Adiposity influences airway wall thickness and the asthma phenotype of HIV-associated obstructive lung disease: a cross-sectional study

Obstructive lung disease, encompassing many phenotypes of both fixed and reversible airflow obstruction, is common in HIV-infected persons [1‐6]. Chronic obstructive pulmonary disease (COPD) in the HIV-infected population is accelerated in smokers and those with poor control of the viral load [1, 3, 7, 8]. Asthma is another commonly diagnosed chronic pulmonary disease in HIV-infected persons [9, 10]. Despite the prevalence of COPD and asthma in HIV-infected persons, little is known about their pathogenesis in this population.

Obesity influences the development of both COPD and asthma in the HIV-uninfected population. Obesity is more prevalent in individuals with mild COPD compared to the general population, however, the causal nature of the relationship between obesity and COPD is unclear [11, 12]. Obesity, central adiposity, and aspects of the metabolic syndrome have been implicated in the pathogenesis of the adult-onset phenotype of asthma [13‐17]. Inflammation related to visceral adipose tissue is thought to drive this association [18]. We have previously shown that doctor-diagnosed asthma in HIV is frequently adult-onset, associated with inflammatory markers common in chronic HIV infection, and 2.5 times more likely in obese compared to normal weight HIV-infected persons [9].

Metabolic effects of HIV and highly-active antiretroviral therapy (HAART) that lead to central adiposity and alterations in inflammation may be relevant to the pathogenesis of airway obstruction in HIV [19, 20]. Long-term HIV infection is associated with chronic inflammation and macrophage activation, measured by high-sensitivity C-reactive protein (CRP) and soluble CD163 (sCD163), respectively [21‐26]. Increased central adiposity is associated with the alteration of systemic adipokine profiles, including higher leptin (a pro-inflammatory cytokine) and lower adiponectin (an anti-inflammatory cytokine). Adiponectin is lower in HIV-infected persons and in chronic inflammation [27]; and reduced levels of adiponectin have been implicated in several HIV-associated co-morbidities such as cardiovascular disease and neurocognitive dysfunction [28‐30]. In the HIV-uninfected population, levels of adiponectin are lower in asthma and paradoxically higher in COPD [31]. The relationship of adiponectin in HIV-associated obstructive lung disease is unknown.

Obstructive lung disease can manifest subjectively as pulmonary symptoms or objectively as pulmonary function changes or airway remodeling detectable on computed tomography (CT) scan [32‐34]. Asthma is often diagnosed by doctors based on episodic, recurrent pulmonary symptoms, such as wheezing. Pulmonary symptoms are more common in HIV-infected individuals with doctor-diagnosed asthma than those without asthma [9]. Airway wall thickening correlates with asthma severity, airflow obstruction, and histopathological changes related to asthma [35‐37]. Airway remodeling quantitatively measured by CT scan has not been assessed in HIV-infected persons.

In this study, our primary objective was to determine the relationship of adiposity and adipose-related inflammation with obstructive lung disease phenotypes in HIV-infected persons. Our secondary objective was to determine the relationship of adiposity and its associated inflammation with airway remodeling, as measured by airway wall thickness on CT imaging, in HIV-infected persons. We hypothesized that visceral (mediastinal) adipose tissue and adipose-associated inflammatory markers (adiponectin and IL-6) would be associated with airway remodeling and the asthma phenotype of airflow obstruction and no association with the COPD phenotype of airflow obstruction, in HIV-infected persons.

A total of 121 HIV-infected participants were evaluated. Overall, the mean (SD) age of all participants was 45.1 (9.8) years, and the majority were men (67.8 %) with a history of smoking (82.6 %) (Table 1). Most participants were on antiretroviral therapy (ART) (87.6 %) with a mean CD4 count of 599.1 cells/μL, and 84 (69.5 %) had an HIV RNA viral level <50 copies/mL. Twenty-three (19 %) participants had the COPD phenotype of airflow obstruction. Thirty-three (27 %) participants were classified as having the asthma phenotype of airflow obstruction; 25 of them had doctor-diagnosed asthma, six of whom also had a bronchodilator response, and eight participants had a bronchodilator response without prior doctor-diagnosed asthma. Twelve (10 %) participants fit both definitions of the COPD and asthma phenotypes, and were included in both groups for analysis.

Participant characteristics differed between those with and without the COPD phenotype of airflow obstruction. Those with the COPD phenotype were older, had heavier smoking history as measured by pack-years, and greater intravenous drug use (Table 1). BMI did not differ between those with and without the COPD phenotype of airflow obstruction (mean [SD] 27.0 kg/m² [7.9] vs. 27.8 kg/m² [6.1], p = 0.59) (Fig. 1). Participant demographics were similar between those with and without the asthma phenotype of airflow obstruction, except for BMI (mean [SD] 30.7 kg/m² [8.1] vs. 26.5 kg/m² [5.3], p = 0.008), and female sex (n, 15 vs. 24, p = 0.06).

Those with the COPD phenotype had lower pre- and post-bronchodilator FEV₁ %predicted and FEV₁/FVC as expected by definition, compared to those without COPD. The participants with the asthma phenotype of obstruction had lower pre-BD FEV₁ (78 %[21] vs. 97 %[16]; p = <0.001) and pre-BD FEV₁/FVC (69 %[12] vs. 77 %[8]; p = <0.001) compared to those without asthma. DL_CO was moderately reduced amongst all participants (mean, 66 %); however, it was lower in participants with the COPD phenotype compared to those without COPD (57 %[15] vs. 68 %[13], p = <0.001). DL_CO was no different in those with the asthma phenotype of obstruction compared to those without asthma (63 %[13] vs. 67 %[14], p = 0.13) (Fig. 1).

Several inflammatory markers were different between the two phenotypes of airflow obstruction (Table 2). CRP levels were higher in participants with both phenotypes of airflow obstruction (COPD p = 0.04, asthma p = 0.04) versus those without airflow obstruction. The total IgE level was elevated in the asthma phenotype group of airflow obstruction (p = 0.03); however, there was no difference in the number of participants with elevated sputum eosinophils. Adiponectin tended to be lower in the asthma phenotype group compared to those without asthma (p = 0.07). There was no difference in the level of sCD163 amongst all participants. There was no difference in the level of IL-6 between those with or without COPD, however, those with the asthma phenotype of obstruction had higher IL-6 levels compared to those without asthma (p = 0.03).

There was excellent inter-reader reliability for subcutaneous (r = 0.955; p < 0.001) and mediastinal (r = 0.995; p < 0.001) adipose measures. There was no difference in the volumes of adipose tissue in the mediastinum or the subcutaneous regions of those with and without the COPD phenotype of obstruction (Table 2). Participants with the asthma phenotype of obstruction had a greater volume of subcutaneous adipose tissue (33.3 cm³ vs. 22.1 cm³, p = 0.004), but no difference in the volume of mediastinal adipose tissue compared to those without asthma (3.9 cm³ vs. 4.0 cm³, p = 0.33).

BMI and CT adipose measurements correlated with several markers of inflammation. Adiponectin correlated inversely with BMI (r = -0.23, p = 0.01), and more strongly with mediastinal adipose tissue (r = -0.41, p < 0.001) than subcutaneous tissue (r = -0.19, p = 0.04)(data not shown). Mediastinal adipose tissue correlated inversely with IgE (r = -0.19, p = 0.04)(data not shown). Both BMI (mean [SD], 30.1 kg/m² [8.2] vs. 25.7 kg/m² [4.0]; p = 0.001) and subcutaneous adipose tissue volume (mean (SD) 17.4√cm³/slice (5.6) vs. 14.6√cm³/slice (4.8); p = 0.008) were greater with high IL-6 (data not shown). There was no correlation between adipose tissue and CRP or sCD163 (data not shown).

Airway wall thickness (WA%) did not differ between those with and without the COPD phenotype of obstruction, but was greater in the group with the asthma phenotype of airflow obstruction (51.0 % vs. 46.8 %; p < 0.001) (Fig. 1). Greater WA% was associated with lower pre-BD FEV₁ (r = -0.19; p = 0.04); there was no association between WA% and pre-BD FEV₁/FVC ratio or DL_CO (data not shown). Greater WA% was associated with BMI, as well as both subcutaneous and mediastinal adipose volumes (Fig. 2). Greater WA% was associated with symptoms of wheezing (p = 0.02) and tended to be associated with phlegm production, but was not associated with cough or dyspnea (data not shown).

We performed multivariable regression models to identify factors associated with the COPD phenotype of airflow obstruction, the asthma phenotype of obstruction, and WA% (Table 3). A list of all variables included in the models can be found in the Methods section. The COPD phenotype was associated with older age and longer smoking history. The asthma phenotype was associated with the female sex, younger age, smoking history, and lower adiponectin values. Greater WA% was associated with greater BMI, younger age, higher sCD163, and higher CD4 counts. People who smoked cigarettes or used cocaine had lower WA%.

This study demonstrates that there are unique associations that may be contributing to the pathogenesis of two airway obstruction phenotypes in HIV-infected persons. We found that the COPD phenotype of airflow obstruction was associated with heavier smoking and older age. In contrast, the asthma phenotype of airflow obstruction was associated with the female sex, younger age, smoking history, and lower adiponectin levels. These findings suggest that, while there is some degree of overlap in the two phenotypes, the association of adiponectin, an important adipose-related mediator of inflammation, with the asthma phenotype of obstruction may be mechanistically important in HIV-associated asthma.

Asthma has several phenotypes of disease in the HIV-uninfected population and may, similarly, have several phenotypes in the HIV-infected population. Similar to the general population, there is strong evidence that the allergic phenotype of asthma is present, with elevated total IgE. Additionally, a subset of asthma in HIV may represent the adult-onset, obesity-mediated phenotype. We have previously shown an association between asthma and BMI in HIV [9], suggesting that the increasingly recognized adult-onset obesity phenotype might also be more prevalent in the HIV-infected population [49, 50]. Our univariate analyses highlight the significant association of increasing volume of subcutaneous tissue with the asthma phenotype of airflow obstruction more than mediastinal adipose tissue as we had hypothesized. Increased deposition of visceral adipose tissue is more strongly associated with metabolic dysregulation and inflammation in the HIV-uninfected population; therefore, subcutaneous adipose tissue in the HIV-infected population may be playing a more significant role in asthma pathogenesis. This study also demonstrated a strong correlation between BMI and adipose measures, both subcutaneous and mediastinal, with increased airway wall thickness, a marker of airway inflammation and remodeling. In multivariable analysis, there was a significant correlation between adiponectin and the asthma phenotype of obstruction, an adipose-mediated inflammatory marker. Several studies have shown a link between adiponectin and asthma in obese asthmatics compared to non-obese asthmatics, but this association has not previously been reported in HIV [14, 17, 18].

Alternatively, HIV-associated asthma may be related to chronic HIV-associated inflammation. HIV leads to immune activation and chronic inflammation, with elevated levels of IL-6, CRP, and D-Dimer compared to the general population [21‐24]. In our study, CRP tended to be associated with greater airway obstruction. We and others have shown that serum IL-6 is associated with adiposity, suggesting that there is increased inflammation related to obesity as well [51]. In our multivariable regression model, airway remodeling was associated with greater BMI, higher CD4 counts, and higher sCD163 levels, suggesting that macrophage activation from chronic immune activation [25, 26] and obesity may both play a role in airway remodeling, but the degree to which they are related is unclear [52, 53].

Inflammation related to lipodystrophy is an important factor causing HIV-associated non-AIDS diseases. Several other HIV co-morbidities, including cardiovascular disease and neurocognitive dysfunction, have been associated with increased adiposity and inflammation [28‐30]. Our findings mirror these associations, suggesting that adipose-related factors, chronic inflammation, and macrophage activation are probable contributors to the increased disease burden of certain types of obstructive lung disease in HIV infection.

It is important to note the global reduction in the DL_CO and high smoking rate amongst all participants within this cohort of HIV-infected persons. Diffusion impairment is commonly reduced in the HIV-infected population compared with the general population, even in never smokers [3, 5]. A significantly lower DL_CO was a feature of the COPD phenotype, but not the asthma phenotype, confirming that the asthma phenotype is a unique phenotype of airflow obstruction within this population. This phenotype may be related to different factors than those that drive diffusion impairment and COPD. Furthermore, airway wall thickness did not correlate with DL_CO, only with BMI and FEV_1.

This cohort of HIV-infected persons is representative of the HIV population in the post-ART era, with the majority of participants well-controlled on ART with an average CD4 count > 500cells/μL and HIV RNA levels <50copies/mL. Although there was a high prevalence of tobacco and drug use in this cohort, which may lead to alterations in some of the measured inflammatory markers, these risk factors did not differ between those with and without airflow obstruction, and are not markedly higher than the U.S. HIV population estimates [54, 55]. Another strength of this study is that we used a variety of methods to assess adiposity and its correlation with asthma, including clinical data, radiographic data, and serum biomarkers.

Our study has limitations that need to be considered. It is cross-sectional, and therefore, it is not possible to attribute cause and effect, particularly between obesity and asthma. However, many studies showing an association between obesity and asthma have been corroborated in longitudinal studies and even an intervention study [56‐58]. We defined the asthma phenotype of obstruction as having either doctor-diagnosed asthma or a bronchodilator response with pulmonary function testing, and only 42.4 % of those with doctor-diagnosed asthma had a bronchodilator response. The absence of bronchodilator response could be attributed to treated or well-controlled asthma not manifested on pulmonary function testing at the time of the study. The composite definition of the asthma phenotype may lead to heterogeneity in our study group, but we have found pulmonary function and biomarker similarities in participants fitting this composite definition [9]. Bronchodilator responsiveness is more classically seen in asthma; however, bronchodilator reactivity can be seen with fixed airflow obstruction, and in fact, may represent a sub-phenotype of COPD (asthma-COPD overlap syndrome (ACOS) rather than asthma. Despite our inclusion of the overlap participants in both the COPD and asthma airflow obstruction groups during analysis, there was a striking difference between participant characteristics suggesting that participants with the overlap are a unique phenotype. We measured mediastinal adipose tissue because we had access to chest CT scans done as part of the parent study [45]. Mediastinal adipose tissue volume has been shown in other populations to have similar characteristics to abdominal visceral adipose tissue and to correlate with subclinical cardiovascular disease in HIV [59]; however, this measure may not be an adequate reflection of visceral adiposity. Dedicated abdominal CT scans may be more useful in this assessment, but were not available. Also, we used adiponectin as our adipose-mediated inflammatory marker as these were non-fasting blood samples; however, leptin has been more strongly correlated with obesity-related inflammation and asthma. Mediastinal adipose tissue and adiponectin have been correlated with cardiovascular disease in HIV, supporting the validity in these measures as important markers in adipose-mediated inflammation. Finally, the level of associations that were seen in this study were interesting and novel, albeit overall low to moderate in biological associations between the variables.

In conclusion, there are different features associated with the COPD phenotype versus the asthma phenotype of obstructive lung disease in HIV-infected persons. Adipose-related factors, chronic inflammation, and macrophage activation are probable contributors to the increased disease burden of obstructive lung disease in HIV infection. These differences point out that there is likely not a single mechanism of airway obstruction in HIV, and a better understanding of the airway disease phenotypes and their pathogenesis is needed to optimize treatment of airflow obstruction in HIV.

Not applicable for this study.

ATS, American Thoracic Society; BMI, Body mass index; COPD, Chronic obstructive pulmonary disease; CRP, C-reactive protein; DL_CO, Diffusing capacity for carbon monoxide; FEV₁, Forced expiratory volume in 1 second; FVC, Forced vital capacity; HAART, Highly active antiretroviral therapy; HIV, Human immunodeficiency virus; IRB, Institutional review board; RNA, ribonucleic acid; SD, Standard deviation; WA%, Airway wall thickness.