The source population consisted of 9,598 male firefighters who were actively employed by the FDNY on 9/11, first arrived at the WTC between 9/11–9/24/2001, and had ≥3 post-9/11 forced expiratory volume at one second (FEV₁) measurements from routine medical monitoring PFTs taken at FDNY.⁵ A subset of this population received at least one clinically-indicated BD-PFT performed according to American Thoracic Society (ATS) standards¹⁵ at a hospital-based pulmonary function laboratory between 9/11/2001 and 9/10/2017. We excluded 57 participants whose BD-PFT occurred before their first post-9/11 medical monitoring examination. The final study population included 2,137 firefighters (Figure 1). Participants provided written informed consent. The Montefiore Medical Center (FWA #00002558)/Albert Einstein College of Medicine (FWA #00023382) Institutional Review Board approved this study.

Demographic data were retrieved from the FDNY employee database. Participants’ height, weight, self-reported smoking status (current, former or never-smoker) and time of initial arrival at the WTC site were assessed during routine medical monitoring examinations at FDNY; both active duty firefighters and WTC-exposed retirees are scheduled to have a monitoring exam once every 12–18 months. We classified individuals as having high (morning of 9/11), moderate (afternoon on 9/11–9/12) or low (9/13–9/24) WTC exposure based on their WTC site arrival time.⁴ Current and former smokers were grouped together as ever-smokers in our analyses. Those who consistently self-reported no cigarette smoking were classified as never-smokers.

Eosinophil concentration was measured from blood drawn shortly after 9/11, during the first post-9/11 monitoring examination. The median first post-9/11 blood draw date was 1/10/2002 (interquartile range: 11/26/2001–12/27/2002). We also had pre-9/11 blood data (eosinophil concentration) for the 1,008 participants who had blood drawn at a pre-9/11 monitoring exam. Serum biomarkers from the first post-9/11 blood draw, including IgE and cytokines, were available for a subgroup of the study cohort (N=215). Serum was stored at −80°C; IL-4 and IFN- γ were assayed with EMD Milipore HSTCMAG28SPMX21 and IgE with HGAMMAG-303E.

Participants’ most recent BD-PFT from the 9/11/2001–9/10/2017 period provided the pre- and post-bronchodilator FEV₁ and forced vital capacity (FVC) measurements used to define our main outcome. A bronchodilator response of a >12% and 200 ml increase from baseline FEV₁ diagnosed asthma.¹⁶ COPD was defined according to the GOLD criteria, which requires FEV₁/FVC ratio<0.7 on a post-bronchodilator PFT.⁷ We classified individuals who had a bronchodilator response and FEV1/FVC≥0.7 as having isolated-asthma, and those who had FEV1/FVC<0.7 and no bronchodilator response as having isolated-COPD. Individuals who met the criteria for both asthma and COPD had asthma/COPD overlap. Total lung capacity (TLC) and residual volume (RV) measurements were also available from the BD-PFT data; these were measured prior to bronchodilator administration. We used spirometric measurements from 22,737 routine monitoring PFTs (always done without BD) taken between 9/11/2001 and 9/10/2017 to assess post-9/11 FEV₁ trajectories. FEV₁ values from post-9/11 monitoring PFTs that occurred prior to the BD-PFT were used to determine whether patients with asthma and/or COPD had accelerated (>64 ml/year) or expected (≤64 ml/year) FEV₁ decline post-9/11, but prior to our outcome determination; for individuals who had neither diagnosis, all post-9/11 FEV₁ values were included in the FEV₁ decline rate calculation. Pre-9/11 FEV₁ and FVC measurements were available from 1,265 spirometries performed at FDNY monitoring in the year prior to 9/11 (9/11/2000–9/10/2001).

Demographic and other characteristics of the study population and serum biomarker subgroup were assessed as proportions and means (±SD), with independent sample t-tests or chi square tests used to evaluate differences, as appropriate. Longitudinal FEV₁ % predicted, FEV₁/FVC ratio and post-9/11 rate of FEV₁ change were estimated in four subsets of the population defined by outcome (isolated-asthma, isolated-COPD, asthma/COPD overlap, or neither condition) using linear mixed effects models with random intercepts. Participants’ age on 9/11, height and race were included as fixed effects in the models with absolute FEV₁ or FEV₁/FVC ratio as the outcome. Mean FEV₁ % predicted and FEV₁/FVC ratio values in the four groups were estimated for each one-year period between 9/11/2000 and 9/10/2017, and mean rates of FEV₁ change were determined using the post-9/11 spirometry data.

We performed log-rank Mantel-Cox tests to examine the univariable associations of post-9/11 FEV₁ trajectory (accelerated vs. expected FEV₁ decline), eosinophil concentration and smoking status with incident asthma/COPD overlap, followed by multivariable marginal Cox regression models for multiple events to evaluate shared and distinct risk factors for isolated-asthma, isolated-COPD, and asthma/COPD overlap. Censoring occurred at the time of the BD-PFT. Blood eosinophil concentration was assessed first as a binary (≥300 cells/μl vs. <300 cells/μl) and then as a continuous variable. We carried out two sensitivity analyses: one that excluded individuals with FEV₁/FVC ratio<0.7 on a pre-9/11 monitoring PFT (N=69), and another that examined the relationship between pre-9/11 eosinophil concentration and the outcomes of interest (N=1,008). Absolute change in eosinophil concentration from pre- to post-9/11 was also investigated. A multivariable-adjusted Cox regression analysis for multiple events data was also performed in the serum biomarker subpopulation (N=215) in order to determine whether log₂ transformed serum IgE and 21 cytokines were associated with isolated-asthma, isolated-COPD, or asthma/COPD overlap. After Bonferroni correction, the significance cut-off for the serum biomarker analyses was set at a two-sided p-value of 0.0024. For all other analyses, reported p-values are two-sided and considered significant at the <0.05 level. Multivariable models included age, race, smoking status, WTC exposure, first post-9/11 FEV₁/FVC ratio and BMI as covariates. Covariates were selected based on theory. Data analyses were performed using SAS version 9.4. We created figures using Prism 7.

Demographic and other characteristics of the 2,137 firefighters with clinically-indicated post-9/11 BD-PFT in the final study population (Figure 1) and those without BD-PFT are presented in Table 1. Compared with WTC-exposed firefighters who did not have a BD-PFT, the study population was slightly different in that it was older, had a higher BMI and post-9/11 blood eosinophil concentration, and a greater proportion of ever-smokers. These differences were more pronounced in those who would develop post-BD FEV₁/FVC ratio<0.70. The serum biomarker subgroup was similar to the study population, with the exception of having a smaller proportion of ever-smokers.

Clinically-indicated BD-PFT diagnosed isolated-asthma in 202 individuals (9.4%), isolated-COPD in 215 (10.1%), and asthma/COPD overlap in 99 (4.6%) (Figure 2). At the time of BD-PFT, the asthma/COPD overlap subgroup had a lower pre-BD FEV₁ % predicted, lower pre-BD FEV₁/FVC ratio and higher RV/TLC ratio than any other diagnostic category (Table 2). Bronchodilator response was similar in asthma/COPD overlap and isolated-asthma patients (22.6±13.3% vs. 19.9±12.8% increase in FEV₁, respectively, p=0.09).

Both the pre-9/11 and first post-9/11 FEV₁ % predicted in each subgroup were on average ≥80% on monitoring PFTs, but were lowest in those who went on to develop asthma/COPD overlap (Figure 3A). FEV₁/FVC ratio on the first post-9/11 monitoring PFT was also lowest in this subgroup (Figure 3B). The annual post-9/11 FEV₁ loss in individuals with asthma/COPD overlap was similar to that of the COPD subgroup (47.6 ml/year, 95% CI: 43.5–51.6, and 47.2 ml/year, 95% CI: 44.7–49.6, respectively), and greater than the rate of FEV₁ loss in those with isolated-asthma (43.4 ml/year, 95% CI: 40.7–46.2) or neither outcome (36.8 ml/year, 95% CI: 35.9–37.6).

Univariable analyses showed that the incidence of asthma/COPD overlap was elevated in participants with post-9/11 eosinophil concentration ≥300 cells/μl (HR: 1.69, 95% CI: 1.00–2.81, p<0.05) (Figure 4A), those with a history of smoking (HR: 1.67, 95% CI: 1.11–2.50, p=0.02) (Figure 4B) and those experiencing post-9/11 accelerated FEV₁ decline (HR: 2.05, 95% CI: 1.22–3.43, p=0.006) (Figure 4C). In multivariable marginal Cox regression models for multiple events, asthma/COPD overlap was predicted by eosinophil concentration ≥300 cells/μl (Table 3a). Eosinophil concentration was not significantly associated with isolated-asthma or isolated-COPD. When isolated-asthma and asthma/COPD overlap were compared directly, asthma/COPD overlap was still associated with elevated eosinophils (Table 3b). Results were similar if analyses were restricted to those who had eosinophils measured less than 15 months after 9/11 (data not shown). A unique risk factor for isolated-asthma was high intensity WTC exposure, and for isolated-COPD was ever-smoking. Post-9/11 accelerated FEV₁ decline was associated with all three outcomes. The observed associations did not change when eosinophil concentration was assessed as a continuous variable (data not shown).

To confirm that elevated post-9/11 eosinophil concentration and accelerated FEV₁ decline were associated with incident asthma/COPD overlap, we carried out a sensitivity analysis excluding patients with a pre-exposure PFT that showed a FEV₁/FVC ratio<0.7 (N=69). First post-9/11 eosinophil concentration ≥300 cells/μl and accelerated FEV₁ decline both remained significant predictors of asthma/COPD overlap (HR: 1.67, 95% CI: 1.03–2.71, p=0.03 and HR: 2.15, 95% CI: 1.35–3.43, p=0.001, respectively). To assess if pre-exposure blood eosinophil levels were indicative of a predisposition to asthma/COPD overlap, we performed another sensitivity analysis using pre-9/11 eosinophil concentration in place of the post-9/11 measurement. The subgroup of participants who had had a pre-9/11 blood draw (N=1,008) had baseline characteristics and lung function similar to those of the full study population (data not shown). We found that pre-9/11 eosinophil concentration ≥300 cells/μl was also associated with the outcome (HR: 1.42, 95% CI: 1.22–1.66, p<0.001). Change in eosinophil concentration from pre- to post-9/11 was not associated with asthma/COPD overlap (data not shown).

In order to gain further insight into the immunological pathways associated with isolated-asthma, isolated-COPD, and asthma/COPD overlap, we examined serum Th1, Th-17 and Th2 biomarkers obtained within six months of 9/11. A multivariable marginal Cox regression analysis for multiple events in the serum biomarker subpopulation (N=215) showed that higher early post-9/11 IgE was associated with incident asthma/COPD overlap, but this result was not significant after adjustment for multiple comparisons (Table 4). We found that elevated IL-4 predicted asthma/COPD overlap and elevated IL-21 predicted both isolated-asthma and isolated-COPD, while elevated IFN-γ was a protective factor for isolated-asthma and isolated-COPD. Early post-9/11 levels of IL-5, IL-13, IL-17, IL-23 and IL-6 were not associated with any of the three mutually exclusive outcomes (data not shown).