Angiotensin Receptor Blockers and Subclinical Interstitial Lung Disease: The MESA Study

To the Editor:

Increased lung attenuation (high-attenuation areas [HAAs]) on quantitative computed tomography (CT) and visually identified interstitial lung abnormalities (ILAs) are two image-based measures of subclinical interstitial lung disease (ILD) that have been used to study early risk factors for ILD in community-dwelling adults (1, 2).

The renin-angiotensin system has been implicated in the pathogenesis of lung fibrosis, attributed mainly to mechanisms of angiotensin II through the angiotensin I (AT1) receptor (3, 4). Local angiotensin II is generated within the lung parenchyma and regulates cell growth and fibrogenesis (3). Angiotensin peptides and angiotensin I and II receptor expression are increased in idiopathic pulmonary fibrosis (IPF) lung tissue (5, 6). Angiotensin-converting enzyme inhibitors and angiotensin receptor blockers (ARBs) attenuate lung fibrosis in animal models by interrupting the angiotensin system (4, 7–13), and have been proposed as therapeutic agents in ILD (14, 15). On the other hand, a recent study showed an increased risk of all-cause mortality among patients with IPF who receive ARBs (16).

We tested the hypothesis that these medications would be associated with less subclinical ILD on CT, lower levels of serum biomarkers of alveolar injury and remodeling, and higher lung function among community-dwelling adults. We secondarily examined the relationships between these outcomes and both β-blocker use and calcium channel blocker use.

Methods

We conducted a cross-sectional analysis of adults in MESA (Multi-Ethnic Study of Atherosclerosis), a National Heart, Lung, and Blood Institute–funded multicenter, prospective cohort study, to investigate the prevalence and progression of subclinical cardiovascular disease. The enrollment criteria and study procedures have been previously described (17). We measured self-reported antihypertensive medication use, serum matrix metalloproteinase-7 (MMP-7) and surfactant protein-A (SP-A) levels, and HAAs (n = 6,769) by CT at exam 1 (2000–2002) and by spirometry at exams 3 and 4 (2004–2006); and ILAs (n = 2,431 as previously defined [1, 18, 19]) at exam 5 (2010–2012). HAA was defined as the percentage of imaged lung volume with CT attenuation values between −600 and −250 Hounsfield units measured using a modified version of the Pulmonary Analysis Software Suite at a single center, as previously described (1, 20). We examined associations between medication use and the above outcomes using linear and logistic regression models with purposeful variable selection informed by a directed acyclic graph. Each model included binary variables for each class of antihypertensive (angiotensin-converting enzyme inhibitor, ARB, β-blocker, and calcium channel blocker) and was adjusted for age, sex, race, current smoking status, cigarette pack-years, self-reported diabetes, hypertension (21), estimated glomerular filtration rate, self-reported asthma, self-reported emphysema, and educational attainment. For consistency with prior work, we adjusted for additional potential confounders in the HAA models, including height, waist circumference, study site, milliampere dose, and imaged lung volume (1, 18, 19, 22). We examined models stratified by smoking status and tested for effect modification in multivariable models using likelihood ratio tests. We performed sensitivity analyses by further adjusting for left ventricular (LV) wall thickness in a subset of participants who underwent cardiac magnetic resonance imaging (23). Statistical significance was defined as a two-tailed P value less than 0.05. We performed statistical analysis using Stata 14.1 (Stata Corp.) and R version 3.4.1 (R Foundation for Statistical Computing). MESA was approved by the institutional review boards of all collaborating institutions, and all participants provided informed consent.

Results

Among the participants with measured ILAs, the mean ± standard deviation age was 60 ± 9.4 years, 53% were women, and 50% were ever-smokers. There were 139 subjects with self-reported ARB use, 282 with angiotensin-converting enzyme inhibitor use, 226 with β-blocker use, and 263 with calcium channel blocker use in overlapping cohorts. We found no strong evidence for associations of any antihypertensive with serum MMP-7, SP-A, or HAA, other than β-blocker use, which was associated with lower HAA overall (data not shown). In longitudinal analyses, ARB use was associated with a 1.9-fold increased odds of having an ILA at the 10-year follow-up (adjusted odds ratio [OR], 1.9; 95% confidence interval [CI], 1.2 to 3.2) and lower percent predicted forced vital capacity (FVC) (adjusted mean difference −2.5 percentage points; 95% CI, −4.9 to −0.2) (Figure 1). Findings were similar among ever-smokers and never-smokers (Figure 1). Although the CIs included the null value, angiotensin-converting enzyme inhibitor use was associated with a 1.5-fold increased odds of developing an ILA (adjusted OR, 1.5; 95% CI, 0.97 to 2.3) overall and among ever-smokers (adjusted OR, 1.8; 95% CI, 1.1 to 3.0; Figure 1). β-blocker use was associated with a lower prevalence of ILA among smokers (Figure 1). Calcium channel blocker use was not associated with subclinical ILD (Figure 1). Similar findings were obtained after further adjustment for LV wall thickness (data not shown).

Figure 1.

Forest plot of multivariable adjusted associations of antihypertensive medication use and interstitial lung abnormalities (ILA) and percent predicted forced vital capacity (ppFVC). Both models are adjusted for age, sex, race, smoking status, self-reported diabetes, self-reported hypertension, glomerular filtration rate, self-reported asthma, self-reported emphysema, and educational attainment. Boxes represent point estimates and whiskers are 95% confidence intervals. ACE = angiotensin-converting enzyme; ARB = angiotensin receptor blocker; CCB = calcium channel blocker; CI = confidence interval; OR = odds ratio.

Discussion

We observed that the use of ARBs was associated with CT-based evidence of subclinical ILD in community-dwelling middle-aged and older adults. Angiotensin-converting enzyme inhibitor use was associated with subclinical ILD among smokers. These findings suggest a possible harmful role of renin-angiotensin inhibition in subclinical ILD. However, these hypothesis-generating findings should be interpreted cautiously by clinicians until further evidence is gathered.

Our primary findings were unexpected and contradict mechanistic studies (3–15) and a study that showed an association between angiotensin-converting enzyme inhibitor and ARB use and slowed progression of emphysema (24), but they align with a recent study that suggested that ARB use was associated with a higher mortality rate among patients with IPF (16). We speculate that ARB-induced compensatory increases in renin levels may lead to higher angiotensin II levels, possibly overcoming AT1 inhibition in the lung (25). Although the data are limited, renin itself promotes collagen synthesis and transforming growth factor β-1 expression through angiotensin II–independent mechanisms (26). Nevertheless, without mechanistic support, our findings should be considered preliminary in nature.

Our study has a number of additional limitations. The HAA analyses were cross-sectional in nature. Because we only studied subclinical ILD, the differences in FVC between groups were not clinically significant. No data were available regarding the dosage of antihypertensive medication used. Furthermore, we performed an observational study of treatment effect, which may have been confounded by indications. We adjusted for these indications, none of which are known to be strong risk factors for the outcome, lessening this concern.

In conclusion, we found that angiotensin-converting enzyme inhibitor use was associated with subclinical ILD in a population-based study of U.S. adults. These findings remain unexplained, but do suggest that ARBs may be risk factors for ILD.