Superior accuracy of mid-regional proadrenomedullin for mortality prediction in sepsis with varying levels of illness severity

BACKGROUND: The use of novel sepsis biomarkers has increased in recent years. However, their prognostic value with respect to illness severity has not been explored. In this work, we examined the ability of mid-regional proadrenomedullin (MR-proADM) in predicting mortality in sepsis patients with different degrees of organ failure, compared to that of procalcitonin, C-reactive protein and lactate. METHODS: This was a two-centre prospective observational cohort, enrolling severe sepsis or septic shock patients admitted to the ICU. Plasma biomarkers were measured during the first 12 h of admission. The association between biomarkers and 28-day mortality was assessed by Cox regression analysis and Kaplan–Meier curves. Patients were divided into three groups as evaluated by the Sequential Organ Failure Assessment (SOFA) score. The accuracy of the biomarkers for mortality was determined by area under the receiver operating characteristic curve (AUROC) analysis. RESULTS: A total of 326 patients with severe sepsis (21.7%) or septic shock (79.3%) were enrolled with a 28-day mortality rate of 31.0%. Only MR-proADM and lactate were associated with mortality in the multivariate analysis: hazard ratio 8.5 versus 3.4 (p < 0.001). MR-proADM showed the best AUROC for mortality prediction at 28 days in the analysis over the entire cohort (AUROC [95% CI] 0.79 [0.74–0.84]) (p < 0.001). When patients were stratified by the degree of organ failure, MR-proADM was the only biomarker to predict mortality in all severity groups (SOFA ≤ 6, SOFA = 7–12, and SOFA ≥ 13), AUROC [95% CI] of 0.75 [0.61–0.88], 0.74 [0.66–0.83] and 0.73 [0.59–0.86], respectively (p < 0.05). All patients with MR-proADM concentrations ≤0.88 nmol/L survived up to 28 days. In patients with SOFA ≤ 6, the addition of MR-proADM to the SOFA score increased the ability of SOFA to identify non-survivors, AUROC [95% CI] 0.70 [0.58–0.82] and 0.77 [0.66–0.88], respectively (p < 0.05 for both). CONCLUSIONS: The performance of prognostic biomarkers in sepsis is highly influenced by disease severity. MR-proADM accuracy to predict mortality is not affected by the degree of organ failure. Thus, it is a good candidate in the early identification of sepsis patients with moderate disease severity but at risk of mortality. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (doi:10.1186/s13613-017-0238-9) contains supplementary material, which is available to authorized users.