Serial change in PFTS is predictive of outcome in systemic sclerosis associated interstitial lung disease.

Abstract

Rationale: Interstitial lung disease (ILD) is a leading cause of mortality in systemic sclerosis (SSc) but is not severe or progressive in all patients. Serial pulmonary function tests (PFTs) can predict outcomes but there are no established data regarding what is clinically meaningful decline. We sought to quantify change in PFTs as a predictor of outcome in SSc-ILD. Method(s): SSc-ILD patients were identified through a nationwide cohort study. Forced expiratory volume in one second (FEV1), forced vital capacity (FVC), diffusing capacity for carbon monoxide (DLCO) and KCO (DLCO/alveolar volume) were collected during follow up. Demographic and disease-related data were prospectively collected. The rate of change over four years in the absolute value for each variable was calculated for each patient using a regression line of best fit. Percentage change in absolute values over the first year of follow-up from diagnosis of SSc-ILD (+/-90 days) was determined. The composite outcome variable was deterioration (need for home oxygen or lung transplantation, or death). Univariable Cox regression models were used to determine the relationship between each of rate of decline and percentage decline in each PFT variable and outcome. Receiver operating characteristics (ROC) curves were used to determine the best cut-off points to predict adverse outcomes. Result(s): Among 264 patients, there were 49 events (38 deaths, 10 prescribed supplemental oxygen, one lung transplant) over a mean (+/-SD) follow-up of 3.0 (+/-1.7) years. The rates of decline over time in each of FVC, DLCO and KCO were significantly predictive of deterioration or death (hazard ratios [HR] of 0.06, 0.637 and 0.06 respectively, p values all <0.0001). Cut-off values of zero (i.e. no overall change in PFTs over four years) gave the optimal sensitivity-specificity trade-off with negative predictive values (NPVs) of 88-96%. The percentage changes in FVC, DLCO and KCO over one year from diagnosis (+/-90 days) were significantly associated with outcome (HRs 1.06 [p<0.04], 1.07 [p<0.0001] and 1.05 [p<0.04] respectively). The best trade-off between sensitivity and specificity was a decline in FVC of 10% and in DLCO and KCO of 15% with NPVs of 92-93%. A decline in KCO of 15% predicted a poor outcome with a likelihood ratio of 7.62. Conclusion(s): The course of SSc-ILD is evident within the first 1-4 years. A decline within the first year in FVC of >/=10% and DLCO or KCO of >/=15% bodes badly and identifies patients who should be monitored more closely and considered for therapy.