Effect of smoking and comorbidities on survival in idiopathic pulmonary fibrosis

A total of 223 patients (91 female, 132 male) with pulmonary fibrosis treated in Kuopio University Hospital between 1^st January 2002 and 31^st December 2012 were collected from medical records of the hospital by using International Classification of Diseases version 10 (ICD-10) codes J84.1, J84.8 and J84.9. Clinical, radiological and histological information of each patient was transferred from medical records to special data collection forms designed for the present study. Pulmonary fibroses with a known etiology, i.e. connective-tissue disease, occupational exposure etc., were excluded. The data concerning comorbidities and survival was updated on 16^th October 2016.

Smoking status was categorized as non-smoker, ex-smoker and current smoker and collected as years, pack-years or both if available. Ex-smokers were defined as having smoked regularly for more than one year, and longitudinal changes in smoking habits after the diagnosis of IPF were not considered. The presence of comorbid diseases i.e. COPD, asthma, hypertension, heart failure for any reason, CAD, cerebral infarction, transient ischemic attacks, OSA, lung cancer, GER, depression, diabetes and other cancers were gathered from the medical records; in this case the time of diagnosis was the cut-off for subdivision into either before or after the diagnosis of IPF. Pulmonary function tests (PFT) included results of spirometry i.e. forced vital capacity (FVC), forced expiratory volume in one second (FEV1) and FEV1/FVC ratio, diffusion capacity i.e. diffusion capacity of carbon monoxide (DLco) and potential value of diffusion capacity per liter of lung volume (DLco/VA). PFT results were expressed as absolute numerical values and percentages of predicted values. Information from histological samples from surgical lung biopsy or autopsy was gathered.

No consents for inclusion into this study were gathered since this was a retrospective study and the majority of the patients are deceased. The study protocol was approved by the Ethical Committee of Kuopio University Hospital (statement 17/2013) and from the National Institute for Health and Welfare (Dnro THL/1052/5.05.01/2013). Permission to use data from death certificates was given by Statistics Finland (Dnro: TK-53-911-13). This study was conducted in compliance with the Declaration of Helsinki.

The PFT were evaluated using the Finnish reference values [20]. CPI was calculated from the PFT results as follows: 91.0 – (0.65 x % predicted DLco) – (0.53 x % predicted FVC) + (0.34 x % predicted FEV1) [21]. The findings of the first and last high resolution computed tomography (HRCT) of each patient were re-evaluated by a radiologist according to the recent statement of the American Thoracic Society (ATS) and European Respiratory Society (ERS) and re-classified as usual interstitial pneumonia (UIP), possible UIP and not UIP [22]. All the patients whose re-analyses were categorized as not definite UIP in HRCT were evaluated by a multidisciplinary discussion with a radiologist, a pathologist and a pulmonologist before being included into the present study. Either death or lung transplantation were considered as end-points in the survival analyses.

Statistical analysis was performed with IBM SPSS Statistics version 21. Group differences were tested by Kruskall-Wallis or Mann-Whitney U-test for continuous variables and by Chi-square testing or Fisher exact test for categorical variables. Data is presented as mean with standard deviation for continuous variables or frequencies with percentages when variables are categorical. Survival curves were estimated by the Kaplan-Meier method and differences in survival were compared using log-rank test, in addition median survival is expressed. Univariate and multivariate survival analysis were computed using Cox regression models. Results from Cox survival analyses are shown in the form of a hazard ratio with 95 % confidence intervals. An ordered logistic regression was performed in order to evaluate the impact of age and smoking status on the number of comorbidities. P-value < 0.05 was considered as statistically significant.

A total of 132 patients with IPF were evaluated in this study; of these, 97 (73.5%) were male and 35 (26.5%) female. The mean age of the patients at the time of diagnosis was 70.5 years and the median survival 45.0 months. When we updated the data (16^th October 2016), a total of 115 patients had deceased and 15 patients were still alive. Three patients (one of whom had died) had undergone lung transplantation. Two males and 2 females had an unknown smoking history. From the 128 (96.7%) patients with a known smoking history, 45 (35.2%) were non-smokers, 66 (51.2%) ex-smokers and 17 (13.4%) were current smokers. There were more female non-smokers (p <0.001) and fewer ex-smokers than their male counterparts (p <0.001) but no gender difference was seen in the proportion of current smokers (p=0.410). Current smokers had smoked for significantly more years and more pack-years than ex-smokers and they were also significantly younger at the time of diagnosis and at death compared to non-smokers and ex-smokers; this remained the case when male patients were analyzed separately. Current smoker females were younger at the time of diagnosis compared to ex-smoking females (0.036) and they died at a younger age than their non-smoking counterparts (p=0.038). Female patients had significantly longer survival time compared to males, 42.2 months and 29.1 months, respectively (p=0.039) and ex-smokers had a shorter survival time compared to current smokers and non-smokers (Table 1, Fig. 1). No significant differences in PFT results, CPI values or HRCT findings were detected between the groups of different smoking histories (Table 1).

In the univariate analyses, DLco% and CPI were significantly related to survival: for DLco% hazard ratio (HR) was 0.97 with 95% confidence interval (95% CI) 0.96 – 0.98 and with p-value <0.001 and for CPI HR was 1.04, 95% CI 1.02 – 1.06 and p-value < 0.001. For that reason DLco% and CPI were used in the severity adjustment in the multivariate analyses.

When survival differences were compared between ex-smokers and current smokers in step-by-step multivariate analyses i.e. adding one factor at a time to the model using DLco % and CPI in severity adjustment, the survival difference in favor of current smokers was reduced to a marginally non-significant level (p=0.098 and p=0.128, respectively). When age at the time of diagnosis was added into the multivariate analyses, smoking history no longer exerted any statistically significant effect on survival (Table 2). When survival differences were compared between ex-smokers and non-smokers, the better survival of non-smokers disappeared after severity adjustment with DLco% and CPI while age remained as a significant predictor of survival (Table 3). Male gender was found to be a significant risk factor for shorter survival when comparing ex-smokers and non-smokers, but not in the comparison between ex-smokers and current smokers (Tables 2 and 3).

Twenty-one (15.9%) of the patients did not have any comorbidities while 36 (27.3%) had one, 30 (22.7%) had two, 20 (15.2%) had three, 21 (15.9) had four and 4 (3.0%) had five comorbidities. The most common comorbidities were cardiovascular diseases (CVDs) (72.7 %) (Fig. 2). Females were more likely than males to suffer from asthma, hypertension or diabetes. Current smokers had significantly more COPD (p=0.000) and lung cancer (p=0.006) compared to ex-smokers, this difference was observed in males, but not in females when the data was subdivided according to genders (Table 4). The multivariate analyses were adjusted for age, gender and smoking status and in addition, DLco % or CPI in two different models (Table 5). In multivariate analysis with DLco %, CVD and COPD were related to poorer survival and in the multivariate analysis with CPI, it was noted that the use of insulin was related to poorer survival (Table 5). In the ordered logistic regression, age, but not smoking history, exerted a significant impact on the number of comorbidities (standardized beta coefficient 1.044, p= 0.003), but the number of comorbidities did not have any statistically significant effect on survival even in multivariate analyses.

Comorbidities had been most often diagnosed before IPF (Fig. 2). CVDs and CAD were more often diagnosed before IPF in ex-smokers compared to current smokers (p <0.001, for both) and non-smokers (p=0.011 and p=0.016, respectively). CVDs were also more likely to be diagnosed before IPF in non-smokers than in current smokers (p=0.044) and in non-smoking males in comparison to male ex-smokers (p=0.002). Non-smokers (p=0.046) and ex-smokers (p=0.001) had received a GER diagnosis more often before the diagnosis of IPF compared to current smokers (p=0.046 and 0.001, respectively). After adjustment for age, smoking status, gender and separately both DLco % and CPI, it was found that the time of the comorbidity diagnosis in relation to the IPF diagnosis did not have any effect on survival.

Pharmacological treatments for IPF are presented in Table 6, symptomatic medical treatment in Table 7 and medications for comorbidities in Table 8.