Associations of Serum Adipokines With Subclinical Interstitial Lung Disease Among Community-Dwelling Adults: The Multi-Ethnic Study of Atherosclerosis (MESA)

Abstract

Background: Adipokines have inflammatory and fibrotic properties that may be critical in interstitial lung disease (ILD). We examined associations of serum adipokine levels with CT imaging-based measures of subclinical ILD and lung function among community-dwelling adults.

Methods: A subset of the original Multi-Ethnic Study of Atherosclerosis cohort (n = 1,968) had adiponectin, leptin, and resistin measured during follow-up visits (2002-2005). We used regression models to examine associations of adiponectin, leptin, and resistin levels with (1) high-attenuation areas (HAAs) from CT scans (2004-2005, n = 1,144), (2) interstitial lung abnormalities (ILAs) from CT scans (2010-2012, n = 872), and (3) FVC from spirometry (2004-2006, n = 1,446). We used -(1/HAA²), which we denoted with H, to model HAA as our outcome to meet model assumptions.

Results: Higher adiponectin was associated with lower HAA on CT imaging among adults with a BMI ≥ 25 kg/m² (P for BMI interaction = .07). Leptin was more strongly associated with ILA among never smokers compared with ever smokers (P for smoking interaction = .004). For every 1-SD increment of log-transformed leptin, the percent predicted FVC was 3.8% lower (95% CI, -5.0 to -2.5). Higher serum resistin levels were associated with greater HAA on CT in a fully adjusted model. For every 1-SD increment of log-transformed resistin there was an increase in H of 14.8 (95% CI, 3.4-26.3).

Conclusions: Higher adiponectin levels were associated with lower HAA on CT imaging among adults with a higher BMI. Higher leptin and resistin levels were associated with lower FVC and greater HAA, respectively.

Keywords: adipokine; chest imaging; epidemiology (pulmonary); interstitial lung disease.

Figure 1

A and B, Continuous associations of serum resistin with high-attenuation areas (n = 1,119) (A) and % predicted FVC (n = 1,420) (B). A, Overall P for association = .01. B, Overall P for association = .045. Models are adjusted for exam 3 age, sex, race/ethnicity, smoking status, cigarette pack-years, missing cigarette pack-years indicator, statin medication use, and BMI. HAA model is also adjusted for study site, radiation dose, percent emphysema, and total lung volume imaged. The x axis is log-scale. The solid line is the overall effect estimate; thin dashed lines represent the 95% confidence interval bands. Each vertical hashmark in the rug plot along the x axis represents one participant.

Figure 2

Continuous associations of serum adiponectin with high-attenuation areas stratified by BMI. Blue line: adults with BMI ≥ 25 kg/m² (n = 793). Red line: adults with BMI < 25 kg/m² (n = 327). P for BMI interaction = .07. Model is adjusted for exam 3 age, sex, race/ethnicity, smoking status, cigarette pack-years, missing cigarette pack-years indicator, statin medication use, percent emphysema, study site, radiation dose, and total lung volume imaged. The x axis is log-scale. The solid line represents the overall effect estimate; thin dashed lines indicate the 95% confidence interval bands. Each vertical hashmark in the rug plot along the x axis represents one participant.

Figure 3

A-C, Continuous associations of serum adiponectin (A), leptin (B), and resistin (C) with percent predicted FVC stratified by sex. A, Red line, women (n = 681); blue line, men (n = 740); P for sex interaction = .06. B, Red line, women (n = 681); blue line, men (n = 735); P for sex interaction = .006. C, Red line, women (n = 681); blue line, men (n = 739); P for sex interaction = .007. Model is adjusted for exam 3 age, race/ethnicity, smoking status, cigarette pack-years, missing cigarette pack-years indicator, statin medication use, and BMI. The x axis is log-scale. The solid line represents the overall effect estimate; thin dashed lines indicate the 95% confidence interval bands. Each vertical hashmark in the rug plot along the x axis represents one participant.

Figure 4

Continuous associations of serum leptin with interstitial lung abnormalities stratified by smoking status. Red line, ever smokers (n = 433); blue line, never smokers (n = 415). P for smoking interaction = .004. Model is adjusted for exam 3 age, sex, race/ethnicity, statin medication use, percent emphysema, and BMI. The x axis is log-scale. The solid line represents the overall effect estimate; thin dashed lines indicate the 95% confidence interval bands. Each vertical hashmark in the rug plot along the x axis represents one participant.