[Predictive value of glycosylated serum protein combined with glycemic variability on secondary persistent inflammatory immunosuppressed catabolic syndrome prediction in elderly septic patients].

OBJECTIVE: To investigate the predictive value of glycosylated serum protein (GSP) combined with glycemic variability (GV) in persistent inflammation immunosuppressive catabolic syndrome (PICS) in elderly septic patients.

METHODS: A retrospective study was conducted. The septic patients aged ≥ 60 years old with 28 days hospitalized duration admitted to geriatric intensive care unit (ICU) of Guangzhou General Hospital of Guangzhou Military Command from January 2014 to December 2017 were enrolled. The patients were divided into two groups according to whether PICS occurred within 14 days after ICU admission according to the PICS diagnostic criteria. General patients' data including gender, age, underlying disease, site of infection, the length of ICU stay were collected, acute physiology and chronic health evaluation II (APACHE II), GSP at 1 day after hospitalization and lymphocyte count (LYM), C-reactive protein (CRP), albumin (ALB), prealbumin (PA) levels at 1 day and 14 days were recorded. The levels of blood glucose on the 1st day and 14th day were observed, the GV was calculated. Data were cross-validated using the random forest method. Receiver operating characteristic (ROC) curve was plotted to evaluate the predictive value of glycemic parameters for PICS.

RESULTS: A total of 315 elderly septic patients were included. Patients with malignant tumors, severe autoimmune diseases, and immunosuppressive therapy or dead within 28 days of hospitalization were excluded. A total of 132 patients were enrolled in the analysis, including 45 in the PICS group and 87 in the non-PICS group. The length of ICU stay in the PICS group was significantly longer than that in the non-PICS group [days: 35.0 (22.0, 49.5) vs. 8.0 (5.0, 23.0), P < 0.01]. No significant difference in the baseline data of gender, age, underlying disease, infection site or APACHE II score between the two groups was found. (1) Parameters for PICS diagnosis: with the prolongation of ICU stay, LYM and PA in the non-PICS group were increased and those in the PICS group were decreased, and CRP and ALB levels were decreased in both groups. LYM, ALB and PA levels in the PICS group were significantly lower than those in the non-PICS group at 14 days after ICU admission [LYM (×109 /L): 0.6 (0.5, 0.7) vs. 1.1 (0.9, 1.6), ALB (g/L): 25.4±2.7 vs. 29.9±4.3, PA (g/L): 0.08 (0.05, 0.14) vs. 0.11 (0.10, 0.21), all P < 0.01], and CRP level was significantly higher than that in the non-PICS group (mg/L: 87.5±56.3 vs. 49.2±49.1, P < 0.01). (2) Glycemic parameters: the GSP level of the PICS group at 1 day after ICU admission was significantly lower than that of the non-PICS group (mmol/L: 2.3±0.6 vs. 2.7±0.6, P < 0.01), but there was no statistically significant difference in the level of blood glucose or GV at 1 day and 14 days after ICU admission as compared with the non-PICS group [blood glucose (mmol/L): 10.0±3.3 vs. 9.4±3.3 at 1 day, 10.8±3.6 vs. 10.4±3.5 at 14 days; GV: (24.2±1.4)% vs. (23.7±1.2)% at 1 day, (24.8±7.8)% vs. (24.7±7.7)% at 14 days, all P > 0.05]. (3) ROC curve analysis: 1-day GSP as well as 1-day and 14-day GV had certain predictive value for PICS secondary to sepsis in the elderly. The predictive value of 1-day GSP combined with 14-day GV was the highest, its area under ROC curve (AUC) was 0.637, with a sensitivity of 95.8% and a specificity of 25.0%, while the positive likelihood ratio was 1.278, the negative likelihood ratio was 0.167, the positive predictive value was 71.9%, and the negative predictive value was 75.0%.

CONCLUSIONS: GSP combined with GV could effectively predict secondary PICS in elderly septic patients.