Utility of procalcitonin in diagnosing early postoperative sepsis after pediatric cardiac surgery in Malaysia
Article information
Abstract
Background
Systemic inflammation following cardiopulmonary bypass (CPB) can interfere with analysis of routine clinical and biochemical parameters. Procalcitonin (PCT) is a potential biomarker for diagnosing early postoperative sepsis in pediatric patients following cardiac surgery utilizing CPB. This study aimed to evaluate the diagnostic accuracy of PCT compared to other biomarkers, especially C-reactive protein (CRP), in this clinical setting.
Methods
A prospective single-center study was conducted over a 10-month period during the coronavirus disease 2019 (COVID-19) pandemic (2021–2022), enrolling 89 pediatric patients postcardiac surgery. PCT, CRP, and complete blood count were analyzed, and area under the curve (AUC) was employed for statistical analysis.
Results
PCT and CRP demonstrated moderate discriminatory ability with AUCs of 0.678 and 0.635, respectively. White cell count exhibited fair discriminatory power, and platelet count performed poorly in distinguishing septic from nonseptic cases (AUC: white cell count, 0.545; platelet, 0.486).
Conclusions
PCT and CRP hold promise as diagnostic markers for early postoperative sepsis in pediatric cardiac surgery patients. However, these biomarkers are not adequate standalone indicators, emphasizing the continued need for clinical judgment supported by multiple diagnostic parameters.
INTRODUCTION
Postoperative sepsis is a serious complication for children after cardiopulmonary bypass (CPB) and may involve noncardiopulmonary organs. In addition to prompt administration of antibiotics in the intensive care setting, diagnosis is often challenging due to systemic inflammatory response syndrome (SIRS) following CPB. SIRS occurs due to the exposure of plasma components to the synthetic surfaces of the circuit, as well as from tissue ischemia [1,2].
Procalcitonin (PCT) is a useful biomarker for bacterial infection in both adults and children, including newborn babies [3,4]. PCT can help identify sepsis in neonates, and to differentiate between noninvasive and invasive infections in ill patients, with greater sensitivity and specificity than C-reactive protein (CRP) [4,5]. Fewer studies have explored the use of PCT in children receiving intensive care, especially following cardiac surgery requiring CPB. PCT may be particularly useful when SIRS following CPB affects the interpretation of routine biochemical and clinical markers. PCT is typically a more definitive early predictor of postoperative sepsis than CRP or white blood count (WBC) in non-CPB surgeries. However, CPB-related inflammation negates this predictive advantage during the first three postoperative days following cardiac surgery [1,2]. Prognostic factors that can clearly differentiate sepsis from SIRS are needed, because early recognition and initiation of antibiotics are crucial for treating sepsis. Therefore, in this study, we sought to explore the potential of PCT in the diagnosis of early postoperative sepsis after pediatric cardiac surgery.
MATERIALS AND METHODS
This study was approved by the National Heart Institute Research Ethics Committee (No. IJNREC/440/2019) before data collection. Consent from parents was obtained before enrollment.
Study Design
This study was designed as a single-center prospective study, collecting data from all patients suspected to have sepsis following cardiac surgery. The study was conducted over a 10-month period at the National Heart Institute, Kuala Lumpur during the coronavirus disease 2019 (COVID-19) pandemic (2021–2022). The population enrolled consisted of patients from 7 days to 18 years old who underwent cardiac surgery with CPB and were suspected to have sepsis within 7 days after surgery. Those with known immunodeficiency conditions, preoperative sepsis within 7 days before surgery, patients on extracorporeal membrane oxygenation postoperative, and those who died within 48 hours postoperative were excluded.
Sepsis was defined as suspected or confirmed infection with potential life-threatening organ dysfunction affecting either respiratory, cardiovascular, neurologic or coagulation system, in alignment with the Phoenix Sepsis Score [6]. A patient was suspected to have sepsis when they developed conditions deemed atypical of routine postoperative course. Changes in ventilatory requirements, tachypnea, or new consolidations on chest radiography; unexplained labile hemodynamics with increased vasopressor support (such as noradrenaline >0.1 μg/kg/min); new onset coagulation abnormalities (such as platelet < 100 ×103/µl) or abnormal leucocyte counts; changes in neurological condition; temperature >38.5 ºC or <36 ºC; feeding intolerance; urine turbidity; and wound erythema all warranted suspicion of sepsis and inclusion in the study.
Once enrolled, biochemical and surgical data including type of cardiac problem, surgical intervention, severity score, length of CPB, aortic cross-clamp time, and postoperative data including hemodynamic values, inotropic support score, length of invasive and noninvasive ventilation, and length of intensive care unit (ICU) stay were collected. Inotropic support scoring is the summation of inotropic doses at the point of sampling=dopamine (μg/kg/min)+dobutamine (μg/kg/min)+100×epinephrine (μg/kg/min)+10×milrinone (μg/kg/min)+10,000×vasopressin (units/kg/min)+100×norepinephrine (μg/kg/min) [7].
Routine septic screening including complete blood count, CRP, and blood cultures as well as endotracheal aspirate culture, urine culture, and wound swabs when suitable, with the addition of PCT, were collected. All investigations were sent to the laboratory for analysis except for PCT. PCT was analyzed using a point-of-care test, the AQT90 FLEX PCT immunoassay test by Radiometer. This assay has previously demonstrated excellent analytical performance and diagnostic agreement compared to other tests [8]. All patients were given standard prophylactic antibiotics (cefazolin for closed chest; teicoplanin and piperacillin/tazobactam for open chest) and were escalated to second-line antibiotics upon diagnosis of clinical sepsis. Type of antibiotics and culture results were recorded. Routine care after cardiac surgery was unaffected. Repeat septic screening was performed based on clinical needs. Patients were then classified into one of two groups: Septic and Nonseptic. The Septic group fulfilled the Phoenix Sepsis Score criterion of two or more points, indicating potentially life-threatening organ dysfunction with clinical infection (after evaluation of microbiological results, biochemical results, and clinical parameters in the ICU) [6]. A positive culture was not necessary for this diagnosis in this cohort of patients as suggested by a 2022 Indian study demonstrating that 16% of pediatric cardiac patients had culture-negative sepsis [9].
Statistical Methods
Categorical variables were expressed as frequencies or percentages. Normally distributed continuous variables were expressed as mean±standard deviation and nonnormally distributed continuous variables as median with interquartile range. Statistical analysis was done using IBM SPSS 29.0.2.0 (IBM Corp.) utilizing Student t-test (normally distributed data), Mann-Whitney U-test (nonnormally distributed data), receiver operating characteristic curve and area under the curve (AUC). A P-value of <0.05 was considered statistically significant.
RESULTS
Baseline Data
This study was conducted over 10 months (May 2021 until February 2022) during the COVID-19 pandemic. In total, 575 patients underwent cardiac surgery at our center during this period. Ninety-one patients were suspected to have postoperative sepsis and were subsequently enrolled into the study. Two were excluded due to out-of-range PCT values, while 89 were suitable for analysis. We reviewed all biochemical, microbiological, and clinical parameters at the end of data collection to determine the number of patients who were truly septic. In total, 79 were concluded to be nonseptic while 10 were found to have sepsis (Figure 1). The clinical parameters, biochemical data, and outcome results are shown in Table 1.
Patient enrollment flowchart. CPB: cardiopulmonary bypass; PCT: procalcitonin.
Demographic and results for sample population (n=89)
In the Septic group, there were four infants, four children 1–5 years old, and two patients 6–18 years old. Five patients underwent Risk Adjustment for Congenital Heart Surgery (RACHS)-1 score 2 surgeries, three underwent RACHS-1 score 3 surgeries, one underwent RACHS-1 score 4 surgery, and one underwent RACHS-1 score 6 surgery. Those in the Septic group had a significantly longer CPB duration (mean 182 minutes), longer cross-clamp time (mean 108 minutes), and five patients (50%) in the Septic group had delayed chest closure compared to seven (8.9%) in the Nonseptic group. Comparison of WBC, platelet, CRP, PCT, heart rate, respiratory rate, temperature, and vasoactive inotropic score revealed no differences between the two groups. One patient in the septic group experienced a neurological insult versus two in the Nonseptic group. No patients in the Septic group had low cardiac output syndrome or acute renal impairment while one patient had low cardiac output syndrome in the Nonseptic group. All patients in the Septic group had positive cultures except those with septicemia, as shown in Table 2. After applying the Phoenix Sepsis Score to the dataset, the Septic group (median, 3) and the Nonseptic group (median, 2) showed no statistically significant difference (P=0.163), indicating no discriminatory ability for sepsis diagnosis in this cohort.
Types of infection
DISCUSSION
Sepsis and SIRS are challenging to differentiate because they share similar clinical and laboratory features such as fever, elevated white blood cell count, raised capillary permeability, and fluid buildup in tissues [2,5]. The likelihood of infection following cardiac surgery is high due to factors like invasive lines and endotracheal tubes, open chest cavity, and transient deficiencies in immunologic function [10]. While features like long CPB time, long cross-clamp time, and delayed chest closure increase the likelihood of sepsis, clinicians still depend on biochemical septic markers for diagnosis and monitoring of sepsis [2].
Typical parameters for sepsis scoring have to be interpreted carefully, as congenital cardiac patients often have abnormal baseline parameters. For example, hypoxemia (PaO2 of 40–50 mm Hg) is normal for a cyanotic patient. Institutional practices, such as the use of low-dose inotropes postoperative in the absence of poor cardiac function, have to be considered as well. Our institution routinely uses milrinone at a low dose of 0.3–0.5 μg/kg/min after congenital cardiac surgery. The Phoenix Sepsis Score was introduced by a task force from the Society of Critical Care Medicine in 2024. Its data-driven criteria replaced older inflammatory response-based definitions and are focused on organ dysfunction. However, its application has not been validated in all pediatric subgroups [11]. Pediatric cardiac surgery patients are at a higher risk for bleeding, hemodynamic instability requiring inotropes, and are almost always ventilated after surgery [12]. These factors potentially confound and impact the Phoenix Sepsis Score, and thus some of the organ dysfunction measures used in the modeling process may not reflect actual organ dysfunction, but rather iatrogenic effects or clinician therapeutic choices [13,14]. Local institutional practices may also affect scoring. To date, the only study that used the Phoenix Sepsis Score in pediatric patients after cardiac surgery found challenges in application of the score, namely the routine use of milrinone leading to overdiagnosis of sepsis, quantification of the degree of respiratory dysfunction in a single ventricle physiology, and patients on continuous sedation, which is common after surgery [15]. Although the Phoenix Sepsis Score criteria have shown enhanced performance in diagnosing pediatric sepsis and septic shock globally, more studies are required to validate this scoring system in pediatric cardiac surgery.
Farias et al. systematically reviewed 21 studies and discovered that PCT is a superior indicator of postoperative sepsis compared with CRP in children undergoing congenital cardiac surgeries utilizing CPB [16]. Nasser et al. [2] suggested trends observed in of infection biomarkers, including PCT, are more crucial than the absolute value when predicting postoperative infection. Another study in 2018 concluded that PCT-directed anti-infective treatment in ICU patients with infection and sepsis leads to better survival and reduces the length of antibiotic prescription [17].
In our study, PCT level spikes during the early postoperative period, i.e., day 1 to day 3 postoperative, were similar to kinetics described by Minami et al. [18], with peak PCT values as high as 72 ng/ml. The erratic nature of PCT levels may suggest that it is less reliable for tracking postoperative inflammation compared to CRP, which showed a more consistent increase [1]. Although repeated PCT samples were taken in certain patients for monitoring purposes, we decided to exclude those samples from analysis as our study is focused on the diagnostic value of PCT in pediatric patients postoperative, rather than monitoring trends in PCT.
To determine the diagnostic value of PCT, AUC provides an inclusive representation of diagnostic accuracy in laboratory tests, with a value closer to one indicating better discrimination (Figure 2, Table 3). The AUC for PCT was found to be 0.678, indicating a moderate ability to discriminate between sepsis and nonsepsis groups. The AUC for CRP was 0.635, suggesting a similarly moderate ability. However, both AUC values are below the commonly accepted threshold of 0.8 for a diagnostic test to be considered accurate. White cell count has a fair ability and platelet count performs poorly in distinguishing between sepsis and nonsepsis cases. These findings suggest that CRP and PCT may be more useful as diagnostic markers for sepsis compared to white cell count and platelet count.
Receiver operating characteristic curves for biomarkers in diagnosing sepsis. CRP: C-reactive protein; PCT: procalcitonin.
Area under the curve
Our study has several limitations. First, the sample size was relatively small compared to similar studies, primarily due to the severe constraints imposed by the COVID-19 pandemic. Specifically, the outbreak severely limited our ability to collect samples within the pediatric ICU. Stringent infection control measures further restricted access to patients and complicated sample collection. Additionally, border closures delayed the acquisition of necessary reagents, which affected participant enrollment. These unique and substantial factors underscore the complexities faced by researchers during the pandemic. Second, we acknowledge potential differences between point-of-care PCT testing and laboratory PCT tests, a comparison which was beyond the scope of this study.
In conclusion, PCT is not a reliable standalone marker for sepsis in the early postoperative period following pediatric cardiac surgery. Sepsis in a cardiac surgical setting still depends on a clinical acumen supported by biochemical, radiographical, and microbiological findings. Further research may be needed to identify additional biomarkers or to refine the use of existing biomarkers in combination with other clinical parameters to improve diagnostic accuracy in distinguishing between sepsis and nonsepsis cases. Although the Phoenix Sepsis Score appears promising, further validation and investigation are needed to confirm its clinical utility in children undergoing cardiothoracic surgery.
KEY MESSAGES
▪ Procalcitonin (PCT) demonstrates a moderate ability to identify sepsis in pediatric patients after cardiac surgery, with an area under the curve of 0.678, suggesting it may be a useful adjunct to traditional diagnostic markers like C-reactive protein (CRP).
▪ Despite the potential utility of PCT and CRP as biomarkers for early postoperative sepsis, this study emphasizes that clinical judgment, supported by biochemical, radiographical, and microbiological findings, is essential for accurate diagnosis.
Notes
CONFLICT OF INTEREST
No potential conflict of interest relevant to this article was reported.
FUNDING
None.
ACKNOWLEDGMENTS
We would like to acknowledge the provision of the AQT90 FLEX PCT immunoassay test by Radiometer Malaysia for use in the study. We are also thankful for statistical support provided during analysis of results by the Research Department of the National Heart Institute.
AUTHOR CONTRIBUTIONS
Conceptualization: KWS, KKT. Methodology: KWS, KKT. Formal analysis: KWS. Data curation: MYMR. Visualization: KWS. Project administration: KWS. Writing – original draft: MYMR. Writing – review & editing: KWS, KKT. All authors read and agreed to the published version of the manuscript.