Physiological predictors of exertional oxygen desaturation in patients with fibrotic interstitial lung disease

Introduction: In patients with fibrotic interstitial lung disease, hypoxaemia on exertion is frequent, and contributes to exercise intolerance, exertional dyspnoea and reduce quality of life. The recent AmbOx trial shown that ambulatory oxygen is associated with improved quality of life in ILD patients with isolated exertional hypoxia. However, few data are available on the predictors of desaturation in ILD and no physiological parameter thresholds have been identified. Methods: We analysed predictors of oxygen desaturation (SpO2≤88%) on a 6MWT in patients with ILD without severe resting hypoxia (SpO2 at rest≥92%), in a derivation cohort (N=146) and a validation cohort (n=154) presenting at a single institution. Univariable logistic regression analyses was used to identify variables predictive of oxygen desaturation (SpO2≤88%) on a 6MWT. Any factors potentially associated on univariable analysis (p < 0.10) were added to the multivariable model. Finally, a backward stepwise selection (p-in < 0.05, p-out > 0.10) was used to determine the factors independently associated with SpO2≤88%. Receiver operator curve (ROC) analyses were performed on significant variables derived from the final logistic regression models, and optimal cut-off points for each variable were identified by using the Youden's index. Results: A total of 300 ILD patients were included in the analysis. Patients from the validation cohort had less severe disease (mean CPI: 44.4 (SD 13) vs 52.9 (SD 10.6), p=0.0001) and were less likely to desaturate on 6MWT (26% vs 63%, p-value<0.001). On univariable analysis, SpO2 at rest, DLCO%, FVC%, FEV1% and CPI were associated with desaturation on 6MWT in the derivation cohort, and were confirmed in the validation cohort. On multivariable analysis, only DLCO and SpO2 remained independently predictive of oxygen desaturation on 6MWT in both cohorts. The optimal predictive threshold of DLCO≤40% and SpO2≤95%, respectively, was identified in the derivation cohort and confirmed in the validation cohort. Integrating the two variables into a “DeOX” score (0-2), we have created a prediction model of desaturation on 6MWT where the presence of one or both variables (DLCO≤40% and/or SpO2≤95%) strongly increased the risk of desaturation in both cohorts separately and in all patients (for both cohort: OR: 8.1 (95%CI: 4.14-15.88) for score 1; OR: 24.8 (95%CI 11.78-57.04) for score 2). Conclusions: We propose a novel scoring system for the risk of oxygen desaturation during 6MWT in patients with ILD, using a combination of two physiological variables, widely available in clinical practice.