Idiopathic Pulmonary Fibrosis (IPF) is a progressive, scarring lung disease with a poor prognosis and median survival of 3 years. It is a heterogeneous disorder with varying rates of progression which presents a challenge for accurate prognostic prediction. The composite physiologic index (CPI) and the Gender, Age and Physiology (GAP) score are validated scoring systems for prognostic determination in IPF. Our data suggest these scoring systems have limited usefulness and we have undertaken a modelling approach to evaluate clinical prognostic parameters.
Methods Gender, age, smoking history, presence of emphysema on HRCT thorax and echocardiogram confirmed pulmonary hypertension were collected retrospectively from 253 IPF patients (in accordance with ATS/ERS criteria and MDT consensus) from a single centre in the UK between 19th April 2007 and 13th November 2014. Lung function including FEV1, FVC, DLco and 6 minute walk test (distance, resting and minimum oxygen saturation and maximum heart rate) were collected at baseline, 6 and 12 months of follow up. Survival data were censored at 1st January 2016. The relationship between GAP or CPI and survival was analysed by Spearman’s correlation, ROC area under the curve and Chi2 analysis. Multivariate analysis and linear regression were used for the modelling.
Results Of the 253 patients included 188 were male (74%) with age 71.4 ± 8.3 years (mean ± SD). There were 164 (64.8%) ex-smokers and 12 (4.7%) current smokers. At presentation 19 patients had pulmonary hypertension and 35 had evidence of emphysema on HRCT thorax. Baseline lung function FEV1 79 ± 22% predicted, FVC 82 ± 19% predicted, DLco 45 ± 15% predicted (mean ± SD). Median survival was 1169 days (3.2 years). The association between survival and CPI (r2 0.59, p < 0.01) or GAP (r2 0.45, p < 0.01) was modest. However ROC curve analysis demonstrated that GAP and CPI were poor predictors of survival. Chi2 analysis shows there is no significant difference between these scoring systems. Multivariate analysis demonstrated that baseline% predicted DLco (r2 = 0.32, p = 6×10–20) (Figure 1) had the strongest association with survival.