Study Design and Participants. The FDNY WTC-health program (WTC-HP) electronic medical record (EMR) will be used to obtain clinical variables such as age, gender, years of FDNY service, WTC site exposure level, and lung function measures, as previously described. 22,27,62–65,71 Our observational study is NYU IRB Approved # 21–00679 and available at clinicaltrials.gov #NCT05216133. Study Definitions and Inclusion/Exclusion Criteria can be found in Table 1.
Study Population: Source Cohort. All participants in the WTC-HP (n = 14,976) were screened, Fig. 2. Inclusion Criteria: i. Actively consented and enrolled member of the WTC-HP. ii. Pre-9/11 spirometry with Forced Expiratory Volume in 1 second (FEV1) ≥ Lower Limit of Normal (LLN) iii. Male Firefighter status on 9/11 with exposure at the WTC-site and entry into WTC-HP before the site closure on 7/24/2002. Exclusion Criteria: i. lung disease prior to 9/11 as defined by positive methacholine or bronchodilator test, or FEV1 < LLN. ii. Not part of initial cohort in data extraction from August 1, 2017.72 After all inclusion/exclusion criteria applied, the baseline cohort consists of n = 4,192. Sub-cohort Development. A representative cohort of 20% was randomly selected (n = 837; SPSS v. 28) from the above baseline cohort, Fig. 2.
Recruited Cohort will be developed to assess for noninvasive biomarkers. We will recruit a subset N = 40/group (i. AHR only ii. GERD only iii. BE iv. GERD/BE and AHR overlap or v. No GERD nor AHR) from the sub-cohort, Fig. 2. Recruitment strategies will include: i. Direct mailings; ii. Email (potential participants will be sent the same IRB-approved recruitment message to their personal emails using end-to-end encryption; iii. Study website will include recruitment messages providing general information on the study and answers to frequently-asked questions. No direct communications will be made with participants through the website, and no PHI will be used or available within the study website; iv. Telephone contact. A description of the study will be provided to potential participants and, upon their expression of interest, the investigator will perform an eligibility screening. In addition to meeting the inclusion criteria as outlined above, participants should: i. have available serum from their first post 9/11 WTC-HP ii. not currently be receiving treatment for malignancy iii. have no limitations to a minimal risk blood draw iv. be willing and able to sign consent; and v. be able to attend a single-visit.
Case Status. WTC-AHR will be defined as having a positive methacholine (PC200 < 16), or a positive bronchodilator response (by ATS/ERS guidelines with improvement of FEV1 by 12% and at least 200mL) at least once post-9/1173,74 and/or EMR diagnosis. GERD will be defined as: biopsy-proven erosive esophagitis LA grade C or D; stricture or Barrett’s esophagus on endoscopy; and/or esophageal acid exposure time > 6% on a pH or pH impedance study. GERD will also be defined on EMR diagnosis and/or PPIs, H2 blockers, antacid, or surface agent use.75 BE, as a subset of GERD, will have any of the following additional inclusion criteria: biopsy-proven columnar epithelium lining ≥ 1cm of the distal esophagus with intestinal metaplasia characterized with goblet cells on histology; diagnosis on EMR, Table 2-3.75 The recruited participants will be consented prior to any research activity and measurement visit via REDCap software or in person.
Measurement Visit. Participant demographic information, medical history and medication history will be obtained. A physician will perform the physical examination, and verify that inclusion/exclusion criteria are met. Enrolled participants will undergo the following assessments.
Blood Sampling. After at least an 8 hour fast, serum and plasma will be obtained, aliquoted and banked. Each stored specimen will be assigned a unique code to ensure proper identification and linkage to the respective participant. Aliquots from the fresh samples will be assayed for complete blood count (with differential) and chemistry panel. These data are already available for the banked samples. For all samples, lipid profile, metabolomics, and protein biomarker profiling will be performed.10,28–30,76,77
Salivary Pepsin Assessment. 30mL sterile plastic tubes with 0.5 ml of 0.01M citric acid, adjusted to a pH of 2.5 (RD Biomed Ltd., Hull, UK), will be used by the participants to collect saliva in the AM (prior to brushing teeth, drinking or eating), 1h after finishing lunch, and 1h after finishing dinner.78,79 Participants will be instructed to cough a few times prior to spitting into the tube to clear saliva from the back of the throat and then spit into the tube. The collected samples will be stored at 4°C and analyzed within 2 days. Salivary Pepsin will be analyzed using Peptest (RD Biomed Ltd., Hull, UK) as previously described.79 Briefly, plastic tubes will be centrifuged at 4,000 rpm for 5 minutes, and 80µL of supernatant will be added to 240µL of migration butter solution for 10 seconds. 80µL of the mixture will be added to the well of the Peptest, which contains two unique human monoclonal antibodies that detect and capture pepsin protein (specific to pepsin-3), with a lower limit of detection of 16 ng/mL and an upper limit of 500 ng/mL. A salivary pepsin level of ≥ 16 ng/mL will be considered positive. The sample will be processed in a Pepcube reader to quantify the pepsin concentration.78
Spirometry will be assessed using a KoKo PFT spirometer (nSpire Health Inc), and lung function assessment will be considered acceptable as per the ATS/ERS guidelines.80 We will select the largest acceptable measures for electronic archiving. Each participant’s predicted percentage (%) will be calculated by NHANES III equations based on their age at examination, height, sex, and race.80,81
FeNO will be quantified using NIOX VERO® (Aerocrine).82,83 Participants will be instructed to inhale to their total lung capacity via mouthpiece for 2–3 seconds. Then, they will exhale at a flow rate of 0.05L/second. The device will provide results in parts per billion (ppb).
Exhaled breath condensate (EBC ) will be collected using RTubes (Respiratory Research, Inc., USA).84 Approximately 1-2mL of EBC sample will be obtained after 10 min of quiet normal breathing.85 PH measurement. EBC pH assay is extremely simple to perform, inexpensive, and robust, and can be easily processed on the day of collection.86 EBC will be de-aerated of CO2 by bubbling free argon gas (350ml/min) under a micro-pH reader (Orion PerpHecT micro-pH electrode) and stabilized pH will be recorded after approximately 3–5 minutes.87 Aliquots are then stored at -80⁰C and thawed only once prior to histamine and biomarker assessment.
Naso/oropharyngeal microbiome. Collection. Trained study team members will collect naso/oropharyngeal samples using commercially available kits (OMR-110 by DNA Genotek, Canada). Each naris will be swabbed in a circular fashion 10 times. The oropharyngeal sample will be collected by swabbing in the back of the throat in 10 circular motion to ensure sufficient swab collection. Each absorbent swab will be placed into a vial containing 1 mL of stabilizing liquid using aseptic technique. The sample will be treated with lyophilized Proteinase K, and incubated in the original vial at 50⁰C for 1 hour in a water bath prior to aliquoting for long-term storage at -80⁰C.
Quality of Life, Aerodigestive Disease and End-Organ Effect Questionnaires.
Gastrointestinal impact will be assessed using with the Patient Assessment of Upper Gastrointestinal Disorders – Quality of Life (PAGI-QoL) and the Patient Assessment of Upper Gastrointestinal Disorders Symptom Severity Index (PAGI-SYM). Both questionnaires use a 6-point Likert scale (MAPI Research Trust).88–91
Respiratory and QoL assessment will utilize the Health-Related Quality of Life measures (HRQL)92, St. George’s Respiratory Questionnaire (SGRQ), and the Short-form-36 (SF-36). HRQL assesses an individual's perceived physical and mental health. The SGRQ is a standardized, self-administered airways disease-specific questionnaire divided into three subscales- symptoms, activity, and impact.93 SF-36 will capture supplemental information about their mental health, general health perception, emotional, and social role functioning.94
Cognition will be assessed using the Montreal Cognitive Assessment (MoCA; version 8.1) and the Mini-Mental State Examination (MMSE). MMSE is a cognitive test used to evaluate early dementia.95,96 Combining MoCA and MMSE can improve diagnostic utility.97 The MoCA will be administered by a trained/certified investigator. Members of our research team have completed MoCA training and certification through a validated MoCA cognition portal98 (https://mocacognition.com/). Similar to the MoCA, the MMSE assesses orientation, memory, visuospatial and language domains. Additionally, the MMSE evaluates comprehension, reading and writing.99 The PI will thoroughly review all scores.
Power Analysis. A sample size of 40 cases for each group of GERD, AHR, AHR/GERD overlap, BE, and non-GERD/non-AHR Controls (all will be subsets of AIM 1 N = 898 randomly selected cohort) achieves 80% power to detect difference as small as 0.78 SD with two-sample t-test at 0.01 significance level to account for multiple comparisons. This will allow us to achieve 80% power and significance of 0.05, based on prior studies with salivary pepsin test (personal communication with Dr. Peter Dettmar of Peptest), Fig. 2.
Statistical Analysis SPSS 28 (IBM) will facilitate database management and statistics. Continuous variables expressed as mean, standard deviation (SD) if normally distributed, and as median, inter-quartile range (IQR) if skewed. Two-sample t-test and ANOVA will compare continuous data. Count and proportions will summarize categorical data and Pearson-χ2 will compare categorical data. Multivariate binary logistic regression will estimate biomarker-disease relationship for case status as a binary outcome while adjusting for confounding. Cox proportional hazards model will evaluate the effects of biomarkers, smoking, and exposure on the hazard of developing WTC-GERD or BE over time. The maximum potential effectiveness of a biomarker will be calculated by Youden Index.100 Goodness of fit, using the Hosmer-Lemeshow test. Survival curves compared by Log-rank test. Pearson 𝜒2-test will compare SABA and LABA usage between GERD, AHR, AHR/GERD overlap, BE, and non-GERD or AHR controls. Significance will be assessed by p < 0.05 for all statistical tests. Graphs will be created using Prism (v.10, GraphPad Software).
Missing data
Variables with missing values in a small proportion of participants will be imputed using multiple imputation methods. To assess the missing at random assumption, we will evaluate the comparability between samples with missing data and those without. Sensitivity analysis will be performed by comparing the results obtained from the complete data analysis to the results obtained from multiple imputation.
Model Building. We have previously identified key biomarkers using a machine learning approach.10,28–30 We have further refined this analysis pipeline and will utilize this methodology to identify AHR, GERD, AHR/GERD overlap, and BE biomarkers. Specifically, we will utilize random forests (RF) of the filtered, normalized biomarkers. Models assessed via a modified hamming distance between variable importance rankings of models with identical hyper-parameters. A refined profile of the top 5% of important biomarkers by MDA will be included in a gradient-boosted tree model (xgboost package, R-Project) to build a classifier of AHR, GERD, AHR/GERD overlap, and BE. A random hyperparameter space search determined a final model that maximized AUCROC.
We will also use linear mixed-effects models will be used to assess the temporal trend of biomarkers with time adjusting for confounders. The longitudinal biomarkers processes will be associated to risk of developing WTC-GERD/BE using the joint modeling technique.101 The joint-modeling approach has become the primary method for analysis of longitudinal biomarker process and time-to-event outcome, and multiple R packages are available to implement the models. We will also consider a single index longitudinal model which enables us to reduce the dimensionality of multiple biomarkers and to evaluate joint effects of multiple biomarkers together to identify key risk factors. The single-index model incorporates longitudinal data to calculate hazard of each parameter as well as personalized dynamic risk for prognostication. Specifically, this will allow us to use a patient’s data from a single clinical exam to identify risk of GERD, AHR, overlap, or BE. Furthermore, this will allow the identification of false negatives and undertreated cases in the entire FDNY cohort.