Superior accuracy of mid-regional proadrenomedullin for mortality prediction in sepsis with varying levels of illness severity
Severity in sepsis depends on the extent of organ failure as evaluated by the SOFA score, which in turn is directly associated with the risk of mortality [15]. Nonetheless, the emergence of an increasing number of biomarkers may provide a new avenue with which to improve prognostic accuracy in a simple and rapid manner. In this regard, our study suggests that MR-proADM may be a promising biomarker. However, previous studies evaluating the prognostic role of MR-proADM in sepsis have provided conflicting results. Christ-Crain et al. [16] found that MR-proADM yielded an AUROC of 0.81 for detecting ICU mortality in a group of 53 patients with sepsis. In contrast, Suberviola et al. [17] found limited value of MR-proADM for predicting hospital mortality in 137 sepsis patients, with an AUROC of 0.62. Yet Marino et al. [18] showed that in 101 patients with sepsis, severe sepsis or septic shock, plasma adrenomedullin was strongly associated with the severity of disease, vasopressor requirement and 28-day mortality. These divergent results on the prognostic role of MR-proADM may be explained by differences in patient characteristics, disease severity, infectious source, surgical versus medical and small sample sizes across the various studies.
In the present study, we demonstrated for the first time that the performance of biomarkers to predict mortality in sepsis strongly depends on the degree of organ failure upon ICU admission. Stratifying patients based on their SOFA score allowed us to demonstrate that MR-proADM was the only biomarker able to identify non-survivors in all the severity groups. This is particularly important for the less severely ill patients (SOFA score ? 6), since this group represents either the earliest presentation in the clinical course of sepsis and/or the less severe form of this disease.
Thus, MR-proADM may be a good candidate, after validation in further studies, to be incorporated in an early sepsis management protocol, since it can provide rapid prognostic value and help to guide diagnostic interventions and treatment decisions, consequently resembling the role of troponin in myocardial infarction or d-dimer in pulmonary embolism. The cut-off value of MR-proADM identified for this group of patients (1.79 nmol/L) could be very useful in this regard. This cut-off is able to detect mortality with a good sensitivity and a high negative predictive value. Thus, MR-proADM may potentially help stratify patients in clinical trials examining novel therapies for sepsis.
MR-proADM showed greater predictive value for the risk of mortality than other more commonly used biomarkers, including lactate, in patients with an intermediate degree of organ failure (SOFA score 7–12). In contrast, both MR-proADM and lactate performed similarly in the most severe patients (SOFA ? 13). Therefore, our results support the importance of considering the degree of organ failure when designing studies for the discovery of prognostic biomarkers in sepsis.
The assessment of organ failure by using the SOFA score was recently proposed by the SEPSIS-3 consensus to identify high risk patients with suspected infection [15]. Our results show that a “positive” MR-proADM value may improve the ability of SOFA to predict mortality in sepsis. Interestingly, a combination of MR-proADM with clinical scores such as PSI or CURB-65 also performed better than the clinical scores alone in patients with Community Acquired Pneumonia (CAP) or lower respiratory tract infections (LRTI) [12, 19–21]. As a result, MR-proADM could be used as a reliable risk-stratification tool with the ability to predict mortality or adverse events and to guide clinical decisions. Further clinical studies evaluating strategies combining MR-proADM with other classical severity scores and/or biomarkers for improving the recognition and prognostication of sepsis are therefore warranted [22, 23].
Finally, we observed that an MR-proADM value lower than 0.88 nmol/L may allow to “rule out” mortality in the 28 days following admission to the ICU. This cut-off may be especially useful for guiding early clinical decisions, when the clinical signs of overt organ failure are not yet apparent.
Indeed, our results are similar to those of previous studies. Albrich et al. found that patients with LRTI and MR-proADM concentrations 0.75 nmol/L had an overall mortality of less than 0.5% (11). Furthermore, Krüger et al. showed that patients with CAP and an MR-proADM concentration of 0.9 nmol/L had a survival probability of 99.3% (12). Bello et al. [24] also found an optimal MR-proADM cut-off for predicting 30-day mortality in patients with CAP of 1.06 nmol/L.
Our study is limited in that we evaluated MR-proADM and other biomarker levels only on the day of ICU admission. As a result, we cannot extrapolate our findings to the emergency department or general ward. MR-proADM monitoring over time may further illustrate a temporal trend, which can indicate the success of specific therapies and consequently increase its outcome predictive value [25]. Finally, in our cohort, MR-proADM levels slightly differed depending on the source of infection. Fungal infections induced the highest levels of MR-proADM, while viral infection induced the lowest. This was likely related to the fact that fungal infections resulted in a higher disease severity (median SOFA score of 12 vs. 9 points in patients with no fungal infection), while viral infections resulted in a milder disease severity (median SOFA score of 6.5 vs. 9 points in patients with no viral infection). The potential influence of the source of infection and the type of microbe on MR-proADM’s ability to predict mortality in sepsis merits further investigation.