The Relationship between Plasma Adiponectin Concentration and Insulin Resistance Is Altered in Smokers

Abstract

Context: Low plasma adiponectin concentrations in smokers may contribute to the adverse consequences that occur in these individuals.

Objective: The objective of the study was to define the relationship among smoking, plasma adiponectin concentrations, insulin resistance, and inflammation.

Design: This was a cross-sectional, observational study with a 2 × 2 factorial design and a prospective longitudinal arm.

Setting: The study was conducted at a general clinical research center.

Participants: Apparently healthy smokers (n = 30) and nonsmokers (n = 30), subdivided into insulin resistant (IR) (n = 15) and insulin sensitive (IS) (n = 15) subgroups participated in the study.

Intervention: Intervention included pioglitazone administration for 3 months to 12 IR smokers and eight IS smokers.

Main Outcome Measures: Measures included fasting plasma adiponectin and C-reactive protein (CRP) concentrations and changes in adiponectin after pioglitazone treatment in IR and IS smokers.

Results: Being either a smoker or having insulin resistance was independently associated with lower adiponectin concentrations (P = 0.046 and 0.001, respectively). The difference in mean adiponectin concentration between smokers and nonsmokers did not depend on the insulin resistance status of the subjects. No difference was detected in average CRP concentrations between smokers and nonsmokers (P = 0.18) and between IR and IS subjects (P = 0.13). CRP concentrations were unrelated to adiponectin in smokers (r = −0.05, P = 0.78) and nonsmokers (r = 0.03, P = 0.86). Finally, pioglitazone treatment increased adiponectin concentrations in both IR (P < 0.001) and IS smokers (P = 0.001).

Conclusions: Plasma adiponectin concentrations are lower in smokers and IR subjects and are unrelated to CRP concentrations. These findings suggest that low levels of adiponectin in smokers may be independent of both insulin resistance and a generalized inflammatory response.

ADIPONECTIN, AN ADIPOCYTE gene product, is secreted in large amounts from adipose tissue and is present in relatively high concentration in the blood. Low circulating concentrations of adiponectin have been associated with obesity, dyslipidemia, essential hypertension, type 2 diabetes, and cardiovascular disease (1–5). It is apparent that the clinical syndromes in which hypoadiponectinemia occurs are all associated with peripheral resistance to insulin-mediated glucose uptake (6). In addition, variations in plasma adiponectin concentration and/or molecular forms have been suggested to modulate insulin sensitivity (2, 7–9).

More recently, plasma adiponectin concentrations have been reported to be low in smokers (10–13). Because results of in vitro studies have indicated that addition of proinflammatory cytokines, such as TNF-α and IL-6, to isolated adipocytes can reduce adiponectin expression (14–16), it is possible that the association between smoking and lower adiponectin concentrations results from inflammation-mediated down-regulation of adiponectin expression in adipose tissue. In support of this notion is evidence of an inverse relationship between markers of inflammation such as C-reactive protein (CRP) with adiponectin (17, 18). In addition, smoking has been identified as a source of reactive oxygen species (19, 20) and is associated with increased levels of inflammatory markers (21–23), further suggesting that smoking-induced inflammation may contribute to lower adiponectin levels in smokers.

On the other hand, because the prevalence of insulin resistance may also be increased in smokers (24, 25), it is not clear whether hypoadiponectinemia in these individuals is due to smoking, and possibly the associated inflammation, or to the coexistence of insulin resistance. The possible contribution of insulin resistance, as estimated by a surrogate indicator, to reduced adiponectin concentrations in smokers was considered in one brief correspondence (12), and the results of multivariate analysis suggested that the relationship between smoking and adiponectin was independent of insulin action. However, we felt that the possible relationship among smoking, hypoadiponectemia, and insulin resistance was important enough to evaluate further, using a specific measure of insulin action, rather than a surrogate estimate. To address these issues, we have in the present study measured adiponectin and CRP concentrations in smokers and nonsmokers, with each group further stratified into insulin-resistant (IR) and insulin-sensitive (IS) subgroups.

If adiponectin is lower in smokers, and this is not simply related to the extent of insulin resistance, identification of therapies that correct this abnormality may be clinically important. In this context, thiazolidinedione compounds have been shown to increase plasma adiponectin concentrations (9, 26), and there is evidence that they can exert a beneficial antiinflammatory effect independent of their ability to enhance insulin sensitivity (27). The ability of such agents to increase adiponectin may be particularly useful to prevent cardiovascular disease in smokers in light of the observation that smokers with low adiponectin concentrations are at greater risk for coronary stenosis than smokers with high concentrations of adiponectin (28). Thus, irrespective of the reasons that adiponectin concentrations are lower in smokers, we thought it would be important to also evaluate the ability of pioglitazone treatment to increase plasma adiponectin concentrations in the two smoking groups.

Subjects and Methods

The study population consisted of 30 smokers and 30 nonsmokers who responded to print advertisements describing our research interest in smoking-associated metabolic abnormalities. Screening inclusion criteria included: healthy individuals between ages 29 and 65 yr; body mass indices (BMIs) between 20 and 35 kg/m²; and no medications known to affect glucose, insulin, or lipoprotein metabolism. In addition, the smokers had to have a history of smoking a minimum of 10 cigarettes per day for at least the past 5 yr. Potential participants were further evaluated by medical history, physical examination, and routine clinical laboratory measurement to exclude individuals with apparent disease, laboratory evidence of type 2 diabetes, anemia, and abnormal liver or kidney function. Renal function was also evaluated by calculating the estimated glomerular filtration rate (GFR) using the Cockcroft-Gault formula (29). Volunteers meeting these eligibility criteria were scheduled for admission to the General Clinical Research Center of the Stanford University Medical Center. The Stanford Human Subjects Committee approved the study, and each subject gave informed consent.

Insulin-mediated glucose disposal was quantified by a modified version (30) of the insulin suppression test as described and validated by our research group (31, 32). After an overnight fast, an iv catheter was placed in each arm of the subject. One arm was used for the simultaneous infusion of octreotide (0.27 μg/m²·min), insulin (32 mU/m²·min), and glucose (267 mg/m²·min) for 180 min, and the other arm was used for collection of blood samples. During the last 30 min of the infusion, blood was sampled at 10-min intervals to measure plasma glucose and insulin concentrations, and the values obtained were averaged to determine the steady-state plasma glucose (SSPG) and insulin concentrations. Because the steady-state plasma insulin concentrations are comparable in all individuals and glucose infusion is identical, the resultant SSPG concentrations provide a direct measure of the ability of insulin to mediate the disposal of a given glucose load; i.e. the higher the SSPG concentration, the more insulin resistant the individual. For the purpose of this study, individuals with a SSPG concentration more than 145 mg/dl were considered IR, whereas those with a SSPG concentration less than 95 mg/dl were classified as IS; and both sets of individuals were included in the study. The cut point for defining IR was based on a prospective study showing that apparently healthy, nonobese individuals with SSPG concentrations greater than 140 mg/dl had a significant increase in the incidence of a number of age-related diseases (33). The cut point for the IS stems from evidence that approximately one third of a large, apparently healthy population had values below this level (34).

Given evidence that administration of thiazolidinedione compounds increased plasma adiponectin concentrations in IR nonsmokers (26), we evaluated the possibility that this intervention would be equally effective in both IR and IS smokers. To accomplish this goal, we enrolled 20 smokers, 12 classified as being IR and eight IS by the criteria outlined above, and treated them for 3 months with pioglitazone. Treatment was initiated with 15 mg/d for 2 wk, increased to 30 mg/d for the next 2 wk, and followed by 45 mg/d for 8 wk. Plasma alanine aminotransferase levels were checked at the end of each month, and the daily dose of pioglitazone increased only in the presence of continued normal liver function. Liver function did not deteriorate in the 20 volunteers studied, and they all completed the treatment period on the full pioglitazone dose.

Plasma glucose and insulin levels were measured as described previously (26). Plasma adiponectin levels were measured with a RIA established by Linco Research, Inc. (St. Charles, MO). This assay has a sensitivity of 0.01 mg/dl and intra- and interassay coefficient of variation of less than 8%. Serum high-sensitivity CRP was measured with a chemiluminescent assay established for use on an Immulite automatic analyzer (Diagnostics Products Corp., Los Angeles, CA). This assay has a sensitivity of 0.01 mg/dl and intra-and interassay coefficients of variation of less than 8%.

Summary statistics are expressed as mean ± sd or number of subjects. Adiponectin and CRP values were log transformed for statistical analyses. Means and the 95% confidence intervals (CIs) of the log-transformed data were calculated, and these values were then back transformed to the original scale. The resultant adiponectin and CRP averages, known as geometric means, are reported along with their 95% CI. One-way ANOVA and Tukey’s post hoc comparison test were used to compare the mean clinical and metabolic variables among the IR smokers, IS smokers, IR nonsmokers, and IS nonsmokers. A χ² test was used to assess the differences in gender distribution among these four subgroups. Two-factor ANOVA was performed to evaluate the main effects of smoking (smoker vs. nonsmoker) and insulin resistance (IR vs. IS) and their interaction on plasma adiponectin and CRP concentrations. Because differences in gender and body fat have been shown to influence adiponectin levels, an analysis of covariance (ANCOVA) was performed to explore the effects of gender and BMI as covariates on plasma adiponectin concentrations in addition to the main effects of smoking and insulin resistance and their interaction. Pearson correlation coefficients were calculated to assess the strength of association between adiponectin and estimated GFR and between adiponectin and CRP concentrations. For the longitudinal pioglitazone treatment study in smokers, differences in baseline characteristics of the IR and IS subjects were compared with unpaired t test, and gender distribution was compared with Fisher’s exact test. The effect of pioglitazone treatment in smokers was evaluated using paired t test.

Results

The study participants were primarily of Caucasian ancestry (n = 44) of whom four were of Hispanic ethnicity; the rest of the volunteers were of Asian (n = 11) and African (n = 5) ancestries. Clinical and metabolic characteristics of the smokers and nonsmokers further divided into IR and IS groups are given in Table 1. All four groups were similar in terms of age, gender distribution, and BMI. The creatinine levels were normal and 1.2 mg/dl or less in all subjects. The estimated GFR of the subjects was 107 ± 19 ml/min per 1.73 m², and there was no significant correlation between adiponectin and the estimated GFR (r = −0.19; P = 0.16). By design, IR individuals, both smokers and nonsmokers, had SSPG concentrations that were approximately three times higher than the values in the two IS groups.

Figure 1A shows adiponectin concentrations in the 30 smokers and the 30 nonsmokers, with the IR and IS individuals in each group identified. The most obvious finding is that the variability of plasma adiponectin concentrations was much greater in the nonsmokers. For example, adiponectin concentrations were 20.9 μg/ml or less in all smokers, whereas that value was exceeded in 20% of nonsmokers. Consistent with prior reports (10–13), the plasma adiponectin concentrations [mean (95% CI)] were lower in smokers [8.6 μg/ml (6.9–10.8)] than nonsmokers [11.7 μg/ml (9.2–15.0)]. Adiponectin levels were also lower in IR individuals [7.8 μg/ml (6.2–9.7)], compared with those who were IS [13.0 μg/ml (10.5–16.2)]. Differences in means among these groups were compared using two-factor ANOVA with interaction. The results showed that smokers had significantly lower (F = 4.2, P = 0.046) average adiponectin concentrations, compared with nonsmokers, after taking into account differences in their insulin resistance status (significant main effect of smoking). IR subjects on average also had significantly lower (F = 11.7, P = 0.001) adiponectin concentrations, compared with IS subjects, after controlling for differences in their smoking status (significant main effect of insulin resistance). Moreover, the differences in mean adiponectin concentrations between smokers and nonsmokers did not depend on the insulin resistance status (F = 0.46, P = 0.50) of the subjects (nonsignificant smoking and insulin resistance interaction effect). When gender and BMI were included as covariates in the ANCOVA, the main effects of smoking and insulin resistance on adiponectin levels remained significant (P = 0.037 and 0.008, respectively), whereas the effects of BMI and the interaction between insulin resistance and smoking were not. This finding indicated that after adjustment for differences in gender, BMI, and insulin resistance status, the mean adiponectin levels continued to be significantly lower in smokers. Furthermore, this analysis confirmed that on average women had higher adiponectin levels, compared with men (P = 0.002), as previously reported (35).

Figure 1B displays the CRP concentrations in smokers and nonsmokers, again subdivided based on their classification as IR or IS. In contrast to the data shown in Fig. 1A, the variability in the individual values seemed comparable in nonsmokers and smokers. The CRP values [mean (95% CI)] were somewhat higher in the smokers [1.61 μg/ml (1.22–2.12)] than nonsmokers [1.19 μg/ml (0.82–1.72)], and there was also a trend toward higher values among the IR subjects [1.65 μg/ml (1.15–2.36)], compared with those who were IS [(1.16 μg/ml (0.87–1.54)]. Two-factor ANOVA showed that the average CRP concentrations were not significantly different between the smokers and nonsmokers (F = 1.8, P = 0.18), even after controlling for differences in their insulin resistance status. There were also no significant differences in average CRP concentrations between the IR and IS subjects despite dissimilarity in their smoking status (F = 2.4, P = 0.13). Finally, there was no significant effect (F = 0.01, P = 0.92) of an interaction between smoking and insulin resistance on CRP levels.

The contribution of inflammation toward adiponectin levels was evaluated by examining the correlation coefficients between CRP and adiponectin concentrations. The results of this analysis showed there was no significant relationship between adiponectin and CRP levels among smokers (r = −0.05, P = 0.78), nonsmokers (r = 0.03, P = 0.86), or the whole group (r = −0.04, P = 0.74) of subjects.

The results in Fig. 2 display the effects of administering pioglitazone to the two groups of smokers. By selection, pretreatment SSPG concentrations in the IR and IS smokers were quite different (198 ± 42 vs. 79 ± 13 mg/dl, P < 0.001). However, there were no differences in the age (50 ± 10 vs. 51 ± 5 yr, P = 0.90), gender distribution (female/male, 4/8 vs. 6/2; P = 0.17), or BMI (29.3 ± 3.6 vs. 26.7 ± 4.1 kg/m², P = 0.15) of the two groups. After pioglitazone therapy, SSPG significantly decreased to 142 ± 70 mg/dl (P = 0.001) in the IR smokers but remained unchanged in the IS smokers (78 ± 30 mg/dl, P = 0.99). Plasma adiponectin levels [mean (95% CI)] increased by 9.5 μg/ml (5.5–16.2, P < 0.001) in IR smokers and 10.0 μg/ml (4.0–25.0, P = 0.001) in IS smokers. It should also be noted that the average posttreatment adiponectin concentrations in both IR smokers [14.7 μg/ml (8.9–24.3)] and IS smokers [22.7 μg/ml (13.7–37.6)] were similar to or greater than the value in IS nonsmokers [15.9 μg/ml (11.2–22.6)].

Discussion

Although not a major goal of this study, we demonstrated that the IR individuals have significantly lower plasma adiponectin concentrations than their IS counterparts, a finding consistent with previous publications (2, 7, 36). Because there is evidence suggesting that the prevalence of insulin resistance is increased in smokers (24, 25), it was possible that reports of lower adiponectin concentrations in smokers (10–13) was not related to smoking, per se, but to the concomitant presence of insulin resistance in smokers. The results of the current study argue against this possibility and provide evidence that plasma adiponectin levels are on the average lower in smokers and demonstrate less interindividual variability than in nonsmokers.

At the simplest level, our results are consistent with previous reports that plasma adiponectin concentrations are lower in cigarette smokers, compared with nonsmokers (10–13). In contrast to these earlier studies, we quantified insulin-mediated glucose disposal with the insulin suppression test (30–32), a method shown to provide essentially identical values for insulin action as the euglycemic, hyperinsulinemic clamp technique (32), thus allowing us to account more precisely for the effect of insulin resistance when assessing the role of cigarette smoking. The fact that mean adiponectin concentrations in smokers were lower by 26% in the present study, a decrease similar to the 20% decrement in adiponectin concentrations in the 311 smokers studied by Iwashima et al. (11), lends further support for the notion that plasma adiponcentin concentrations are lower in smokers.

Our data provide substantial evidence that differences in the degree of insulin sensitivity have a powerful impact on plasma adiponectin concentrations with the levels in IS individuals essentially 1.5-fold higher than those in IR subjects. Smoking also appears to have a distinct negative effect on plasma adiponectin concentrations because these levels were lower in smokers than nonsmokers. Furthermore, the lower plasma adiponectin concentrations in smokers were not dependent on the insulin resistance status of the subjects. This indicates that the effect of smoking independently attenuates the differences in plasma adiponectin concentrations normally present between IR and IS individuals. Finally, when we evaluated the effects of gender and BMI on adiponectin in ANCOVA, the mean adiponectin levels continued to be significantly lower in smokers, compared with nonsmokers, even after controlling for differences in gender, BMI, and insulin resistance status.

In contrast to the impact of smoking on plasma adiponectin concentrations, we could not discern an adverse effect of smoking on CRP concentration. Thus, although the CRP concentrations in smokers were somewhat higher than in nonsmokers, the differences were not statistically significant, even after controlling for differences in their insulin resistance status. Moreover, there was no relationship between concentrations of CRP and adiponectin in smokers. Although these data suggest that lower adiponectin concentrations in IR and IS smokers are not a consequence of greater systemic inflammation, we cannot rule out the possibility that a relationship between CRP and adiponectin concentrations might have emerged if our study population had been larger. Furthermore, we recognize that other inflammatory factors such as TNF-α and IL-6 have been associated with lower adiponectin levels (37, 38) and may directly reduce adiponectin secretion (14–16); however, our data would appear to indicate that if these factors are indeed relevant, they may be operating not through systemic inflammation but via local induction of adipose tissue inflammation.

Because cigarette smoke contains thousands of potentially bioactive constituents, including free radicals, it is quite possible that one or more of these factors may lower adiponectin production or release from adipocytes. For example, several studies demonstrated that nicotine, a major component of cigarette smoke, promotes inflammation and appears to have direct effects on adipose tissue, inducing adipose tissue lipolysis via enhanced release of catecholamines (39–42). Consistent with this, addition of nicotine (as well as hydrogen peroxide) to 3T3-L1 adipocytes reduced expression of adiponectin in a dose-dependent fashion (11).

Finally, the result of administrating pioglitazone to smokers once again shows that thiazolidinedione compounds will enhance insulin sensitivity when given to nondiabetic, insulin-resistant individuals (26, 27), even if, as in this instance, they continue to smoke. Not surprisingly, there was no significant change in SSPG concentrations in the IS smokers. In contrast, plasma adiponectin concentrations increased significantly in both IS and IR smokers, indicating that this therapy, whether by antiinflammatory effects on adipose tissue or direct regulation of adiponectin gene expression (43), can overcome the detrimental effects of smoking on adiponectin concentrations.

In summary, at a pathophysiological level, the data presented support the concept that adiponectin concentrations are lower in smokers (10–13), and this phenomenon does not seem to be due to the concomitant presence of insulin resistance. These data, and other examples of discordance between adiponectin concentrations and extent of IR (26, 44–47), must be kept in mind when considering adiponectin as a marker of IR. At a clinical level, it is obvious that smoking cessation is the most effective way to overcome the harmful effects of smoking. On the other hand, there are individuals who are either unwilling or unable to stop smoking, and our findings raise the possibility that thiazolidinedione administration may be of some clinical benefit in these situations. This may be particularly important if future studies confirm recent findings indicating that individuals who smoke and have low adiponectin levels are at a substantially greater risk for cardiovascular disease than individuals with either of these risk factors alone (28).

Acknowledgments

This work was supported by the Tobacco-Related Disease Research Program (12RT-0159), the National Insitutes of Health/Stanford Vascular Biology and Medicine Training Grant (5 T32 HL07708), Grant RR-00070 from the National Institutes of Health, and resources and use of the facilities at the Carl T. Hayden Veterans Affairs Medical Center (Phoenix, Arizona).

Disclosure statement: The authors have no potential conflicts of interest to disclose.

Abbreviations:

• ANCOVA,

  Analysis of covariance;

• BMI,

  body mass index;

• CI,

  confidence interval;

• CRP,

  C-reactive protein;

• GFR,

  glomerular filtration rate;

• IR,

  insulin resistant;

• IS,

  insulin sensitive;

• SSPG,

  steady-state plasma glucose.