Time-dependent Cox regression: Serial measurement of the cardiovascular biomarker proadrenomedullin improves survival prediction in patients with lower respiratory tract infection

BackgroundThe aim of this study was to determine the prognostic utility of serial measurement of the cardiovascular biomarker midregion proadrenomedullin (MR-proADM) in patients admitted with lower respiratory tract infection. In a previous trial in dyspneic patients (BACH trial) we could show that serial measurement of MR-proADM proves useful for risk assessment and patient monitoring.Models designed to evaluate serial biomarker measurements usually fail to answer two fundamental questions necessary to judge their clinical relevance: whether serial measurements provide additional information on top of the first measurement, and, if yes, at which time point a re-evaluation may be clinically useful.MethodWe apply an adapted time-dependent Cox model to data from the ProHosp trial, a prospective trial, which was observational in regards to application of prognostic biomarkers, where blood draws for biomarker evaluation were collected at day of patient inclusion, days 3, 5 and 7. In this trial, the cardiovascular biomarker MR-proADM was evaluated for its ability to predict survival in comparison to clinical risk scores.ResultsWith the adapted time-dependent Cox model, we could demonstrate a significant added value of the follow up measurements on top of that obtained on admission. Despite a high correlation between serial measurements, the gain can be observed as early as 3 days after inclusion. We illustrate the added prognostic value and clinical relevance of re-evaluation via Kaplan-Meier plots.ConclusionWe could demonstrate that the prognostic biomarker MR-proADM can potentially serve as a outcome monitoring marker in patients admitted with lower respiratory tract infections.