Demographic information and clinical characteristics
The study included 319 patients diagnosed with CAP. According to the tertile of serum MANF level, CAP patients were divided into three groups: Tertile 1 (T1) group, Log MANF lower than 3.15; T2 group, Log MANF from 3.15 to 3.33; T3 group, Log MANF higher than 3.33. As illustrated in Table 1, there was no significant differences of age, body mass index, the numbers of complications, the counts of white blood cell (WBC), neutrophil, lymphocyte, monocyte, eosinophil, and basophil among three groups. Moreover, the etiology was analyzed. There was no difference of streptococcus pneumonae, legionella pneumophila, respiratory virus, or others in CAP patients with different tertiles of serum MANF (Table 1). In addition, the number of male patients were increased across with serum MANF among CAP patients. In addition, the obvious differences of creatinine, D-dimer, C-reactive protein (CRP), and interleukin-6 (IL-6) were observed among three subgroups (Table 1).
Level of serum MANF in CAP patients
The levels of serum MANF in CAP patients and control group were quantified using ELISA. As shown in Supplemental Fig. 1, the level of serum MANF was obviously higher in CAP patients than those in healthy participants. Moreover, the level of serum MANF was gradually elevated in cases in accordance with CURB-65 score (Fig. 1A). In addition, serum MANF was compared in CAP cases with different PSI score. The results indicated that serum MANF was highest in V grade among CAP cases (Fig. 1B). Additionally, the content of serum MANF was elevated in >10 scores grade compared with < 4 scores and 4 ~ 6 scores of APACHE II (Fig. 1C). Lastly, the level of serum MANF was compared in cases with different SMART-COP score. We found that the concentration of serum MANF was higher in 5 ~ 6 and 7 ~ 8 scores than other classifications (Fig. 1D).
The level of serum MANF in CAP patients with different severity. The level of serum MANF was detected using ELISA. (A-D) The differences of serum MANF was compared in CAP patients with different scoring criteria. (A) The level of serum MANF in CAP cases of different CURB-65 scores. (B) The level of serum MANF in CAP cases of different PSI scores. (C) The level of serum MANF in CAP cases of different APACHE II scores. (D) The level of serum MANF in CAP cases of different SMART-COP scores. *P < 0.05, **P < 0.01.
Associations between serum MANF and scoring criteria
The relationships of serum MANF with different scoring criteria were evaluated by multivariate linear regression analyses. The results hinted that each 1-unit increase of Log MANF was associated with the elevations of 1.050 score in CURB-65, 45.969 score in PSI, 2.602 score in SMART-COP, and 8.260 score in APACHE II, respectively (Table 2). In addition, CAP patients were allocated into three subgroups in terms of the tertiles of serum MANF. Multivariate logistic regression analyses revealed that the odds ratio (OR) was 3.731 (95%CI: 1.279, 10.880) of PSI, and 3.041 (95%CI: 1.104, 8.379) of APACHE II in T3 group compared with T1 group (Table 2).
Associations between serum MANF and prognostic outcomes
The level of serum MANF on admission was compared in CAP cases with different clinical outcomes. Although there was no difference of serum MANF in CAP cases with hospitalization duration (Fig. 2D), the level of serum MANF was dramatically higher in cases who underwent mechanical ventilation, death, ICU admission during hospitalization (Fig. 2A-C). Besides, the links of serum MANF with different prognostic outcomes were determined. Chi-square test found that the relative risks (RRs) of mechanical ventilation, ICU admission, and 30-day mortality were sensibly elevated in line with the increased serum MANF among CAP cases (Table 3). The potential confounding factors were controlled. Multivariate logistic regression analysis indicated that there were positive relationships between serum MANF and the poorly clinical outcomes among CAP cases during hospitalization (Table 3).
The level of serum MANF in CAP patients with different prognosis. The level of serum MANF was detected using ELISA. (A-D) The differences of serum MANF was compared in CAP patients with different prognosis. (A) The level of serum MANF in CAP cases with and without mechanical ventilation. (B) The level of serum MANF in CAP cases with and without ICU admission. (C) The level of serum MANF in CAP cases with and without death. (D) The level of serum MANF in CAP cases with different hospital length. **P < 0.01.
Predictive powers of serum MANF for poorly prognostic outcomes
The predictive powers of serum MANF for different clinical outcomes were estimated by ROC curve. The area under the curve (AUC) for different prognosis were as follows: Mechanical ventilation, 0.83 (95%CI: 0.75, 0.90); ICU admission, 0.83 (95%CI: 0.77, 0.89); 30-Day mortality, 0.84 (95%CI: 0.71, 0.96) (Fig. 3). The cutoff values of serum MANF for different prognosis were shown below: Mechanical ventilation, 3230.6 pg/mL; ICU admission, 4200.5 pg/mL; 30-day mortality, 5105.9 pg/mL (Fig. 3).
The predictive capacities of serum MANF for different prognosis. (A-C) The predictive capacities of serum MANF for different prognostic outcomes were analyzed by ROC curve. (A) The predictive power for mechanical ventilation. (B) The predictive power for ICU admission. (C) The predictive power for death.