Ethnic Differences in the Relationship Between Adiponectin and Insulin Sensitivity in South Asian and Caucasian Women

  1. Marlene Martin, BA1,
  2. Latha P. Palaniappan, MD, MS2,
  3. Andrea C. Kwan, BA3,
  4. Gerald M. Reaven, MD4 and
  5. Peter D. Reaven, MD5
  1. 1Stanford University School of Medicine, Stanford, California
  2. 2Department of Health Care and Policy Research, Palo Alto Medical Foundation Research Institute, Palo Alto, California
  3. 3University of California, San Diego, California
  4. 4Division of Cardiovascular Medicine, Falk CVRC, Stanford University, Stanford, California
  5. 5Department of Medicine, Carl T. Hayden Veterans Affairs Medical Center, Phoenix, Arizona
  1. Address correspondence and reprint requests to Gerald M. Reaven, MD, Stanford University, Falk CVRC, CV 287, Stanford, California 94305. E-mail: greaven{at}


OBJECTIVE—To assess whether lower adiponectin concentrations in South Asian Indians may be responsible for their greater degree of insulin resistance.

RESEARCH DESIGN AND METHODS—Insulin-mediated glucose uptake and plasma total and high molecular weight (HMW) adiponectin concentrations were quantified in 52 women of South Asian (SA) and Caucasian (CAU) ancestry and compared.

RESULTS—Mean ± SD total (2,965 ± 1,278 vs. 4,235 ± 160 ng/ml) and HMW (1,001 ± 352 vs. 1,591 ± 854 ng/ml) adiponectin were lower in SAs than CAUs (P < 0.005). Insulin-resistant CAUs (CAU-IR) had lower total (2,665 ± 1,040 vs. 5,133 ± 1,086 ng/ml) and HMW (987 ± 479 vs. 1,935 ± 838 ng/ml) adiponectin than insulin-sensitive CAUs (CAU-IS) (P < 0.01), but there were no significant differences between insulin-resistant (SA-IR) and insulin-sensitive (SA-IS) SAs. HMW adiponectin did not differ between SA-IR and CAU-IR, but SA-IS had significantly lower adiponectin concentrations than CAU-IS.

CONCLUSIONS—Insulin resistance status is not associated with significantly lower levels of adiponectin in these SA women, in contrast to the CAU women.

Insulin sensitivity (13) and circulating adiponectin (3,4) concentrations have been reported to be reduced in individuals of South Asian (SA) ancestry, leading to the theory that lower adiponectin values may explain the increased prevalence of insulin resistance in SAs. However, previous cross-sectional studies have not controlled for differences in insulin resistance within and between ethnic groups, and prior comparisons of adiponectin between SAs and Caucasians (CAUs) have not included measurement of circulating high molecular weight (HMW) adiponectin, possibly the multimer most closely related to insulin sensitivity (5).

The current analysis was initiated to address both of these unresolved issues and involved comparing total plasma and HMW adiponectin concentrations and related metabolic variables in volunteers of CAU and SA ancestry stratified into insulin-resistant (IR) and sensitive (IS) subgroups.


The study population consisted of 52 women volunteers of SA (n = 30) and CAU (n = 22) ancestry. Participants were 30–65 years of age, nondiabetic, healthy, and not taking medication known to affect carbohydrate or lipid metabolism. After obtaining informed consent, the participants were admitted to the general clinical research center, and plasma glucose and lipid concentrations were determined after an overnight fast, as described previously (6). Steady-state plasma glucose (SSPG) concentrations were determined by a modified version (7) of the insulin suppression test (8,9). The results of the insulin suppression test were used to stratify volunteers into IR and IS subgroups based on the results of a previous study (10); subjects whose SSPG concentrations were in the upper 40% were classified as IR, and those in the lowest 40% as IS.

Plasma total and HMW adiponectin levels were measured with ELISA (Alpco Diagnostics, Salem, NH) (effective range 0.075–4.8 ng/ml, samples diluted appropriately), and the coefficient of variation was <15%.

Data are expressed as the mean ± SD, and statistical significance (P < 0.05) of comparisons was made by appropriate independent t and χ2 tests. All statistical analyses, including multivariate modeling, were performed using SPSS version 13.0.


Table 1 compares variables in SAs vs. CAUs, as well as variables stratified into IR and IS subgroups. SAs and CAUs were not different in terms of age, BMI, waist circumference (only available on 77% of the SA and 95% of the CAU group, data not shown), or most other variables measured, including SSPG, except for smoking and total cholesterol. Both total and HMW adiponectin concentrations were significantly lower in the SA group, despite having similar BMI and SSPG values.

Within-subgroup comparisons show that SSPG concentration (by selection) was higher in SA-IR than in SA-IS individuals, associated with higher triglycerides and lower HDL cholesterol concentrations. There were no differences between the IR and IS subgroups of SA women in any other variables. Despite the difference in SSPG concentrations, total adiponectin, HMW adiponectin concentrations, and HMW–to–total adiponectin ratio (data not shown) were not significantly lower in SA-IR compared with SA-IS individuals.

The relationship between SSPG and adiponectin concentrations was quite different in CAU individuals. Higher SSPG concentrations in the CAU-IR group were associated with significantly lower total and HMW adiponectin concentrations than in the CAU-IS group. Post hoc power analysis in the SA group suggests that this study had 99% power to detect an adiponectin difference of similar magnitude (2,468 ng/ml) and 80% power to detect half the difference (1,285 ng/ml) between the SA-IR and SA-IS individuals that we observed between the CAU-IR and CAU-IS individuals. We had limited power (36%) in the SA group to detect a quarter (702 ng/ml) of the observed difference in the CAU-IR versus CAU-IS groups.

As BMI varied between the two CAU groups, total and HMW adiponectin concentrations were compared between the eight CAU-IS individuals with the highest BMI values and the eight CAU-IR subjects with the lowest BMI values. When this was done, the BMI values in these subgroups were no longer significantly different (31.6 ± 4.7 vs. 29.0 ± 0.95 kg/m), but the IR group still had significantly lower values for total (2,665 ± 1,040 vs. 5,104 ± 1,402 ng/ml, P = 0.014) and HMW (987 ± 479 vs. 2017 ± 1,012 ng/ml, P = 0.02) adiponectin concentrations. Comparable results were present when subjects were matched on waist circumference instead of BMI.

These results suggest that total and HMW adiponectin concentrations vary as a function of SSPG concentrations in women of CAU ancestry but not in SA women. Total and HMW adiponectin were not significantly different when comparing IR women in the two ethnic groups (P = 0.85), whereas IS women of SA ancestry had significantly lower (P < 0.001) total and HMW adiponectin concentrations when compared with CAU-IS individuals. In multivariate modeling, even after fully adjusting for age, BMI, waist circumference, HDL cholesterol, smoking status, ethnicity, and insulin resistance status, the interaction term for insulin resistance status and ethnicity was a significant independent predictor (P = 0.04) of adiponectin levels. Similar results for these data were seen when insulin resistance was used as a continuous measure.


The data presented are consistent with those of previous studies, demonstrating that adiponectin concentrations are lower in SA individuals (3,4), and show for the first time that this is also true for HMW adiponectin. However, these findings do not necessarily mean that lower total or HMW adiponectin concentration accounts for the increased prevalence of insulin resistance in SA individuals. The difference between the SA and CAU groups occurs because, in contrast to the presence of higher values in CAU-IS compared with -IR women, total and HMW adiponectin concentrations were not higher in IS than IR individuals of SA ancestry. This lack of a difference between the two SA groups occurred despite the fact that the SSPG concentrations were threefold lower in the IS individuals. Thus, the ethnic difference in adiponectin concentrations is due to the fact that adiponectin concentrations remain low in SAs, irrespective of the degree of insulin sensitivity, a finding that does not appear to support the idea that lower adiponectin concentrations are responsible for greater insulin resistance in SA. Finally, the lack of a relationship between insulin sensitivity and circulating adiponectin concentrations in SA women is not unique and is similar to the disassociation between measures of insulin sensitivity and plasma adiponectin concentrations previously noted to occur in smokers (11) and following moderate amounts of weight loss (12), exercise training (13), and administration of cis-retinoic acid (14).

These data indicate that both total and HMW adiponectin concentrations are lower in IR than IS women of CAU ethnicity. Since these values are lower in SA women, irrespective of their degree of insulin sensitivity, it appears that the relationship between insulin resistance and total and HMW adiponectin concentrations is different in SA compared with CAU women. The strength of our findings is the use of a specific accurate method (rather than an estimate based on surrogate markers) to quantify insulin action and measurement of both total and HMW adiponectin.

An obvious weakness is the relatively few subjects enrolled in the study. As noted above, we had limited power (36%) to detect the 702 ng/ml that was actually observed between the SA-IR and -IS groups. However, it should be noted that this difference in adiponectin level (702 ng/ml) between the SA-IR and -IS (16 vs. 14 subjects) groups was considerably smaller than the statistically significant difference observed in adiponectin in the CAU-IR (n = 8) and -IS (n = 14) groups (2,468 ng/ml). Despite the greater number of subjects in the SA group, it is possible that our study was not sufficiently powered to assess the smaller difference noted in this group. However, the results of our multivariate modeling did reveal a significant interaction between ethnicity and insulin resistance status in predicting adiponectin levels, providing additional evidence suggesting that the relationship between insulin resistance and adiponectin varies by ethnicity.

Subject selection may also be considered a limitation. We chose to contrast the extremes of insulin resistance in each ethnic group and thus compared subjects in the highest 40% and lowest 40% of the insulin-resistance distribution. The decision to compare the most insulin resistant and insulin sensitive in each ethnicity allows us to more closely examine the interaction of insulin resistance and ethnicity on adiponectin levels. For this purpose, the most important contrast is between the IR and IS women in each ethnic group, as it is already well known that SAs generally have lower adiponectin levels than CAUs (3,4). Despite these potential limitations, these data suggest that the increased prevalence of insulin resistance in SAs is not explained by lower total and HMW adiponectin concentrations and emphasize that comparisons of adiponectin concentrations between different ethnic groups must be made between individuals of comparable degrees of insulin sensitivity.

Table 1—

Comparison of demographic and metabolic variables in SA and CAU women and IR and IS subgroups of these ethnicities


This research was supported by the general clinical research center at Stanford University (RR000070). L.P.P. received support from a National Institute of Child Health and Human Development K12 Award (5K12HD043452-02)


  • Published ahead of print at on 17 January 2008. DOI: 10.2337/dc07-1781.

    The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

    • Accepted January 4, 2008.
    • Received September 10, 2007.


| Table of Contents

This Article

  1. Diabetes Care vol. 31 no. 4 798-801
  1. All Versions of this Article:
    1. dc07-1781v1
    2. 31/4/798 most recent