Impact of gastroesophageal reflux on longitudinal lung function and quantitative computed tomography in the COPDGene cohort

Rationale Gastroesophageal reflux disease (GERD) is a common comorbidity in chronic obstructive pulmonary disease (COPD) and has been associated with increased risk of acute exacerbations, hospitalization, emergency room visits, costs, and quality-of-life impairment. However, it remains unclear whether GERD contributes to the progression of COPD as measured by lung function or computed tomography. Objective To determine the impact of GERD on longitudinal changes in lung function and radiographic lung disease in the COPDGene cohort. Methods We evaluated 5728 participants in the COPDGene cohort who completed Phase I (baseline) and Phase II (5-year follow-up) visits. GERD status was based on participant-reported physician diagnoses. We evaluated associations between GERD and annualized changes in lung function [forced expired volume in 1 s (FEV1) and forced vital capacity (FVC)] and quantitative computed tomography (QCT) metrics of airway disease and emphysema using multivariable regression models. These associations were further evaluated in the setting of GERD treatment with proton-pump inhibitors (PPI) and/or histamine-receptor 2 blockers (H2 blockers). Results GERD was reported by 2101 (36.7%) participants at either Phase I and/or Phase II. GERD was not associated with significant differences in slopes of FEV1 (difference of − 2.53 mL/year; 95% confidence interval (CI), − 5.43 to 0.37) or FVC (difference of − 3.05 mL/year; 95% CI, − 7.29 to 1.19), but the odds of rapid FEV1 decline of ≥40 mL/year was higher in those with GERD (adjusted odds ratio (OR) 1.20; 95%CI, 1.07 to 1.35). Participants with GERD had increased progression of QCT-measured air trapping (0.159%/year; 95% CI, 0.054 to 0.264), but not other QCT metrics such as airway wall area/thickness or emphysema. Among those with GERD, use of PPI and/or H2 blockers was associated with faster decline in FEV1 (difference of − 6.61 mL/year; 95% CI, − 11.9 to − 1.36) and FVC (difference of − 9.26 mL/year; 95% CI, − 17.2 to − 1.28). Conclusions GERD was associated with faster COPD disease progression as measured by rapid FEV1 decline and QCT-measured air trapping, but not by slopes of lung function. The magnitude of the differences was clinically small, but given the high prevalence of GERD, further investigation is warranted to understand the potential disease-modifying role of GERD in COPD pathogenesis and progression. Clinical trials registration NCT00608764.

Cross-sectional studies that evaluated the relationship between GERD and lung function have revealed conflicting resultssome studies observed worse airflow obstruction [2,18,19] in those with GERD, while other studies showed no significant relationship between GERD and lung function [12,[20][21][22]. Due to the crosssectional design of these studies, we cannot derive definitive conclusions about the causal associations between GERD and COPD disease progression. Therefore, to evaluate if GERD is associated with COPD disease progression as measured by lung function or quantitative chest imaging, we analyzed the data from a large, longitudinal, multicenter cohort study. To our knowledge, this is the first study to longitudinally assess both lung function and quantitative chest imaging over a five-year period to evaluate the association between GERD and COPD disease progression.

Patient selection
COPDGene (ClinicalTrials.gov Registration # NCT0060 8764) is an ongoing multicenter, longitudinal study designed to investigate the genetic and epidemiologic characteristics of smoking-related lung disease. A complete description of the protocol has been published previously [18]. Briefly, the primary inclusion criteria are: self-identified racial/ethnic category of non-Hispanic white or African-American, 45-80 years old, with a minimum of 10 pack-year smoking history (except for a small number of non-smoking controls). For the current analysis, we selected from the full cohort of 10,720 enrolled participants and included participants who were former or current smokers and completed both baseline (Phase I, 2008-2011) and 5-year follow-up (Phase II, 2012-2016) study visits. The research protocol was approved by the institutional review board at each participating institution and all participants provided written informed consent.

Diagnosis of GERD
Our primary predictor variable was the presence or absence of GERD. Standardized medical history and medication inventories were administered by research staff. GERD diagnosis was assessed by asking participants, "Have you ever been told by a physician that you have gastroesophageal reflux?" [16,19]. To obtain an accurate medication list, participants were instructed to bring all current medications to the study visit. All medications were captured on the medications questionnaire, including GERD-related medications such as protonpump inhibitors (PPI) and histamine receptor-2 blockers (H2 blockers).

Lung function
Our primary outcome variable was rate of lung function decline, as assessed by post-bronchodilator spirometry. Participants underwent spirometry (EasyOne™ spirometer; ndd, Andover, MA) before and after administration of 180 μg of albuterol (via Aerochamber Activis, Parsippany, NJ) at Phase I and Phase II. Percent predicted and lower limit of normal (LLN) values were obtained using National Health and Nutrition Examination III reference equations for spirometry [20]. COPD severity was assessed using spirometry criteria outlined by the Global Initiative for Obstructive Lung Disease (GOLD) guidelines [21]. GOLD 0, defined as forced expiratory volume in 1 s/forced vital capacity (FEV 1 /FVC) ≥0.7 and an FEV 1 < 80% predicted in current or former smokers without COPD, is not currently included in the GOLD guidelines, but was used previously [22]. Preserved ratio and impaired spirometry (PRISm) was defined as a postbronchodilator FEV 1 /FVC ≥0.7 and an FEV 1 > 80% predicted in former or current smokers [23,24]. Rates of FEV 1 and FVC changes per year were calculated by dividing the differences between Phase I and Phase II by the number of years between visits.

Quantitative computed tomography (QCT) imaging measurements
Our secondary outcomes were QCT-based measures of lung disease. Participants included in this analysis had high-resolution CT scans at full inspiration at Phase I and Phase II to assess emphysema and airway disease. Quantitative imaging analysis were performed using VIDA (VIDA Diagnostics, Iowa City, IA; http://www. vidadiagnostics.com) software and Thirona (https:// thirona.eu/) software. QCT outcomes included airway wall thickness (AWT)-Pi10 (square root of the wall area of a theoretical airway of 10 mm luminal perimeter), airway wall area (100 X wall area/total bronchial area), air trapping (percent of lung with attenuation values less than − 856 HU on expiratory CT) [25], emphysema (percent of voxels on inspiratory CT with attenuation values less than − 950 Hounsfield Units (HU), and Perc15 lung density (the 15th percentile point defined as the HU below which the 15% of voxels with the lowest density are distributed, adjusted for CT-based lung volumes) [26]. Rates of QCT imaging measurement changes per year were calculated by dividing the differences between Phase I and Phase II by the number of years between visits.

Statistical analysis
Demographic, clinical, and lung health characteristics were compared between those with and without GERD using descriptive statistics. GERD was assessed identically at the Phase I and Phase II visits, and in our primary analysis, we categorized those with GERD as reporting GERD at either visit and compared outcomes to those with no GERD at either visit. Associations between GERD and longitudinal changes in spirometry and QCT chest measurements were assessed using linear regression models with sandwich standard errors. We present the outcome data using three models: Model 1 is unadjusted; Model 2 covariates included age, sex, race, whether the patient smoked between Phase I and Phase II, body mass index (BMI), clinical center, and FEV 1 % predicted at Phase I; and Model 3 included covariates in Model 2 and whether or not the patient had ≥1 acute exacerbation of COPD between Phase I and Phase II.
Secondary analyses compared changes in spirometry between [1] those with 'persistent GERD' (GERD at both Phase I and Phase II) vs. no GERD at either visit, [2] those with 'incident GERD' (no GERD at Phase I, but GERD at Phase II) vs. no GERD, and [3] those with 'resolved GERD' (GERD at Phase I, but not Phase II) vs. no GERD. Adjustments were made for the same covariates as Model 3 in the primary analyses. Additionally, logistic regression models were used to estimate the odds of having rapid FEV 1 decline (defined as FEV 1 decline of ≥40 mL/year) for those with vs. without GERD after adjustment for the same covariates. Lastly, we explored associations between GERD treatment (PPI and/or H 2 blocker at Phase I and/or Phase II) and changes in lung function using linear regression models.
Among our 5728 study participants, pharmacologic treatment with PPIs was reported by 990 (24%) and H 2 blockers by 260 (6.5%), of whom most (81%) reported GERD at either visit, though 19% did not report GERD at either visit. Among those with GERD, treatment with PPI and/or H 2 blocker at either Phase I and/or Phase II was associated with faster decline in lung function (Table 5). Among participants with GERD, the decline in both FEV 1 (difference of − 6.61; 95% CI, − 11.9 to − 1.36) and FVC (difference of − 9.26; 95% CI, − 17.2 to − 1.28) were faster in those receiving PPI and/or H2 blocker compared to those who are not receiving either medication. Among those without GERD, PPI and/or H 2 blocker treatment was not associated with lung function decline, though these estimates had less precision than in those with GERD due to the smaller sample.
No significant differences in the slopes of change of the QCT chest measures was found in those taking PPI and/or H 2 blockers compared to those who were not taking medications (Supplemental Table 2S).

Discussion
Data from our large, multicenter, longitudinal cohort suggest that GERD may contribute to progressive loss of lung function and increases in air trapping over time. Although other studies have found cross-sectional Table 2 Linear regression models of the association between gastroesophageal reflux disease (GERD) and slopes of lung function and Quantitative CT measures of lung disease. ß coefficients reflect the mean differences in the row outcome of interest between those with GERD compared to those without GERD  associations between GERD and lung health, to our knowledge, our study is the first to use a longitudinal study design over a five-year period and assess both lung function and quantitative imaging measurements.
Despite several statistically significant associations between GERD and longitudinal changes in spirometry and QCT measures, we note that the magnitude of the effect sizes were clinically small, with point estimates of 2-5 mL/year faster FEV 1 decline among those with GERD. Although this degree of faster lung function decline might not be expected to lead to significant clinical problems, we note that the effect size of cigarette smoking has been estimated at 4-27 mL/year faster decline, so an additional 2-5 mL/year might still contribute to disease progression, in combination with other factors that might contribute to lung function decline such as non-cigarette smoke exposures, respiratory infections, and abnormal inflammatory responses [27,28]. Although we have little data to guide us in categorizing more significant rates of QCT changes over time, for spirometry, we applied a common definition of rapid FEV 1 decline of ≥40 mL/year [29]. In this categorical logistic regression analysis, we saw that GERD was associated with a 20-33% increased odds of rapid decline. Although we must advise caution in interpreting this secondary analysis, these results suggest there might be a subgroup of persons more susceptible to pulmonary effects of GERD, but further research is needed to explore this hypothesis.
We included acute exacerbations as a covariate in our models as episodes of acute exacerbations have been known to accelerate lung function decline [30]. We found that rapid FEV 1 decline and progression in QCTmeasured air trapping were associated with GERD, but the estimates of slopes of FEV 1 and FVC were Table 3 Multivariable linear regression models of the association between gastroesophageal reflux disease (GERD) and slopes of lung function. ß coefficients reflect the mean differences in the row outcome of interest between those with GERD compared to those without GERD   Table 4 Multivariable logistic regression models of the association between gastroesophageal reflux disease (GERD) and rapid FEV 1 decline (FEV 1 decline of ≥40 mL/year, n = 2572). Adjusted odds ratios reflect the relative odds of rapid FEV 1 decline between those with GERD, compared to those without GERD  Table 5 Multivariable linear regression models of the association between treatment with proton-pump inhibitor (PPI) and/or H 2 blocker and slopes of lung function, in those with and without gastroesophageal reflux disease (GERD). ß coefficients reflect the mean differences in the row outcome of interest between those with treatment with PPI and/or H 2 blocker, compared to those not receiving treatment attenuated compared to the model that did not include acute exacerbations as a covariate. We evaluated loss of lung function both as continuous variables (mL/year) and as a categorical variable (FEV 1 decline ≥40 mL, yes vs. no), then stratified GERD into 'persistent', 'incident', and 'resolved.' Rapid decline in FEV 1 is a strong predictor of mortality and COPDrelated hospitalization [31]. This current study suggests that GERD is an independent predictor of rapid FEV 1 decline, using a multivariate logistic regression model controlling for age, sex, race, smoking status, BMI, FEV 1 % predicted at baseline, and acute exacerbations. Participants with 'resolved GERD' do not appear to have increased odds of rapid FEV 1 decline raising the question about the potential role of GERD treatment in slowing lung function decline, which will need to be addressed in future clinical trials.
Smaller cross-sectional studies that have evaluated the relationship between lung function severity and GERD have shown mixed results. Mokhlesi et al. [32] found that symptomatic GERD was more prevalent in COPD patients with FEV 1 ≤ 50% compared to those with FEV 1 > 50% (23% vs. 9%, respectively; p = 0.08), while Rogha et al. [2] showed that patients with GERD have more severe COPD compared to those without GERD (GOLD stage ≥2 or higher: 88% vs. 67%, respectively; p = 0.005) supporting our findings. In contrast, several other studies found no association between lung function and GERD, possibly due to relatively small sample sizes [12,[33][34][35]. Our present study expands the literature on the relationship between lung function and GERD by adding temporal dimension and a larger sample size.
In addition to adding a temporal dimension in the assessment of the impact of GERD in lung function, we also evaluated whether GERD contributes to the progression of small airway disease and emphysema over time using QCT. Small airway obstruction and emphysematous lung destruction reflect abnormalities in lung function [36]. Airway changes using QCT in the context of aspiration have been evaluated previously [37][38][39][40][41][42], but these studies are also limited to small cross-sectional studies. Hiller et al. found that patients with recurrent aspiration of gastric contents had QCT evidence of bronchial wall thickening (95%) and air trapping (44%) [42]. Similarly, Cardasis et al. found increased airway wall thickening on QCT of patients with pathologicallyconfirmed chronic occult aspiration of whom 96% had diagnosis of GERD [41]. We found that the rate of air trapping progression over 5 years was faster in those with GERD compared to those without GERD. This could represent the development of distal small airway disease as a result of ongoing pulmonary micro-aspiration of refluxed gastric material and/or vagally-mediated reflex bronchoconstriction in GERD [4,5]. However, the slopes of the other QCT measurements of small airway disease (AWT-Pi10 and airway wall area) and emphysema (% emphysema and Perc15 lung density) were not different between those with and without GERD. We hypothesize that air trapping in the setting of GERD is a possible early measurable imaging manifestation of small airway disease, possibly an imaging finding that can be seen prior to the other QCT measurements of small airway disease and emphysema. This hypothesis will need to be addressed in future longitudinal studies.
The association between GERD and lung health bring into question the role of anti-reflux treatment in management of COPD. We observed that pharmacologic treatment with PPI and/or H 2 blocker among participants with GERD was associated with an accelerated decline in FEV 1 and FVC. Using the same COPDGene cohort, Martinez et al. showed that the use of PPI was associated with improved SGRQ total score, but also increased exacerbations highlighting the possibility of confounding-by-indication [16]. The significant decline in lung function with PPI and/or H 2 blocker use in this cohort is also likely due to confounding-by-indication. Xiong et al. suggested that treatment of GERD with PPI in patients with COPD is associated with delayed deterioration of FEV 1 after 1-year follow-up [43]. However, several other studies evaluating the efficacy of anti-reflux medications in COPD did not report on the impact of these pharmacologic therapies on lung function [12,32,[44][45][46]. A limitation to the pharmacologic treatment of GERD is that anti-reflux medications do not target nonacid reflux and weakly acidic reflux. Surgical intervention with fundoplication, on the contrary, impacts both acid and non-acid reflux. Most of the literature on anti-reflux surgery focuses on lung transplant. Although the evidence is conflicting, a systematic review by Robertson et al. suggested that anti-reflux surgery provided benefit in lung function among lung transplant patients [47]. We did not find studies specifically addressing antireflux surgery in COPD and lung function. The effects of anti-reflux therapies in COPD outcomes, specifically lung function, are unclear highlighting the need for carefully-designed clinical trials.
Our study has limitations. GERD was based on self-report of a physician diagnosis, not on validated reflux questionnaires, pH monitoring, or esophageal manometry. Therefore, GERD misclassification might have affected our results. Because this is an observational study, we cannot establish causal inferences.

Conclusion
GERD was associated with faster COPD disease progression as measured by rapid FEV 1 decline and QCTmeasured air trapping, but not by slopes of lung function. The magnitude of the differences was clinically small, but given the high prevalence of GERD, further investigation is warranted to understand the potential disease-modifying role of GERD in COPD pathogenesis and progression.
Additional file: Supplemental Figure 1S. Participant flow diagram. Supplemental Table 1S. Cohort characteristics at Phase I by availability of quantitative CT (QCT) measurements at both visits. Supplemental Table 2S. Multivariable linear regression models of the association between treatment with proton pump inhibitor (PPI) and/or H 2 blocker (n = 960) and slopes of quantitative CT (QCT) measures of lung disease among those with gastroesophageal reflux disease (GERD).