Use of procalcitonin as a biomarker for sepsis in pediatric burns

Infection and sepsis continue to be the leading cause of morbidity and death in burn injuries. Diagnosing sepsis in burns is challenging as signs and symptoms of sepsis are not specific and overlap with those related to the burn injury. While the use of procalcitonin (PCT) as a biomarker is recommended for diagnosing sepsis in burns, evidence for children with burns is scarce. In this study, we aimed to investigate the role of PCT in distinguishing sepsis in pediatric burns. A prospective observational study was conducted in a single pediatric burn unit. Children hospitalized with burns ≤ 30% of total body surface area were included while patients with chemical burn, inhalation injury, or concomitant chronic diseases were excluded. Patients were classified into three groups for sepsis, systemic inflammatory response syndrome (SIRS), or controls using the American Burn Association (ABA) criteria. The predictive role of C-reactive protein (CRP) and PCT was investigated for distinguishing sepsis. Seventy-two patients were included in the study. The median total body surface area (TBSA) size was 12% (2.0–28.5%), and the median abbreviated burn severity index (ABSI) score was 3 (2–7). The median length of burn unit stay was 9.5 days (1–59 days). Sepsis was diagnosed in 11 patients (15.2%), and SIRS was present in 23 patients (40.0%), whereas 38 patients (52.8%) had neither SIRS nor sepsis (control group). Receiver operating characteristic analysis revealed that CRP and PCT levels distinguished sepsis patients from non-sepsis patients while PCT had a higher positive predictive value (50.0% vs. 45.0%). Optimal cutoff values of CRP and PCT for distinguishing sepsis were 66.75 mg/L and 0.95 ng/mL. Conclusions: PCT levels could distinguish sepsis in children with burn injuries, performing better than CRP levels. Confirmatory studies are needed to evaluate the development of sepsis and the role of PCT in diagnosing sepsis in pediatric burn patients. What is Known: • Even though there are excellent criteria for the diagnosis of infection and sepsis in children and several clinical parameters and biomarkers are being studied, it's difficult to diagnose burn wound sepsis in children. What is New: • Data from this study showed that procalcitonin levels performed better than CRP levels as a biomarker for distinguishing sepsis from systemic inflammatory response syndrome (SIRS) in children with burn injuries. What is Known: • Even though there are excellent criteria for the diagnosis of infection and sepsis in children and several clinical parameters and biomarkers are being studied, it's difficult to diagnose burn wound sepsis in children. What is New: • Data from this study showed that procalcitonin levels performed better than CRP levels as a biomarker for distinguishing sepsis from systemic inflammatory response syndrome (SIRS) in children with burn injuries.


Introduction
Infections constitute a major reason for morbidity and mortality in burn patients as sepsis leads to multi-system organ failure and death [1,2]. While prompt initiation of systemic antibiotic therapy is crucial to reduce mortality in septic patients [3], diagnosing infection in burn patients is challenging as underlying pathophysiology and hypermetabolism due to extensive burn injury result in clinical and laboratory findings which are practically indistinguishable from those of systemic inflammatory response syndrome (SIRS) or sepsis, as defined by conventional criteria [4,5] Therefore, a new definition of sepsis was proposed specific to adult and pediatric burn injuries by American Burn Association (ABA) in 2007 [6]. The new sepsis definition in burn patients, however, requires the documentation of infection defined as a positive culture, a pathologic tissue source, or a clinical response to antimicrobials, a procedure which is not free from uncertainty and delay [7]. Prophylactic use of antibiotics in burns is not recommended [8], while empirical approach for diagnosing infection in burns may cause overuse of antibiotics and bacterial resistance [9,10].
Procalcitonin (PCT), a protein and member of the calcitonin family, has emerged as a promising biomarker indicating systemic bacterial and fungal infections. The role of PCT as a timely diagnostic and monitoring biomarker in sepsis has been researched extensively [11]. The use of PCT for diagnosing sepsis in burns was considered a viable option in a recent meta-analysis, yet this analysis involved only three studies conducted with pediatric burn patients. These three pediatric studies were also conducted before the introduction of ABA criteria for diagnosing sepsis in burns [12].
This study was designed to evaluate the use of PCT as a biomarker for distinguishing sepsis as diagnosed by the current ABA criteria in pediatric burn patients to contribute to the growing body of evidence exploring biomarkers to diagnose sepsis accurately and timely in pediatric burns.

Methods
We conducted a single-center prospective observational study. As our pediatric burn unit could not provide intensive care support, only patients with burns ≤ 30% of total body surface area (TBSA) were admitted. Consecutive patients between the ages of 1 month and 16 years who were hospitalized in the pediatric burn unit between March 2015 and April 2016 were eligible for inclusion. Patients with chemical burns, inhalation injury, or concomitant chronic disease (cystic fibrosis, congenital heart disease, immune deficiency disorders, or chronic lung disease) were excluded. Standard care for burn injuries was provided to all patients in accord with institutional guidelines, and patients were followed up during their stay in the burn unit. Written informed consent from patients' parents and written or oral assent from patients, where possible, were obtained before study participation. The study design was reviewed and approved by the institutional research ethics committee (approval number: 449/ 03032015).
Parameters for demographics, burn characteristics, treatments and interventions, diagnostic results, and length of stay were prospectively recorded with case report forms. The severity of burn at baseline was assessed with Abbreviated Burn Severity Index (ABSI) [13].
SIRS is considered to be present when a patient demonstrates two or more of the following [14]: 1. Temperature above 38.5 °C or below 36 °C 2. Tachycardia, defined as a mean heart rate > 2 standard deviation (SD) above normal for age 3. Tachypnea, mean respiratory rate > 2 SD above normal for age 4. White blood cell count (WBC), count > 12000/mm 3 or < 4000/mm 3 , or left shift defined as > 10% bands Sepsis was diagnosed according to the pediatric American Burn Association (ABA) criteria by burn specialist surgeons and pediatric physicians with consideration of each patient's entire clinical picture coupled with at least three of the following findings [6]: 1. Temperature > 39° or < 36.5 °C 2. Progressive tachycardia: < 2 SD above age-specific norms 3. (85% age-adjusted max heart rate) 4. Progressive tachypnea: children < 2 SD above age-specific norms (85% age-adjusted max respiratory rate) 5. Thrombocytopenia (will not apply until 3 days after initial resuscitation): < 2 SD below age-specific norms 6 In addition, it is required that a documented infection be identified, defined as culture-positive infection, pathologic tissue source identified, or clinical response to antimicrobial drugs.
Patients were then classified into diagnostic groups of sepsis, SIRS, or neither (control group) to determine biomarker performance. An exploratory analysis indicated an agreement not strong enough between ABA and the American College of Chest Physicians/Society of Critical Care Medicine (ACCP/SCCM) definitions for sepsis (kappa coefficient = 0.65, data not shown) [6,15]. This finding was in line with ABA recommendations that the ACCP/SCCM criteria should not be used in burn patients [6].
Peripheral blood samples were taken for routine blood tests of complete blood count, biochemistry, and C-reactive protein (CRP) as well as PCT at initial burn unit admission (baseline) and when development of sepsis was suspected. As the measurement of blood CRP level for suspected cases of infection was recommended by our institutional guidelines, we compared the predictive powers of baseline PCT and CRP for diagnosing infection in burns along with platelets, WBC, and absolute neutrophil count (ANC). Elecsys BRAHMS PCT assay (Roche Diagnostics, Mannheim, Germany) was used on a Cobas c601 device (Roche Diagnostics, Mannheim, Germany) with electrochemiluminescence immunoassay for detection of PCT levels while CRP levels were determined with immunoturbidimetric method on a Cobas c501 device (Roche Diagnostics, Mannheim, Germany) per manufacturer's instructions. When a diagnosis of sepsis was suspected, skin swabs were taken from suspected burn wounds along with sterile blood and urine samples for microbiological cultures to document infectious pathogens as per local institutional guidelines.
Statistical analyses were performed with SPSS 15.0 for Windows (IBM Corporation, Armonk, NY, USA). Assumptions for normality and parametric approach were not fulfilled, and a non-parametric approach was used for analyses. Categorical variables were expressed numerically with frequencies, whereas continuous data were presented as mean with standard deviation with or without median with range. Comparisons between independent groups were performed with Mann-Whitney U and Kruskal-Wallis analysis of variance by rank tests as appropriate. Mann-Whitney U test with Bonferroni correction was used for post hoc tests. Categorical comparisons were carried out with the chi-squared test. Discriminating powers of WBC, ANC, PCT, and CRP were evaluated with receiver operating characteristic (ROC) analysis with the area under curve calculation for an optimal cutoff point. Alpha level was considered statistically significant when below 0.017 for Bonferroni-corrected tests and below 0.05 for others.

Results
Seventy-two patients were included in the study group. Thirty patients (41.7%) were female, and the median age was 22 months (range: 6-168 months). Contact with hot liquid was the most common (n = 66, 91.7%) cause of burns. The median burn size (TBSA) was 12% (2.0-28.5%). The median ABSI score was 3 (2-7), and 40 patients (55.6%) had ABSI scores of 1 to 3. The median length of burn unit stay was 9.5 days (1-59 days). Closed burn wound dressing was used for all patients while a surgical intervention was performed for 38 patients (52.8%). No in-hospital mortality or septic shock was observed. Patient characteristics at admittance and clinical features are presented in Table 1.
When patients were classified into diagnostic groups, 38 patients (52.8%) had neither SIRS nor sepsis, whereas SIRS was present in 23 patients (40.0%). Sepsis was diagnosed in 11 patients (15.2%), and the median time from admittance to sepsis diagnosis was 5 days (2- [n = 1] in blood cultures). In two of three sepsis patients that could not identify an infectious agent were burn wound infection, and one of them was pneumonia. Sepsis was diagnosed with the source of pathological tissue and a clinical response to antimicrobials in burn wound infections. Pneumonia was diagnosed with the use of physical examination, chest radiography, and clinical response to antimicrobials. While diagnostic groups were different by rank for burn size, ABSI score, and length of stay, post hoc analyses revealed that sepsis patients had significantly larger burn area (p < 0.001), ABSI score (p = 0.025), and length of stay in the burn unit (p = 0.002) while the SIRS group had significantly larger burn area (p = 0.002) when these groups were compared to control patients. Other baseline and clinical features were similar between diagnostic groups (p > 0.05) ( Table 1).
Baseline WBC, ANC, CRP, and PCT, but not platelets, were significantly different by rank between diagnostic groups. In post hoc analyses, however, SIRS and control groups were similar while the sepsis group had significantly lower WBC and PNL and higher CRP and PCT levels when compared to SIRS or control group (p = 0.004 for PNL level between the sepsis and SIRS groups, p ≤ 0.001 for all the others; Table 2). The sepsis group's baseline Hb, Plt, WBC, ANC, CRP, and PCT were significantly different from the parameters seen in sepsis time (Table 3). Logistic regression and ROC analysis were used in patients to assess the relationship between laboratory findings and distinguishing sepsis. A negative association was seen between PCT, CRP, and sepsis patients. ROC analysis revealed that only CRP and PCT levels strongly distinguished sepsis patients from non-sepsis patients with areas under the curve being 0.907 and 0.901, respectively. Optimal cutoff values of CRP and PCT for distinguishing sepsis were 66.75 mg/L and 0.95 ng/mL, respectively (Table 4). Sensitivity and specificity of PTC were found to be 90.9% (95% CI 58.7-99.8%) and 85.9% (95% CI 75.6-93.0%), respectively ( Table 5). Details of ROC analysis findings are presented in Tables 4 and 5.

Discussion
This prospective study was designed to evaluate the role of PCT as a biomarker for distinguishing sepsis as diagnosed by the current ABA criteria in pediatric burn patients and showed that PCT levels performed better than CRP levels  as a biomarker. We found that the majority of the burn patients were males with scalding being the most common cause of burn injuries. Male sex is an identified risk factor for burns, and previous studies in children similarly reported that males constituted a higher proportion of burn patients [16][17][18] Scalding has also been reported to be the most common type of burn injury in children [16,19]. As we excluded cases with inhalation injury or burn size over 30% of TBSA, the median ABSI score was relatively low in our patient cohort, and no death was observed. Overall, the baseline and clinical features of our patient group were similar to those reported in previous studies [16,17,19]. Burn patients have an increased risk for the development of infection due to impaired skin integrity and immune suppression [20] while the diagnosis of sepsis in burn patients is challenging as burn injury causes hypermetabolism and systemic inflammation, mimicking a septic condition [21]. ABA criteria have been proposed to provide better guidance for diagnosing sepsis in burn patients, yet studies are scarce regarding sepsis in children with burns [6]. Using the ABA criteria, we found that burn size, ABSI score, and length of stay in the hospital differed when burn patients were classified into the sepsis, SIRS, and control groups. The sepsis and SIRS groups, however, were similar when compared in post hoc analysis. Full-thickness burn size has been reported to be associated with sepsis [22] while the ABSI score has been found to be predictive of the length of hospital stay [23]. With these findings in light, we consider that the severity of burns could be related with infectious and clinical course of burn injury in children, yet systemic inflammation may not be distinguished from sepsis even when ABA criteria are used. Larger pediatric studies with a specific purpose of validating ABA definitions of sepsis and SIRS are needed for further conclusions.
Blood biomarker analysis indicated that WBC, ANC, CRP, and PCT levels were significantly different in the sepsis group while the SIRS and control groups had similar levels, reflecting the presence of change in systemic inflammatory markers in burn patients regardless of a diagnosis of SIRS [24]. SIRS has been defined by ABA as non-specific for clinical use in burn-related injury, and routine diagnosis of burn patients with SIRS is not recommended [6]. WBC and ANC were not found to be predictive of infection development in burns. A similar finding has been reported previously for adult burn patients [25]. While we found that CRP and PCT had values of high sensitivity and specificity for distinguishing burn patients with sepsis from non-sepsis patients, PCT had a higher positive predictive value. In a recent meta-analysis, PCT was found to be a strong diagnostic marker for predicting sepsis in burn patients [12]. In this analysis, the area under the curve was calculated to  [26][27][28][29][30][31][32][33][34][35][36][37]. While this meta-analysis recommended the clinical use of PCT for identifying sepsis in burn patients, caution should be advised as the studies included had a high degree of heterogeneity, mostly related to small sample sizes, different patient age groups, diverse PCT sampling times, measurement methods and cutoff values, and different criteria for sepsis diagnosis [12]. Two studies using the ABA criteria have been published since, investigating the predictive role of PCT for diagnosing sepsis in children with burns. These studies have found that PCT could better distinguish sepsis in pediatric burn patients when compared with CRP, albeit with diverse proposed values of PCT cutoff point (0.5 to 4 ng/mL), sensitivity (60.0% to 100%), and specificity (18.8% to 67.0%) [7,38]. Another study investigating the association between PCT levels and septicemia has reported that a serum PCT level above 0.5 ng/mL had 100% sensitivity and 83% specificity for predicting septicemia, which was confirmed with a positive blood culture [39]. Routine use of a PCT level above 2 ng/mL for the assessment of sepsis in children with burn injuries has been reported as well [40]. While these reports are not conclusive for the optimal use of PCT for diagnosing sepsis in pediatric burns, our results are in line with available evidence, supporting the continued investigation of PCT as a sepsis biomarker in pediatric burns. Our study had several limitations. As admission to our burn unit was limited to burn size being equal to or below 30% of TBSA, our findings could be only relevant for mild to moderate burns. Additionally, while we conducted a prospective study, a confirmatory sample size was not calculated, and therefore, our findings reflect an exploratory investigation. Nevertheless, the findings of this study could be of use for future meta-analyses conducted specifically for the role of PCT in diagnosing sepsis in children with burn injuries.
As a conclusion, we found that PCT had a high negative and a moderate positive predicting value for distinguishing sepsis, which was diagnosed using the specific ABA criteria in pediatric burns. Further studies are needed to investigate the development of sepsis and the use of PCT for sepsis diagnosis in pediatric burn patients.