The Impact of Leptin and Adiponectin on Incident Type 2 Diabetes Is Modified by Sex and Insulin Resistance

Abstract

Background:

Adiponectin and leptin and their ratio have been associated with incident type 2 diabetes (T2DM), although the data presented are conflicting and the populations studied have been small. In this large, prospective, nested, case referent study, we hypothesized that these associations are sex specific and may be modified by insulin resistance.

Methods:

Men and women aged 30–60 years with incident T2DM (n=640) and a prior health survey within the Västerbotten Intervention Programme (VIP) and matched referents (n=1564) were identified. Using conditional logistic regression analyses, we tested whether baseline plasma adiponectin and leptin levels and their ratio independently predicted incident T2DM, stratified for gender and insulin resistance.

Results:

Adjusted for traditional risk factors, fourth-quartile levels of adiponectin were associated with a reduced risk of T2DM in men [odds ratio (OR) 0.55 (0.36–0.86)] and women [OR 0.47 (0.27–0.83)]. Quartile four of the leptin/adiponectin ratio predicted T2DM in both men [OR 3.08 (1.68–5.67)] and women [OR 3.31 (1.56–7.03)], whereas quartile-four levels of leptin predicted T2DM only in men [OR 2.30 (1.32–4.02)]. When stratified for insulin sensitivity and adjusted for body mass index (BMI), log_e-transformed leptin predicted T2DM in insulin-sensitive men [OR 1.56 (1.13–2.17)] but not in insulin-resistant men [OR 1.03 (0.76–1.39)]. The effect of adiponectin and the leptin/adiponectin ratio was not influenced by the insulin sensitivity status.

Conclusions:

Leptin in men and adiponectin in both sexes were independent predictors of T2DM. The association was modified by the degree of insulin sensitivity. The leptin/adiponectin ratio may add predictive information beyond the separate hormones.

Introduction

T he relation between obesity and subsequent risk of type 2 diabetes (T2DM) is well established,¹ but the mediators are incompletely understood.^2,3 The adipocyte-derived hormones leptin and adiponectin have emerged as possible candidates, owing to multiple effects on glucose and lipid metabolism.^4

–7

Leptin may have a detrimental effect on β-cells^8,9 and may decrease insulin sensitivity,⁴ but prospective studies have been inconsistent about whether leptin independently predicts T2DM.^{10

–18} Notably, leptin-associated risk of T2DM and cardiovascular disease may be sex specific,^11,19 possibly owing to differences in leptin signaling as shown in mice.²⁰

Adiponectin has both antiinflammatory and insulin-sensitizing properties and may decrease the risk for diabetes through several mechanisms. High adiponectin levels have consistently been associated with a decreased risk of T2DM.²¹ Low adiponectin levels were recently shown to be associated with T2DM only in insulin-resistant and not in insulin-sensitive persons.²²

The leptin/adiponectin ratio may better classify obesity-related aberrations than either of the two hormones alone,²³ and it has been suggested to better reflect insulin resistance than a homeostasis model assessment (HOMA).^24,25 We hypothesize in this study that leptin and adiponectin and their ratio independently predict T2DM, and that these associations are sex specific and may be modified by insulin resistance.

Methods

Study subjects

The Västerbotten Intervention Programme (VIP) was initiated in 1985 with the aim of preventing cardiovascular disease and diabetes in Västerbotten County in northern Sweden. The methodology has been described previously.^26,27 In summary, all inhabitants were invited to a health survey at their nearest health care center in the year they turned 30, 40, 50, or 60 years old for a health examination, including measurements of height, weight, and blood pressure. Participants were offered a standardized oral glucose tolerance test (OGTT) after fasting overnight, and their plasma lipids were measured. They were asked to complete a questionnaire about lifestyle factors, sociodemographic characteristics, heredity, and quality of life. Moreover, participants were asked to donate blood samples for research; the samples were stored at −80°C at the Northern Sweden Medical Biobank. Participants differed only marginally from nonparticipants in social conditions. In all, 96% of the participants were born in Sweden and were presumptively Europids, although ethnicity was not registered.

The diabetes registry

In 2002, when over 60,000 unique individuals had participated in the VIP, the VIP registry was linked to the registry for inpatient and outpatient care in the county to identify subjects with a diabetes diagnosis. Altogether, 1923 subjects were identified, and of these, 68% consented to participate in the registry. Together with 5% deceased patients, this gave 1413 eligible diabetes patients. Research nurses visited all health institutions in the county to collect data from medical records. Each case was reviewed by a specialist in internal medicine and classified according to the 1999 World Health Organization (WHO) recommendations. The study population and the validation process for the diagnosis up to the year 2002 have been described previously.²⁸ The same procedure was repeated in 2007 when the VIP register was additionally linked to the pharmaceutical registry at the National Board of Health and Welfare to identify subjects using antidiabetic drugs in Västerbotten. At this time point, over 85,000 unique individuals had participated in the VIP and 86% had consented to participate in research and donate blood to the Biobank. At the time of assembly of this study, 1501 cases were available. Of these, 42 were classified as diabetes type 1 (2.8%), 1342 as type 2 (89.4%), and 117 were unclassifiable (7.8%). The year of diagnosis of T2DM was known, and June 30 was chosen as the day of diagnosis. Patients diagnosed <1 year after the VIP survey date were excluded, leaving 371 men and 269 women with T2DM for this study. Mean duration between the date of the survey and the date of diagnosis was 6.4 years (range 1.0–15.5 years). A referent population of 1564 VIP participants without a diabetes diagnosis was selected and group-matched for sex, age, geographic region, and examination date.

This study was approved by the research ethics committee at Umeå University, Umeå, Sweden, and informed consent was obtained from participants.

Baseline measurements

Physical activity was measured as frequency of leisure-time physical exercise in changed clothing. For the analysis, two groups were identified, those who had exercised once or more per week during the last 3 months versus those with less frequent exercising. Smoking was classified as daily cigarette smoking. Many Swedes use tobacco in the form of “snus,” a moist form of snuff, and users were those taking “snus” daily. Height and weight were measured in light clothing, and body mass index (BMI) was calculated. Blood pressure was measured in the supine position after 5 min of rest, and hypertension was defined as systolic blood pressure (SBP) ≥140 mmHg and/or diastolic blood pressure (DBP) ≥90 mmHg, and/or on antihypertensive medication.

The majority of subjects had a 2-h 75 g OGTT. Fasting and postload glucose levels were determined by a Reflotron benchtop analyzer (Roche Diagnostics, Basel, Switzerland) until 2004; from 2005, a HemoCue benchtop analyzer (Quest Diagnostics, Madison, NJ) was used. Plasma cholesterol and triglycerides were analyzed on the Reflotron analyzer. Samples from participants with increased cardiometabolic risk were tested for lipid fractions at the nearest hospital from 1993 onward. In this study, triglycerides were analyzed in 79% and high-density lipoprotein (HDL) (omitted from further analysis in this study) in 30% of the participants.

The following analyses were performed on stored samples from the Biobank. Leptin and adiponectin were analyzed with double-antibody radioimmunoassays (RIAs) (Millipore, Billerica, MA). Total coefficient of variation (CV) for leptin was 4.7% at both low (2–4 ng/mL) and high (10–15 ng/mL) levels. For adiponectin, the total CV was 15.2% at low (2–4 μg/mL) and 8.8% at high (26–54 μg/mL) levels. Fasting levels of insulin and C-peptide in serum were analyzed with Roche Elecsys assays on a Modular E170 analyzer (both Roche Diagnostics, Basel, Switzerland). Total CV for insulin was <5% at 20 mIU/L and <4% at 100 mIU/L. Total CV for C-peptide was <6% at both 0.6 nmol/L and 3.2 nmol/L.

Insulin resistance and β-cell function estimated by HOMA2 (HOMA2-IR and HOMA2-B, respectively) were calculated using an online calculator (www.dtu.ac.uk/homa). Both plasma insulin and C-peptide were used for the HOMA2 estimations; however, C-peptide was omitted because the results were nearly identical with those obtained using insulin.

Statistical analysis

Nonnormally distributed data were log_e-transformed prior to analysis. Descriptive statistics are presented as means, and skewed data as geometric means, with 95% confidence intervals (CIs). Differences in means between cases and referents were tested using the Student unpaired t-test. Continuous variables were categorized into quartiles by the distribution of the referent values, separately for men and women. Missing values for categorical variables were treated as a separate category (not included in the tables).

The influence of studied variables on incident T2DM, stratified for sex and insulin sensitivity status, was tested by univariate and multivariable logistic regression analysis using the conditional maximum likelihood routine designed for matched analysis to estimate odds ratios (ORs) and 95% CIs. The sex-specific distribution among referents of HOMA2-IR was used to define insulin sensitivity status, forming two groups with a cutoff between quartiles 3 and 4. Values below cutoff were defined as insulin sensitive and above cutoff as insulin resistant.

The distribution of cases and controls over quartiles was tested with a chi-squared test for linear trend. The association between log_e-transformed values of leptin, adiponectin, and BMI was explored with a BMI-adjusted, sex-specific, partial correlation in referents, using a two-tailed test.

Possible interaction between leptin and adiponectin was studied by construction of combination terms, including high and low levels (below and above the median value for referents, stratified for sex). Synergy indices with 95% CIs were calculated.²⁹ A synergy index >1.00 was taken to indicate positive interaction, whereas a value <1.00 was seen as a sign of antagonism. The impact of leptin, adiponectin, and their ratio (dichotomized into high and low values based on cases only) on time to diagnosis was tested using Kaplan–Meier analysis (log-rank test).

A P value <0.05 was considered statistically significant. All calculations were performed with the statistical program SPSS, version 18 (SPSS Inc., Chicago, IL).

Results

The baseline characteristics of the study population stratified for sex and diabetes status are presented in Table 1. Most traditional risk factors for T2DM as well as leptin and adiponectin differed substantially between diabetic patients and referents in both men and women. Women with future T2DM had a lower educational level than women without. Family history of diabetes was more common in men and women developing T2DM.

Table 1.

Baseline Characteristics According to Outcome Status Stratified for Sex

	Men						Women
	Referents		T2DM		P	Missing	Referents		T2DM		P	Missing
n	775		371				789		269
Age (years)	53.1	(52.5–53.7)	51.2	(50.4–52.1)	0.001	0/0	52.1	(51.5–52.7)	51.2	(50.2–52.2)	0.139	0/0
BMI (kg/m2)	25.9	(25.6–26.1)	29.3	(28.9–29.6)	<0.001	18/3	25.4	(25.1–25.7)	30.3	(29.8–30.9)	<0.001	9/6
Fasting glucose (mmol/L)	5.3	(5.2–5.4)	6.3	(6.2–6.4)	<0.001	14/4	5.2	(5.2–5.3)	6.2	(6.1–6.3)	<0.001	12/6
Postload glucose (mmol/L)	6.3	(6.1–6.4)	8.6	(8.4–8.9)	<0.001	27/45	6.8	(6.7–6.9)	9.3	(9.1–9.6)	<0.001	31/28
Diabetes heredity, first–degree (%)	17.5	(14.5–20.5)	33.7	(29.4–38.1)	<0.001	39/27	25.0	(21.7–28.3)	45.3	(39.6–50.9)	<0.001	65/26
HOMA2-B (%)^*	81.4	(78.4–84.4)	86.7	(82.1–91.5)	0.057	25/20	79.4	(76.8–82.0)	88.3	(83.3–93.5)	0.002	25/11
HOMA2-IR^*	0.93	(0.89–0.97)	1.66	(1.55–1.78)	<0.001	25/20	0.85	(0.81–0.89)	1.62	(1.50–1.75)	<0.001	25/11
Systolic blood pressure (mmHg)	134	(132–135)	141	(139–143)	<0.001	14/8	129	(128–131)	141	(139–143)	<0.001	20/8
Diastolic blood pressure (mmHg)	83	(82–84)	89	(87–90)	<0.001	14/8	79	(79–80)	85	(84–87)	<0.001	20/8
Hypertension (%)	35.6	(32.2–39.1)	54.8	(49.8–59.8)	<0.001	14/8	30.2	(26.8–33.5)	52.9	(47.2–58.6)	<0.001	20/8
Cholesterol (mmol/L)	5.9	(5.9–6.0)	6.1	(6.0–6.2)	0.078	15/8	6.0	(5.9–6.0)	6.2	(6.0–6.3)	0.015	17/8
Triglycerides (mmol/L)^*	1.3	(1.3–1.4)	2.0	(1.9–2.1)	<0.001	171/111	1.2	(1.1–1.2)	1.8	(1.7–1.9)	<0.001	123/50
Leptin (ng/mL)^*	2.8	(2.6–2.9)	4.7	(4.4–5.1)	<0.001	0/0	9.4	(9.0–9.8)	14.1	(13.0–15.2)	<0.001	0/2
Adiponectin (μg/mL)^*	8.5	(8.2–8.8)	5.9	(5.6–6.3)	<0.001	4/1	12.9	(12.4–13.4)	8.3	(7.8–8.8)	<0.001	0/2
Leptin/adiponectin ratio (ng/μg)^*	0.33	(0.30–0.35)	0.80	(0.73–0.87)	<0.001	4/1	0.73	(0.68–0.77)	1.70	(1.53–1.89)	<0.001	0/3
Insulin (mIU/L)^*	6.2	(5.9–6.5)	10.6	(9.8–11.3)	<0.001	11/15	5.7	(5.4–5.9)	10.3	(9.6–11.2)	<0.001	14/6
Physical excercise ≥1/week (%)	27.1	(24.0–30.2)	17.0	(12.4–21.5)	<0.001	44/35	30.8	(27.5–34.0)	20.2	(14.5–25.8)	0.001	77/26
Daily cigarette smokers (%)	27.5	(24.2–30.7)	31.4	(26.6–36.1)	0.181	21/17	25.4	(22.3–28.5)	26.6	(21.2–31.9)	0.704	28/13
Uses “snus” (moist snuff ) (%)	19.4	(16.5–22.4)	22.3	(18.1–26.5)	0.274	59/21	0.4	(0.0–1.1)	1.7	(0.1–2.8)	0.049	88/35
At least high school education (%)	29.4	(26.2–32.7)	27.8	(23.1–32.5)	0.567	21/11	37.3	(33.9–40.7)	27.5	(21.6–33.4)	0.005	31/18
University education (%)	13.4	(11.0–15.8)	10.8	(7.4–14.3)	0.228	21/11	21.0	(18.2–23.8)	12.4	(7.5–17.2)	0.002	31/18

Values given are numbers (n).

Means (^* indicates geometric mean) with 95% confidence interval (CI) and for categorical proportions % and differences between groups.

Numbers (n) with missing data are presented for referents/type 2 diabetes mellitus.

HOMA2–B and HOMA2–IR is homeostasis model asessment 2 β-cell function and insulin resistance, respectively.

Hypertension was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg and/or being on blood pressure medication.

Physical exercise was defined as excercise in changed clothing once or more per week the last 3 months.

T2DM, type 2 diabetes mellitus; BMI, body mass index.

Leptin correlated strongly with BMI in both men (r=0.66) and in women (r=0.73) (all P<0.0001), whereas the correlation between adiponectin and BMI was weaker in both men (r=−0.10, P<0.005) and in women (r=−0.17, P<0.001). Leptin did not correlate with adiponectin in men and only weakly in women (r=−0.08, P<0.05) after adjustment for BMI.

Table 2 presents univariate and BMI-adjusted ORs for risk factors for incident T2DM. Most risk factors were associated with similar T2DM risk in men and women. High leptin levels predicted T2DM in both men and women in the univariate analysis (all P<0.001), but after adjustment for BMI, leptin still predicted diabetes in men (P<0.001) but not in women (P=0.3). Leptin per BMI unit (leptin/BMI ratio) showed similar results (P<0.001 and P=0.4, respectively). High adiponectin levels predicted decreased risk of T2DM in all models in both men and women (all P<0.001).

Table 2.

Risk Factors for Future Type 2 Diabetes Mellitus

	Men				Women
	Univariate		BMI adjusted		Univariate		BMI adjusted
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
BMI (kg/m²)	1.12	(1.10–1.15)			1.12	(1.10–1.15)
Fasting glucose (mmol/L)	1.26	(1.21–1.32)	1.24	(1.19–1.30)	1.73	(1.59–1.89)	1.57	(1.43–1.73)
Postload glucose (mmol/L)	1.25	(1.20–1.29)	1.21	(1.17–1.25)	1.27	(1.22–1.32)	1.21	(1.16–1.26)
Diabetes heredity yes/no	1.67	(1.33–2.09)	1.48	(1.18–1.86)	1.89	(1.46–2.44)	1.59	(1.23–2.07)
HOMA2–B ^ln	1.13	(0.92–1.38)	0.80	(0.65–0.99)	1.31	(1.01–1.70)	0.76	(0.57–1.01)
HOMA2–IR ^ln	2.19	(1.89–2.55)	1.71	(1.44–2.04)	2.73	(2.26–3.29)	1.91	(1.51–2.41)
Systolic blood pressure (mmHg)	1.02	(1.01–1.03)	1.01	(1.01–1.02)	1.02	(1.02–1.03)	1.02	(1.01–1.02)
Diastolic blood pressure (mmHg)	1.03	(1.03–1.04)	1.02	(1.01–1.03)	1.04	(1.03–1.05)	1.03	(1.01–1.04)
Hypertension yes/no	2.03	(1.64–2.53)	1.61	(1.29–2.00)	2.18	(1.68–2.82)	1.58	(1.21–2.07)
Cholesterol (mmol/L)	1.05	(0.96–1.15)	1.05	(0.96–1.14)	1.13	(1.02–1.25)	1.08	(0.98–1.20)
Triglycerides ^ln	2.36	(1.88–2.96)	1.88	(1.48–2.39)	3.42	(2.63–4.45)	2.33	(1.74–3.12)
Leptin ^ln	2.04	(1.75–2.36)	1.47	(1.21–1.78)	2.39	(1.93–2.96)	1.17	(0.88–1.57)
Adiponectin ^ln	0.48	(0.40–0.57)	0.55	(0.46–0.67)	0.40	(0.34–0.49)	0.50	(0.40–0.61)
Leptin/adiponectin ratio ^ln	1.87	(1.68–2.09)	1.59	(1.39–1.83)	2.08	(1.83–2.38)	1.69	(1.42–2.01)
Leptin/BMI ratio ^ln	2.08	(1.75–2.47)	1.45	(1.20–1.76)	2.20	(1.70–2.85)	1.14	(0.85–1.53)
Insulin ^ln	2.02	(1.74–2.33)	1.54	(1.30–1.83)	2.55	(2.12–3.06)	1.75	(1.39–2.20)
Physical excercise no/yes	1.65	(1.23–2.20)	1.49	(1.11–2.00)	1.50	(1.09–2.06)	1.30	(0.94–1.79)
Regular cigarette smoking yes/no	1.17	(0.93–1.47)	1.13	(0.90–1.42)	1.03	(0.78–1.36)	1.13	(0.85–1.50)
Regular snus use yes/no	1.11	(0.86–1.43)	1.14	(0.88–1.48)	2.49	(0.87–7.13)	1.78	(0.51–6.30)
University education no/yes	1.27	(0.90–1.77)	1.25	(0.89–1.76)	1.54	(1.05–2.25)	1.49	(1.00–2.22)

Figures given are univariate and BMI adjusted odds ratios (OR) with 95% confidence intervals (95% CI) for future type 2 diabetes mellitus (T2DM), stratified for gender.

Diabetes heredity refers to first–degree heredity.

Hypertension was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg and/or taking blood pressure medication.

Physical exercise was defined as excercise in changed clothing once or more per week the last 3 months.

ln, Log_e transformed; HOMA2–B and HOMA2–IR, homeostatic model asessment 2 β-cell function and insulin resistance, respectively; leptin/BMI, leptin per body mass index (BMI unit).

The risk estimates for future diabetes related to quartiles of leptin, adiponectin, and the leptin/adiponectin ratio are presented in Table 3. High leptin levels were associated with an increased risk of T2DM in both men and women. In all multivariable models tested, the increased risk for the upper leptin quartiles remained in men, but not in women, once adjusted for BMI. When BMI was removed from models 2–5 in women, the OR for quartile four was 3.37 (1.84–6.18), 2.48 (1.29–4.78), 3.37 (1.77–6.39), and 2.83 (1.53–5.22), respectively. Adding triglycerides or “snus” to models 2–5 did not alter the risk estimates associated with leptin (data not shown). The risk associated with quartile 4 of leptin was 1.87 (1.03–3.41) in men and 1.56 (0.77–3.19) in women when insulin was used instead of HOMA2-B in model 3. The leptin/BMI ratio gave similar results as for leptin alone in both the univariate and multivariable models (data not shown).

Table 3.

Relative Risk for Future Diabetes According to Quartiles of Leptin, Adiponectin, and Leptin/Adiponectin Stratified for Sex

	Men					Women
	Quartile 1	Quartile 2	Quartile 3	Quartile 4	P_trend	Quartile 1	Quartile 2	Quartile 3	Quartile 4	P_trend
Leptin
Cutoff value (ng/mL)	≤1.7	1.8–3.0	3.0–4.5	≥4.5		≤6.3	6.3–10.1	10.2–14.9	≥14.9
Number of cases/controls	25/196	53/192	93/195	200/192	<0.0005	20/197	49/198	71/197	127/197	<0.0005
Univariate	1.00	1.92 (1.19–3.09)	2.94 (1.88–4.59)	4.78 (3.14–7.28)		1.00	2.23 (1.32–3.76)	3.23 (1.96–5.32)	4.86 (3.01–7.83)
Model 1	1.00	1.59 (0.98–2.59)	2.16 (1.36–3.41)	2.66 (1.67–4.23)		1.00	1.62 (0.95–2.76)	1.79 (1.06–3.03)	1.69 (0.97–2.96)
Model 2	1.00	1.55 (0.86–2.80)	1.99 (1.14–3.46)	2.30 (1.32–4.02)		1.00	1.86 (0.98–3.53)	1.69 (0.88–3.25)	1.81 (0.92–3.56)
Model 3	1.00	1.49 (0.81–2.75)	1.92 (1.08–3.40)	2.15 (1.19–3.87)		1.00	1.77 (0.91–3.46)	1.62 (0.82–3.21)	1.65 (0.81–3.35)
Model 4	1.00	1.39 (0.75–2.56)	1.77 (1.00–3.14)	1.88 (1.03–3.43)		1.00	1.70 (0.87–3.33)	1.56 (0.79–3.08)	1.56 (0.76–3.17)
Model 5	1.00	1.54 (0.86–2.77)	1.88 (1.08–3.25)	2.25 (1.29–3.92)		1.00	1.86 (0.98–3.54)	1.74 (0.91–3.32)	1.76 (0.90–3.45)
Adiponectin
Cutoff value (μg/mL)	≤6.2	6.2–8.8	8.8–12.1	≥12.1		≤9.2	9.3–13.9	14.0–18.4	≥18.4
Number of cases/controls	195/193	101/193	31/193	43/192	<0.0005	149/197	76/198	21/197	21/197	<0.0005
Univariate	1.00	0.69 (0.54–0.88)	0.28 (0.19–0.41)	0.35 (0.25–0.49)		1.00	0.62 (0.47–0.82)	0.24 (0.15–0.37)	0.21 (0.14–0.34)
Model 1	1.00	0.75 (0.58–0.97)	0.35 (0.23–0.51)	0.45 (0.32–0.64)		1.00	0.71 (0.53–0.95)	0.32 (0.20–0.52)	0.33 (0.21–0.54)
Model 2	1.00	0.83 (0.62–1.12)	0.40 (0.25–0.64)	0.52 (0.34–0.81)		1.00	0.70 (0.50–0.98)	0.38 (0.22–0.64)	0.44 (0.25–0.76)
Model 3	1.00	0.86 (0.63–1.17)	0.40 (0.24–0.65)	0.55 (0.36–0.86)		1.00	0.70 (0.49–0.99)	0.39 (0.22–0.67)	0.47 (0.27–0.83)
Model 4	1.00	0.89 (0.65–1.20)	0.41 (0.25–0.68)	0.58 (0.37–0.91)		1.00	0.72 (0.51–1.02)	0.40 (0.23–0.71)	0.50 (0.28–0.88)
Model 5	1.00	0.85 (0.63–1.15)	0.41 (0.26–0.66)	0.53 (0.34–0.83)		1.00	0.69 (0.49–0.97)	0.37 (0.22–0.64)	0.44 (0.25–0.76)
Leptin/adiponectin
Cutoff value (ratio)	≤0.18	0.18–0.34	0.34–0.61	≥0.61		≤0.39	0.39–0.74	0.74–1.38	≥1.38
Number of cases/controls	18/193	41/191	78/195	233/192	<0.0005	12/197	27/198	59/197	168/197	<0.0005
Univariate	1.00	2.14 (1.23–3.74)	3.50 (2.09–5.87)	6.75 (4.16–10.94)		1.00	2.50 (1.26–4.96)	4.38 (2.35–8.17)	9.10 (5.05–16.40)
Model 1	1.00	1.77 (1.01–3.11)	2.55 (1.50–4.34)	4.16 (2.78–6.98)		1.00	2.13 (1.07–4.23)	3.08 (1.63–5.83)	4.80 (2.53–9.10)
Model 2	1.00	1.58 (0.80–3.12)	2.28 (1.22–4.26)	3.08 (1.68–5.67)		1.00	1.79 (0.81–3.99)	2.26 (1.07–4.78)	3.31 (1.56–7.03)
Model 3	1.00	1.41 (0.70–2.84)	2.18 (1.16–4.10)	2.85 (1.53–5.30)		1.00	1.90 (0.82–4.39)	2.34 (1.06–5.15)	3.27 (1.46–7.34)
Model 4	1.00	1.36 (0.68–2.73)	2.03 (1.08–3.83)	2.56 (1.36–4.82)		1.00	1.88 (0.81–4.34)	2.22 (1.01–4.91)	3.02 (1.34–6.81)

Figures given are cutoff values for leptin, adiponectin, and leptin/adiponectin ratio, numbers (cases/controls), odds ratios (95% confidence interval), and test for linear trend (P value).

Hypertension was defined as systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg and/or being on blood pressure medication.

Physical exercise was defined as excercise in changed clothing once or more per week the last 3 months.

Model 1 adjusted for body mass index (BMI).

Model 2 adjusted for BMI, cholesterol, hypertension, regular cigarette smoking, physical activity, university education, first–degree diabetes heredity, fasting, and postload glucose.

Model 3 adjusted as model 2+for log_e transformed HOMA2–B (homeostatic model assesment 2, β-cell function).

Model 4 adjusted as model 2+for log_e transformed HOMA2–IR (homeostatic model assesment 2, insulin resistance).

Model 5 adjusted as model 2+for log_e transformed adiponectin or log_e transformed leptin (for quartiles of leptin and adiponectin, respectively).

High adiponectin levels expressed as quartiles three and four were associated with decreased risk of T2DM in both men and women in all tested models. An increasing leptin/adiponectin ratio was strongly associated with increased risk of T2DM in both men and women. Notably, the ratio remained associated with diabetes in women despite adjustments.

After stratification for insulin sensitivity and adjustment for BMI, log_e-transformed leptin (continuous) remained associated with future T2DM only in insulin- sensitive men (Table 4). By contrast, in both insulin-sensitive and insulin-resistant men and in women, a high adiponectin level decreased the risk while a high leptin/adiponectin ratio increased the risk of T2DM.

Table 4.

Relative Risk for Future Type 2 Diabetes Mellitus According to Leptin, Adiponectin, and Leptin/Adiponectin Stratified for Sex and Insulin Resistance

	Men				Women
	Insulin sensitive		Insulin resistant		Insulin sensitive		Insulin resistant
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
n (cases/referents)	119/563		232/187		86/574		170/190
Leptin ^ln
Univariate	2.06	(1.51–2.79)	1.27	(1.01–1.60)	2.17	(1.49–3.17)	1.36	(0.96–1.92)
BMI adjusted	1.56	(1.13–2.17)	1.03	(0.76–1.39)	1.12	(0.68–1.86)	0.82	(0.54–1.23)
BMI and adiponectin ^ln adjusted	1.63	(1.16–2.28)	1.03	(0.76–1.39)	1.06	(0.64–1.74)	0.93	(0.62–1.40)
Adiponectin ^ln
Univariate	0.45	(0.32–0.63)	0.70	(0.55–0.90)	0.39	(0.28–0.54)	0.54	(0.40–0.73)
BMI adjusted	0.46	(0.32–0.66)	0.71	(0.56–0.92)	0.47	(0.33–0.65)	0.59	(0.43–0.81)
BMI and leptin ^ln adjusted	0.45	(0.32–0.65)	0.71	(0.56–0.92)	0.47	(0.34–0.66)	0.60	(0.44–0.82)
Leptin/adiponectin ratio ^ln
Univariate	2.13	(1.68–2.70)	1.33	(1.12–1.57)	2.16	(1.63–2.50)	1.66	(1.31–2.10)
BMI adjusted	1.88	(1.45–2.42)	1.23	(1.02–1.50)	1.70	(1.30–2.23)	1.37	(1.04–1.80)

Figures shown are univariate and adjusted odds ratios (OR) with 95% confidence interval (95% CI) for type 2 diabetes mellitus (T2DM) stratified for sex and insulin resistance.

Insulin sensitivity and resistance were defined by the distribution of homeostasis model assessment of insulin resistance (HOMA2–IR) among controls with quartiles 1–3 defined as insulin sensitive, and quartile 4 as insulin resistant.

Cutoffs for insulin resitance were for men 1.35 and for women 1.27, respectively.

BMI, body mass index; ln, Log_e transformed.

At each level of leptin, decreasing levels of adiponectin increased the risk for future diabetes in both men and women (Fig. 1). The unadjusted synergy index for the combination of high leptin and low adiponectin was 1.24 (0.88–1.74) in men and 1.45 (0.98–2.15) in women, indicating interaction. These estimates were, however, attenuated after adjustment for BMI, being 1.09 (0.72–1.64) and 1.01 (0.62–1.63), respectively.

FIG. 1.

Body mass index (BMI)-adjusted odds ratios (ORs) for incident type 2 diabetes mellitus (T2DM) in men and women for tertiles (T1–T3) of adiponectin versus tertiles of leptin. Cutoffs were based on the distribution among referents stratified for sex for adiponectin in men ≤7.0 and ≥10.9 μg/mL and in women ≤10.7 and ≥16.6 μg/mL, respectively. For leptin in men, they were ≤2.1 and ≥3.9 ng/mL and in women ≤7.6 and ≥12.8 ng/mL, respectively.

The median time between the survey and diagnosis, shown in Fig. 2, was shorter for men with a high leptin/adiponectin ratio compared with those with a low ratio [66 (57–75) and 86 (75–97) months, respectively (log rank 0.046)]. The same tendency, although not significant, was seen in women (log rank 0.154). Dichotomized groups for leptin and for adiponectin did not differ in time to diagnosis (data not shown).

FIG. 2.

Kaplan–Meier curves showing median time from survey to diagnosis of type 2 diabetes mellitus (T2DM) in men and women stratified for a high and low leptin/adiponectin ratio (above and below median in cases only). Cum, cumulative.

Discussion

We report that the risk for future T2DM linked to leptin, adiponectin, and their ratio differs between men and women and between insulin-sensitive and insulin-resistant individuals. Leptin predicts T2DM only in insulin-sensitive men. This has not been shown previously. We also demonstrate that high adiponectin levels were associated with decreased risk of T2DM in both men and women and in both insulin-sensitive and insulin-resistant individuals. In contrast to our findings, a recent study demonstrated that low adiponectin levels predicted T2DM only in insulin-resistant individuals.²² Together these results indicate the possibility of a yet incompletely understood interaction of adipocyte hormones with insulin sensitivity status. Whether this potential difference in action of leptin and adiponectin reflects different pathogenesis leading to T2DM or different time frames within the same pathogenetic pathway needs further investigation.

The divergent results as to whether leptin actually predicts T2DM^{10

–18} can partially be explained by a varying degree of insulin resistance in the different populations. There are other possible explanations. For example, Snijder et al.¹⁶ argue that adjustment for any metabolic factor could be problematic, as they all may be mediators in potential causative pathways, including leptin. Obesity is linked to many causative pathways leading to diabetes and strongly correlates with leptin. To adjust for obesity (e.g., using BMI) may therefore be questionable. In a previous report based on 177 cases also included in this study, we demonstrated that a composite “obesity factor” including leptin was associated with T2DM in both men and women and that this “obesity factor” was strongly correlated to other explanatory factors.¹⁸ Furthermore, leptin levels for a specific level of obesity differ between ethnic groups.³⁰ The use of different adjustment models in different studies also makes comparisons difficult.

An adipo-insular axis for the metabolic communication between the adipose tissue and the β-cell has been postulated, with insulin and leptin as plausible mediators in a bidirectional mode.^8,31 Insulin is the main regulator of leptin production, and the β-cell expresses receptors for leptin, but the effect of leptin on insulin secretion is uncertain, because both inhibitory and stimulatory effects have been demonstrated.⁸ A reduced peripheral insulin sensitivity has also been associated with hyperleptinemia, although presented studies report inconsistent results.⁴ Adiponectin has several antidiabetogenic features, such as acting as a peripheral insulin sensitizer, increasing glucose uptake in muscles, and decreasing gluconeogenesis in the liver.³² Adiponectin lowers glucose levels in diabetic mice,³³ and reduced adiponectin levels precede hyperinsulinemia in Rhesus monkeys, suggesting a causative role for adiponectin.⁵ In addition, similarly to leptin, insulin regulates adiponectin expression and release in adipose tissue.³⁴ Both adiponectin and leptin are related to adiposity and increased free fatty acid levels in blood, which may promote leptin resistance and reduce circulating adiponectin levels.³⁵

For the leptin/adiponectin ratio, where previous studies are inconsistent,^14,17 we demonstrate a strong association with incident T2DM. The ratio remained significantly predictive in both men and women despite adjustments. Men with a high ratio had a shorter time until diagnosis and this could not be demonstrated for the separate hormones. This may indicate a predictive capacity for the ratio beyond that of the separate hormones, for which the predictive capacity only marginally exceeds the predictive ability of traditional risk scores.³⁶

This study has several strengths, such as the large number of well-defined diabetic patients, in combination with extensive data from the baseline examination, the high participation rate, the nested prospective case–referent design, and the long follow-up time. Possible limitations are the identification of incident cases through routine care, not including an OGTT. As this will underreport the incidence, we assume that it may have reduced our ability to detect an effect of the studied variables. Further limitations are the lack of a standardized lipid analysis and the lack of waist measurement, because this was not included in the VIP protocol at this stage. However, it should be noted that leptin was associated with T2DM in Mauritian men despite adjustment for waist.¹¹

To conclude, this large prospective study demonstrates that adiponectin in both men and women and leptin in men are independently associated with future T2DM and that the association differs according to the insulin sensitivity status. We also propose that the leptin/adiponectin ratio may add further prognostic information.

Footnotes

Acknowledgments

We are indebted to the Västerbotten Intervention Programme and the Medical Biobank and the funds supporting them. This study was supported by grants from the Swedish Heart and Lung Foundation, the county councils of Northern Sweden (Visare Norr) and the Faculty of Medicine at Umeå University. Dr. Lilja received support from the Research and Development Unit of Jämtland County Council, Östersund, Sweden. Dr. Söderberg was supported by grants from the Västerbotten County Council (ALF) and Sjölunds Foundation. The authors are grateful to Paul Franks for valuable input and work with the initial data file, and to Margareta Danielsson and Åsa Ågren for assistance in technical matters. We also express our gratitude to the Swedish National Research School in General Practice.

Author Disclosure Statement

No competing financial interests exist.

References

Ferrannini

, Camastra

. Relationship between impaired glucose tolerance, non-insulin-dependent diabetes mellitus and obesity. Eur J Clin Invest, 1998; 28,Suppl 2:3–6; discussion 6–7.

Cameron

, Boyko

, Sicree

et al. Central obesity as a precursor to the metabolic syndrome in the AusDiab study and Mauritius. Obesity (Silver Spring), 2008; 16:2707–2716.

Cameron

, Zimmet

, Söderberg

et al. The metabolic syndrome as a predictor of incident diabetes mellitus in Mauritius. Diabet Med, 2007; 24:1460–1469.

Margetic

, Gazzola

, Pegg

et al. Leptin: A review of its peripheral actions and interactions. Int J Obes Relat Metab Disord, 2002; 26:1407–1433.

Trujillo

, Scherer

. Adiponectin—journey from an adipocyte secretory protein to biomarker of the metabolic syndrome. J Intern Med, 2005; 257:167–175.

Hynes

, Jones

. Leptin and its role in lipid metabolism. Curr Opin Lipidol, 2001; 12:321–327.

Yamauchi

, Kamon

, Minokoshi

et al. Adiponectin stimulates glucose utilization and fatty-acid oxidation by activating AMP-activated protein kinase. Nat Med, 2002; 8:1288–1295.

Kieffer

, Habener

. The adipoinsular axis: Effects of leptin on pancreatic beta-cells. Am J Physiol Endocrinol Metab, 2000; 278:E1–E14.

Matarese

, Sanna

, Lechler

et al. Leptin accelerates autoimmune diabetes in female NOD mice. Diabetes, 2002; 51:1356–1361.

10.

McNeely

, Boyko

, Weigle

et al. Association between baseline plasma leptin levels and subsequent development of diabetes in Japanese Americans. Diabetes Care, 1999; 22:65–70.

11.

Söderberg

, Zimmet

, Tuomilehto

et al. Leptin predicts the development of diabetes in Mauritian men, but not women: A population-based study. Int J Obes (Lond), 1997; 31:1126–1133.

12.

Welsh

, Murray

, Buckley

et al. Leptin predicts diabetes but not cardiovascular disease: results from a large prospective study in an elderly population. Diabetes Care, 2009; 32:308–310.

13.

Wannamethee

, Lowe

, Rumley

et al. Adipokines and risk of type 2 diabetes in older men. Diabetes Care, 2007; 30:1200–1205.

14.

Thorand

, Zierer

, Baumert

et al. Associations between leptin and the leptin/adiponectin ratio and incident type 2 diabetes in middle-aged men and women: Results from the MONICA/KORA Augsburg study 1984–2002. Diabet Med, 2010; 27:1004–1011.

15.

Schmidt

, Duncan

, Vigo

et al.

Leptin and incident type 2 diabetes: Risk or protection?

Diabetologia, 2006; 49:2086–2096.

16.

Snijder

, Dekker

, Bouter

et al. Comment on: Schmidt MI, Duncan BB, Vigo A, et al. (2006) Leptin and incident type 2 diabetes: Tisk or protection? Diabetologia 49:2086–2096. Diabetologia, 2007; 50:234–236; author reply 237–238; discussion 239–240.

17.

Ley

, Harris

, Connelly

et al. Adipokines and incident type 2 diabetes in an Aboriginal Canadian [corrected] population: The Sandy Lake Health and Diabetes Project. Diabetes Care, 2008; 31:1410–1415.

18.

Norberg

, Stenlund

, Lindahl

et al. Components of metabolic syndrome predicting diabetes: No role of inflammation or dyslipidemia. Obesity (Silver Spring), 2007; 15:1875–1885.

19.

Söderberg

, Stegmayr

, Stenlund

et al. Leptin, but not adiponectin, predicts stroke in males. J Intern Med, 2004; 256:128–136.

20.

Meyers

MG.

Jr.

Outstanding scientific award lecture 2010: Deconstructing leptin: From signals to circuits. Diabetes, 2010; 59:2708–2714.

21.

, Shin

, Ding

et al. Adiponectin levels and risk of type 2 diabetes: A systematic review and meta-analysis. JAMA, 2009; 302:179–188.

22.

Hivert

, Sullivan

, Shrader

et al. Insulin resistance influences the association of adiponectin levels with diabetes incidence in two population-based cohorts: The Cooperative Health Research in the Region of Augsburg (KORA) S4/F4 study and the Framingham Offspring Study. Diabetologia, 2011; 54:1019–1024.

23.

Zhuo

, Wang

, Fu

et al. Comparison of adiponectin, leptin and leptin to adiponectin ratio as diagnostic marker for metabolic syndrome in older adults of Chinese major cities. Diabetes Res Clin Pract, 2009; 84:27–33.

24.

Zaletel

, Barlovic

, Prezelj

. Adiponectin-leptin ratio: A useful estimate of insulin resistance in patients with type 2 diabetes. J Endocrinol Invest, 2010; 33:514–518.

25.

Finucane

, Luan

, Wareham

et al. Correlation of the leptin:adiponectin ratio with measures of insulin resistance in non-diabetic individuals. Diabetologia, 2009; 52:2345–2349.

26.

Hallmans

, Ågren

, Johansson

et al. Cardiovascular disease and diabetes in the Northern Sweden Health and Disease Study Cohort—evaluation of risk factors and their interactions. Scand J Public Health, 2003; 31,Suppl 61:18–24.

27.

Norberg

, Wall

, Boman

et al. The Västerbotten Intervention Programme: Background, design and implications. Glob Health Action, 2010; 3:4643.

28.

Franks

, Rolandsson

, Debenham

et al. Replication of the association between variants in WFS1 and risk of type 2 diabetes in European populations. Diabetologia, 2008; 51:458–463.

29.

Hallqvist

, Ahlbom

, Diderichsen

et al. How to evaluate interaction between causes: a review of practices in cardiovascular epidemiology. J Intern Med, 1996; 239:377–382.

30.

Lilja

, Rolandsson

, Shaw

et al. Higher leptin levels in Asian Indians than Creoles and Europids: A potential explanation for increased metabolic risk. Int J Obes (Lond), 2010; 34:878–885.

31.

Seufert

. Leptin effects on pancreatic beta-cell gene expression and function. Diabetes, 2004; 53,Suppl 1:S152–S158.

32.

Sheng

, Yang

. Adiponectin and its association with insulin resistance and type 2 diabetes. J Genet Genomics, 2008; 35:321–326.

33.

Berg

, Combs

, Du

et al. The adipocyte-secreted protein Acrp30 enhances hepatic insulin action. Nat Med, 2001; 7:947–953.

34.

Hajri

, Tao

, Wattacheril

et al. Regulation of adiponectin production by insulin: Interactions with tumor necrosis factor-α and interleukin-6. Am J Physiol Endocrinol Metab, 2011; 300:E350–E360.

35.

Zhang

, Zhang

. Adipose “talks” to distant organs to regulate insulin sensitivity and vascular function. Obesity (Silver Spring), 2010; 18:2071–2076.

36.

Buijsse

, Simmons

, Griffin

et al. Risk assessment tools for identifying individuals at risk of developing type 2 diabetes. Epidemiol Rev, 2011; 33:46–62.