Clinical value of circulating bioactive adrenomedullin for prediction of outcome and hydrocortisone response in sepsis patients-a post-hoc analysis of the HYPRESS trial

Purpose: Sepsis requires stratification for host-directed therapies through the discovery of adequate biomarkers enabling prediction of outcomes and treatment responses. Adrenomedullin has previously demonstrated potential for prognostic enrichment. This study aimed to assess associations of bioactive adrenomedullin (bio-ADM) levels at ICU admission and sepsis outcomes and to evaluate the potential of bio-ADM as marker to identify subgroups of patients with moderate disease severity that might benefit from hydrocortisone treatment.

Methods: We used data from the HYPRESS trial (NCT00670254) to investigate, if bio-ADM is useful to predict sepsis outcomes (septic shock, 90- and 180-day mortality) and benefit or harm by hydrocortisone treatment. Optimal cut-offs for outcome predictions were determined by Youden's index. Logistic regression was used to assess bio-ADM subgroups and treatment interaction.

Results: Bio-ADM levels differed significantly in patients with or without septic shock within 14 days (p = 0.011). While the area under the ROC curve (AUC) was only 0.603 (CI 0.531-0.676), patient subgrouping using bio-ADM levels showed significantly higher cumulative incidence of septic shock within 14 days in the subgroup of patients with bio-ADM levels ≥ 37 pg/mL (p < 0.001). The odds ratio for the development of septic shock in this group was 4.67 (95% CI 1.53, 20.3, p = 0.016). A bio-ADM cut-off of ≥ 136 pg/mL was predictive for 90-day (OR 8.21, 95% CI 2.46-27.9, p < 0.001) and 180-day mortality (OR 4.87, 95% CI 1.49-16.0, p = 0.008). Hydrocortisone therapy did not reduce the incidence of septic shock (OR 1.59, 95% CI 0.37-8.15, p = 0.54), 90-day (OR 1.53, p = 0.23) or 180-day mortality (OR 1.41, p = 0.25), regardless of bio-ADM stratification (interaction term p = 0.58 for septic shock; p = 0.31 for 90-day mortality; p = 0.51 for 180-day mortality).

Conclusions: Whereas bio-ADM levels are associated with sepsis outcomes, our data do not indicate usefulness of the marker to identify patients potentially benefitting from hydrocortisone therapy.