptors on the cell surface and subsequently activating downstream signaling pathways (29). Through posttranscriptional alternative RNA splicing, several isoforms of leptin receptors with identical extracellular and transmembrane domains but variable intracellular domains are expressed in humans (30). The long form of leptin receptor with the full length of intracellular domain is expressed primarily in the hypothalamus, whereas the short forms of leptin receptor (OB-R_S) are expressed primarily in peripheral tissues (30). sOB-R, a special leptin receptor with the extracellular domain only, is formed by ectodomain shedding of leptin receptors on the cell surface (31), and provides the primary binding capacity in human circulation (15). The function of sOB-R is not entirely clear but believed to delay the clearance of leptin from the circulation and, thus, increase leptin's availability (16). In addition, there is evidence suggesting that sOB-R not only alters the clearance of leptin but also potentiates leptin action (32). This is an analogy to the observation that the action of IL-6 can be boosted by binding to its soluble receptor (33), which shares a homologous structure with leptin receptor (34). Furthermore, because sOB-R levels are highly correlated with the cell surface expression of leptin receptors (r = 0.80) (35), especially OB-R_S, sOB-R may represent the total amount or biological activity of OB-R_S expressed in peripheral tissues. Unlike long form of leptin receptor, OB-R_S does not contain the intracellular motifs required to activate the Janus kinase/signal transducers and activators of transcription signaling pathway that mediates the energy homeostasis effects of leptin (36). However, accumulating evidence suggests that OB-R_S may mediate the effects of leptin on insulin sensitivity and other peripheral effects through a distinct pathway involving insulin receptor substrate/phosphatidylinositol-3-OH kinase pathway (37,,,–41). Although more data are needed to further elucidate sOB-R's functions, these mechanisms may underlie the inverse association with diabetes risk for sOB-R observed in the current study.

Consistent with a previous study (20), we found a significant, positive correlation between sOB-R and HMW adiponectin in the current study. Both biomarkers are correlated with lower levels of adiposity and decrease after weight loss (17,19), and their short-term variations are nearly identical (42), suggesting these two biomarkers may share some common regulatory factors. However, the inverse associations of these two biomarkers were independent of each other, and women with high levels of both biomarkers had dramatically lower risk of developing diabetes. Whether these two molecules share the same pathway or affect diabetes risk through distinct pathways warrants further investigation.

Although intervention studies conducted in leptin- deficient animal models and humans supported a beneficial effect of leptin on insulin sensitivity or type 2 diabetes (43,,,–47), prospective epidemiologic data on circulating leptin levels and future risk of type 2 diabetes have been mixed in subjects with normal leptin secretion ability (6,,,,,,,–14). In humans, leptin levels are strongly correlated with subcutaneous adiposity, reflecting leptin resistance in obese individuals. To remove the confounding effects of BMI, we examined BMI-adjusted leptin levels in relation to diabetes risk. In our study, BMI-adjusted leptin levels were not significantly associated with diabetes risk. However, our data suggest that the association between leptin and type 2 diabetes was modulated by BMI. Leptin levels tended to be associated with a lower risk among relatively obese subjects. Two previous studies also found a similar association (8,14). One possible explanation for the positive association between leptin and diabetes risk in lean participants could be that high leptin levels reflected the amount of adipose tissue that was not captured by BMI, although further adjustment for waist circumference did not change our observation. The beneficial effects of high leptin levels on diabetes risk in overweight or obese people may be related to the peripheral effects of leptin on insulin sensitivity rather than its effects on weight regulation (5) because in the cerebrospinal fluid free leptin levels are already saturated at low levels of circulating leptin levels (48). Whether high leptin levels in lean people represent a deteriorating metabolic status needs further examination.

The current study has several strengths. The prospective study design made it unlikely that disease status or treatment may influence the leptin or sOB-R levels (i.e., that reverse causation occurred). In addition, we used BMI-adjusted residuals of leptin biomarkers in the current analysis to more completely control for strong confounding by BMI. Other strengths include a large sample size, long follow-up period, validated approach for confirming type 2 diabetes case subjects, and adjustment for a multitude of risk factors for type 2 diabetes.

Several limitations merit discussion as well. First, in observational studies, residual confounding, especially the confounding by adiposity not captured by BMI and waist circumference in the current study, cannot be entirely ruled out. In addition, some covariates were assessed at different time points, which may lead to potentially incomplete controlling for confounding by these factors. Second, a single baseline measurement of leptin or sOB-R levels may not represent the long-term levels of these markers. However, leptin levels have been shown to be quite stable over time (49). Similar to adiponectin, the short-term variation of sOB-R levels was relatively small (50). Third, because our study sample included only middle-aged women who were predominantly white, it is unclear whether the results can be generalized to men and other ethnic groups.

In conclusion, we found a strong, inverse association between circulating soluble leptin receptor levels and risk of type 2 diabetes, independent of obesity and leptin levels. On the other hand, the association between leptin and diabetes risk may be modified by adiposity. Biological mechanisms underlying these novel observations need to be further elucidated.

This work was supported by the National Institutes of Health and the Intramural Research Program of the National Institute of Child Health & Human Development (Grants CA-87969, DK-58845, and DK-58785). Q.S. was supported by a Unilever Postdoctoral Fellowship. J.B.M. was supported by a Career Development Award from the American Diabetes Association and Grant K24 DK-080140 from the National Institute of Diabetes and Digestive and Kidney Diseases. C.S.M. was supported by discretionary grants from the Tanita Corporation and Beth Israel Deaconess Medical Center.

J.B.M. currently has research grants from GlaxoSmithKline and sanofi-aventis and has served on consultancy boards for sanofi-aventis and Interleukin Genetics. No other potential conflicts of interest relevant to this article were reported.

Parts of this study were presented in abstract form at the Scientific Sessions of the American Heart Association, Orlando, Florida, 14–18 November 2009.