Obesity, adipokines and asthma

Authors


Dr Tuomas Jartti
Department of Pediatrics
Turku University Hospital
PO Box 52
FIN-20520 Turku
Finland

Abstract

Background:  The prevalence of asthma and obesity is increasing concomitantly, but many aspects of this link are unclear. Our objective was to examine whether obesity is associated with asthma in three time points of life, and whether immunomodulatory adipokines, leptin and adiponectin are linked to overweight-associated asthma.

Methods:  We studied the association between obesity and asthma at ages 3–18 years [mean (SD), 10 years (5), = 3582, year 1980], 9–24 years [16 years (5), = 2764, 1986] and 24–39 years [32 years (5), = 2620, 2001] in a prospective cohort study and further tested for associations with serum leptin and adiponectin concentrations. Data on allergy status, smoking and other laboratory values (serum insulin, plasma C-reactive protein and serum lipid values) were also analyzed.

Results:  Allergy and parental asthma were significantly associated with asthma at all ages. At ages 24–39 years, but not earlier, body mass index (BMI) (odds ratio, OR 1.05; = 0.019) and female gender (OR 1.56; = 0.031) were independently associated with asthma. Increase in BMI was also associated with incident asthma during adulthood (OR 1.08; = 0.030). Levels of leptin, adiponectin or any other obesity-related biomarker were not independently associated with asthma.

Conclusions:  Asthma is linked with obesity in adults, but our results do not support a significant role for leptin, adiponectin or any other obesity-related biomarker studied in this association. Other factors should be sought for better understanding the connection between obesity and asthma.

Epidemiological studies have shown that the prevalence of asthma and obesity is increasing concomitantly (1). Moreover, obesity has been associated with measures of asthma severity (2). The reasons for these associations are unknown. Suggested mechanisms include sedentary lifestyle, dietary factors, systemic inflammation and reduced chest wall compliance by obesity, insulin resistance, co-morbidities and common genetic predisposition (3–7). Other possible mechanisms include changes in immune function by obesity-associated adipokines, such as leptin and adiponectin (3, 4).

Adipose tissue propagates inflammation both locally and systemically. Leptin, which acts on the hypothalamus to induce satiety and to increase metabolism, has proinflammatory properties. Leptin could aggravate asthma by increasing proliferative responses of CD4+ T cells, activation of mast cells, activating transcription factors AP-1 and nuclear factor κB, or by boosting interferon gamma responses (8–10). In the obese, serum leptin concentrations are markedly increased probably due to leptin resistance and hypertrophic and hyperplastic adipocytes (3, 11). In contrast to leptin, plasma adiponectin levels are decreased in subjects with obesity, and levels increase after weight loss (3). Adiponectin is one of the most abundant gene products of adipose tissue and induces the anti-inflammatory cytokine IL-10 and the endogenous IL-1 receptor antagonist, inhibits nuclear factor κB signaling, expression of the adhesion molecules, as well as proliferation and migration of cultured vascular smooth muscle cells (12–17). Thus, leptin and adiponectin could have a role in the pathogenesis of asthma, as supported by animal studies (10, 18, 19). There are, however, only few studies in humans, and these studies have not yet convincingly established the link between these adipokines and asthma (20–25).

Here, we report findings from the Young Finns Study in which subjects were followed up to 21 years from childhood to adulthood (26, 27). We first examined in this cohort at three successive time points (1980, 1986 and 2001) whether obesity is independently associated with asthma when the subjects were 3–18 years, 9–24 years and 24–39 years old respectively. Second, we examined whether the adipokines leptin and adiponectin are associated with overweight and asthma. The data were analyzed with respect to allergy status, parental asthma, smoking habits and other obesity-associated biomarkers, i.e. serum insulin, plasma C-reactive protein (CRP) and serum lipids.

Methods

Subjects and study design

The Cardiovascular Risk in Young Finns Study is an ongoing five-center follow-up study of atherosclerosis risk factors in Finnish children and adolescents described in detail elsewhere (26, 27). The individuals were randomly chosen from the population register of the Social Insurance Institution of Finland. The original sample size was 4320 children and adolescents aged 3, 6, 9, 12, 15 or 18 years. The first cross-sectional survey of this cohort was conducted in 1980; 3596 subjects with available data (83% of those invited) participated at that time. Further follow-up studies were conducted 3 years apart, in 1983 (data not reported here) and 1986 with 2799 participants. The last re-examination of the subjects took place in 2001 when the individuals (2624 participated) had reached an age between 24 and 39 years. The loss of participants was 22% and 27% after 6 and 21 years respectively.

Here, we report the association of obesity and asthma at ages 3–18 years in 1980 (= 3582), 9–24 years in 1986 (= 2764) and 24–39 years (= 2620) in 2001 in this prospective cohort study. All subjects with information on the presence or absence of asthma were included in this analysis. Further, we report the association of asthma with serum leptin and adiponectin concentrations, as well as with allergy status, smoking habits and other obesity-related biomarkers (serum insulin, CRP and lipid values). Participants gave written informed consent, and the study was approved by local ethics committees.

Clinical data

The data on current physician-diagnosed asthma and self-reported allergies and parental asthma were based on a parental interview in participants aged ≤18 years and a questionnaire in participants aged >18 years in each cross-sectional survey. Body mass index (BMI) was calculated as participants’ weight in kg divided by the square of their height in meters. Normal weight for subjects aged ≤18 years was defined according to BMI cut-offs introduced by Cole et al. and for those aged >18 years by BMI < 25 kg/m2; for the purpose of this study all subjects above these cut-off points were classified as overweight (28). Weight status at the last time point was also assessed by waist circumference as a continuous variable. Smoking habits were determined using a questionnaire. Subjects smoking on a daily basis at the time of each cross-sectional survey (asked if aged ≥12 years) were defined as smokers.

Biochemical analyses

Serum leptin and adiponectin concentrations were analyzed with a radioimmunoassay (Human Leptin and Adiponectin RIA kits respectively; Linco Research, Inc, MO, USA; interassay coefficient of variations, respectively, 2.5–9.2% and 10.8–14.97%). The fasting plasma CRP concentrations were analyzed by a high-sensitive latex turbidometric immunoassay (Wako Chemicals GmbH, Neuss, Germany). Subjects with febrile illness within 2 weeks prior to sampling were excluded from the CRP analysis except in 1980 when such information was not available. Serum insulin was measured with time-resolved fluoroimmunoassay method using Wallac AutoDELFIA analyzer (Wallac, Turku, Finland). Standard enzymatic methods were used for measuring levels of serum total cholesterol, triglycerides, high-density lipoprotein cholesterol, apolipoprotein A-1 and apolipoprotein B. Low-density lipoprotein cholesterol concentration was calculated by using the Friedewald formula (29). Details of these methods have been described previously (30, 31). Venous blood samples were drawn always in the morning after an overnight fast. Menstrual phase was recorded but it was not associated with the levels of obesity-related biomarkers (data not shown).

Statistics

Logistic regression models were used to assess the relationship of obesity, laboratory values and asthma in both univariable and multivariable models as shown. Differences between clinical characteristics in asthma groups were calculated with chi-squared test or Fisher’s exact test as appropriate. anova and Tukey’s multiple comparisons were used to compare overweight/asthma groups. A nominal value of P < 0.05 was regarded statistically significant. The statistical analyses were carried out using sas/stat(r) software, Version 9.1.3 SP4 of the SAS System for Windows; SAS Institute Inc., Cary, NC, USA.

Results

Analysis of dropouts

The representativeness of the present study cohort in 2001 was tested by comparing its baseline (1980) characteristics with that of the dropouts. As the participants where older than dropouts (10.7 vs 10.0 years, < 0.0001), age-adjusted analysis was used. There were no differences in the study variables between dropouts and participants (e.g. the differences for the prevalence of asthma, = 0.74, and for the prevalence of allergy, = 0.59).

Subject characteristics

Clinical and laboratory data at the three different time points during the follow-up are given in Table 1. The prevalence of asthma increased from 1.8% to 4.6% when the subjects became older. Similarly, the mean BMI increased from 17.8 to 25.1 kg/m2 during follow-up. Eight percent of the subjects reported allergy at ages 24–39 years. The prevalence of parental asthma was 4% at each time point of this follow-up cohort study. Twenty-five percent reported smoking during adulthood. Smoking was not associated with asthma whether it was analyzed by smoking on a daily basis or not, current vs ex- vs lifetime nonsmokers or ever vs never smoker (data not shown).

Table 1.   Patient characteristics
 Ages 3–18 years (year 1980)Ages 9–24 years (year 1986)Ages 24–39 years (year 2001)
AllAsthmaNo asthmaAllAsthmaNo asthmaAllAsthmaNo asthma
  1. Data are presented as mean (SD) or no. of cases (%).

  2. *Data available: asthma+ (= 64), asthma− (= 3489) (1980), asthma+ (= 75), asthma− (= 2379) (1986), asthma+ (= 112), asthma− (= 2164) (2001).

  3. †Data available: = 2253 (2001). asthma+ (= 112), asthma− (= 2141) (2001).

  4. ‡Recorded only for subjects aged ≥12 years.

No.358264 (1.8%)3518 (98.2%)2764101 (3.7%)2663 (96.3%)2620121 (4.6%)2499 (95.4%)
Body mass index (kg/m2)*17.8 (3.1)18.4 (3.2)17.8 (3.1)20.0 (3.5)20.2 (3.5)20.0 (3.5)25.1 (4.4)25.9 (5.2)25.0 (4.4)
Waist (cm)†84.1 (12.3)85.7 (12.9)84.1 (12.3)
Mean age (years)10.4 (5.0)11.2 (4.3)10.4 (5.0)16.1 (5.0)16.4 (4.8)16.1 (5.0)31.5 (5.0)31.3 (4.7)31.6 (5.0)
Male (%)1756 (49%)32 (50%)1724 (49%)1301 (47.1%)47 (46.5%)1254 (47.1%)1174 (44.8%)41 (33.9%)1133 (45.3%)
Allergy (%)256 (7.2%)33 (51.6%)223 (6.3%)156 (5.7%)17 (20.5%)139 (5.2%)202 (7.7%)19 (15.7%)183 (7.3%)
Parental asthma (%)146 (4.1%)13 (20.3%)133 (3.8%)122 (4.4%)18 (17.8%)104 (3.9%)109 (4.2%)10 (8.3%)99 (4.0%)
Active smoking (%)‡219/1684 (13%)6/33 (18.2%)213/1651 (12.9%)356/1953 (18.2%)8/60 (13.3%)348/1893 (18.4%)641/2547 (25.2%)22/120 (18.3%)619/2427 (25.5%)

Of the subjects with both asthma and weight status available, 320/3553 (9.0%) subjects at ages 3–18 years and 215/2454 (8.8%) subjects at ages 9–24 years were considered overweight according to study criteria. BMI at ages 24–39 years (= 2276) was distributed as follows: BMI <18.5 kg/m2 (= 46, 2.0%); BMI ≥18.5 but <25 kg/m2 (= 1241, 55%); BMI ≥25 but <30 kg/m2 (= 709, 31%) and BMI ≥ 30 kg/m2 (= 280, 12%). In these BMI categories, the prevalence of asthma was as follows: 2/46 (4.4%), 55/1241 (4.4%), 30/709 (4.2%) and 25/280 (8.9%) respectively.

Association between clinical characteristics and asthma

Allergy and parental asthma were associated with asthma at all ages but less when the whole cohort had reached adulthood (Table 2). However, at ages 24–39 years, also BMI and female gender were independently associated with asthma; these associations were absent at earlier time points (Table 2, stratified analysis by sex is shown in Table S1). Increase in BMI (1986 vs 2001) was also associated with incident asthma at ages 24–39 years (asthma absent in 1980 and 1986, asthma present in 2001, = 55) compared with those without asthma (asthma absent in 1980, 1986 and 2001, = 1669) both in univariable (odds ratio, OR 1.08; 95% confidence interval, CI 1.01–1.15; = 0.030) and multivariable models (full model including gender, age, smoking, allergy and parental asthma, OR 1.08; 95% CI 1.01–1.17; = 0.030). Similar association did not exist when change in BMI between ages 3–18 years and 9–24 years was compared between subjects with first-time asthma diagnosis in the visit at ages 9–24 years (= 18) and subjects without asthma at both occasions (= 2404, = 0.36 in univariable model). Menarche age was not associated with asthma at ages 24–39 years (data not shown).

Table 2.   Association between clinical characteristics and asthma at ages 3–18, 9–24 and 24–39 years
 nUnivariable*Multivariable*†
Odds ratio95% CIPOdds ratio95% CIP
  1. *In the stratified analysis by sex, BMI (odds ratio; 1.05; 95% CI 1.01–1.10; = 0.021), waist circumference (respectively, 1.03; 1.001–1.04; = 0.0060) and parental asthma (2.75; 1.26–6.01; = 0.011) of female subjects at ages 24–39 years were associated with asthma in univariable models. The associations between asthma and BMI (1.06; 1.01–1.11; = 0.0002) and asthma and parental asthma (2.49; 1.06–5.88; = 0.037) persisted in multivariable models (Table S1). In male subjects at ages 24–39 years, allergy was associated with asthma both in univariable (3.58; 1.64–8.27; = 0.0016) and multivariable models (3.61; 1.58–8.28; = 0.0024). Otherwise, no significant differences were found.

  2. †Including body mass index, age, sex, allergy, parental asthma and active smoking (smoking for ages 24–39 years only): = 3553 (ages 3–18 years), = 2453 (ages 9–24 years), = 2191 (ages 24–39 years).

Ages 3–18 years
 Body mass index (kg/m2)35531.050.98–1.130.171.040.95–1.150.50
 Age (years)35821.030.98–1.080.240.980.91–1.050.50
 Male35821.040.64–1.710.870.990.59–1.660.97
 Allergy358215.739.46–26.16<0.000114.768.70–25.06<0.0001
 Parental asthma35826.493.44–12.22<0.00015.082.55–10.10<0.0001
 Active smoking16841.50.61–3.680.38
Ages 9–24 years
 Body mass index (kg/m2)24541.010.95–1.080.741.0040.93–1.090.91
 Age (years)27641.010.97–1.060.511.020.96–1.080.59
 Male27640.980.66–1.460.931.160.72–1.850.55
 Allergy27464.682.67–8.19<0.00014.502.42–8.38<0.0001
 Parental asthma27645.343.09–9.21<0.00015.252.73–9.84<0.0001
 Active smoking19530.680.32–1.450.32
Ages 24–39 years
 Body mass index (kg/m2)22761.041.00–1.080.0501.051.01–1.090.019
 Waist (cm)22531.011.00–1.030.17
 Age (years)26200.990.95–1.0250.541.000.94–1.020.23
 Male26200.620.42–0.910.0140.640.43–0.960.031
 Allergy26202.361.41–3.940.0012.291.35–3.870.0021
 Parental asthma26202.191.11–4.300.0241.970.95–4.090.069
 Active smoking25470.660.41–1.050.0800.720.44–1.190.20

Association between obesity-related biomarkers and asthma

Of the obesity-related biomarkers (leptin, adiponectin, insulin and CRP), only leptin was associated with asthma, both in univariable and multivariable models, at ages 24–39 years but not at earlier time points (= 2279, Table 3, stratified analysis by sex is shown in Table S2). However, when also clinical data were included in the model, the association between leptin and asthma was abolished (= 0.36, Table 4; in backward step-wise elimination of the least significant factors, only BMI, gender and allergy remained significant). Similarly, no association was found at ages 24–39 years when leptin and incident asthma were considered (full model = 0.69; in step-wise elimination model, only BMI, gender and allergy remained significant). Interestingly, leptin levels were associated with allergy status in univariable models (= 2279; OR 1.017; 95% CI 1.003–1.032; = 0.019). The association did not persist in multivariable models including clinical factors (= 2177, = 0.27, data not shown), and there was no association between leptin and asthma in allergic subjects (= 180, = 0.76, data not shown). Leptin/adiponectin ratio was also associated with asthma at ages 24–39 years (= 2279; OR 1.18; 1.07–1.30; = 0.0010), but not at ages 9–24 years (= 574; = 0.99).

Table 3.   Association between obesity-related biomarkers and asthma at ages 3–18, 9–24 and 24–39 years
 nUnivariable*Multivariable*†
Odds ratio95% CIPOdds ratio95% CIP
  1. CRP, C-reactive protein.

  2. *In the stratified analysis by sex, insulin levels were associated with asthma (odds ratio 1.03; 95% CI 1.00–1.06; = 0.018) in female subjects at ages 24–39 years in univariable model, but this association did not persist in multivariable models (online Table S2). No other significant differences were found.

  3. †Including the variables specified in the table for each age category; = 205 (ages 3–18 years), = 574 (ages 9–24 years), = 2165 (ages 24–39 years).

Ages 3–18 years
 Leptin (ng/ml)2641.0520.80–1.380.721.0810.80–1.470.62
 CRP (mg/l)22391.0180.93–1.110.700.7650.27–2.190.62
 Insulin (mU/l)35211.0160.98–1.060.430.9540.81–1.120.56
Ages 9–24 years
 Adiponectin (μg/ml)5740.9430.83–1.080.390.9330.81–1.070.33
 Leptin (ng/ml)5780.9750.87–1.100.671.0050.89–1.140.94
 Insulin (mU/l)24600.9740.93–1.020.270.9420.83–1.070.36
Ages 24–39 years
 Adiponectin (μg/ml)22800.990.95–1.030.580.980.94–1.030.49
 Leptin (ng/ml)22791.021.01–1.040.00951.021.00–1.040.032
 CRP (mg/l)21701.010.96–1.060.790.980.92–1.050.57
 Insulin (mU/l)22821.020.99–1.050.131.010.98–1.050.51
Table 4.   An association between clinical characteristics including leptin levels and asthma at ages 24–39 years in multivariable analysis including all variables (= 2200)
FactorOdds ratio95% CIP *
  1. *In the stratified analysis by sex, leptin levels were not associated with asthma either in female or male subjects (> 0.2, data not shown).

Body mass index (kg/m2)1.071.01–1.130.024
Age (years)0.970.94–1.010.18
Male0.540.31–0.940.030
Allergy2.301.36–3.900.0020
Parental asthma1.970.95–4.100.069
Active smoking0.710.43–1.180.19
Leptin (ng/ml)0.990.96–1.020.36

When the link between leptin and overweight-associated asthma was tested at ages 24–39 years, obese asthmatics had higher leptin levels than obese nonasthmatics in univariable models (= 0.045, median 19.8 vs 11.7 ng/ml, = 1816, Table 5). The difference was, however, mainly due to the preponderance of female subjects among adult asthmatics (Table S3). After adjusting for gender and age both significantly affecting leptin (median serum leptin concentrations, female vs male, 13.4 vs 4.3 ng/ml, < 0.0001; age, r = 0.052, = 0.013; use of oral contraceptives had no effect), the link between leptin levels and overweight-associated asthma did not persist (Table 5). Similar association was found between leptin/adiponectin ratio and overweight-associated asthma (Table 5). None of the lipid values studied (total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein triglycerides, apolipoprotein A-1 and apolipoprotein B) could be linked to asthma at ages 24–39 years even if analyzed by gender (data partly shown in Table S4).

Table 5.   Association between leptin values and overweight-associated asthma at ages 24–39 years
FactorGroupMedian (interquartile range)Univariable*Multivariable*†
Overall PP (between groups)Overall PP (between groups)
ow−/as−ow−/as+ow+/as−ow−/as−ow−/as+ow+/as−
  1. ow, overweight; as, asthma. Only children with normal weight and no asthma in 1986 were included. Grouping according to status in 2001: ow−/as− (= 1003); ow−/as+ (= 34), ow+/as− (= 749) and ow+/as+ (= 30).

  2. *In the stratified analysis by sex, leptin levels or leptin/adiponectin ratios were not associated with overweight-associated asthma either in female or male subjects (> 0.6, data not shown).

  3. †Including gender and age (both significantly affecting leptin levels).

  4. ‡The univariable geometric mean values (95% confidence intervals) for leptin values in groups 3 and 4 are 11.05 (10.42–11.71) and 16.42 ng/ml (12.26–22.00) respectively. In the multivariable model, these values are 12.53 (12.05–13.03) and 12.61 ng/ml (10.40–15.30) respectively. Using a cut-off of ≥30 kg/m2 for obesity, these values are 18.41 (16.43–20.63) and 19.16 ng/ml (12.67–28.12) in univariable models (= 1.00) and 18.97 (17.47–20.60) and 15.62 ng/ml (11.59–21.05) in multivariable models (= 0.61) respectively.

Leptin (ng/ml)ow−/as−6.30 (3.3–11.5)<0.0001   <0.0001   
ow−/as+7.25 (4.5–13.7) 0.63   0.70  
ow+/as−11.7 (5.7–21.0)‡ <0.00010.012  <0.0001<0.0001 
ow+/as+19.8 (8.9–26.0)‡ <0.00010.00030.045 <0.0001<0.00011.00
Leptin (ng/ml)/adiponectin (μg/ml)ow−/as−0.64 (0.36–1.13)<0.0001   <0.0001   
ow−/as+0.80 (0.59–1.66) 0.32   0.35  
ow+/as−1.64 (0.85–2.86) <0.0001<0.0001  <0.0001<0.0001 
ow+/as+2.05 (1.25–3.95) <0.0001<0.00010.050 <0.0001<0.00010.78

Discussion

We studied the link between obesity, obesity-associated biomarkers and asthma in individuals who were followed 21 years from childhood to adulthood. We found that BMI was independently associated with higher prevalence and new diagnoses of asthma in adulthood (at ages 24–39 years) but not in adolescence/young adulthood (at ages 3–18 and 9–24 years). BMI was also independently associated with incident asthma during adulthood. Leptin, adiponectin, insulin and CRP were used as obesity-related biomarkers but only leptin was associated with asthma in adulthood; the association existed in the whole population and in the subgroup of obese subjects. These associations, however, were secondary to the preponderance of female gender among adult-onset asthmatics, and did not persist in multivariable models.

Our results agree with many previous epidemiological studies showing that obesity is associated with asthma in adults (1). In our study, the association was modest, i.e. the risk of asthma increased 20–25% when BMI increased from 25 to 30 kg/m2 in adults. Although many of the previous studies have had methodological limitations, such as cross-sectional design, use of self-reported asthma as the primary outcome and not including information on airway hyper-responsiveness, there are also prospective and weight loss studies that have further supported the link between obesity and asthma in adults (1, 32). Furthermore, the association seems to be stronger in female than in male subjects as also found in our study (25, 33). This association may be related to hormonal effects, excess proportion of fat to overall weight or common genes in female subjects (4, 34). Furthermore, a recent report found that the association between BMI and the severity of asthma was stronger in women with early menarche supporting the view that hormonal factors may be involved (35). We could not, however, corroborate the association between age of menarche and asthma.

Studies in children and adolescents are less numerous and their results less coherent, but support for an association between body weight and future risk of asthma has been found (36). Three prospective studies evaluating the influence of gender on the association of weight status and asthma ended up with opposite results (37–39). Furthermore, timing of obesity during childhood may be a relevant factor, i.e. the earlier the obesity develops, the higher the risk for later development of asthma (39). We did not find support for the link between obesity and asthma in school-aged children or for the incident asthma between childhood and adolescence/young adulthood. Our study, however, included only limited number of young children. Instead, individual and parental allergies were strongly associated with school-age asthma as expected (40).

Some studies have suggested a role for leptin in obesity-associated asthma. In a cross-sectional study by Guler et al., higher leptin levels were found in asthmatic than in nonasthmatic children, but this difference was confined entirely to boys (20). Interestingly, leptin was a predictive factor for asthma, whereas gender, age or BMI were not. In another cross-sectional study, Sood et al. found, especially in women, support for an association of asthma with both leptin and BMI but after adjustment only the association with BMI remained significant (22). Considering that in the former study by Guler et al. the adjustment for serum leptin levels did not affect the association between BMI and asthma, both of these studies may imply that the relationship between obesity and asthma is not mediated via leptin, or that leptin is an independent predictor of asthma (20, 22). In a Swedish study, leptin levels were twice as high in obese asthmatic children than in obese nonasthmatic children, but the difference was not statistically significant (21). Furthermore, an alternate day calorie restriction has improved pulmonary function and asthma-related symptoms and reduced levels of leptin in overweight adults, but correlations between leptin levels and pulmonary function or asthma symptoms were not reported (41). Most recently, Kim et al. did not find link between leptin and asthma in children but leptin showed association with pulmonary function (23). This may be explained by the finding that girls and nonatopic children may have stronger associations between leptin levels and asthma than in boys and atopic children (24). Although we found an association between leptin levels and allergy status similar to one previous report (42), we did not find a connection between leptin levels and asthma in allergic subjects. Causality between leptin infusion and airway hyper-responsiveness have, however, been shown in mice (10, 18, 43), but according to recent results this association is unlikely to be due to a direct effect of leptin on airway smooth muscle (44). Thus, there are some evidence for the link between leptin and asthma, but the evidence is still weak and not convincing for the link between leptin and obesity-associated asthma. The strong associations between leptin and obesity, gender and age make designing and analysis of clinical studies challenging. In our case, these other covariates were associated stronger than leptin with asthma. Furthermore, BMI was clearly more strongly associated with asthma than leptin in multivariable models. Our study does not support a significant role for leptin in the development of adult onset of asthma with or without obesity.

Reports are scarce on the link between obesity-associated asthma and adiponectin, other obesity-related biomarkers or lipid levels. Serum adiponectin is reduced during pulmonary allergic reactions, and adiponectin attenuated allergic airway inflammation and airway hyper-responsiveness in mice (19). Although decreased serum concentrations of adiponectin observed in the obese could contribute to the development of airway hyper-responsiveness, evidence is limited in humans. High serum adiponectin concentrations may protect against current asthma in premenopausal women, but does not seem to explain the association between asthma and adiposity (25). Very recently, Kim et al. did not find any link between adiponectin and asthma in children but adiponectin showed association with pulmonary function (23). This again may be explained by the finding that girls and nonatopic children may have stronger associations between adiponectin levels and asthma than in boys and atopic children (24). High-sensitivity CRP is a marker of subclinical inflammation, and CRP levels have been shown to be associated with airflow obstruction and airway inflammation (45). Hancox et al. found inverse associations between pulmonary function and CRP at ages 26 and 32 years (46). The associations were similar in men and women and were independent of smoking, asthma and BMI. In the obese subjects, however, higher CRP levels are most likely due to increased insulin resistance (47). To our knowledge, there is only one report showing that obesity may be related to an increased risk of allergic asthma through mechanisms involved in the development of insulin resistance (6). Although calorie restriction has improved asthma symptoms and reduced lipid values, and hypercholesterolemia has been suggested as a potential independent risk factor for asthma, an intervention with simvastatin did not exhibit anti-inflammatory effects or clinical improvements in asthma despite reduced lipid values (41, 48, 49). We did not find any significant link between asthma or obesity-associated asthma and adiponectin, insulin, CRP or lipid levels.

Strengths of our population-based study include large sample size, partly longitudinal design and careful and comprehensive obesity-related laboratory measurements and adjustments for confounding factors. There are also limitations. Physician-diagnosed asthma and allergies were self-reported and airway hyper-responsiveness was not measured. Individual and parental allergies were, however, closely associated with asthma at all ages, and leptin was associated with allergy, which supports correct asthma and allergy diagnoses in our study (40). One may argue that our results could be influenced by changes in the overall prevalence of asthma during the last decades (50). This, however, is not likely to bias our results, as changes in general awareness of diseases or diagnostic practices are not expected to be limited to selective subgroups. Moreover, statistical analysis showed that there were no significant differences in the prevalence of asthma during the study period when comparing 24-year-old subjects in 1986 with 24-year-old subjects in 2001 (prevalence 3.2% and 3.9% respectively). Although power analysis was not performed beforehand, the adjusted difference in leptin values between asthmatics with and without overweight/obesity had no clinical relevance.

In conclusion, asthma is inevitably linked with obesity in adults. Despite some previous reports, we did not find support for the role of leptin or adiponectin in this association. Other factors should be sought for better understanding the pathogenesis of obesity-associated asthma.

Conflict of interest

None.

Acknowledgments

Supported by the Academy of Finland (grants 77841, 114034 and 210283), Research Funds from Turku University Hospital, the Finnish Cultural Foundation, the Turku University Foundation, the Foundation for Pediatric Research, the Paulo Foundation, the Finnish Foundation for Cardiovascular Research, the Juho Vainio Foundation and the Yrjö Jahnsson Foundation.

Ancillary