Serum activin A and B levels predict outcome in patients with acute respiratory failure: a prospective cohort study

INTRODUCTION: 30 day mortality in patients with Acute Respiratory Failure (ARF) is approximately 30%, defined as patients requiring ventilator support for more than 6 hours. Novel biomarkers are needed to predict patient outcomes and to guide potential future therapies. The activins A and B, members of the Transforming Growth Factor β family of proteins, and their binding protein, follistatin, have recently been shown to be important regulators of inflammation and fibrosis but no substantial data are available concerning their roles in ARF. Our objectives were to evaluate whether the serum levels of activin A, B and follistatin are elevated in 518 patients with ARF from the FINNALI study compared the concentrations in 138 normal subjects that form a reference range. METHODS: Specific assays for activin A, B and follistatin were used and the results analyzed according to diagnostic groups as well as according to standard measures in intensive care. Multivariable logistic regression was used to create a model to predict death at 90 days and 12 months from the onset of the ARF. RESULTS: Serum activin A and B were significantly elevated in most patients and in most of the diagnostic groups. Patients who had activin A and/or B concentrations above the reference maximum were significantly more likely to die in the 12 months following admission [either activin A or B above reference maximum: Positive Likelihood Ratio [LR+] 1.65 [95% CI 1.28-2.12, P = 0.00013]; both activin A and B above reference maximum: LR + 2.78 [95% CI 1.96-3.95, P < 0.00001]. The predictive model at 12 months had an overall accuracy of 80.2% [95% CI 76.6-83.3%]. CONCLUSIONS: The measurement of activin A and B levels in these patients with ARF would have assisted in predicting those at greatest risk of death. Given the existing data from animal studies linking high activin A levels to significant inflammatory challenges, the results from this study suggest that approaches to modulate activin A and B bioactivity should be explored as potential therapeutic agents.