Sepsis is one of the leading causes of death in the ICU. The pathogenesis of sepsis remains incompletely understood, thereby impeding the development of therapeutics, diagnostics and biomarkers to predict outcomes [1]. Our previous studies have proved that miR-122, miR-193b*, miR-483-5p and miR-574-5p were all differentially expressed between sepsis survivors and non-survivors, differentiated by 28-day mortality [2],[3]. However, whether these biomarkers related to patients with both sepsis and acute respiratory distress syndrome (ARDS) remains unclear. Here we evaluate the levels of these four microRNAs (miRNAs) along with C-reactive protein (CRP), procalcitonin (PCT), Sequential Organ Failure Assessment (SOFA) score, and Acute Physiology and Chronic Health Evaluation (APACHE) II score to determine the ideal biomarkers for sepsis patients.

Serum samples were collected from 232 sepsis patients who were admitted to ICUs of the Chinese PLA General Hospital. All the patients met the definition of sepsis developed in 2003 [4]. Inclusion and exclusion criteria are described in Table 1. Another 24 normal individuals were also included in this study. Serum levels of miRNAs, CRP and PCT were analyzed using methods as described in detail previously [3]. This study was approved by the ethics committee of the Chinese PLA General Hospital. Appropriate informed consent was obtained from each patient and normal individual.

Table 1 Inclusion and exclusion criteria

The clinical data of these 232 patients are shown in Table 2. After comparison of the levels of the four miRNAs in three pairs of groups (normal individuals and sepsis patients, survivors and non-survivors, sepsis without ARDS and sepsis plus ARDS), only the cycle threshold of mir-122 was differentially expressed in all three (P < 0.01) (Figure 1). Univariable and multivariable regression analyses were then used to evaluate the association between miR-122 and 28-day mortality in different ICUs. After adjustment using clinical data and additional parameters (SOFA score, APACHE II score and ARDS), the odds ratio of miR-122 association with 28-day mortality was around 0.376 to 0.868 (P < 0.05) in the different ICUs. The area under the curve for the predictive value of miR-122 was around 0.706 to 0.770 (P < 0.01) with high sensitivity and specificity (Table 3). As a result, only miR-122 can be used as a biomarker with regards to patients with both sepsis and ARDS. miR-122 is a liver-specific miRNA and levels of it in serum were correlated with drug-induced liver injury [5]. We reported that miR-122 correlated with coagulation disorders in sepsis patients and serum levels of miR-122 correlated with serum antithrombin III levels [6]. Our study reveals a potential novel target to develop a biomarker for sepsis prognosis and therapeutic strategies.

Table 2 Clinical characteristics of the 232 sepsis patients
Figure 1
figure 1

Cycle thresholds of the four microRNAs (miRNAs) in the three pairs of groups. ARDS, acute respiratory distress syndrome.

Table 3 The association between miR-122 levels and 28-day mortality in sepsis patients