The Role of Cardiopulmonary Exercise Test in IPF Prognosis

Abstract

Background. In IPF, defects in lung mechanics and gas exchange manifest with exercise limitation due to dyspnea, the most prominent and disabling symptom. Aim. To evaluate the role of exercise testing through the 6MWT (6-minute walk test) and CPET (cardiopulmonary exercise testing) in the survival of patients with IPF. Methods. This is a prospective, observational study evaluating in 25 patients the relationship between exercise variables through both the 6MWT and CPET and survival. Results. By the end of the observational period 17 patients were alive (33% mortality). Observation ranged from 9 to 64 months. VE/VCO2 slope (slope of relation between minute ventilation and CO2 production), VO2 peak/kg (peak oxygen consumption/kg), VE/VCO2 ratio at anaerobic threshold, 6MWT distance, desaturation, and DLCO% were significant predictors of survival while VE/VCO2 slope and VO2 peak/kg had the strongest correlation with outcome. The optimal model for mortality risk estimation was VO2 peak/kg + DLCO% combined. Furthermore, VE/VCO2 slope and VO2 peak/kg were correlated with distance and desaturation during the 6MWT. Conclusion. The integration of oxygen consumption and diffusing capacity proved to be a reliable predictor of survival because both variables reflect major underlying physiologic determinants of exercise limitation.

Figure 1

The cumulative Kaplan-Meier survival plot. (a) Survival of 25 IPF patients followed till death (uncensored: n = 8) or reporting of the study (censored: n = 17), combined by a plot (b) where the number of patients at risk is shown.

Figure 2

Overall death risk through the Cox proportional hazards models. In box plot (a), the Cox model includes only VE/VCO₂ slope, in box plot (b), the Cox model includes VO₂ peak/kg, and in box plot (c), the Cox model includes VO₂ peak/kg + DLCO%. A bidirectional stepwise model selection method minimizing the Bayesian information criterion (BIC) was utilized for selecting the optimal model which was identified as VO₂ peak/kg + DLCO%. Data are described using standard box plots with medians (interquartile range). Risk was found to be significantly differentiated between dead and alive with P = 0.0028, P = 0.008, and P < 0.0001, respectively.

Figure 3

A threshold of mortality was identified. A threshold VO₂ peak/kg of 14.2 mL/min/kg was associated with an increased risk of mortality according to optimum Cox proportional hazards model.