Background: Idiopathic pulmonary fibrosis (IPF) is the most common interstitial lung disease with a median survival time of 2.5 years without treatment. For this rare disease, there is still a lack of validated biomarkers for mortality risk assessment. Over the last few years, body composition analysis (BCA) has emerged as a predictive tool in various diseases.

Aim: To inverstigate whether fully automated BCA derived markers correlate with overall survival (OS) in patients with IPF.

Patients and Methods: We retrospectively studied 79 IPF patients (19 % female) with a CT scan of the thorax (38 % at diagnosis). A fully automated BCA was performed using a 3D Multi-res U-Net. The BCA features were aggregated and combined into a Sarcopenia, Fat, and Myosteatosis index. The expression of the indices were compared by splitting the cohort into patients who died within two years (n=36) and those who did not (n=33).

Results: Median survival time was 23 months. In total 53 patients (67 %) died, 36 (45 %) within two years. A higher (> median) Sarcopenia and Fat index and lower (< median) Myosteatosis index were associated with longer survival time (35 vs 16 months for high vs low Sarcopenia index, p=0.0658; 44 vs 14 months for high vs low Fat index, p < 0.0001; 33 vs 14 months for low vs high Myosteatosis index, p=0.0056). Only the Fat index was an independent predictor (HR=0.394, 95%-CI=0.205-0.758, p=0.005) of survival compared to other clinical parameters.

Conclusion: The fully automated BCA may provide biomarkers with a  predictive value for the overall survival in patients with IPF.