Assessment of Serum Interleukins 18 and 22 in Predicting Neonatal Sepsis and Mortality: A Case-Control Study

Abstract

Background:

Sepsis is a leading cause of neonatal morbidity and mortality, but its early diagnosis remains challenging. Neonatal sepsis (NS) induces an immune response with the release of Inflammatory cytokines, which may serve as early diagnostic biomarkers. Limited data indicate interleukin (IL)-18 and IL-22 levels are elevated in NS and associated with mortality.

Objectives:

To investigate the diagnostic performance of serum IL-18 and IL-22 levels in predicting NS and mortality.

Design:

Case-control study.

Methods:

We enrolled 55 newborns with culture-positive sepsis and 34 age- and sex-matched healthy controls. Serum IL-18 and IL-22 levels were measured using enzyme-linked immunosorbent assay. Receiver operating characteristic (ROC) curve analysis was used to assess the diagnostic performance of IL-18 and IL-22 in predicting NS and mortality.

Results:

Compared with controls, newborns with NS had significantly higher median levels of IL-18 (910 vs 256 pg/mL, P < .001) and IL-22 (96 vs 20 pg/mL, P < .001). Among septic neonates, non-survivors had significantly higher median levels of IL-18 (1820 vs 873 pg/mL, P < .001) and IL-22 (139 vs 91 pg/mL, P < .001) than survivors. ROC curve analysis demonstrated excellent performance of IL-18 and IL-22 in predicting NS (area under the curve [AUC] 0.990 and 0.998, respectively) and mortality (AUC 0.942 and 0.994, respectively).

Conclusion:

IL-18 and IL-22 are potentially promising biomarkers for identifying NS and predicting neonatal mortality.

Introduction

Neonatal sepsis (NS) represents a severe bloodstream infection accompanied by a systemic inflammatory response that occurs during the first month of life.^1,2 This condition imposes a substantial medical and socioeconomic burden worldwide.^3-5 The estimated incidence of NS is much higher in low- and middle-income countries than in high-resource settings (28-39 vs 0.3-0.8 per 1000 live births) as is the NS-related mortality (18% vs 0-8%).^5,6 Management guidelines emphasize the importance of early identification and treatment of NS in order to improve clinical outcomes. However, diagnosis of NS remains challenging due to its non-specific clinical manifestations.^7,8 The gold standard for diagnosis is a positive microbial culture from normally sterile body fluids (eg, blood, cerebrospinal fluid, urine), but this method has a poor positive yield and requires a long turnaround time.^5,9 Moreover, commonly used laboratory tests to support clinical diagnosis of NS, such as C-reactive protein (CRP) and hematological counts, have limited accuracy, particularly for single-timepoint measures.^10-12 Therefore, more reliable and earlier biomarkers for NS are highly needed.

Interleukins (ILs) are a subgroup of cytokines that play a central role in immunoinflammatory response to infection.^10,13,14 The initial host response depends on the identification of microbial antigens by immune cells through pathogen recognition receptors, particularly Toll-like receptors (TLRs).^2,5 The pathogenesis of NS involves an early phase of systemic inflammatory response syndrome (SIRS), characterized by excessive production of proinflammatory cytokines (eg, IL-1β, IL-6, IL-8, IL-12, IL-18, interferon-γ [INF-γ] and tumor necrosis factor-α [TNF-α]), followed by immune suppression and the release of anti-inflammatory cytokines (eg, IL-4, IL-10, IL-11, IL-13).^13,15,16 Moderate increases in circulatory cytokines seem protective and enhance antimicrobial defense, whereas exaggerated release contributes to sepsis-associated multiorgan damage and death.^2,5,17-19 Since proinflammatory interleukins typically rise more rapidly than acute phase reactants (eg, CRP, procalcitonin), their levels could serve as earlier and more accurate predictors of NS.^10,11,13

Among interleukins, IL-18 and IL-22 are key mediators of the immunoinflammatory response during sepsis, with evidence of complex interplay between each other.^20,21 An accumulating body of evidence from experimental and human adult studies indicates that serum IL-18 and IL-22 levels are elevated in sepsis and associated with poor outcomes.^22-46 Some studies have reported higher IL-18 levels in serum and urine of newborns with sepsis, particularly among non-survivors,^47-50 although these findings were not replicated in other studies.^51,52 Most of these neonatal studies had methodological limitations, including variable definitions of NS and possible confounding factors, which limit the validity of their findings. The role of IL-22 in NS has been investigated in only one study, which showed no significant association between cord blood IL-22 levels and early onset sepsis (EOS); however, that study included only 7 cases with culture-confirmed sepsis. ⁵³

Considering this background, we aimed to investigate serum levels of IL-18 and IL-22 in NS. The present study highlights the potential clinical utility of IL-18 and IL-22 for early identification of NS and prediction of neonatal mortality.

Patients and Methods

Design and Setting

This case-control study was performed between March and October 2024 in the neonatal intensive care unit (NICU) of Sohag University Hospital and included 2 independent groups: 55 cases with NS and 34 age- and sex-matched healthy control newborns.

Participants

Eligible cases were term newborns ⩽28 days old with culture-positive sepsis who were admitted to the NICU during the study duration. Controls were age- and sex-matched healthy neonates presenting for routine health checkups at the same facility. Exclusion criteria were major congenital anomalies, hematological diseases, malignancies, autoimmune disorders, need for cardiopulmonary resuscitation within the past 6 hours, prior antibiotic therapy before admission, and failure to get informed consent.

Procedures

Study participants underwent comprehensive history-taking and clinical assessment, including gestational age, mode of delivery, sex, postnatal age, and weight. We measured hemodynamic parameters using Vista 120 hemodynamic monitor (Dräger Medical Inc., Lübeck, Germany). Complete blood count was conducted by Sysmex XN-1000TM Hematology Analyzer (Sysmex Corp., Kobe, Japan). C-reactive protein was measured through a latex agglutination test (Reactivos GPL, Barcelona, Spain) according to manufacturer’s instructions. The severity of organ dysfunction was evaluated using the neonatal Sequential Organ Failure Assessment (nSOFA) score, which contains respiratory, cardiovascular, and hematological components, with a total score ranging from 0 to 15. ⁵⁴

Blood samples for cultures were obtained from each neonate before starting antibiotics. Two samples (at least 1 mL for each) were drawn from different peripheral veins 30 minutes apart. Prior to venipuncture, the skin was disinfected by cleansing with 70% isopropyl alcohol swabs for 10 seconds, repeated 3 consecutive times, followed by a minimum of 30 seconds of drying. Samples were inoculated into BACT/ALERT^® Pediatric FAN^® PLUS bottles (BioMérieux Diagnostics Inc., Durham, NC, USA) and incubated at 37°C for 7 days. Bottles that remained negative after 5 days were subcultured onto blood, chocolate, and MacConkey agar plates and kept at 37°C for 48 hours. Bacterial isolates were determined using conventional biochemical and serological procedures according to Clinical Laboratory Standard Institute (CLSI) criteria. ⁵⁵ Antibiotic susceptibility testing was performed using the Kirby-Bauer disk diffusion method and interpreted following CLSI guidelines. ⁵⁶

NS definition relied on the presence of at least 2 SIRS criteria, one of which has to be an abnormal body temperature or leukocyte count: (a) core body temperature of >38.5°C or <36°C; (b) tachycardia [mean heart rate > 180 beat/min] or bradycardia [mean heart rate < 100 beat/min] in the absence of external/painful stimulus or causative medications; (c) tachypnea [mean respiratory rate > 50 cycle/min in first week of life or >40 between 1 and 4 weeks of life] or connection to mechanical ventilation in the absence of general anesthesia or neuromuscular disease; and (d) leukocytosis [>34 ×10⁹/L in first week of life or >19.5 × 10⁹/L between 1 and 4 weeks of life ), leukopenia [<5 × 10⁹/L], or >10% immature neutrophils. ⁵⁷ The current study included only neonates with proven infection, as confirmed by positive blood culture. According to the age at onset of symptoms, NS was classified as EOS (⩽72 hours of life) and late-onset sepsis (LOS, >72 hours of life).^1,5

For measuring serum IL-18 and IL-22 levels, 2 mL of blood were obtained from peripheral veins of suspected cases with NS within the first 2 hours after admission to the NICU. Blood was collected in plain tubes, and sera were separated by centrifugation at 3000×g for 20 minutes and kept at −20°C. Only samples from cases with a positive blood culture were analyzed. Just before analysis, stored sera were thawed and mixed well, and hemolyzed or lipemic specimens were excluded. Serum IL-18 and IL-22 levels were assessed using human Enzyme-Linked Immunosorbent Assay kits (SinoGeneClon Biotech CO., Ltd, Hangzhou, China), according to the manufacturer’s instructions. In brief, serum samples, standards, and quality controls were added into 96-well plates precoated with purified monoclonal antibodies specific for human IL-18 or IL-22. This allows binding of the target IL to their corresponding fixed antibodies. After washing, horseradish peroxidase (HRP)-labeled anti-human IL-18 or IL-22 antibodies were added to form an antibody-antigen-enzyme-antibody “sandwich” complex. Following a thorough washing, we put tetramethylbenzidine substrate solution into wells, producing a blue color due to catalysis by the HRP in proportion to the bound IL; a stop solution was then added to terminate the reaction. Optical density was measured at 450 nm, and IL concentrations were determined by comparison with the standard curve. Samples exceeding the highest standard were diluted and re-assayed. The detection ranges of the IL-18 and IL-22 kits were 31.25-2000 and 2-40 pg/mL, respectively.

Newborns with suspected sepsis were managed according to institutional guidelines. Management included general supportive care to maintain normal body temperature, oxygenation, and perfusion, as well as fluid, electrolyte, and acid-base balance. Empirical antibiotic therapy was initiated immediately for newborns with clinically suspected sepsis using ampicillin and amikacin; cefotaxime and vancomycin were administered for suspected meningitis and staphylococcal infection, respectively. The choice and duration of antibiotic treatment were subsequently guided by the clinical condition and the results of blood culture and sensitivity testing. ⁵⁸

Statistical Analysis and Sample Size Calculation

Sample size was calculated using STATA/BE 17 (StataCorp, College Station, TX, USA) to detect a mean difference of 200 pg/mL in serum IL-18 between the 2 independent groups (cases with NS and healthy controls) at 80% power and .05 alpha level. This calculation resulted in 34 subjects per group, which was increased to 55 participants in the NS group to allow for meaningful subgroup comparison between survivors and non-survivors.

Data were analyzed using STATA/BE 17. Quantitative data were presented as mean ± SD for normally distributed variables and median (IQR) for non-normally distributed variables. Categorical data were expressed as frequency (%). Comparisons between the 2 independent groups were performed using the Student t-test for normally distributed quantitative data, the Mann-Whitney test for non-normally distributed data, and the Chi-square/Fisher exact test for categorical data. The abilities of IL-18, IL-22, and other parameters for predicting NS and mortality were investigated using receiver operating characteristic (ROC) curve analysis by comparing the area under the ROC curves (AUCs). In general, higher AUC measures indicate better predictive performance, with AUC ⩾ 0.8 considered good discrimination and AUC < 0.5 implying no predictive ability. Optimal cutoff values for IL-18, IL-22, and CRP were determined from the ROC curve analysis, followed by calculation of their sensitivity, specificity, positive predictive value, negative predictive value, and accuracy. The DeLong method was used to compare the predictive abilities of these parameters. A P-value less than .05 was deemed statistically significant.

Results

The study included 55 neonates with culture-confirmed sepsis (median age 7 days; 56% males) and 34 age- and sex-matched healthy controls. Among septic newborns, most had LOS (n = 39, 71%) and gram-positive bacterial growth in blood cultures (n = 37, 67%); 7 cases (12.7%) died during hospitalization. Table 1 summarizes the clinical and laboratory features of the sepsis and control groups. Compared with controls, neonates with sepsis had significantly higher median serum levels of IL-18 (910 vs 256 pg/mL), IL-22 (96 vs 20 pg/mL), CRP (18 vs 3 mg/dL), neutrophil count (7.7 vs 3.8 ×10⁹/L), and total leukocyte count (11.2 vs 8.5 ×10⁹/L), as well as lower platelet count (136 vs 194 ×10⁹/L) and lymphocyte count (3.6 vs 4.8 ×10⁹/L).

Table 1.

Comparison of Characteristics Between Neonates With Sepsis and Healthy Controls.

Characteristics	Sepsis group (n = 55)	Control group (n = 34)	P-value
Age (days), median (IQR)	7 (4-15)	7 (4-11)	.735 a
Gestational age (weeks), mean ± SD	38.6 ± 0.81	38.6 ± 0.76	.814 b
Cesarean section vs vaginal delivery, frequency	42/13	21/13	0.141 c
Male sex, frequency (%)	31 (56.4%)	19 (55.9%)	0.965 c
Weight (kg), mean ± SD	3.2 ± 0.29	3.3 ± 0.20	.125 b
Early vs late onset sepsis, frequency	16/39	NA	NA
Gram −ve vs +ve blood culture, frequency	18/37	NA	NA
nSOFA score, median (IQR)	4 (2-6)	NA	NA
Death, frequency (%)	7 (12.7%)	NA	NA
Hemoglobin level (g/dL), mean ± SD	13.5 ± 2.06	13.7 ± 2.24	.641 b
Total leukocytic count (×109/L), median (IQR)	11.2 (10.2-12.7)	8.5 (7.4-9.5)	<.001 a
Neutrophil count (×109/L), median (IQR)	7.7 (6.3-8.8)	3.8 (3.1-4.1)	<.001 a
Lymphocyte count (×109/L), median (IQR)	3.6 (3.02-4.35)	4.8 (4.3-5.3)	<.001 a
Platelet count (×109/L), median (IQR)	136 (64-177)	194 (175-246)	<.001 a
C-reactive protein (mg/dL), median (IQR)	18 (6-48)	3 (0-6)	<.001 a
Interleukin 18 (pg/mL), median (IQR)	910 (738-1550)	256 (196-294)	<.001 a
Interleukin 22 (pg/mL), median (IQR)	92 (86.4-98)	20.3 (16-26)	<.001 a

Abbreviation: nSOFA, neonatal sequential organ failure assessment.

^aMann-Whitney test, ^bStudent t-test, ^cPearson Chi-Square test.

The comparison between survivors and non-survivors of septic newborns is shown in Table 2. Compared with survivors, non-survivors had significantly higher median serum levels of IL-18 (1820 vs 873 pg/mL), IL-22 (139 vs 91 pg/mL), CRP (48 vs 12 mg/dL), neutrophil count (9.7 vs 7.6 ×10⁹/L), and total leukocyte count (13.1 vs 11.2 ×10⁹/L) as well as lower platelet count (45 vs 146 ×10⁹/L) and younger age (2 vs 7 days).

Table 2.

Comparison Between Survivors and Non-Survivors Among Neonates With Sepsis.

Characteristics	Non-survivors (n = 7)	Survivors (n = 48)	P-value
Age (days), median (IQR)	2 (1-4)	7 (5-15)	.003 a
Gestational age (weeks), mean ± SD	39.0 ± 0.82	38.5 ± 0.80	.163 b
Cesarean section vs vaginal delivery, frequency	5/2	37/11	0.664 c
Male sex, frequency (%)	4 (57.1%)	27 (56.3%)	1.000 c
Weight (kg), mean ± SD	3.0 ± 0.18	3.2 ± 0.29	.035 b
Early onset sepsis, frequency (%)	5 (71.4%)	11 (22.9%)	0.018 c
Gram −ve blood culture, frequency (%)	6 (85.7%)	12 (25.0%)	0.004 c
nSOFA score, median (IQR)	9 (8-10)	3 (2-5)	<.001 a
Hemoglobin level (g/dL), mean ± SD	13.9 ± 1.57	13.5 ± 2.13	.640 b
Total leukocytic count (×109/L), median (IQR)	13.1 (11.1-14.5)	11.2 (9.6-12.3)	.020 a
Neutrophil count (×109/L), median (IQR)	9.7 (8.9-10.2)	7.6 (6.3-8.2)	<.001 a
Lymphocyte count (×109/L), median (IQR)	3.4 (2.2-4.4)	3.8 (3.1-4.3)	.495 a
Platelet count (×109/L), median (IQR)	45 (35-130)	146 (96-179)	.030 a
C-reactive protein (mg/dL), median (IQR)	48 (24-48)	12 (6-24)	.009 a
Interleukin 18 (pg/mL), median (IQR)	1820 (1705-1855)	873 (718-1273)	<.001 a
Interleukin 22 (pg/mL), median (IQR)	139 (135-157)	91 (86-96)	<.001 a

Abbreviation: nSOFA, neonatal sequential organ failure assessment.

^aMann-Whitney test, ^bStudent t-test, ^cFisher’s Exact test.

As shown in Table 3, serum IL-18 and IL-22 levels showed a significantly positive correlation with each other (r .62, P < .001). Additionally, IL-18 levels showed significant correlations with nSOFA (r .833, P < .001), CRP (r .481, P < .001), and neutrophil count (r .36, P .008). Furthermore, IL-22 levels significantly correlated with nSOFA (r .689, P .001), neutrophil count (r .304, P .024), and platelet count (r −.280, P .038).

Table 3.

Correlation of Serum Interleukin-18 and -22 Levels With Other Features in Neonates With Sepsis (n = 55).

Characteristics	Interleukin-18 r (P) a	Interleukin-22 r (P)
Age (days)	−0.09 (.503)	−0.18 (.199)
Weight (kg)	−0.06 (.657)	−0.185 (.177)
nSOFA score	0.833 (<.001)	0.689 (<.001)
Hemoglobin level (g/dL)	−0.01 (.935)	0.01 (.945)
Total leukocytic count (×109/L)	0.19 (.175)	0.18 (.185)
Neutrophil count (×109/L)	0.36 (.008)	0.304 (.024)
Lymphocyte count (×109/L)	−0.196 (.152)	−0.120 (.383)
Platelet count (×109/L)	0.018 (.895)	−0.280 (.038)
C-reactive protein (mg/dL)	0.481 (<.001)	0.265 (.051)
Interleukin 22 (pg/mL)	0.62 (<.001)	–

Abbreviation: nSOFA, neonatal sequential organ failure assessment.

^aPearson correlation test.

ROC curve analyses of serum IL-18 and IL-22 levels for predicting NS revealed AUC values of 0.990 and 0.998, respectively (Figure 1). At an optimal cutoff value of 610 pg/mL, IL-18 demonstrated 94.6% sensitivity and 97.1% specificity for predicting NS. Similarly, IL-22 showed 96.4% sensitivity and 100% specificity at a cutoff value of 67 pg/mL. Both IL-18 and IL-22 outperformed CRP in predicting NS (Table 4).

Figure 1.

Receiver operating characteristic (ROC) curve of serum interleukin-18 and -22 in predicting neonatal sepsis. AUC, area under the ROC curve.

Table 4.

Summary of the Abilities of Different Parameters for Predicting Neonatal Sepsis.

Score	Interleukin-18	Interleukin-22	C-reactive protein
AUC (95% CI)	0.9896 (0.9735-1.000)	0.9984 (0.9951-1.000)	0.9310 (0.8850-0.9770)
Optimal cutoff	610 (pg/mL)	67 (pg/mL)	3 (mg/dL)
Sensitivity (%)	94.6	96.4	96.4
Specificity (%)	97.1	100	70.6
PPV (%)	98.1	100	84.1
NPV (%)	91.7	94.4	92.3
Accuracy (%)	95.5	97.8	86.5

Abbreviations: AUC, area under the receiver operating characteristics; NPV, negative predictive value; PPV, positive predictive value.

Regarding the prediction of neonatal mortality, ROC curve analyses of IL-18 and IL-22 showed AUC values of 0.942 and 0.994, respectively (Figure 2). At 1540 pg/mL cutoff, IL-18 achieved 100% sensitivity and 83.3% specificity. In addition, IL-22 showed 100% sensitivity and 95.8% specificity for predicting neonatal mortality at 106.6 pg/mL cutoff. Both IL-18 and IL-22 performed better than CRP in predicting neonatal mortality but showed comparable predictive abilities with nSOFA (Table 5).

Figure 2.

Receiver operating characteristic (ROC) curve of serum interleukin-18 and -22 in predicting neonatal mortality. AUC, area under the ROC curve.

Table 5.

Summary of the Abilities of Different Parameters for Predicting Neonatal Mortality.

Score	Interleukin-18	Interleukin-22	C-reactive protein	nSOFA
AUC (95% CI)	0.9420 (0.8785-1.000)	0.9940 (0.9798-1.000)	0.8036 (0.6770-0.9302)	0.9866 (0.9634-1.000)
Optimal cutoff	1540 (pg/mL)	106.6 (pg/mL)	24 (mg/dL)	6
Sensitivity (%)	100	100	71.4	100
Specificity (%)	83.3	95.8	79.2	95.8
PPV (%)	46.7	77.8	33.3	77.8
NPV (%)	100	100	95.0	100
Accuracy (%)	85.5	96.4	78.2	96.4

Abbreviations: AUC, area under the receiver operating characteristics; NPV, negative predictive value; nSOFA, neonatal sequential organ failure assessment; PPV, positive predictive value.

Discussion

Newborns are more susceptible to infections due to their underdeveloped immune system. ⁵ Underdiagnosis of NS may result in life-threatening overwhelming infection and long-term neurodevelopmental impairment, whereas overdiagnosis may lead to inappropriate use of antibiotics, which might disturb the neonatal microbiota and the development of innate and adaptive immunity.^7,59 Timely diagnosis and treatment of NS are critical for improving clinical outcomes. However, the identification of NS is challenging due to non-specific clinical manifestations and the lack of timely and accurate diagnostic tools.^8,10,11,13 Microbial culture remains the gold standard for diagnosing NS, as it allows definitive pathogen recognition and susceptibility testing to guide appropriate antibiotic therapy. Nevertheless, its clinical utility is limited by long turnaround times and low pathogen detection percentage, particularly in cases of prior antibiotic exposure, low-colony-count bacteremia, or inadequate blood sample volumes.^5,7,9,12 Accordingly, the decision to initiate antibiotics is usually based on clinical assessment and risk factor evaluation, supported by laboratory tests such as CRP and hematological indices.^5,10 However, these commonly used biomarkers have limited diagnostic accuracy, particularly for single-timepoint measures, and are influenced by several confounders, including mode of delivery, gestational and postnatal age, intercurrent conditions (eg, perinatal asphyxia, intraventricular hemorrhage), and the timing and method of blood sampling.^10-12,60 A recent meta-analysis reported poor diagnostic performance of CRP in LOS, with 62% sensitivity and 74% specificity. ⁶¹ Therefore, research has focused on identifying novel biomarkers that can precisely discriminate between infected and non-infected neonates at an early stage, thereby enhancing timely management and improving clinical outcomes.^5,7

In the present study, serum IL-18 levels were significantly increased in newborns with sepsis, particularly among non-survivors, correlated with nSOFA, and outperformed CRP in predicting both NS and mortality. Our findings are consistent with Li et al, ⁴⁹ who reported significantly higher serum IL-18 levels in neonates with sepsis compared with healthy controls, with the highest levels observed among non-survivors. In our study, ROC curve analyses of IL-18 revealed AUC values of 0.990 and 0.942 for predicting NS and mortality, respectively. These values are higher than those reported by Li et al., ⁴⁹ who found AUCs of 0.77 and 0.80 for NS and mortality prediction, respectively. The superior diagnostic performance of IL-18 in our study may be attributed to the inclusion of a homogeneous group of term newborns with culture-positive sepsis. In contrast, Li et al. ⁴⁹ enrolled neonates based on SIRS criteria without microbiological confirmation, which may have included false-positive cases. Furthermore, in Li et al. ⁴⁹ study, the percentage of preterm newborns was higher in the sepsis group than in the control group (13.2% vs 3.2%). Preterm newborns typically have higher levels of inflammatory cytokines as part of response to several non-infectious conditions, such as respiratory distress syndrome, chronic lung disease, and intracranial hemorrhage, which might have confounded the diagnostic performance of IL-18 for NS prediction in preterm babies.^62,63

Other studies have also demonstrated elevated IL-18 levels in newborns with sepsis, measured in peripheral blood, ⁴⁸ urine,^47,50 and cord blood. ⁶⁴ Notably, Higazi et al ⁴⁷ showed that urinary IL-18 levels have better accuracy than CRP in predicting NS. Barekatain et al ⁶⁵ also found higher salivary IL-18 levels among newborns with culture-positive EOS compared with healthy controls, although the differences did not reach statistical significance, likely due to the small sample size. In contrast, Kumar et al ⁵² observed no significant difference in serum IL-18 levels between very low birth weight neonates with LOS and controls, and Bender et al ⁵¹ concluded that IL-18 levels can not predict EOS. Beyond the neonatal populations, several adult studies have also documented an association between elevated serum IL-18 levels and both sepsis and septic thrombocytopenia.^27,30,43,46

On the other hand, multiple studies have demonstrated an association of higher IL-18 levels with sepsis severity and poor prognosis in both neonatal ⁴⁷ and adult^28,29,31,36 populations. Okuhara et al, ³⁷ using a mouse model, showed that IL-18 is associated with sepsis-induced cardiac dysfunction, which was attenuated by IL-18 gene deletion, resulting in improved survival. In another murine model of NS, Wynn et al ⁶⁶ demonstrated that IL-18 administration enhanced the systemic inflammatory response and increased mortality, whereas IL-18 deletion markedly improved survival. In the same vein, other studies reported that IL-18 deletion or IL-18 neutralization guards against sepsis-induced organ injuries.^34,35,44 Conversely, Mastroeni et al ⁶⁷ demonstrated that pretreatment with recombinant IL-18 in mice infected with Salmonella typhimurium reduced bacterial burden in the liver and spleen, while pretreatment with anti-IL-18 antibodies increased hepatic and splenic bacterial count and decreased survival.

IL-18, a member of the IL-1 family, is a proinflammatory cytokine produced by various hematopoietic (eg, macrophages) and non-hematopoietic cell types upon activation, serving as part of the immune defense mechanism against infections.^68,69 It plays a critical role in the stimulation of Th1 or Th2 cells, depending on the immunological context. IL-18 induces the production of IFN-γ and other chemokines (eg, MIP-1α, MIP-1β, and MCP-1), which recruit macrophages and monocytes to sites of infection and promote the inflammatory response.^20,69 Furthermore, IL-18 can promote the generation of IL-2 and granulocyte-macrocyte colony-stimulating factor by activated T cells, boost natural killer cell cytotoxicity, increase phagocytosis by polymorphonuclear cells, and enhance expression of Fas ligand in Th1 cells.^68,69 While moderate elevations of circulatory IL-18 seem protective and enhance antimicrobial immunity, excessive IL-18 secretion can exacerbate inflammation, resulting in myocardial, and renal injury and poor clinical outcome.^16,35 Joshi et al ⁷⁰ elegantly described the dual role of IL-18 in a lipopolysaccharide-induced sepsis animal model. Low-dose lipopolysaccharide induced a moderate, transient increase in serum IL-18 and IFN-γ levels, promoting antibacterial defense with protective effects. Conversely, high-dose lipopolysaccharide led to high serum IL-18 and IFN-γ levels, impairing antibacterial defense and increasing mortality. In the latter scenario, anti-IL-18 therapy was protective and restored antibacterial activity. ⁷⁰ These findings indicate that serum IL-18 may serve not only as a useful biomarker for NS diagnosis but also as a prognostic marker for sepsis severity and related mortality. ¹⁶

A novel aspect of the current study is the evaluation of serum IL-22 levels in NS. Similar to IL-18, IL-22 was significantly elevated in septic newborns, particularly among non-survivors, correlated with nSOFA, and showed better diagnostic accuracy than CRP for identifying NS and predicting mortality. Our findings disagree with the only previous study investigating IL-22 levels in NS, which reported no significant association between cord blood IL-22 levels and EOS in 474 neonates; however, that study included only 7 cases of culture-confirmed sepsis. ⁵³ In adults, elevated IL-22 levels have been observed in patients with surgical abdominal sepsis ²⁵ and sepsis-induced ARDS. ⁴¹

Further evidence from animal studies supports the crucial role of IL-22 in sepsis. Several studies have demonstrated the protective effects of IL-22 in different animal models. For instance, Trevejo-Nunez et al ³⁹ showed that IL-22 expression is upregulated in the lungs of mice with pneumococcal pneumonia, and hepatic IL-22R1 deficiency led to higher lung bacterial loads. In a model of Klebsiella pneumoniae infection, Aujla et al ²² found increased pulmonary IL-22 levels, which enhanced antibacterial activity and reduced bacterial burden and mortality. In the same vein, Broquet et al ²⁶ demonstrated that IL-22 neutralization increased predisposition to infection, lung neutrophil accumulation, and lung damage in a murine model of Pseudomonas aeruginosa pneumonia. Additionally, Yu et al ⁴² found that administration of IL-22 to mice with endotoxemia improved survival and reduced pro-inflammatory cytokines (IL-6, TNF-α, IL-1β, MCP-1). Other mouse studies have also shown that IL-22 administration mitigates sepsis-induced injury in the lung ⁷¹ and liver.^33,38 Treerat et al ⁷² showed that IL-22 is upregulated and mitigates the progression of Mycobacterium tuberculosis infection. Similarly, Tripathi et al ⁴⁰ demonstrated that recombinant IL-22 can control tuberculosis by reducing alveolar neutrophil infiltration and alleviating pulmonary epithelial cell damage. Furthermore, researchers demonstrated that IL-22 is upregulated and exerts protective effects against murine intestinal infection with Citrobacter rodentium.^23,24,45 Finally, Hasegawa et al ³² showed that IL-22 deficiency led to increased mortality of C. difficile infection in mice. In contrast, some studies have reported harmful effects of IL-22. In a murine model of polymicrobial peritonitis, IL-22 levels were highly elevated in the liver and spleen, and IL-22 blockade decreased bacterial burden and organ dysfunction. ¹⁹ Another study identified elevated serum IL-22 level as a risk factor for poor prognosis in sepsis. ⁴¹

IL-22 belongs to the IL-10 cytokine family and is mainly produced by Th22 cells as well as other immune cells, such as Th1, Th17, CD8⁺ T cells, and natural Killer T cells.^73-75 IL-22 exerts immunoregulatory effects through interaction with the IL-22 receptor complex, activating downstream signaling pathways involving STAT1/3/5, MAPK, and NF-κB.^71,74 IL-22 acts as a double-edged sword, exhibiting both pro- and anti-inflammatory effects and producing protective or harmful consequences depending on the infection or disease model.^42,75 It mediates innate immunity by enhancing the secretion of antibacterial proteins from epithelial cells, promoting epithelial cell fluidity, stimulating mucin production, inducing the release of other cytokines, protecting the epithelial barrier, and promoting cell proliferation and wound healing. ⁷⁵ Nevertheless, under pathological conditions, excess IL-22 can shift from a protective to a harmful factor, disrupting immune defense through proinflammatory effects.^25,76

Notably, IL-18 and IL-22 likely operate in opposite directions during the immunoinflammatory process: IL-18 primarily signals from mononuclear phagocytic and epithelial cells to lymphocytes, whereas IL-22 predominantly communicates from lymphocytes to epithelial cells. ²⁰ A mounting body of evidence indicates complex crosstalk between IL-18 and IL-22 during infection and inflammation.^20,21 IL-18 could promote the biological activity of IL-22 by enhancing IL-22 expression through group 3 innate lymphoid cells ⁷⁷ and downregulating IL-22 binding protein expression. ⁷⁸ Vice versa, IL-22 can upregulate IL-18. Muñoz et al ²¹ demonstrated that IL-22 augments the production of IL-18 during intestinal infection, and Kim et al ⁷⁹ found that IL-22 increases the expression of pro-IL-18 in human keratinocytes. Furthermore, Chaing et al ⁸⁰ showed that IL-22 regulates IL-18 expression to maintain the intestinal epithelial barrier, and administration of IL-18 to IL-22-deficient mice with adherent-invasive E. coli restored IFN-γ production by T cells and rescued lysozyme⁺ Paneth cell function.

The strengths of this study include its case-control design and strict eligibility criteria, with inclusion limited to neonates with blood culture-positive NS. However, we also acknowledge some limitations. First, the relatively small sample size in the non-survivor group may limit the reliability and generalizability of the mortality-related findings, highlighting the need for larger studies. Moreover, the study included only term newborns who were appropriate for gestational age. Future research should include preterm neonates and those with intrauterine growth restriction, who often present with non-infectious conditions (eg, respiratory distress syndrome, chronic uterine hypoxia) that may increase inflammatory cytokines and potentially affect the diagnostic accuracy of IL-18 and IL-22 for NS diagnosis.^62,63 Another limitation is the assessment of serum IL-18 and IL-22 levels at a single time point (upon admission). Serial measurements would be valuable to evaluate their dynamic changes and could improve diagnostic accuracy. It would also be useful to investigate changes in IL-18 and IL-22 levels after antibiotic initiation and their correlation with clinical recovery. Furthermore, although multivariate analysis could theoretically assess independence of these biomarkers, it was not feasible in this study due to the matched case-control design, strong collinearity between IL-18 and IL-22 with each other and with other predictors (eg, nSOFA, CRP, neutrophil count, platelets), and the relatively small sample size and number of events, which would lead to model instability and overfitting. ⁸¹ Despite these limitations, the exceptional discriminatory power of IL-18 and IL-22 (AUCs > 0.98 for sepsis diagnosis, and >0.94 for mortality prediction) underscores their potential as biomarkers, significantly outperforming CRP. Finally, we did not perform feasibility and cost-benefit analyses for cytokine assays, which are more technically demanding and expensive than widely available acute phase reactants such as CRP.

Conclusion

Serum levels of IL-18 and IL-22 are markedly elevated in newborns with sepsis and correlate with increased risk of mortality. These findings indicate that serum IL-18 and IL-22 are promising biomarkers for both NS diagnosis and prognosis. Future studies should further investigate their diagnostic utility using serial measurements, larger and more inclusive populations, and implementation and cost-benefit analyses.