(372) Measures of Lung Injury and ICU Mortality in Acute Respiratory Failure After Stem Cell Transplant

Introduction: Acute respiratory failure (ARF) after stem cell transplantation (SCT) is associated with 42-61% mortality in those who require invasive mechanical ventilation (IMV). There are limited data correlating objectives measures of lung injury with outcomes in this population. We sought to further delineate the relationship between gas exchange derangements early after initiation of IMV and ICU mortality.

Methods: Retrospective cohort study of patients aged 0-32 years with ARF requiring IMV within one year of SCT. Data included patient and disease-specific variables, respiratory support, and measures of lung function: oxygenation index (OI), alveolar to arterial oxygen gradient (Aa), PaO2/FiO2 (P/F) ratio, and dead space fraction over 48 hours after initiation of IMV. Receiver operator characteristic (ROC) curves assessed the association of gas exchange measures and mortality. Optimal cut points for each gas exchange value were identified.

Results: 28 IMV events for ARF were analyzed. Cohort age was 15.8  7.5 years, lymphoma was the most common primary diagnosis (82%), and 79% had allogeneic SCT. Survival to ICU discharge was 25%. Patient characteristics were similar between survivors and non-survivors. Non-survivors more commonly had allogeneic SCT, pneumonia, and a non-CMV viral illness. Nearly all measures of gas exchange were significantly worse in non-survivors. Thresholds associated with 100% mortality were identified for Aa, OI, P/F ratio, and dead space fraction at 48 hours are 273 (AUC 0.84, sensitivity 65%, specificity 100%), 11 (AUC 0.91, sensitivity 75%, specificity 100%), 173 (AUC 0.85, sensitivity 86%, specificity 50%) and 0.16 (AUC 0.78, sensitivity 89%, specificity 71%), respectively.

Conclusions: Nearly all measures of gas exchange are worse in non-survivors in this small series. Critical thresholds of gas exchange associated with mortality are notably lower than criteria at which ECMO is recommended in patients with ARDS. More work is needed to identify disease specific thresholds at which ECMO should be considered to support ARF following HSCT. Predictive analytics and modeling that factor primary disease, transplant type and concomitant post-SCT diagnoses in addition to measures of lung injury may provide better insight on prognosis.