Early detection of soluble CD27, BTLA, and TIM-3 predicts the development of nosocomial infection in pediatric burn patients

Abstract

Thermal injury induces concurrent inflammatory and immune dysfunction, which is associated with adverse clinical outcomes. However, these effects in the pediatric population are less studied and there is no standard method to identify those at risk for developing infections. Our goal was to better understand immune dysfunction and identify soluble protein markers following pediatric thermal injury. Further we wanted to determine which early inflammatory, soluble, or immune function markers are most predictive of the development of nosocomial infections (NI) after burn injury. We performed a prospective observational study at a single American Burn Association-verified Pediatric Burn Center. A total of 94 pediatric burn subjects were enrolled and twenty-three of those subjects developed a NI with a median time to diagnosis of 8 days. Whole blood samples, collected within the first 72 hours after injury, were used to compare various markers of inflammation, immune function, and soluble proteins between those who recovered without developing an infection and those who developed a NI after burn injury. Within the first three days of burn injury, innate and adaptive immune function markers (ex vivo lipopolysaccharide-induced tumor necrosis factor alpha production capacity, and ex vivo phytohemagglutinin-induced interleukin-10 production capacity, respectively) were decreased for those subjects who developed a subsequent NI. Further analysis of soluble protein targets associated with these pathways displayed significant increases in soluble CD27, BTLA, and TIM-3 for those who developed a NI. Our findings indicate that suppression of both the innate and adaptive immune function occurs concurrently within the first 72 hours following pediatric thermal injury. At the same time, subjects who developed NI have increased soluble protein biomarkers. Soluble CD27, BTLA, and TIM-3 were highly predictive of the development of subsequent infectious complications. This study identifies early soluble protein makers that are predictive of infection in pediatric burn subjects. These findings should inform future immunomodulatory therapeutic studies.

Keywords: immune checkpoint inhibitors; immune dysfunction; inflammation; nosocomial infection; pediatric thermal injury; soluble proteins.

Figure 1

Plasma cytokines levels within the first 72 hours following pediatric thermal injury. Burn injury, regardless of infection (NI, n=23 and No NI, n=61) significantly decreased median plasma levels of IL-10 (A), IL-6 (B), IL-8 (C), and IL-17 (D) compared to healthy controls (n=17, except IL-8 n=11). Subjects who went on to develop NI had higher plasma levels of IL-10 (A), IL-6 (B), and IL-8 (C) compared to those who recovered from burn injury without developing an infection. There were no significant differences in the plasma levels of IL-2 (E) and IL-12 (F) between groups in the first 72 hours. Statistical analysis was performed using one-way ANOVA plus Dunn’s test. Lines represent median and interquartile range; dots represent the individual data points. NI, nosocomial infection (red); No NI, no infection (blue); HC, healthy control (green).

Figure 2

Immune cell presence within the first 72 hours following pediatric thermal injury. Adaptive immune cells as measured through absolute CD4+ lymphocytes (A), CD4 regulatory T cells (B) and natural killer cells (C) were decreased in subjects who went on to develop an infection (n=21, 21, 19, respectively) relative to those who did not (n=48, 47, 46, respectively) and healthy controls (n=11). Innate immune cells as measured through CD66b+ neutrophils (D) and CD14+ monocytes (E) were elevated in children who recovered without developing an infection (n=46, 47, respectively) relative to healthy controls (n=11) and those who did develop an infection (n=21). CD14+ monocyte HLA-DR expression (F) was significantly lower in children that went on to develop NI (n=21) compared to those patients that did not (n=46) and healthy controls (n=11). Burn injury regardless of infection (NI, n=18 and No NI, n=48) significantly decreased median percent PD-1 expression on CD4+ lymphocytes (G) relative to healthy controls (n=11) but displayed no difference between burn groups. There was no significant difference in the percent CTLA-4 expression (H) between groups (NI, n=17, No NI, n=42, HC, n=10). Statistical analysis was performed using one-way ANOVA plus Dunn’s test. Lines represent median and interquartile range; dots represent the individual data points. NI, nosocomial infection (red); No NI, no infection (blue); HC, healthy control (green).

Figure 3

Immune function within the first 72 hours following pediatric thermal injury. Innate and adaptive immune function were measured by ex vivo LPS and PHA stimulation, respectively. Within the first 72 hours of injury, LPS induced TNFα (A) was decreased for all burn subjects (NI, n=23 and No NI, n=66) compared to healthy controls (n=15). PHA-induced cytokine-production capacities (B–D) were decreased for all burn subjects (NI, n=23 and No NI, n=67) compared to healthy controls (n=17). Subjects who went on to develop an infection had significantly lower LPS induced TNFα (A) and PHA induced IL-10 (B) compared to those who recovered from burn injury without developing an infection. Statistical analysis was performed using one-way ANOVA plus Dunn’s test. Lines represent median and interquartile range; dots represent the individual data points. NI, nosocomial infection (red); No NI, no infection (blue); HC, healthy control (green).

Figure 4

Soluble protein in unstimulated plasma within the first 72 hours following pediatric thermal injury. Soluble BTLA (A), TIM-3 (B), and CD27 (C) were all significantly higher in children that went on to develop nosocomial infection (n=22) compared to those patients that did not (n=59) within the first 72 hours after injury. Soluble CD27 (C) was also significantly higher from healthy controls (n=12). Statistical analysis was performed using one-way ANOVA plus Dunn’s test. Lines represent median and interquartile range; dots represent the individual data points. NI, nosocomial infection (red); No NI, no infection (blue); HC, healthy control (n=12).

Figure 5

Receiver operating characteristics curve analysis within the first 72 hours following pediatric thermal injury. The area under the receiver operating characteristic curves was highest for LPS-induced TNFα production capacity (0.82; purple) and soluble CD27 (0.81; black). This was followed by soluble BTLA (0.78; green), unstimulated plasma IL-10 (0.78; red), and soluble TIM-3 (0.77; blue). The remaining markers (CD4+, Tregs, CD14+, %HLA-DR expression, plasma IL-6, plasma IL-8, and PHA IL-10) had an AUC ≤ 0.75 (Supplemental Table 1). The optimal threshold for the prediction of the development of a nosocomial infection was for <505.8 pg/mL for LPS-induced TNFα, >1129 pg/mL for soluble CD27, >1353 pg/mL for soluble BTLA, and >705.3 pg/mL for soluble TIM-3, and >11.59 pg/mL for plasma IL-10.