Study design and subjects
From January 2014 to December 2018 an observational prospective cohort study was carried out. We enrolled all consecutive patients (n = 115) referred to the Center for rare lung disease of the University Hospital of Modena and to the Pneumological clinic of the Federico II University Hospital of Napoli with a new diagnosis of IPF. Diagnosis of IPF was performed according to the 2011 criteria of the American Thoracic Society/European Respiratory Society [1].
Data were registered in an ad hoc database. At baseline, we collected for each patient socio-demographic characteristics [age, sex, smoking status, amount smoked (pack-years), educational level, lifetime occupational history, exposure to asbestos] and clinical parameters [body mass index (BMI), dyspnea, time of onset of respiratory symptoms, comorbidities, pharmacological treatment, disability perceived in relation to health conditions and psychological distress]. We also recorded pulmonary function tests (PFTs) and calculated the prognostic indices Gender, Age, Physiology (GAP) and the Composite Physiologic Index (CPI).
Of the 115 patients recruited, 101 were treated with pirfenidone or nintedanib and of those 89 agreed to perform PFTs after 12 months of therapy. The study was concluded when the last recruited patient completed the 12 month treatment period.
All the 115 patients were censored for death and the date of long-term oxygen therapy (LTOT) initiation for the entire study period.
A written informed consent was provided by all participants before recruitment. The study was conducted in accordance to the Declaration of Helsinki and approved by the institutional ethics committees of University Hospitals of Ferrara, Modena and Napoli (N.160494).
Measurements
Sociodemographic and clinical features
BMI was calculated by dividing weight (Kg) by height squared (m2). The number of pack years was figured as number of cigarettes smoked per day x number of years smoked/20. Level of dyspnea was assessed by the modified Medical Research Council (mMRC) scale. Disability perceived in relation to health conditions and psychological distress were measured using the World Health Organization Disability Assessment Schedule (WHODAS) 12 items version [9] and the Hospital Anxiety and Depression Scale (HADS) [10], respectively.
Occupational exposure
We collected a complete occupational history, including a job activity checklist and a specific checklist of occupational dust exposure related to IPF (organic dusts; stone, sand, or metal dusts; and wood dust) [6].
The recorded information included job title, tasks performed, detailed description of the activity, use of individual protection devices, substances contact, years spent in each job and in each occupational dust exposure. Occupational exposure was defined as occupational exposure to dusts related to IPF by 10 or more years before diagnosis [5].
Lung function
PFTs were performed according to international criteria [11]. To assess the possible influence of occupational exposure on respiratory function and disease progression, forced vital capacity percent of predicted (FVC % pred.) and diffusing capacity of the lungs for carbon monoxide percent of predicted (DLCO % pred.) at diagnosis and at 12 month follow-up were used.
Prognostic predictors
Gender, Age, Physiology (GAP) index is a validated, multidimensional tool that predicts mortality in IPF. The calculation of the score encompasses gender (G), age (A) and two lung physiology variables (P) (FVC % pred. and DLCO % pred.). Points are assigned for each variable to obtain a total range from 0 to 8. According to this score, patients are classified in stage I (0–3 points), stage II (4–5 points), or stage III (6–8 points) [12]. As the GAP increases, the probability of mortality rises.
Composite Physiologic Index (CPI) is a validated, multidimensional index that correlates with the extent of pulmonary fibrosis and mortality and thus predicts IPF progression [13, 14]. CPI is computed as follows: CPI = 91.0 - (0.65 × DLCO % predicted) - (0.53× FVC % predicted) + (0.34 × FEV1% predicted). Higher CPI scores indicate more severe fibrosis and poorer prognosis [14].
Both GAP and CPI index were calculated at diagnosis and at 12 month follow-up, although GAP has not been circumstantially validated at 1 year.
Initiation of long-term oxygen therapy (LTOT)
LTOT can be defined as oxygen used for at least 15 h per day in chronically hypoxaemic patients [15]. In patients with IPF, LTOT initiation may be a marker of poor prognosis as it predicts a median survival of less than 18 months [13]. The date of long term oxygen initiation was recorded for each participant.
Data analysis
IPF patients were classified according to exposure in two groups: exposed (≥ 10 years) and not exposed (< 10 years).
First, at baseline, we investigated whether the two groups differed in selected demographic variables, clinical characteristics and lung functional parameters (FVC; FEV1; DLCO; GAP and CPI index), using chi-squared and Kruskal-Wallis tests for categorical and continuous variables, respectively.
Second, we evaluated whether the two groups of patients differed in lung function measurements and prognostic predictors measured after 12 month therapy.
Overall, a total of eight multiple regression models were fit. However, as FVC and FEV1 at diagnosis were highly collinear (Spearman rho = 0.96), only the analyses related to FVC were reported to avoid redundancy.
In all models, covariates were included in a stepwise forward process using the following criteria: clinical relevance, with gender, age at symptoms onset, smoke and occupational dust exposure forced to entry. Occupational dust exposure was treated either as continuous or ordinal variable, including the above mentioned two groups of exposure (< 10 years and ≥ 10 years) as dummy variables.
The validity of final regression models was assessed as follows: the assumption of constant error variance was checked graphically, plotting Pearson residuals vs. fitted values, and formally, using the Cook-Weisberg test for heteroskedasticity. High leverage observations were identified by computing Pearson, standardized and studentized residuals, and Cook’s D influence. In all models, we found less than 10 high-leverage observations, excluding which we noted no substantial changes.
As a separate, additional evaluation, we tested with Cox proportional hazard analysis whether there was any evidence that starting oxygen therapy depended on: (a) previous work exposure to dusts lasting ≥10 years; (b) number of cigarette pack-years; (c) baseline FEV1; (d) IPF stage at baseline (separately assessed using GAP and CPI index). We selected all covariates a priori, and, in order to avoid overfitting, we fitted two separate models, each including one of the two IPF scoring systems, with all other covariates remaining stable. Finally, we used Schoenfeld’s test to check the validity of proportional hazards assumption for both models.
Statistical significance was defined as a two-sided p-value< 0.05, and all analyses were carried out using Stata, version 13.1 (Stata Corp., College Station, Texas, USA, 2013).