Role of biomarkers in pediatric sepsis: What evidence says?

Abstract

Sepsis is common in hospitalized pediatric patients, leading to increased morbidity, including multiple organ dysfunction and mortality. Fluid resuscitation and antibiotic administration are the primary protective mechanisms for sepsis. However, not all infections are bacterial, and unnecessary antibiotic use increases the risk of developing multidrug resistance; therefore, it is essential to distinguish bacterial from viral or other infections. Routine laboratory investigations cannot always identify the cause of diseases, but assessing different biomarker levels can help identify these infections and treat sepsis accordingly. This mini-review aims to critically analyze the available evidence supporting the use of biomarkers in pediatric sepsis. We have used the search engines PubMed, Cochrane Library, and Google Scholar to retrieve relevant information. We reviewed studies evaluating various biomarkers used for sepsis diagnoses, like C-reactive protein, ferritin, lactate, procalcitonin, tumor necrosis factor-alpha, etc. Apart from the diagnosis, trials are being conducted to assess the role of these biomarkers in monitoring and guiding antibiotic therapy to promote early recovery. The sensitivity of each biological marker varied in different studies, and no single biomarker can identify all types of infections. More robust studies are necessary to compare the roles of various biomarkers in diagnosing and guiding the appropriate therapy.

Key Words: Biomarker; C-reactive protein; Interleukin-8; Serum lactate; Procalcitonin; Sepsis' septic shock; Children

Core Tip: Sepsis is common in hospitalized pediatric patients, leading to increased morbidity and mortality. Routine laboratory investigations cannot always identify the cause of diseases, and many biomarkers have been used to detect these infections. This review analyses the available evidence supporting the use of biomarkers, such as C-reactive protein, ferritin, lactate, procalcitonin, and tumor necrosis factor-alpha. The sensitivity of each marker varied in different studies, and no single biomarker can be used to identify all infections. Apart from the diagnosis, studies have assessed the role of these biomarkers in optimizing antibiotic therapy to promote early recovery.

INTRODUCTION

Sepsis is a fatal condition and a major public health concern. Most fatalities occur in pediatric patients suffering from refractory shock and/or multiple organ dysfunction (MOD), and these patients die within the first 48-72 hours of treatment[1]. Recent international consensus criteria for pediatric sepsis and septic shock defined pediatric sepsis as a Phoenix Sepsis Score (PSS) of ≥ 2 points in children with suspected infection. It indicates cardiovascular, respiratory, coagulation, and/or neurological dysfunction, which can be potentially life-threatening. Septic shock is identified in the presence of sepsis along with cardiovascular dysfunction, which is indicated by at least 1 cardiovascular point in the PSS (severe hypotension for age, blood lactate > 5 mmol/L, or need for vasoactive drugs)[2]. Bradycardia is a sign of sepsis only in the newborn age group; older children do not exhibit bradycardia except as a near-terminal event in shock. A standard 16-bed pediatric intensive care unit (PICU) is expected to manage at least one critically ill child diagnosed with sepsis[3]. The primary sepsis pathophysiology involves the infection activating host response, innate immunity, coagulation abnormalities, organ impairment, vascular endothelium dysfunction, anti-inflammatory mechanisms, and immunosuppression. It is a challenge to identify which pediatric patients will develop sepsis based on the early signs and symptoms[4]. Routine laboratory investigations cannot always identify the cause of diseases, but assessing different biomarker levels can help identify these infections and treat sepsis accordingly[5].

The National Institute of Health defined a biomarker as a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention[6]. Diagnostic biomarkers establish the presence or absence of a disease process, while monitoring disease biomarkers allows physicians to assess the effectiveness of a given therapy. Surrogate biomarkers can monitor the disease to predict clinical outcomes[7]. Biomarkers are commonly evaluated by their sensitivity, specificity, and receiver operating characteristic curves to reflect physiologic states or pathological processes[8]. An ideal biomarker will have 100% sensitivity and specificity. A diagnostic test for sepsis that uses novel biomarkers in combination with or without clinical signs would contribute to prompt and appropriate targeting of monitoring and treatment for children at the highest risk[9].

This mini-review comprehensively evaluates the key biomarkers for pediatric sepsis, emphasizing their clinical utility, advantages, and limitations. By synthesizing current evidence, we aim to enhance the understanding of various biomarkers and patient outcomes in pediatric sepsis.

LITERATURE REVIEW

An electronic search was done using PubMed, Google Scholar, and Cochrane Library to collect relevant evidence. All cohort or interventional studies, randomized trials, systematic reviews, and meta-analyses on sepsis biomarkers are retrieved. Trials involving guiding the antibiotic therapy based on the biomarker profile to avoid antibiotic resistance are included. The search was limited to studies published from 2005 to March 2025 to ensure the inclusion of the most up-to-date research in the field. Studies reporting biomarkers for sepsis in pediatric populations were included, while studies focusing solely on adult and neonatal populations were excluded. Keywords used in the search are “sepsis”, “septic shock”, “biomarker”, prognostic marker”, “pediatric”, and “children”, with the limitation of “humans” combined using Boolean operators (“AND” and “OR”). This mini-review included studies on pediatric patients aged from 1 month to 18 years, excluding neonates.

DISCUSSION

The included studies reported on various biomarkers used for sepsis diagnosis, like C-reactive protein (CRP), ferritin, lactate, procalcitonin (PCT), tumor necrosis factor-alpha (TNF-α), interleukin (IL)-1β, IL-6, IL-8, and interferon-alpha (IFN-α). An overview of common biomarkers is presented in Table 1[7,9-11]. Different studies reported variable sensitivity of each biological marker in predicting sepsis in children (Table 2)[5,12-16]. Most studies were focused on CRP and PCT levels, which showed the highest correlation, making them the most reliable markers for sepsis identification and treatment.

Table 1 Overview of biomarkers in pediatric sepsis.

Biomarker	Type	Reference range	Advantages	Limitations	Ref.
C-reactive protein	Acute phase protein	< 10 mg/L	Cost-effective, high availability, helpful for fungal infection detection, unaffected by immune suppression effects	Low accuracy, variable sensitivity and specificity for detecting bacterial infection	Lim et al[9]
Procalcitonin	Peptide hormone	0.5-2.0 ng/mL	High specificity and sensitivity, more specifically, for bacterial infection, moderate prognostic value	Costly, variable threshold values for various infections, altered serum levels in cases of renal dysfunction, variable sensitivity, and specificity	Lim et al[9]
Ferritin	Acute phase protein	20-200 ng/mL	Potential use for real-time treatment adjustments	Low specificity and sensitivity, elevated in various inflammatory and liver conditions, not specific to sepsis	Lim et al[9], Tonial et al[10]
IL-6	Cytokine	< 7 pg/mL	Promising use in pediatric patients with cancer and febrile neutropenia, key proinflammatory cytokine in the immune response	Few studies in the pediatric population. Low availability, high cost, low specificity, and sensitivity	Z Oikonomakou et al[7], Esposito et al[11]
IL-8	Cytokine	≤ 220 pg/mL	Exhibits prognostic value for pediatric sepsis	Complex interpretation, variability among patients, moderate sensitivity, and specificity	Z Oikonomakou et al[7], Esposito et al[11]
IL-10	Cytokine	< 5 pg/mL	Demonstrates good sensitivity and specificity	limited by variability in expression	Z Oikonomakou et al[7], Esposito et al[11]
Lactate	Metabolic marker	0.5-2.2 mmol/L	Elevated levels suggest impaired oxygen delivery and utilization	Low accuracy for sepsis detection, levels can be elevated in conditions other than sepsis, such as trauma or liver dysfunction	Tonial et al[10], Esposito et al[11]
Leukocytes	Hematological parameter	5000-15000 cells/μL	Inexpensive and quick to obtain	Changes in white blood cell count can occur due to various non-infectious conditions	Z Oikonomakou et al[7]
TNF-related apoptosis-inducing ligand	Apoptosis regulator	40-60 pg/mL	Play a role in prognostication	Limited pediatric-specific data, variability and standardization issues	Z Oikonomakou et al[7], Tonial et al[10]
TNF-α	Cytokine	< 8 pg/mL	Helpful in the early detection of sepsis	Lack of specificity, variable sensitivity	Z Oikonomakou et al[7], Tonial et al[10]

IL: Interleukin; TNF: Tumor necrosis factor.

Table 2 Study details depicting various biomarkers for pediatric sepsis.

Country	Sample size	Age group	Biomarker	Main results			Cut-off value	Remarks	Ref.
				AUC value	Sensitivity (%)	Specificity (%)
India	124 patients	From 1 month to 15 years	PCT (day 0)	0.604	81.8	80.8	> 3.0 ng/mL	This study highlights PCT as a more reliable predictor than CRP for pediatric sepsis, especially in determining disease severity and guiding clinical management	Tyagi et al[12], 2024
			PCT (day 3)	0.993	86.2	89.8
			CRP (day 0)	0.848	63.6	61.0	10 mg/dL
			CRP (day 3)	0.85	84.7	62.8
Canada	20 sepsis patients	Median age 13 years	IL-6, IL-8, IL-10	1.00	-	-	95%CI: 1.00-1.00	P < 0.001	Leonard et al[5], 2024
			CRP	0.804	49.2	89.5	≥ 2.0 mg/dL	-
			PCT	0.746	54.1	87.5	≥ 0.3 ng/mL
Latvia	165 patients	From 1 month to 18 years	CRP	CRP of 0.799, PCT and IL-6 < CRP	83.0	65.0	69.9 mg/mL	The combination of CRP and sFAS enhanced sepsis prediction sensitivity over CRP alone, while the multi-biomarker panel of CRP, PCT, IL-6, sFAS, and sVCAM-1 achieved the highest AUC among all tested models	Rautiainen et al[13], 2022
			PCT		87.0	57.0	0.43 ng/mL
			G-CSF		28.0	83.0	61.58 pg/mL
			Eotaxin		51.0	69.0	61.77 pg/mL
			IL-10		51.0	68.0	23.03 pg/mL
			IL-8		45.0	77.0	21.90 pg/mL
			IL-6		86.0	46.0	18.30 pg/mL
			sVCAM-1		83.0	33.0	868 pg/mL
			sFAS		79.0	44.0	2538.29 pg/mL
United States	194 patients	From 1 month to 18 years	IL-8	0.68	72%	74%	43.5-209.5 pg/mL (survivors), 101-436 pg/mL (non-survivors)	Strongest significance of P < 0.001	Zinter et al[14], 2017
Multicenter, 8 European countries	38,480 children (17082 with WBC values)	0-18 years	WBC, ANC, CRP	0.71, 0.84, 0.71	56%, 32%, 87%, 55%, 55%	74%, 91%, 59%, 91%, 75%	WBC > 15000, WBC > 20000, CRP > 20, CRP > 80, ANC > 10	WBC is significantly associated with SBI; however, WBC does not have a diagnostic benefit than CRP in identifying children with SBI	Kemps et al[15], 2025
Brazil	350 patients	From 6 months to 18 years	CRP	0.648	-	-	> 6.5 mg/mL	In pediatric sepsis patients over six months old, ferritin, lactate, and CRP individually demonstrated strong prognostic value for mortality, and when combined, they predicted fatal outcomes in 75% of cases, whereas total leukocyte count lacked prognostic utility	Tonial et al[16], 2020
			Ferritin	0.785			> 135 ng/mL
			Lactate	0.762			> 1.7 mmol/L
			Leukocytes	0.508			-

ANC: Absolute neutrophil count; AUC: Area under the curve; CRP: C-reactive protein; IL: Interleukin; PCT: Procalcitonin; SBI: Serious bacterial infections; sFAS: Soluble FAS; sVCAM-1: Soluble vascular cell adhesion molecule-1; WBC: White blood cell.

The paediatric population caters to a wide age range from neonates to adolescents. Physiological variations related to age primarily affect the baseline values and cut-off thresholds of different biomarkers. For example, ferritin levels are potentially influenced by the maternal iron stores in infants under six months, while the reference values of leukocytes vary with age[17]. Therefore, age stratification is mandatory when interpreting biomarker values.

CRP

Macleod and Avery successfully isolated C-reactive material, identifying it as a protein that required calcium ions to interact with C-polysaccharide (CPS). They introduced the term "acute phase" to describe its behaviour, while the designation "C-reactive protein" was established to show that this protein is responsible for forming a precipitate with CPS[7]. It is a member of the pentraxin family, a group of plasma proteins that require calcium for ligand binding. CRP is a non-specific acute-phase biomarker used in pediatric sepsis. In healthy individuals, the levels are in trace amounts (< 0.3 mg/L); however, it can increase up to 1000-fold within 48 hours in response to acute conditions such as severe bacterial infections (SBIs) (> 200 mg/L)[18], viral infections (10-40 mg/L), trauma, tissue necrosis, inflammation, parasitic invasion, malignant neoplasia, burns, and aging. The level elevates within 4-6 hours of an inflammatory trigger, doubling every 8 hours, peaking at 36-50 hours, and declining rapidly with a 19-hour half-life, especially in invasive bacterial infections[19].

Non-specific CRP is the most frequently employed biomarker to differentiate sepsis from non-infectious systemic inflammatory response syndrome (SIRS) in pediatrics. A study investigated the clinical significance of high CRP levels (≥ 30 mg/dL) in 435 hospitalized children and found that bacterial infectious diseases were the most common diagnoses. The high CRP correlated with low albumin and with a crude mortality rate of 17.6%[20]. Sanders et al[21] conducted a systematic review of six studies assessing the diagnostic accuracy of CRP in differentiating SBIs from benign infections in children presenting with fever. They showed that CRP can help rule out SBIs; however, it may not exclude all bacterial infections due to its moderate sensitivity of 77%. The diagnostic accuracy of CRP alone for severe sepsis in children with febrile neutropenia was lower than that of PCT and IL-6[22].

Serial CRP measurements at multiple time points improve sepsis detection compared to a single test. A prospective cohort study conducted on 103 septic children reported the use of sequential CRP assessment in the early identification of patients with poor prognosis. A persistently elevated CRP level was associated with worse outcomes[23]. Palalıoğlu et al[24] assessed the impact of antibiotic treatment on levels of pentraxin-3, CRP, and IL-6 in paediatric patients suffering from sepsis and septic shock and found that CRP alone is a good alternative for monitoring treatment. Similarly, using CRP with other biomarkers increases the predictive values of these biomarkers. Santolaya et al[25] validated a risk prediction model based on variables such as age ≥ 12 years, CRP ≥ 90 mg/L, and IL-8 ≥ 300 pg/mL within the first 24 hours of admission to detect severe sepsis in children with high-risk febrile neutropenia.

Findings on diagnostic accuracy between studies vary due to sampling and cut-off thresholds. The suboptimal diagnostic accuracy can be due to potential false-positive readings in conditions such as inflammation secondary to extravasation, cholestasis, or gastrointestinal pathology, triggering an increase in CRP levels[9]. These studies suggest that while CRP demonstrates moderate diagnostic accuracy, its reliability varies across different populations, emphasizing the need for context-specific validation and potential enhancement through additional biomarkers.

PCT

PCT, a precursor to calcitonin, is a 116-amino-acid protein. PCT is generally not found in detectable levels in healthy people (< 0.1 ng/mL); however, it rises swiftly during bacterial infections because of the extensive expression of the CALC-1 gene, which is stimulated by proinflammatory cytokines. The CALC-1 gene regulates PCT production in various cells when triggered by pro-inflammatory cytokines such as TNF-α, IL-1β, and IL-6 during SBIs[26].

PCT is the Food and Drug Administration-approved biomarker used for diagnostic purposes[6]. PCT was first introduced as a diagnostic marker for sepsis in the early 1990s. Since then, substantial evidence has accumulated, demonstrating its efficacy in diagnosing early sepsis. Early studies showed that PCT levels were elevated in patients with SBIs, and subsequent research has consistently supported its use in various clinical settings. A prospective cohort study of 64 PICU patients with SIRS showed that PCT can identify bacterial infections better than CRP as a single biomarker; however, the performance of PCT alone was only moderate[27].

Many systematic reviews and meta-analyses evaluated the role of PCT as a diagnostic tool and found that PCT has variable diagnostic performance depending on the clinical context, infection type, and population. A summary of recent systematic reviews and meta-analyses evaluating the diagnostic accuracy of PCT and other biomarkers is listed in Table 3[28-37]. A meta-analysis of studies evaluating the predictive value of serum biomarkers in assessing and managing fever during neutropenia in children concluded that PCT at a threshold of 0.5 ng/mL appears to be the most suitable admission biomarker to predict adverse outcomes[28].

Table 3 Overview of systematic review and meta-analysis in pediatric sepsis.

Number	Ref.	Biomarkers studies	Number of studies (patients)	Study population	Main results	Conclusion	Limitations
1	Norman-Bruce et al[29], 2024	PCT and CRP	14 studies (n = 7755)	Children aged ≤ 90 days, with fever or history of fever within the preceding 48 hours	For detection of IBI, pAUC was higher for PCT than CRP (0.72 vs 0.28; P = 0.016), but PCT and CRP had similar pAUC values (0.55 vs 0.54; P = 0·92) for detection of SBI	PCT (cutoff of 0·5 ng/mL) had better diagnostic accuracy for IBI than CRP (cutoff of 20 mg/L), and it was similar for SBI	High heterogeneity for SBI studies, lack of a universal SBI definition, and potential bias
2	Qi et al[30], 2024	PCT	5 studies (n = 148)	Children with osteomyelitis	Pooled sensitivity and specificity of PCT were 0.58 (95%CI: 0.49-0.68) and 0.92 (0.90-0.93), respectively	PCT had the greatest AUC at 0.80 for the diagnosis of osteomyelitis in children	Small sample size, variable nature of included studies
3	Kim et al[31], 2021	PCT	18 studies (n = 1462)	Children with bacterial meningitis	pSn, pSp, and DOR of PCT for detecting bacterial meningitis were 087, 0.85, and 35.85, respectively. AUC was 0.921	Blood PCT has high diagnostic accuracy in detecting bacterial meningitis in children	Variable methodologies and small sample sizes
4	Boon et al[32], 2021	20 urine biomarkers and 4 blood biomarkers (CRP, PCT, WBC, absolute neutrophil count)	54 studies (n = 117531. UTI, 628-pyelonephritis, and 6320- bacteraemia)	Children with UTI presenting to ambulatory care	CRP and PCT had low accuracy for cystitis (AUC of 0.75 and 0.71). CRP < 20 mg/L might be useful for ruling out UTI, and PCT ≥ 2 ng/mL for ruling in pyelonephritis	CRP and PCT have low accuracy for cystitis, but can be used for pyelonephritis	Heterogeneous patient selection criteria
5	Shaikh et al[33], 2020	PCT, CRP, and ESR	25 studies (PCT, n = 1000; CRP, n = 189; ESR, n = 1910)	Children aged 0-18 years with culture-confirmed UTI	For cut-off values of 0.5 ng/mL for PCT, 20 mg/L for CRP, and 30 mm/hour for ESR, pSn were 081, 0.93, and 0.83, and pSp were 076, 0.37, and 0.57, respectively	All three tests were sensitive but not very specific for ruling in pyelonephritis	High heterogeneity, limited number of studies per test
6	Tsou et al[34], 2020	PCT	25 studies (n = 2864)	Children with bacterial pneumonia	For a cut-off of 0.5 ng/mL and 2 ng/mL, PCT had a pSn of 0.68 and 0.59, pSp of 0.60 and 0.71, and AUC of 0.68 and 0.71. Elevated PCT did not suggest bacterial pneumonia (odds ratio: 1.36, P = 0.18).	Moderate diagnostic accuracy (AUC = 0.74); best cut-off around 0.5-2 ng/mL	Variability in cutoffs and definitions. Variable timings of PCT measurement
7	Cui et al[35], 2019	PCT	7 studies (504 confirmed AA and 368 controls)	Children with AA and complicated appendicitis	pSn and pSp of PCT for the diagnosis of AA were 062 and 0.86. DOR was 21.4, and AUC was 0.955. PCT was more accurate in diagnosing complicated appendicitis (pSn of 0.89, pSp of 0.90, DOR of 76.73)	PCT was more accurate for complicated appendicitis (pSn of 0.89, pSp of 0.90) than for AA	Small number of studies; moderate heterogeneity; potential publication bias
8	Yoon et al[36], 2019	Presepsin, CRP, PCT	4 studies (n = 308)	Children aged from 1 month to 18 years with sepsis	pSn and AUC of presepsin (0.94 and 0.925) were higher than that of CRP (0.51 and 0.715) and PCT (0.76 and 0.820), whereas pSp of presepsin (0.71) was lower than that of CRP (0.81) and PCT (0.76).	Presepsin has higher sensitivity and diagnostic accuracy, but lower specificity, in detecting sepsis in children	Small sample size, differences in the reference standards
9	Arif and Phillips[28], 2019	30 different biomarkers. The most common were PCT, CRP, IL-6, and IL-8	41 studies (n = 4842)	Febrile neutropenia in children with cancer	The pSn and pSp for different biomarkers to detect any adverse outcome: CRP pSn of 40%, pSp of 65%; PCT pSn of 60%, pSp of 75%; IL-6 pSn of 65%, pSp of 70%; and IL-8 pSn of 70%, pSp of 60%	PCT > 0.5 ng/mL best predicted bacteraemia and severe sepsis: Sensitivity of 0.67, and specificity of 0.73	Inconsistencies in methodology and reporting of outcomes
10	Trippella et al[37], 2017	PCT	12 studies (n = 7260)	Children with fever without an apparent source	For IBI, sensitivity was 0.82 and 0.61, and specificity was 0.86 and 0.94 at PCT levels of 0.5 ng/mL and 2 ng/mL, respectively. For SBI, PCT had lower sensitivity (0.55 and 0.30) and specificity (0.85 and 0.95)	High diagnostic accuracy for IBI (AUC > 0.9) but poor for SBI, especially at higher PCT cutoffs	Inconsistent definitions, variable cut-offs, and heterogeneity in study populations

AA: Acute appendicitis; AUC: Area under the curve; CRP: C-reactive protein; DOR: Diagnostic odds ratio; ESR: Erythrocyte sedimentation rate; IBI: Invasive bacterial infections; IL: Interleukin; pAUC: Partial area under the curve; PCT: Procalcitonin; pSn: Pooled sensitivity; pSp: Pooled specificity; SBI: Serious bacterial infections; WBC: White blood cell; UTI: Urinary tract infections.

PCT concentrations rise more quickly than CRP levels in patients with SBI. PCT levels elevate within 2 hours of infection onset, peaking in serum between 24 hours and 36 hours. PCT should be utilized alongside other findings in treating children with pneumonia, as it has moderate diagnostic accuracy[38]. A prospective study conducted on a large cohort of febrile children aged from 1 month to 16 years of age, attending the emergency department (ED), showed that CRP and PCT were strong predictors of SBI[39].

PCT remains a promising but incompletely understood biomarker for the rapid detection of SBIs in critically ill children. While some studies support its use in patients with neutropenia, its effectiveness in other immunocompromised conditions, especially in critically ill children, has not been thoroughly investigated. PCT may also fail to identify clinically significant localized infections. Additionally, its reliability in fungal infections is limited, as PCT levels tend to be lower than those in bacterial infections. Therefore, the role of PCT remains unclear in detecting sepsis of nonbacterial origin in high-risk children[26].

Heat-shock protein

Heat shock proteins (HSPs) are widely recognized for their critical role in maintaining cellular homeostasis, primarily by functioning as molecular chaperones that facilitate the proper folding of nascent polypeptides and stabilization or refolding of damaged proteins[40]. Traditionally, HSPs were considered strictly intracellular, but HSP70 was found to be released into the extracellular environment by thermally stressed cultured rat embryo cells. The extracellular HSPs actively participate in immune regulation and engage both innate and adaptive immune cells, suggesting a dual role of cellular protection and immunomodulation. Despite this progress, our understanding of the differential roles of intracellular and extracellular HSPs in the context of sepsis remains incomplete[41].

In a study conducted on 20 children with sepsis (median age: 13 years), HSP70 levels positively correlated with mortality [area under the curve (AUC): 0.975, 95%CI: 0.937-1.000] and negatively correlated with the need for inotropic medication, suggesting a greater hemodynamic stability throughout the illness in children with increased HSP70 levels[5]. An observational study found significantly elevated levels of extracellular HSP90α in children with severe sepsis and SIRS[42]. Given their measurable elevation in pediatric patients with severe systemic inflammation, extracellular HSPs - particularly HSP72 and HSP90α - are emerging as potential diagnostic biomarkers for pediatric sepsis. Their ability to reflect underlying immune activation and cellular stress makes them promising candidates for early detection and risk stratification in critically ill children. However, further research is needed to validate their diagnostic utility and define specific thresholds relevant to the pediatric population[42].

Ferritin

Ferritin is widely recognized as a marker of iron storage and, to a lesser extent, as an acute-phase reactant. During infection, elevated ferritin levels reflect the acute-phase response, likely due to the sequestration of serum iron as a host defense mechanism[9]. In children older than six months, ferritin becomes a reliable biomarker, as earlier levels are influenced by maternal iron stores and the physiological transition from fetal to adult hemoglobin. A cohort study of 350 children with sepsis admitted to the PICU, most of whom had iron deficiency anemia, reported a five-fold increase in mortality associated with a ten-fold rise in serum ferritin levels[16]. Markedly elevated ferritin levels (above 1000 ng/mL) have been observed in a range of conditions, including infections, autoimmune diseases, malignancies, and renal disorders, with the highest concentrations reported in hemophagocytic lymphohistiocytosis, ranging from 994 ng/mL to 189721 ng/mL[43].

In one study of children with severe sepsis and septic shock, optimal thresholds to predict mortality were identified as CRP ≥ 4.08 mg/dL and ferritin ≥ 1980 ng/mL[44]. Another study involving 70 PICU-admitted children (aged 1 month to 12 years) assessed ferritin levels within the first five days of illness and found that a cutoff of 558 ng/mL predicted mortality with 74.1% sensitivity and 67.7% specificity. Notably, ferritin retained its prognostic value even in anemic children, with an AUC of 0.764 (95%CI: 0.634-0.894)[45]. Further reinforcing these findings, Sarkar et al[46] conducted a prospective study on 132 children with septic shock or severe sepsis. A ferritin level of ≥ 2375 ng/mL demonstrated excellent predictive value for mortality, with 96.7% sensitivity and 88% specificity, and correlated positively with established severity scores such as Pediatric Risk of Mortality (PRISM) III and perovskite light-emitting diodes.

Leukocytes

Among the many biomarkers studied for the early detection of sepsis in pediatric patients, leukocytes have long served as primary indicators of infection and systemic inflammation. Total white blood cell (WBC) count remains a foundational component in sepsis diagnosis; however, its clinical specificity is limited[9]. Abnormal leukocyte counts reflect the immune system’s response to infection, and in children, leukopenia (WBC < 5000 cells/μL) is more suggestive of sepsis than leukocytosis (WBC > 11000 cells/μL). Neutropenia (absolute neutrophil count < 1500 cells/mm³) can be classified as mild (1000-1500 cells/mm³), moderate (500-1000 cells/mm³), or severe (< 500 cells/mm³). Decreasing cell count is associated with an increased risk of infection, its severity, and duration, and febrile neutropenia carries substantial morbidity and mortality[47]. While these parameters are widely used, they are influenced by factors like age, underlying conditions, and co-existing viral or bacterial infections, which can result in false positives or negatives. Moreover, leukocyte counts may not consistently correlate with infection severity in immunocompromised children. Thus, while leukocyte-based assessments remain useful initial indicators, they should be interpreted along with other biomarkers and clinical parameters to improve diagnostic accuracy.

The researchers have investigated more specific biomarkers, such as the neutrophil CD64 (nCD64) index, to enhance the diagnostic precision at an early stage. A prospective, single-center study by Cao et al[48] evaluated the utility of the nCD64 index in 201 children admitted to a PICU in China. The children were classified into infection and non-infection groups, and biomarkers including nCD64, CRP, PCT, and WBC counts were measured within the first 24 hours of admission. The nCD64 index demonstrated superior diagnostic value, with a sensitivity of 68.8%, specificity of 90.7%, and overall accuracy of 80.5% at a cut-off value of 0.14, outperforming CRP and PCT. It also effectively distinguished infection from SIRS in children with postoperative fever, underscoring its potential role in clinical decision-making. However, given its single-center design and modest sample size, broader validation is needed before general clinical application[48].

Supporting the value of neutrophil-related biomarkers, a prospective cohort study from Egypt included 120 children admitted to the PICU and 40 healthy controls. They found that serum neutrophil gelatinase-associated lipocalin levels were significantly elevated among septic patients compared to controls (P < 0.001). It also showed discriminatory value in patients with SIRS without sepsis and those without SIRS (P = 0.04 and P < 0.001), providing added value for early diagnosis and prognostication in critically ill children[49].

More recently, machine learning approaches have enabled the development of high-precision gene expression-based diagnostic models focused on neutrophil function and lysosomal activity. In a study by Zhang and Zhang[50], a five-gene diagnostic model based on neutrophil extracellular trap (NET)-related genes - MME, BST1, S100A12, FCAR, and ALPL - was constructed using Gene Expression Omnibus datasets and validated with clinical samples. This model demonstrated excellent diagnostic performance with AUC values of 1.00, 0.932, and 0.966 across three datasets, and 0.898 in clinical validation. It also outperformed conventional markers such as PCT, CRP, WBC, and neutrophil, and found a strong association with NET formation[50].

Similarly, a 2023 study by Yang and Zhang[51] identified four lysosome-related genes - STOM, VNN1, SORT1, and RETN - as reliable diagnostic markers for pediatric sepsis. This model demonstrated high predictive accuracy (AUCs of 1.00, 0.989, and 0.971 in datasets; 0.917 in clinical samples) and a robust correlation with neutrophil infiltration, suggesting a mechanistic link between lysosomal gene expression and neutrophil-driven immune responses.

These findings highlight the evolving landscape of pediatric sepsis biomarkers. From basic leukocyte counts to advanced indicators like the nCD64 index and gene expression models, diagnostic tools are becoming increasingly accurate and mechanistically informative. These innovations not only enhance early diagnosis but also support risk stratification and personalized treatment strategies for critically ill children.

Cytokines and chemokines

Cytokines are small proteins secreted by immune cells that regulate immunity, inflammation, and hematopoiesis, and chemokines are a subclass of chemotactic cytokines that guide the migration of immune cells to sites of infection or inflammation. The common cytokines include ILs (IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12p70, IL-17, IL-18, IL-1β), TNF-α, inducible protein-10, and IFNs (IFN-α and IFN-γ)[19]. Prominent pro-inflammatory cytokines include ILs (IL-6, IL-12, IL-17) and TNF-α, while IL-10 and transforming growth factor-beta (TGF-β) are key anti-inflammatory mediators. During sepsis, the balance between pro-inflammatory and anti-inflammatory cytokines becomes dysregulated, contributing significantly to disease progression[52].

The inflammatory cascade is initiated by the release of chemokines such as IL-8 and macrophage inflammatory protein from antigen-presenting cells and endothelial cells, which in turn promote neutrophil recruitment and the release of cytokines like IL-6 and TNF-α. TNF-related apoptosis-inducing ligand also shows pathogen-specific expression patterns and aids in distinguishing bacterial from viral infections, making it a potential biomarker for diagnosis and treatment decision-making[53].

Several cytokines have shown good predictive value in diagnosing sepsis in critically ill children. Angurana et al[54] reported significantly elevated IL-6, IL-12p70, IL-17, and TNF-α levels in non-survivors compared to survivors. However, they showed weak positive correlations with severity scores such as PRISM III, Sequential Organ Failure Assessment, and Vasoactive Inotrope Score, and none could diagnose sepsis as a standalone marker. Leonard et al[5] identified IL-6, IL-8, monocyte chemoattractant protein-1 (MCP-1), and IL-1 receptor antagonist as top predictive biomarkers of sepsis, each demonstrating AUC values exceeding 0.9, indicating higher diagnostic accuracy than the conventional biomarkers like CRP and PCT.

Finnerty et al[55] conducted a study to determine if the cytokine profile in children with burns can identify patients at high risk of developing sepsis and subsequent death. They found that a combination of elevated IL-6 and IL-12 p70 and lower TNF had an elevated risk of dying of sepsis. Wong et al[56] demonstrated that children with septic shock having serum IL-8 levels ≤ 220 pg/mL within 24 hours of admission had a higher likelihood of survival. Zeng et al[57] found significantly elevated levels of CRP, PCT, and IL-6 in children with SIRS and hyperinflammatory state (AUCs ≥ 0.9 for all three), and increased PCT, IL-8, and IL-10 levels in septic children with organ dysfunction [AUCs: (1) PCT of 0.807; (2) IL-8 of 0.711; and (3) IL-10 of 0.860].

Compared to traditional markers, cytokines enhance specificity and may facilitate early risk stratification, severity assessment, and guide therapeutic decision-making. Their clinical utility is enhanced when measured in combination and interpreted in conjunction with patient condition and other laboratory findings. Despite their diagnostic potential, cytokines are not routinely used in clinical practice due to high cost, variability in assay protocols, and limited accessibility. Research is essential to standardize cytokine measurement techniques and validate their utility across larger and more diverse patient cohorts[5].

Lactate

Lactate has become an increasingly recognized biomarker in pediatric sepsis for early diagnosis, risk stratification, and outcome prediction. Lactate is produced as a byproduct of anaerobic glycolysis, where pyruvate is converted into lactate under low oxygen conditions. Typically, a lactate level of < 2 mmol/L (or 18 mg/dL) within the first four hours of presentation is considered normal. Sepsis-induced tissue hypoperfusion leads to increased anaerobic metabolism and lactate production, making elevated serum lactate levels a potential surrogate for circulatory dysfunction correlating with poor outcomes in pediatric septic shock[58].

In a cohort study of children with severe sepsis or septic shock, Gorgis et al[59] found that early lactate levels (measured within three hours of sepsis onset) were not directly predictive of mortality but correlated significantly with PRISM III score, thus supporting its use as an early severity marker rather than a mortality predictor. Scott et al[60] reported that initial lactate levels > 36 mg/dL were independently associated with increased 30-day mortality in children admitted to the ED with suspected sepsis, with a limited sensitivity (20%) but high specificity (92.3%), making it a useful tool for risk stratification. Another study found that early hyperlactatemia (≥ 4.0 mmol/L within 24 hours of admission) was associated with a fivefold increased risk of organ dysfunction in children admitted with SIRS[61].

Lactate clearance (reduction in lactate over time) and lactate trends by serial measurement of serum lactate showed even greater promise[62]. Choudhary et al[63] found significantly lower lactate clearance in non-survivors than survivors, with a clearance of less than 10% at 24 hours strongly predicting in-hospital mortality (positive predictive value of 83.1%) in children with septic shock. Jaiswal et al[64] reinforced these findings, showing that lactate levels at six hours (cut-off: 2.5 mmol/L) provided better predictive accuracy than initial levels, with a sensitivity of 85% and specificity of 74%. Nazir et al[65] found that 24-hour lactate clearance was superior to 6-hour lactate clearance in predicting mortality in children with septic shock, and the optimal lactate clearance was associated with lower in-hospital and 60-day mortality. These studies support the incorporation of early lactate measurement and serial lactate clearance into pediatric sepsis protocols. However, variability in timing, methodology, and cut-off thresholds across studies underscores the need for standardization.

Newer biomarkers

Advancements in proteomics and genomics have led to efforts to discover novel diagnostic markers, which can prevent sepsis progression at the initial stages or in those susceptible to serious infections. Recent studies found dysregulation of micro-RNA in response to viral and bacterial pathogens. They can be termed as perfect biomarkers due to the stability in many body fluids and ease of measurement by reverse transcriptase polymerase chain reaction, and are sensitive and specific for diagnosing disease conditions, including sepsis[66]. DNA and RNA microchips created the possibility of complex investigations at the genomics level, along with the mRNA pool. A cell induction with in vitro lipopolysaccharide to examine the cellular response is an elegant way of proteomic analysis. Studies on newer coagulation markers, including soluble thrombomodulin (sTM) and tissue plasminogen activator inhibitor complex (t-PAIC), suggest their diagnostic value in pediatric sepsis. Significantly higher levels of thrombin-antithrombin complex (TAT), α2-plasmin inhibitor-plasmin complex (PIC), and t-PAIC were found in children with severe sepsis, correlating with poor outcomes [AUC: (1) TAT of 0.862; (2) PIC of 0.759; and (3) The t-PAIC of 0.851][67].

Presepsin, a soluble variant of CD14 found on the surface of macrophages and monocytes, plays a role in activating toll-like receptor 4 when stimulated by lipopolysaccharides, leading to the release of TNF-α. During sepsis, soluble CD14 is released. Its exact physiological function is not completely understood, but it is associated with the phagocytosis and lysosomal breakdown of pathogens. The initial increase in serum concentrations during bacteremia, before any rise in PCT or IL-6, suggests it could serve as a valuable early indicator of sepsis[68]. Several studies have demonstrated its potential superiority over CRP and PCT in terms of early rise and specificity for bacterial infections. Presepsin levels start increasing within two hours and peak in three hours with a halftime of eight hours, which is earlier than PCT and CPR. Most studies reported more than 90% sensitivity with a cut-off of 650 pg/mL for the detection of sepsis[69].

A meta-analysis of four studies, including 308 children, found that presepsin had a higher pooled sensitivity (0.94) but lower specificity (0.71) than that of PCT (0.76 and 0.76) and CRP (0.51 and 0.81) for identifying sepsis. The AUC of presepsin (0.925) was higher than that of PCT (0.820) or CRP (0.715)[36]. Despite its diagnostic promise, the routine clinical use of presepsin remains limited due to a lack of test standardization, high cost, limited availability, and widely variable cut-off values (range: 450-850 pg/mL). Furthermore, age-specific reference ranges are not well established, and more studies are required to validate cost-effectiveness and to establish age-specific cut-offs.

Multi-biomarker approach for pediatric sepsis

A multi-biomarker approach has emerged as a valuable tool for enhancing the diagnosis, risk stratification, and prognosis of pediatric sepsis. Unlike single biomarkers, which often lack sensitivity or specificity, combining multiple biomarkers helps clinicians diagnose complex and heterogeneous sepsis conditions at an early stage. Their combination reflects the multiple host-pathogen interactions in pediatric sepsis, thereby enhancing risk prediction beyond a single biomarker. This strategy enables a comprehensive assessment of the patient’s immune response and disease severity, improving diagnostic accuracy and informing timely therapeutic interventions.

One of the most notable developments in this area is the Pediatric Sepsis Biomarker Risk Model (PERSEVERE), a biomarker-based stratification tool for assessing the patient’s baseline risk of mortality from sepsis. PERSEVERE uses five biomarkers identified through transcriptomic analyses, including C-C chemokine ligand 3 (CCL3), HSP70 kDa 1B (HSPA1B), IL-8, granzyme B (GZMB), and matrix metalloproteinase 8 (MMP-8). It was designed to predict 28-day mortality in children with septic shock[70]. In validation studies, PERSEVERE demonstrated a good sensitivity and specificity for mortality prediction. Later iterations, such as PERSEVERE-II and PERSEVERE-XP, incorporated additional clinical and molecular variables, such as platelet count, protein, and mRNA biomarkers, leading to improved predictive performance[71,72].

These models increase the predictive strength by utilizing synergistic mechanisms and integrating biomarkers of distinct but complementary biological pathways. Inflammatory cytokines (e.g., CCL3, IL-8, TNF-α) reflect pro-inflammatory chemokine signalling; GZMB reveals the cytotoxicity of the immune cells, HSPA1B indicates a cellular stress mechanism, and MMP-8 suggests neutrophil-driven tissue remodelling. Similarly, other combinations use endothelial and vascular injury markers (e.g., angiopoietins, sTM) to capture microcirculatory dysfunction, presepsin to assess immune activation and dysregulation, and lactate and proadrenomedullin to evaluate metabolic stress and tissue hypoxia.

Tonial et al[10] reported that a combination of ferritin, lactate, and CRP predicted mortality in approximately 43% of pediatric sepsis patients, and it increased to 76% when the Pediatric Index of Mortality 2 score was added. Another study found that adding soluble FAS (sFAS) to CRP increased diagnostic sensitivity from 83% to 88%. A broader panel including CRP, IL-6, sFAS, and soluble vascular cell adhesion molecule-1 achieved an AUC of 0.830, outperforming CRP alone (AUC of 0.799)[13]. In a study, combinations of IL-8 and IL-6, MCP-1 and HSP70, and hyaluronan and IL-6 achieved perfect discrimination between sepsis patients and healthy controls (AUC of 1.00), underscoring the potential of multi-biomarker panels in enhancing diagnostic accuracy[5]. A combination of CRP, PCT, and IL-6 improves diagnostic precision for systemic inflammation, while PCT, IL-8, and IL-10 are useful for the early detection of organ dysfunction[57]. Another study demonstrated improved diagnostic performance (AUC of 0.92) of a combination of PCT (AUC of 0.83), high-sensitivity CRP (AUC of 0.76), and pancreatic stone protein (AUC of 0.73) in assessing sepsis risk than any of them alone[73].

Role of biomarkers in optimizing antimicrobial therapy

The timely initiation of appropriate antibiotics improves the outcomes in children admitted with sepsis; however, prolonged antibiotic use increases the risk of developing antimicrobial resistance. A recent study conducted on febrile children attending EDs of nine European countries found that a high proportion of children (80.3%) with a viral phenotype received ≥ 1 systemic antibiotic[74]. In this context, sepsis biomarkers have the potential to optimize antibiotic therapy, from initiation to escalation and discontinuation, in children. A recent study demonstrated the relationship of elevated WBCs with SBI in children. However, WBC does not have an additional diagnostic benefit over CRP in identifying children with SBI[15]. Many studies have demonstrated the utility of CRP and PCT in differentiating bacterial from viral infections and evaluated biomarker-guided antibiotic strategies in pediatric sepsis[15,75].

Although CRP and PCT are the most studied biomarkers, newer biomarkers, such as presepsin, ILs, and gene expression panels, are being investigated. Many studies evaluated the role of CRP in optimizing antibiotic therapy in children. A prospective cohort study [optimizing antibiotic strategies in sepsis (OASIS) study], conducted on 85 children, found that a combination of biomarkers (CRP, serum amyloid A, and PCT) can identify children with SIRS in whom antibiotics might be discontinued at 48 hours[76]. CRP use has been shown to reduce antibiotic prescriptions in studies conducted on adult and neonatal populations, as well as its cost-effectiveness[77-79]. Similarly, PCT-guided antibiotic stewardship has been shown to reduce the duration of antibiotic therapy and hospital stay in children admitted with sepsis or lower respiratory infections, or in stopping antibiotics in febrile neutropenic children with cancer[80,81]. A meta-analysis of four pediatric RCTs (1313 patients with infectious disease) found a significantly shorter duration of antibiotic therapy in the PCT-guided group than in the control group (P < 0.05)[82].

On the contrary, the OASIS II study failed to demonstrate a reduction in overall antibiotic exposure among uninfected PICU patients, following implementation of a biomarker-based algorithm (CRP < 4 mg/dL and PCT < 1 pg/dL)[83]. A large multicenter RCT (the BATCH trial) found that PCT-guided management did not significantly reduce the duration of antibiotics compared to standard care in children treated with intravenous antibiotics; however, it was not inferior to the usual care[84]. Therefore, the initiation or duration of antibiotics cannot be guided solely by biomarkers. Biomarkers should be used as adjuncts to clinical judgment, considering other factors such as microbiological data, healthcare setting, and adherence to antibiotic stewardship guidelines. Combinations of multiple biomarkers and dynamic monitoring of biomarkers are more useful than single values.

Limitations and future directions

Despite the promise of multi-biomarker models, several limitations remain. The global burden of pediatric sepsis underscores the need for standardized, validated protocols for biomarker use. Most studies assessing the role of different biomarkers in sepsis had small sample sizes, heterogeneous study populations, and variable sample collection timings. Current biomarkers cannot differentiate between causative pathogens, predict antimicrobial resistance, or distinguish between sterile and non-sterile inflammatory responses. Moreover, biomarkers alone do not account for critical components of sepsis pathophysiology, such as gastrointestinal barrier dysfunction, which can lead to microbial translocation and MOD. Trials investigating the use of probiotics to modulate cytokines such as IL-6 and TGF-β1 in PICU patients are ongoing, exploring the gut-immune axis as a therapeutic target[85].

A translational "bedside-bench-bedside" approach is advocated to bridge clinical observations and molecular insights. Such a strategy may yield pragmatic diagnostic tools and personalized interventions for children with sepsis[86]. Further research should focus on large-scale, prospective trials to refine biomarker panels, validate their clinical utility, and integrate them into routine practice for precision-based care. Additionally, technological advancements such as point-of-care testing and integration of patient dynamics with predictive analytics or machine learning models could facilitate real-time clinical decision-making[87].

CONCLUSION

Current evidence highlights the value of biomarkers such as CRP, PCT, ferritin, lactate, and cytokines in the early identification and monitoring of pediatric sepsis. These markers, particularly CRP and PCT, show high sensitivity in distinguishing sepsis from non-infectious conditions. Elevated biomarker levels often correlate with clinical findings and disease severity, making them essential tools for early screening and guiding timely interventions. Immunologic assays enable the quantification of immune dysregulation, facilitating personalized treatment strategies tailored to individual sepsis phenotypes. While promising, biomarker-based diagnostics must be further validated through hypothesis-driven research to ensure widespread clinical application. Ultimately, integrating multi-biomarker panels into sepsis protocols could significantly enhance early diagnosis, improve outcomes, and reduce mortality in pediatric populations.