Adipokines, proteins produced by adipose tissue, may regulate systemic inflammation and play a role in asthma (Sood et al., 2006, 2008). Adiponectin is one such adipokine with predominantly anti-inflammatory effects (Kumada et al., 2004; Wulster-Radcliffe et al., 2004). Recent studies suggest that adiponectin may also have potential pro-inflammatory effects, as in the synovial tissue of patients with arthritis where it induces interleukin (IL)-6 and matrix metalloproteinase-1 secretion (Ehling et al., 2006).

Although visceral adipocytes are the most important source of adiponectin, serum adiponectin concentrations are reduced in obese subjects. This may be explained by hypoxia-related necrosis of adipocytes in obese subjects, resulting in activation of macrophages (Cinti et al., 2005). These activated macrophages produce tumor necrosis factor-alpha (TNF-α) and IL-6 which in turn may inhibit the local production of adiponectin in a paracrine fashion (Bruun et al., 2003). Adiponectin and all of its known receptors are expressed in the lung (Miller et al., 2009). Adiponectin is primarily transported from blood into the alveolar lining fluid via the T-cadherin molecule on the endothelium (Hug et al., 2004).

The causative role of low adiponectin in lung disease has been better established in mice than in humans. A study of lean (BALB/cJ) mice of both sexes suggests that exogenous administration of adiponectin attenuates allergen-induced airway hyperreactivity and eosinophilic influx (Figure 1; Shore et al., 2006). Genetically induced adiponectin deficiency in the lungs of normal-weight mice results in increased expression of TNF-α in alveolar macrophages (Summer et al., 2008). Although human data remain inconclusive, a cross-sectional study shows that higher serum adiponectin concentrations may be associated with decreased risk for asthma in women (Sood et al., 2008). It is however not known if serum adiponectin affects asthma control/severity outcomes in adults.

The objective of this study was to assess whether serum adiponectin was associated with asthma control/severity outcomes in men and women separately. If systemic adiponectin affects asthma control/severity outcomes, measures that modify it may lead to newer therapies to treat subjects with asthma.

In this study, higher serum adiponectin concentrations may be associated with adverse clinical outcomes of disease only among men with asthma. Additionally, BMI may be associated with adverse clinical outcomes of disease only among women with asthma. The association of serum adiponectin with adverse clinical outcomes of asthma in men is interestingly in contrast to our previously reported findings that serum adiponectin may be associated with decreased prevalence of asthma in women (Sood et al., 2008) as well as to murine studies showing attenuation of airway hyperreactivity and eosinophilic influx by adiponectin (Shore et al., 2006). Taking these studies together, our data suggest that the mechanisms of the associations between obesity, adipokines and asthma are complex and are different between men and women. Further, the findings from the mouse model (Shore et al., 2006) may not be easily translated to humans and particularly men.

We confirm our cross-sectional results in Table 2 by performing additional longitudinal analyses in Table 3 that suggest that greater serum adiponectin concentrations may worsen the clinical outcomes of asthma among men and not vice versa. Further confirmation of the temporal sequence was provided by our previously published study that showed that serum adiponectin concentrations were unaffected by inhaled allergen-induced bronchoprovocation in sensitized subjects with asthma (Sood et al., 2009). Further, we did not find either prevalent asthma or inhaled corticosteroid use at year 10 examination in the CARDIA study to predict serum adiponectin concentration at year 15. This suggests that there are no acute or chronic effects of asthma on serum adiponectin concentrations.

Although counter-intuitive, there are various explanations for the possibly adverse effects of serum adiponectin on clinical asthma outcomes in men. There may be a gender-related difference in asthma phenotype that may explain our findings. There may be a threshold above which serum adiponectin may have no effect on asthma – concentrations in men usually fall below that threshold and thus demonstrate an association whereas those in women usually fall above that threshold and demonstrate no association. Of note, our graphic exploration of data did not however suggest such a non-linear association. Further, systemic testosterone may either blunt the anti-inflammatory airway effect of adiponectin or augment its little-known pro-inflammatory effect. The high-molecular weight (HMW) isoform of systemic adiponectin may be the beneficial form for asthma, which is disproportionately reduced in men as compared to women (Page et al., 2005; Andersen et al., 2007). Finally, lungs in men may have receptor-mediated resistance to the beneficial effects of adiponectin. It should be noted that the sex-related differences in adiponectin’s anti-inflammatory function have been similarly described in previous studies of markers of cardiovascular disease (Lawlor et al., 2005; Onat et al., 2008). If it is true that greater serum adiponectin concentrations worsen clinical control/severity of disease among men with asthma, pharmacologic measures to “blindly” increase concentrations of all isoforms of serum adiponectin in future clinical trials may in fact be harmful among men with asthma.

Interestingly, a recent study of 14-year old boys by Kattan et al. (2010) showed that greater systemic adiponectin was associated with better asthma control, opposite of what we describe in middle-aged men. It is possible that these conflicting results may be explained by testosterone-related change in adiponectin isoform distribution (toward a lower proportion of HMW isoform) that occurs in men (Page et al., 2005; Andersen et al., 2007). Recent studies suggest that the HMW isoform is the most biologically active form of adiponectin in regulating insulin resistance (Hara et al., 2006). Whether the same is true for asthma is not known but the airway preponderance of the HMW isoform in mice suggests that may be the case (Zhu et al., 2010). The study by Kattan et al. (2010) also differed from ours in its use of clinic-based recruitment; exclusion of mild asthma; and lack of multivariable analyses and of longitudinal pulmonary function data.

A New Zealand-based birth cohort study (Sutherland et al., 2009b) showed that serum adiponectin concentrations were negatively correlated with fractional exhaled nitric oxide levels among men (−0.16; 95% CI − 0.29, −0.04; p = 0.009) in an adjusted analysis. Of note, these data included both men with and without asthma diagnosis. Our study, on the other hand, was restricted to those with asthma diagnosis alone and did not measure exhaled nitric oxide levels. Further, Sutherland’s unadjusted analysis showed that high serum adiponectin concentrations were associated with greater, not lower odds for prevalent reversible airflow obstruction among men. The latter data suggest a pro-inflammatory effect of adiponectin in men and contradict the exhaled nitric oxide data which suggest an anti-inflammatory effect of adiponectin. Finally, unlike our study, Sutherland et al. did not report significant sex-specific interactions for asthma-related outcomes.

Additionally, since adiponectin inhibits proliferation of vascular smooth muscle cells (Ouchi et al., 1999; Kondo et al., 2002; Okamoto et al., 2002), it has been hypothesized that adiponectin may inhibit airway smooth muscle proliferation (Shore, 2006), thus improving airway remodeling in chronic asthma. Our study however did not find any significant effect of serum adiponectin on time-dependent decrease in FEV₁. We therefore conclude that it is unlikely that systemic adiponectin may improve in vivo airway remodeling in human asthma. This is consistent with recent reports that adiponectin did not affect airway smooth muscle proliferation in mice or in vitro human experiments (Shin et al., 2008; Medoff et al., 2009).

Prior studies of the association between obesity and asthma outcomes have yielded conflicting results (Tantisira et al., 2003; The Enfumosa Study Group, 2003; Wenzel and Busse, 2007; Mosen et al., 2008; Taylor et al., 2008; Vortmann and Eisner, 2008; Sutherland et al., 2009a). Our study demonstrates that obesity is associated with adverse clinical outcomes of asthma but additionally report these findings only in women. We further show that serum adiponectin does not explain away the effect of obesity on clinical outcomes of asthma in women, suggesting that the effect of obesity may be mediated via multiple pathways including both mechanical and inflammatory.

The strengths of our study include its sex-specific stratified analysis, well-defined study population set within a cohort structure, and its clinical translational character, based on the recently published data on the role of systemic adiponectin in mouse and human asthma (Shore, 2006; Shore et al., 2006; Sood et al., 2008; Nagel et al., 2009). Further, the results from our cross-sectional analyses are supported by our longitudinal analyses. Another unique feature of our study is its community-based, rather than clinic-based recruitment.

The study however, has several limitations. One significant limitation is the use of self-report of provider-diagnosed asthma. However, self-report is the standard definition used in epidemiological studies, despite the misclassification bias relative to the “gold standard” of airway hyperresponsiveness (Aaron et al., 2008). Although asthma misclassification has not been studied with respect to adiponectin, this misclassification is non-differential between obese and non-obese subjects (Appleton et al., 2006; Aaron et al., 2008; Lang et al., 2009). With non-differential misclassification errors, the bias is typically in a predictable direction (i.e., toward lack of association) and unlikely to produce a spurious effect. Additionally, self-report may miss subjects with mild asthma severity. However, this is unlikely, given that most subjects in our community-based study were of mild intermittent severity, as suggested by their relatively low rates of inhaled corticosteroid use shown in Table 1 (National Heart Lung and Blood Institute, 2007a). This distribution is however consistent with another community-based study where intermittent disease was also the most prevalent asthma type (Fuhlbrigge et al., 2002). Not surprisingly, the proportion of subjects with moderate or severe asthma in our study was smaller than most clinic-based studies – this may affect the generalizability of our findings. We cannot entirely rule out the possibility of the misdiagnosis of chronic obstructive pulmonary disease (COPD) subjects as asthma in our study. However, this possibility is less likely given the relatively young age of subjects (40.2 ± 3.6 years); high proportion of never smokers (59.3%); and rare use of anti-muscarinic inhaler drugs (two subjects using ipratropium and another four using ipratropium–albuterol combination) at CARDIA year 15 examination as well as the high levels of year 10 mean prebronchodilator FEV₁/FVC ratio (76.8 ± 8.1%). Further, we do adjust for smoking behavior, the most important risk factor for COPD, without affecting our results. Further, current definitions of asthma control/severity are less than ideal, particularly for population research (Ungar et al., 2002; Reddel et al., 2009). Although it is reasonable to study medication use as a measure of asthma control/severity (Cockcroft and Swystun, 1996), our assumption that the need for asthma medication can be equated with report of taking the medication may not be entirely true. The lowest dose of maintenance inhaled corticosteroids required after asthma control is achieved, a key measure in assessing clinical disease severity among treated patients (National Heart Lung and Blood Institute, 2007a,b), was not available in this population study. Additionally, data assessing symptomatic asthma control using standardized asthma questionnaires were not collected in this population study. On the other hand, we analyzed a number of respiratory symptoms that may be relevant for several cardiopulmonary conditions including asthma and systemic conditions including obesity. It should nevertheless be pointed out that there were no cases with clinical atherosclerotic cardiovascular disease among the 411 subjects studied at year 15 and that we adjusted for obesity in our models. Longitudinal analysis of the effect of year 15 serum adiponectin on subsequent change in lung function requires the baseline lung function to be assessed at year 15. Unfortunately, spirometry was not performed at CARDIA year 15 examination and therefore spirometry at year 10 was instead used as baseline (Figure 2).

Selection bias may occur if subjects with adiponectin measurements were not representative of the total asthma population. However, we found that subjects with asthma with or without adiponectin measurements did not differ with respect to either obesity or asthma outcomes measures (data not shown). Further, data on atopy were not collected at CARDIA year 15 examination. However, adjustment for self-reported atopic status at year 0 examination did not affect the results. This study did not measure the serum concentrations of various isoforms of adiponectin, other adipokines, or reproductive hormones. It is possible that the various isoforms have varying in vivo activity for asthma. It is also not clear whether there is a correlation between serum and airway concentrations of total adiponectin and its isoforms. Further, a single assessment of a biomarker may be susceptible to short-term variation and may not reflect long-term exposure. However, studies suggest that serum concentrations of adiponectin are stable (Pischon et al., 2003; Lukanova et al., 2006; Kaplan et al., 2007) and therefore, it is a good candidate for long-term risk assessment. Our data in Table 2 are cross-sectional which limits our ability to infer causality. However, the cross-sectional findings are supported by longitudinal data in Table 3 that allow for inference of the direction of association. It is still however premature to conclude a causal association between serum adiponectin and asthma control/severity.

In summary, our study of predominantly mild asthma from a community-based middle-aged population shows that systemic adiponectin concentrations may have opposing associations with clinical control/severity outcomes of asthma between men and women. Among men with asthma, higher systemic adiponectin concentrations may adversely affect clinical outcomes of disease. Future studies involving the modification of systemic adiponectin concentrations may open up newer ways to treat asthma in men.

Authors have no relationships financial, commercial or otherwise that might be perceived by the academic community as representing a potential conflict of interest. The research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.