Pulmonary function and survival in idiopathic vs secondary usual interstitial pneumonia

Abstract

Background: The usual interstitial pneumonia (UIP) pattern of lung injury may occur in the setting of connective tissue disease (CTD), but it is most commonly found in the absence of a known cause, in the clinical context of idiopathic pulmonary fibrosis (IPF). Our objective was to observe and compare longitudinal changes in pulmonary function and survival between patients with biopsy-proven UIP found in the clinical context of either CTD or IPF.

Methods: We used longitudinal data analytic models to compare groups (IPF [n = 321] and CTD-UIP [n = 56]) on % predicted FVC (FVC %) or % predicted diffusing capacity of the lung for carbon monoxide (Dlco %), and we used both unadjusted and multivariable techniques to compare survival between these groups.

Results: There were no significant differences between groups in longitudinal changes in FVC % or Dlco % up to diagnosis, or from diagnosis to 10 years beyond (over which time, the mean decrease in FVC % per year [95% CI] was 4.1 [3.4, 4.9] for IPF and 3.5 [1.8, 5.1] for CTD-UIP, P = .49 for difference; and the mean decrease in Dlco % per year was 4.7 [4.0, 5.3] for IPF and 4.3 [3.0, 5.6] for CTD-UIP, P = .60 for difference). Despite the lack of differences in pulmonary function, subjects with IPF had worse survival in unadjusted (log-rank P = .003) and certain multivariable analyses.

Conclusions: Despite no significant differences in changes in pulmonary function over time, patients with CTD-UIP (at least those with certain classifiable CTDs) live longer than patients with IPF--an observation that we suspect is due to an increased rate of mortal acute exacerbations in patients with IPF.

Figure 1

– A, B, Change in FVC% (A) or DLCO% (B) over time for sample stratified on clinical context of UIP. Points represent observed values, with fitted (predicted) functions for groups (blue = IPF, red = CTD-UIP) superimposed; predicted means are solid, 95% confidence bands are dashed. Predicted values were obtained using mixed-model fits, using knots (ie, allowing change points) at −0.5, 0, 0.5, 1, and 2 y from diagnosis. A spatial power covariance structure was included in the model to account for repeated measures within subjects; random effect terms for subjects were also included, as described in e-Appendix 1. There were no significant differences between groups for segments between comparable time points. The plots demonstrate sharper declines near time of diagnosis for both groups. CTD-UIP = connective tissue disease-related usual interstitial pneumonia; DLCO% = % predicted diffusing capacity of the lung for carbon monoxide; FVC% = % predicted FVC; IPF = idiopathic pulmonary fibrosis.

Figure 2

– Spaghetti plots for FVC% or DLCO% by diagnosis group (IPF or CTD-UIP) and last observation (< 100 wk or ≥ 100 wk). See Figure 1 legend for expansion of abbreviations.

Figure 3

– A-B, Estimated mean FVC% (A) and DLCO% (B) by diagnosis group (IPF = blue, CTD-UIP = red) and last time of follow-up (< 100 wk, solid; ≥ 100 wk, dashed) based on linear mixed-model fits. Plots demonstrate that those with earlier last follow-up tended to have steeper declines in lung function. See Figure 1 legend for expansion of abbreviations.

Figure 4

– Kaplan-Meier survival curves for sample stratified on clinical context of usual interstitial pneumonia (IPF or CTD-UIP). See Figure 1 legend for expansion of abbreviations.

Figure 5

– Kaplan-Meier survival curves for sample stratified on clinical context for usual interstitial pneumonia, including IPF and CTD-UIP subgroups. Other group contained five subjects with antisynthetase syndrome/dermatomyositis/polymyositis, one with mixed CTD, five with primary Sjögren syndrome, and two with systemic lupus erythematosus. RA = rheumatoid arthritis; SSc = systemic sclerosis; UCTD = undifferentiated connective tissue disease. See Figure 1 legend for expansion of other abbreviations.