Inflammatory cytokines as biomarkers for disease severity in pediatric viral pneumonia: a systematic review and meta-analysis

Introduction

Viral pneumonia is one of the most common respiratory infections in children, with a high incidence and hospitalization rate, especially in children under the age of 5 years (1). Although the prognosis of most children is good, some may progress to severe illness, which is associated with respiratory failure, pulmonary complications, and even life-threatening outcomes (2). The identification of the inflammatory cytokines closely related to the progression to severe disease is critical for early risk stratification and accurate intervention (3). However, there is still a lack of systematic integration analyses and risk prediction models for inflammatory factor levels in current clinical practice.

The immune response mechanism of viral pneumonia in children is complex and involves the synergistic regulation of multiple inflammatory factors [e.g., interleukin-6 (IL-6), interleukin-8 (IL-8), tumor necrosis factor alpha (TNF-α), interferon gamma (IFN-γ), C-reactive protein (CRP), and procalcitonin (PCT)] (4,5). A study suggests that these inflammatory factors may serve as biomarkers for disease progression in addition to reflecting the intensity of the inflammatory response (6). However, single studies are often affected by sample size limitations, incomplete detection indicators, and study design heterogeneity, resulting in inconsistent results. For example, while some studies have identified elevated IL-6 as a significant correlate of severe illness, others have found no such significant association (7,8). Furthermore, different virus types [e.g., respiratory syncytial virus (RSV), adenovirus, and influenza virus] elicit distinct immune responses, further contributing to the existing discrepancies in the literature (9). A systematic synthesis is therefore required to resolve these explicit discrepancies and clarify the prognostic utility of these cytokines.

Given these inconsistencies, a systematic synthesis of existing evidence is necessary to clarify the association between inflammatory cytokines and disease severity in children with viral pneumonia. In particular, it is important to evaluate cytokine levels across two clinically relevant comparisons: healthy controls versus affected children, and mild disease versus moderate-to-severe disease. In addition, beyond assessing group-level differences, evaluating the diagnostic performance of key cytokines for identifying children with more severe disease may provide clinically actionable information. Therefore, the present study conducted a systematic review and meta-analysis to comprehensively assess the association between inflammatory cytokines and pediatric viral pneumonia, and to evaluate the diagnostic accuracy of IL-6 in distinguishing moderate-to-severe disease. We present this article in accordance with the PRISMA and CHARMS reporting checklists (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0081/rc).

Methods

Study inclusion and exclusion criteria

The inclusion criteria were as follows:

• Studies involving children under 18 years of age diagnosed with viral pneumonia;
• Studies reporting levels of one or more inflammatory cytokines, including but not limited to IL-6, IL-8, tumor necrosis factor-α (TNF-α), CRP, and PCT;
• Studies comparing inflammatory cytokine levels either between healthy controls and children with viral pneumonia, or between children with mild disease and those with moderate-to-severe disease;
• Studies providing extractable quantitative data, such as mean and standard deviation (SD), median and interquartile range (IQR), or effect estimates that could be converted for meta-analysis;
• Observational study designs, including prospective or retrospective cohort studies and case-control studies.

The exclusion criteria were as follows:

• Studies including adults or mixed populations in which pediatric data could not be independently extracted;
• Studies not reporting inflammatory cytokine data relevant to disease presence or severity;
• Reviews, conference abstracts, case reports, animal experiments, or in vitro studies;
• Studies with substantial data loss or insufficient methodological quality, defined as a Newcastle-Ottawa Scale (NOS) score <5;
• Duplicate publications or studies with overlapping populations.

All articles were independently screened and cross-checked by two reviewers. Discrepancies were resolved through discussion, and when consensus could not be reached, a third reviewer was consulted to ensure the rigor and consistency of the study selection process.

Literature retrieval strategies

The following databases were searched to retrieve relevant articles from database inception to January 25, 2026: PubMed, Web of Science, Embase, the Cochrane Library, China National Knowledge Infrastructure (CNKI), Wanfang, and China Science and Technology Journal Database (VIP). The search strategy combined Medical Subject Headings and free-text terms related to “viral pneumonia”, “children”, “pediatrics”, “inflammatory cytokines”, “biomarkers”, “severity”, and “progression”, using Boolean operators (AND/OR) to maximize sensitivity. In addition, the reference lists of all eligible studies and relevant reviews were manually screened to identify additional potentially eligible articles. Literature retrieval was independently performed by two reviewers. EndNote 20 software (Clarivate, London, UK) was used for reference management and duplicate removal. Titles and abstracts were screened for relevance, followed by full-text review of potentially eligible studies to determine final inclusion. This strategy ensured a systematic, comprehensive, and reproducible literature search.

Literature screening

The literature screening comprised four steps: initial examination, de-duplication, title and abstract screening, and full-text evaluation. First, two researchers independently searched PubMed, Web of Science, Embase, Cochrane Library, CNKI, Wanfang, and VIP, and imported the results into EndNote 20 for standardized management. Subsequently, the duplicate articles were removed, and the remaining articles underwent title and abstract screening. The initial screening of the titles and abstracts was independently performed by the two researchers, and articles that were not related to viral pneumonia in children, contained no data on inflammatory factors, or had unclear subjects were excluded, while articles that appeared to meet the criteria were retained. After the preliminary screening, the full texts of the remaining articles were read to further evaluate the methodological quality and data extractability according to the established classification criteria. For articles with incomplete data or uncertainty, attempts were made to contact the original author(s) for additional information, which were excluded in the absence of a response. If any disagreements arose among the researchers during the screening, a third researcher participated in the discussion and made the final decision to ensure the objectivity and consistency of the screening process. All the screening procedures were documented for review and quality assessment.

Data extraction process

A standardized data extraction form was designed and pre-piloted before formal extraction. Two reviewers independently extracted data from all eligible studies. Disagreements were resolved through discussion, and when consensus could not be achieved, a third reviewer was consulted for adjudication. For studies with missing, unclear, or non-extractable information, the corresponding author(s) were contacted by email to request additional data. When multiple measurement time points were available, data at admission or early diagnosis were preferentially extracted to reflect early inflammatory status, consistent with the objective of early risk prediction for severe viral pneumonia.

Data items

The following data were extracted from each included study:

• Study characteristics: first author, year of publication, country or region, and study design (prospective or retrospective);
• Participant characteristics (PICOS-P): sample size, age, and sex distribution;
• Inflammatory cytokines (exposure variables): biomarkers related to the inflammatory response, including IL-6, IL-8, interleukin-10 (IL-10), interleukin-1β (IL-1β), TNF-α, interferon-γ (IFN-γ), CRP, PCT, serum amyloid A (SAA), transforming growth factor-β (TGF-β), and monocyte chemotactic protein-1 (MCP-1). Data were extracted as mean ± SD, median with IQR, or other extractable quantitative measures that could be converted for meta-analysis;
• Outcomes (PICOS-O): disease presence and disease severity, defined as comparisons between healthy controls and children with viral pneumonia, as well as between mild and moderate-to-severe disease. Severe disease was defined according to the original studies and typically included intensive care unit (ICU) admission, need for invasive mechanical ventilation, respiratory failure, sepsis, death, or clinically defined severe pneumonia;
• Additional information: whether multivariable adjustment was performed, the range of adjusted covariates (if applicable), and methodological quality assessment using the NOS.

Research indicators

The research indicators were defined as follows:

• Basic demographic and clinical characteristics: age, sex, sample size, country or region, and study design (prospective or retrospective).
• Inflammatory cytokines (exposure variables): biomarkers related to the inflammatory response in viral pneumonia, including IL-6, IL-8, IL-10, IL-1β, TNF-α, IFN-γ, CRP, PCT, and other inflammation-related markers such as SAA, TGF-β, and MCP-1. When inflammatory cytokine levels were reported at multiple time points, data obtained at hospital admission or at the early stage of diagnosis were preferentially extracted to reflect the early inflammatory status of the disease. Cytokine data were extracted as mean ± SD, median with IQR, or other quantitative measures that could be converted for meta-analysis.
• Outcome indicators (disease severity): disease severity was defined according to the criteria used in the original studies and generally included ICU admission, requirement for invasive mechanical ventilation, respiratory failure, sepsis, death, or clinically defined severe pneumonia. Children meeting these criteria were classified as the moderate-to-severe group, while the remaining patients were classified as the mild group.
• In addition, information regarding multivariable adjustment, the range of adjusted covariates (if available), and methodological quality assessed using the NOS was recorded for sensitivity and subgroup analyses.

Statistical analysis

All statistical analyses were performed using Stata version 17.0 (StataCorp, College Station, TX, USA). Continuous variables, including inflammatory cytokine levels, were pooled using weighted mean differences (WMDs) with corresponding 95% confidence intervals (CIs). For studies reporting medians and IQRs, data were converted to means and SDs using established methods when appropriate. Meta-analyses were conducted separately for comparisons between healthy controls and children with viral pneumonia, as well as between children with mild disease and those with moderate-to-severe disease. Statistical heterogeneity among studies was assessed using the Cochran Q test and quantified with the I² statistic. An I² value >50% or a Q test P value <0.10 was considered indicative of substantial heterogeneity, in which case a random-effects model (DerSimonian-Laird method) was applied; otherwise, a fixed-effects model was used. Results are presented graphically using forest plots. Subgroup analyses were performed to explore potential sources of heterogeneity according to pathogen type, age group, and sample size. Sensitivity analyses were conducted using a leave-one-out approach, whereby each study was sequentially removed to evaluate the robustness of the pooled estimates. Publication bias was assessed visually using funnel plots and statistically using Begg’s rank correlation test and Egger’s regression test. All statistical tests were two-sided, and a P value of <0.05 was considered statistically significant. In addition, a diagnostic test accuracy meta-analysis was conducted to evaluate the ability of IL-6 to identify moderate-to-severe disease. Pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio, and summary receiver operating characteristic (SROC) curves with corresponding area under the curve (AUC) were calculated using a bivariate random-effects model.

Results

Literature screening results

A total of 2,087 records were initially identified from PubMed, Web of Science, Embase, the Cochrane Library, CNKI, Wanfang, and VIP databases. After removal of 892 duplicate records using EndNote 20, 1,195 records remained for title and abstract screening. Of these, 469 records were excluded for irrelevance, leaving 726 reports sought for full-text retrieval. Among the reports sought for retrieval, 38 full-text articles could not be obtained. The remaining 688 articles were assessed for eligibility through full-text review. Of these, 669 articles were excluded for the following reasons: absence of severity stratification data, overlapping study populations, non-extractable data, inappropriate study design, or non-target populations. Ultimately, 19 studies met the inclusion criteria and were included in the meta-analysis. In addition, reference lists of relevant articles were manually screened, yielding 31 additional records, of which 2 studies met the eligibility criteria and were included in the final analysis. All screening procedures were independently performed by two reviewers, with disagreements resolved by discussion or consultation with a third reviewer. The study selection process is summarized in the PRISMA flow diagram (Figure 1).

Figure 1 Screening flow chart. CNKI, China National Knowledge Infrastructure; VIP, China Science and Technology Journal Database.

Risk assessment and quality evaluation of included studies

The methodological quality of the included studies was assessed using the NOS. All 19 included studies were retrospective observational studies, with NOS scores ranging from 6 to 7, indicating an overall moderate risk of bias (Table 1). Most studies showed acceptable quality in study selection and outcome assessment. However, limited inter-group comparability was a common limitation, as most studies adjusted only for basic demographic variables such as age and sex, with insufficient multivariable adjustment for potential confounders. Consequently, residual confounding represents the primary source of bias. Inflammatory biomarkers were measured using routine clinical laboratory methods, resulting in relatively low measurement bias. Nevertheless, the predominantly single-center retrospective design, small sample sizes in some studies, and lack of blinding procedures contributed to an overall moderate risk profile. In addition, incomplete reporting of participant recruitment and non-response rates may have introduced selection bias. These limitations should be considered when interpreting the pooled results.

Analysis of research indicators

Baseline data of patients

A total of 19 studies involving 2,759 children with viral pneumonia were included in the meta-analysis. The included studies compared inflammatory biomarkers between healthy controls and children with viral pneumonia, as well as between children with mild disease and those with moderate-to-severe disease. Across studies, children in the moderate-to-severe group tended to be younger than those in the mild group. RSV was the most frequently reported pathogen, followed by adenovirus and influenza virus. The proportion of moderate-to-severe cases varied among studies, reflecting differences in viral etiology and disease severity definitions. Detailed baseline characteristics are presented in Table 2.

Inflammatory factor indicators

IL-6 level

Forest plot results indicate that six studies were included in the comparison between the case group and the control group (Figure 2). The random-effects model meta-analysis revealed a statistically significant difference in IL-6 levels between the two groups (WMD =−16.75, 95% CI: −22.60 to −10.90). The negative value indicates significantly higher IL-6 levels in the case group than in the control group, with High heterogeneity among studies (I²=99.5%, P<0.001). In the comparison between the mild and moderate-to-severe groups, eight studies were included (Figure 3). The pooled results similarly showed a statistically significant difference in IL-6 levels between the two groups (WMD =−13.07, 95% CI: −16.88 to −9.26), with the negative effect size indicating significantly higher IL-6 levels in the moderate-to-severe group compared to the mild group. High heterogeneity existed among studies (I²=98.7%, P<0.001), necessitating the use of a random-effects model for all analyses.

Figure 2 Forest plot of IL-6 levels between case and control groups. CI, confidence interval; IL-6, interleukin-6; WMD, weighted mean difference.

Figure 3 Forest plot comparing the IL-6 levels between the mild group and the moderate-to-severe group. CI, confidence interval; IL-6, interleukin-6; WMD, weighted mean difference.

CRP level

Forest plot results indicate that five studies were included in the comparison of CRP levels between the case group and the control group (Figure 4). The random-effects model meta-analysis revealed a statistically significant difference in CRP levels between the two groups (WMD =−5.51, 95% CI: −8.90 to −2.11). The negative value indicates significantly higher CRP levels in the case group than in the control group. Heterogeneity among studies was high (I²=99.6%, P<0.001). In the comparison of CRP levels between mild and moderate-to-severe groups, six studies were included (Figure 5). The pooled results similarly showed a statistically significant difference in CRP levels between the two groups (WMD =−10.16, 95% CI: −17.20 to −3.12). The negative effect size indicated significantly higher CRP levels in the moderate-to-severe group compared to the mild group. High heterogeneity existed among studies (I²=99.2%, P<0.001), necessitating the use of a random-effects model for all analyses.

Figure 4 Forest plot of CRP levels between case and control groups. CI, confidence interval; CRP, C-reactive protein; WMD, weighted mean difference.

Figure 5 Forest plot comparing the CRP levels between the mild group and the moderate-to-severe group. CI, confidence interval; CRP, C-reactive protein; WMD, weighted mean difference.

PCT level

Forest plot results indicate that five studies were included in the comparison of PCT levels between the case group and the control group (Figure 6). The random-effects model meta-analysis revealed a statistically significant difference in PCT levels between the two groups (WMD =−0.72, 95% CI: −1.03 to −0.41). The negative value indicates significantly higher PCT levels in the case group compared to the control group. High heterogeneity existed among studies (I²=99.5%, P<0.001). In the comparison of PCT levels between mild and moderate-to-severe groups, five studies were included (Figure 7). The pooled results showed a statistically significant difference in PCT levels between mild and moderate-to-severe groups (WMD =−0.37, 95% CI: −0.59 to −0.15). The negative effect size indicates that PCT levels in the moderate-to-severe group were significantly higher than those in the mild group, with high heterogeneity among studies (I² =98.8%, P<0.001). Therefore, the random-effects model was used for all analyses described above.

Figure 6 Forest plot of PCT levels between case and control groups. CI, confidence interval; PCT, procalcitonin; WMD, weighted mean difference.

Figure 7 Forest plot comparing the PCT levels between the mild group and the moderate-to-severe group. CI, confidence interval; PCT, procalcitonin; WMD, weighted mean difference.

TNF-α level

Forest plot results indicate that two studies were included in the comparison of TNF-α levels between the case group and the control group (Figure 8). The random-effects model meta-analysis revealed a statistically significant difference in TNF-α levels between the two groups (WMD =−17.30, 95% CI: −30.30 to −4.31), with a negative effect size indicating significantly higher TNF-α levels in the case group than in the control group. Heterogeneity among studies was high (I² =99.6%, P<0.001). In the comparison of TNF-α levels between mild and moderate-to-severe groups, three studies were included (Figure 9). The pooled results showed a statistically significant difference in TNF-α levels between the two groups (WMD =−1.08, 95% CI: −1.65 to −0.51). The negative value indicates significantly higher TNF-α levels in the moderate-to-severe group compared to the mild group. Due to high heterogeneity among studies (I² =95.0%, P<0.001), the random-effects model was used for all analyses.

Figure 8 Forest plot of TNF-α levels between case and control groups. CI, confidence interval; TNF-α, tumor necrosis factor alpha; WMD, weighted mean difference.

Figure 9 Forest plot comparing the TNF-α levels between the mild group and the moderate-to-severe group. CI, confidence interval; TNF-α, tumor necrosis factor alpha; WMD, weighted mean difference.

Results of sensitivity analysis and assessment of publication bias

To evaluate the robustness of the meta-analysis results and the potential for publication bias, sensitivity analysis and publication bias assessment were further performed. Sensitivity analysis was conducted using a leave-one-out approach, whereby each included study was sequentially removed and the pooled effect size was recalculated. The results showed that the direction and magnitude of the pooled estimates did not change substantially after exclusion of any single study. The pooled effect sizes consistently remained within the 95% CI of the overall estimate, and no reversal of effect direction was observed, indicating good robustness of the findings. No individual study was found to exert a decisive influence on the overall results, suggesting that the conclusions were not driven by any single study. Although substantial heterogeneity was observed in some analyses, sensitivity analysis indicated that this heterogeneity did not materially affect the stability of the pooled estimates (Figure 10A).

Figure 10 Sensitivity analysis and assessment of publication bias. (A) Leave-one-out sensitivity analysis, in which each included study was sequentially removed to evaluate the robustness of the pooled estimates. (B) Begg’s funnel plot used to assess potential publication bias among the included studies. CI, confidence interval; s.e., standard error; WMD, weighted mean difference.

Publication bias was initially assessed by visual inspection of Begg’s funnel plot, which showed an approximately symmetrical distribution of the included studies, with no obvious asymmetry. Further assessment using Begg’s rank correlation test revealed no statistically significant evidence of publication bias (Z=0.49, P=0.62; continuity-corrected Z=0.37, P=0.71). In addition, Egger’s regression test also indicated no significant publication bias (intercept t=−0.05, P=0.96). These statistical results were consistent with the visual assessment of the funnel plot, suggesting the absence of significant publication bias. Overall, the combined results of sensitivity analysis and publication bias assessment support the stability and reliability of the meta-analysis findings (Figure 10B).

Diagnostic accuracy of IL-6

Figure 11A-11C summarizes the diagnostic performance of IL-6. Figure 11A shows the pooled sensitivity and specificity, with a sensitivity of 0.80 (95% CI: 0.68–0.89) and a specificity of 0.78 (95% CI: 0.64–0.87), indicating good diagnostic discrimination. Figure 11B presents the pooled likelihood ratios, with a positive likelihood ratio of 3.64 (95% CI: 2.24–5.92) and a negative likelihood ratio of 0.25 (95% CI: 0.15–0.42), suggesting moderate rule-in and good rule-out ability. Figure 11C shows a pooled diagnostic odds ratio of 14.35 (95% CI: 6.88–29.90) and a diagnostic score of 2.69 (95% CI: 1.93–3.40). Despite heterogeneity among studies, the pooled results support the potential clinical utility of IL-6 in disease diagnosis.

Figure 11 Diagnostic accuracy of IL-6 for identifying moderate-to-severe disease. (A) Forest plot of pooled sensitivity and specificity of IL-6. (B) Forest plot of pooled positive likelihood ratio and negative likelihood ratio of IL-6. (C) Forest plot of pooled diagnostic odds ratio and diagnostic score of IL-6. CI, confidence interval; df, degree of freedom; DLR, diagnostic likelihood ratio; IL-6, interleukin-6.

The SROC curve and likelihood ratio analysis of the diagnostic efficacy of IL-6

Figure 12A presents the SROC curve for IL-6 diagnostic performance. The curve is located toward the upper left corner, with a pooled sensitivity of 0.80 (95% CI: 0.68–0.89), pooled specificity of 0.78 (95% CI: 0.64–0.87), and an AUC of 0.86 (95% CI: 0.83–0.89), indicating good overall diagnostic accuracy. Most studies were distributed within the confidence region, although the relatively wide prediction region suggests inter-study heterogeneity. Figure 12B shows the pooled likelihood ratios, with a positive likelihood ratio of approximately 3.64 and a negative likelihood ratio of 0.25, suggesting moderate rule-in and good rule-out capability. Overall, SROC and likelihood ratio analyses support the clinical utility of IL-6 in disease diagnosis.

Figure 12 Summary diagnostic performance of IL-6. (A) SROC curve of IL-6 with corresponding 95% confidence interval and prediction interval. (B) Scatter plot of pooled positive and negative likelihood ratios of IL-6. AUC, area under the curve; IL-6, interleukin-6; LLQ, left lower quadrant; LRN, likelihood ratio negative; LRP, likelihood ratio positive; LUQ, left upper quadrant; RUQ, right upper quadrant; SENS, sensitivity; SPEC, specificity; SROC, summary receiver operating characteristic.

Subgroup analysis

Subgroup analysis results are presented in Table 3, with all comparisons based on the mild group versus the moderate-to-severe group. For IL-6 levels, the moderate-to-severe group demonstrated significantly higher IL-6 concentrations than the mild group across all pathogen types, age strata, and sample size strata. Stratified by pathogen type, the pooled effect size for the RSV infection subgroup was −25.06 (95% CI: −37.78 to −12.75), while that for the mixed viral infection subgroup was −21.32 (95% CI: −33.03 to −9.60), with statistically significant differences between subgroups (P<0.001). Stratified by age, IL-6 elevation was more pronounced in the moderate-to-severe group among children ≤5 years old (WMD =−39.85, 95% CI: −50.66 to −29.04), while the combined effect size for children >5 years old was −14.75 (95% CI: −25.83 to −3.67), with statistically significant differences between subgroups (P<0.001). Stratified by sample size, studies with ≤100 participants (WMD =−25.24, 95% CI: −40.39 to −10.09) and those with >100 participants (WMD =−20.83, 95% CI: −41.62 to −0.03) both showed significantly elevated IL-6 levels in the moderate-to-severe group, with the difference between subgroups approaching statistical significance (P=0.05). Regarding CRP, stratified by pathogen type, CRP levels in the moderate-to-severe group were significantly higher than in the mild group in the mixed-virus infection subgroup (WMD =−10.24, 95% CI: −18.55 to −1.93, P=0.02), whereas no statistically significant difference was observed in the RSV infection subgroup (P=0.30). Stratified by age, CRP levels were significantly elevated in the severe group among children ≤5 years old (WMD =−9.55, 95% CI: −17.65 to −1.45, P=0.02), whereas no statistically significant difference was observed in children >5 years old. Results stratified by sample size showed consistent trends with age stratification. Overall, IL-6 demonstrated relatively stable severity discrimination across different subgroups, whereas CRP variability may be influenced by multiple factors.

Discussion

Understanding the mechanisms underlying the severe progression of viral pneumonia in children has long been a core challenge in clinical studies. Recent research has shown the pivotal role of inflammatory cytokines in early risk stratification and in the mechanistic elucidation of severe viral pneumonia, particularly in cases involving RSV and adenovirus. For example, Tan et al. reported a significant positive correlation between disease severity and bronchoalveolar lavage fluid (BALF) levels of IL-6, IL-10, TNF-α, IL-17A, and MCP-1 in children with RSV pneumonia. Notably, the combined detection of IL-6 and IL-10 yielded superior diagnostic performance, achieving a sensitivity of 87.0% and a specificity of 96.2% (29). Similarly, a 2025 retrospective study on adenovirus pneumonia identified serum IL-6 and IL-10 as optimal biomarkers for severity prediction (AUC =0.88), with IL-6 achieving 93% sensitivity at the optimal cutoff (30). These findings reinforce the robustness and reproducibility of these markers across different viral etiologies.

Notably, distinct cytokine profiles may correspond to specific pathological patterns and clinical phenotypes. Lin et al. observed that in children with pneumonia-associated lung consolidation, elevated BALF IL-8 (≥6,615 pg/mL) was closely linked to hypoxemia risk, whereas IL-1β and IL-2R were correlated with structural complications such as multi-lobar consolidation and atelectasis (31). This suggests that variations in proinflammatory profiles not only reflect inflammation intensity, but also reflect specific pathological processes, such as local airway injury, exudation, and tissue remodeling. Further, broader screening has highlighted the potential of chitinase-3-like protein 1 (CHI3L1), interleukin-1 receptor type α (IL-1Rα), and granulocyte colony-stimulating factor (G-CSF) in predicting length of stay and ICU admission (32), pointing toward a complex prognostic network of immune signaling.

These emerging data align closely with our meta-analysis, corroborating the reliable early predictive value of core cytokines like IL-6 and IL-10 across diverse viral subtypes. This systematic review and meta-analysis revealed the predictive value and potential pathophysiological relevance of inflammatory cytokines in the progression to severe disease. The evidence chain showed that the core inflammatory axes, comprising IL-6, PCT, and CRP, had significant predictive efficacy, supporting the cytokine cascade theory proposed previously (33). A previous single-center study has reported an association between severity and IL-6; however, the role of IL-6 in the causal pathway remains unclear (34). Mechanistically, IL-6, as a key mediator of the T helper 17 cell pathway, promotes the formation of neutrophil extracellular traps by activating STAT3 signaling, which leads to microvascular endothelial injury (35,36). This finding refines the IL-6-sepsis hypothesis proposed in a previous study and provides a theoretical basis for the treatment of severe pneumonia with anti-IL-6 receptor monoclonal antibodies (37).

Pathogen-specific immune responses may contribute to the heterogeneity observed in inflammatory biomarker levels among children with viral pneumonia. A previous study has suggested that adenovirus infection is often accompanied by a more pronounced systemic inflammatory response compared with RSV or influenza virus. Experimental evidence indicates that adenovirus-associated proteins, such as E1A, can activate the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling pathway and may act synergistically with endotoxin to enhance proinflammatory cytokine and acute-phase protein production, including PCT (38). This mechanism may partly explain the higher PCT levels and worse clinical outcomes reported in children with severe adenovirus pneumonia in previous clinical studies.

Age-related differences in immune responses may further modulate inflammatory profiles in pediatric viral pneumonia. Infants and young children exhibit immature mucosal and systemic immune regulation, which may predispose them to exaggerated inflammatory responses. An experimental and clinical study suggests that disruption of epithelial barriers during viral infection may facilitate translocation of microbial products, such as lipopolysaccharide (LPS), thereby amplifying systemic inflammation through the liver-lung axis (39). These age-dependent immune characteristics provide a plausible biological explanation for the stronger associations between inflammatory cytokines and disease severity observed in younger children.

From a pathophysiological perspective, elevated levels of IL-6, PCT, and CRP may reflect distinct but interrelated inflammatory processes during severe viral pneumonia. IL-6 plays a central role in acute-phase responses, immune cell activation, and endothelial dysfunction, whereas PCT is associated with systemic inflammatory stress and may increase in the context of severe infection or secondary complications. CRP reflects downstream hepatic acute-phase activation and systemic inflammatory burden (40-43). Rather than acting as direct causal drivers, these biomarkers likely represent different dimensions of the host inflammatory response that collectively mirror disease severity.

Compared with traditional clinical scoring systems, inflammatory biomarkers may provide complementary biological information; however, their interpretation should remain cautious. Observed associations between cytokine levels and severe outcomes do not imply causality, and thresholds proposed in individual studies require external validation (44,45). Consequently, inflammatory biomarkers should be viewed as adjunctive tools to support clinical assessment rather than standalone predictors.

Several limitations should be acknowledged. RSV and adenovirus accounted for the majority of included cases, whereas data on less common viral pathogens were limited. Variability in laboratory measurement methods may have introduced measurement bias. In addition, most available evidence was derived from retrospective observational studies, restricting causal interpretation. Future research should focus on prospective multicenter studies, standardized cytokine measurement, and dynamic monitoring of inflammatory markers to clarify their clinical utility and biological significance in pediatric viral pneumonia (46).

Conclusions

This study conducted a systematic review and meta-analysis to synthesize the available evidence on the association between inflammatory factors and severe progression of viral pneumonia in children. The results demonstrated that multiple inflammatory markers, including IL-6, CRP, PCT, and TNF-α, were significantly associated with disease severity, indicating that dysregulated inflammatory responses play an important role in severe disease progression. Subgroup analyses showed that IL-6 consistently discriminated disease severity across different pathogen types, age groups, and sample sizes, while sensitivity analyses confirmed the robustness of the pooled results. These findings support the potential value of inflammatory biomarkers in early risk stratification and prognostic assessment. Although the present study does not establish causal relationships, the observed associations provide clinically relevant evidence that may assist in identifying children at higher risk of severe viral pneumonia and in optimizing early clinical management strategies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the PRISMA and CHARMS reporting checklists. Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0081/rc

Peer Review File: Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0081/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0081/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Wang X, Li Y, Shi T, et al. Global disease burden of and risk factors for acute lower respiratory infections caused by respiratory syncytial virus in preterm infants and young children in 2019: a systematic review and meta-analysis of aggregated and individual participant data. Lancet 2024;403:1241-53. [Crossref] [PubMed]
2. Lin J, Tang M, Yuan S, et al. Local Versus Systemic Cytokine Dynamics in Pediatric Pneumonia. Pediatr Pulmonol 2025;60:e71093. [Crossref] [PubMed]
3. Sayegh R, Tang L, Yun KW, et al. Nasal Mucosal Cytokines as Potential Biomarkers for Assessing Disease Severity and Class of Pathogen in Children With Community-Acquired Pneumonia. J Infect Dis 2025;231:1031-40. [Crossref] [PubMed]
4. Wang LP, Hu ZH, Jiang JS, et al. Serum inflammatory markers in children with Mycoplasma pneumoniae pneumonia and their predictive value for mycoplasma severity. World J Clin Cases 2024;12:4940-6. [Crossref] [PubMed]
5. Cao L, Lin M, Sheng F, et al. Efficacy and inflammatory levels (IL-6, IL-10) of adjuvant application of vitamin-A in the treatment of pediatric Mycoplasma pneumoniae pneumonia: a systematic review and meta-analysis. Front Pediatr 2024;12:1345458. [Crossref] [PubMed]
6. Park H, Cho HK, Kim DR, et al. Cytokine and Whole-Genome Sequence Analysis in Korean Patients With Multisystem Inflammatory Syndrome in Children. J Korean Med Sci 2025;40:e319. [Crossref] [PubMed]
7. Ramírez P, Ferrer M, Martí V, et al. Inflammatory biomarkers and prediction for intensive care unit admission in severe community-acquired pneumonia. Crit Care Med 2011;39:2211-7. [Crossref] [PubMed]
8. Yong SJ, Halim A, Halim M, et al. Inflammatory and vascular biomarkers in post-COVID-19 syndrome: A systematic review and meta-analysis of over 20 biomarkers. Rev Med Virol 2023;33:e2424. [Crossref] [PubMed]
9. Ferrarese R, Boutahar S, Genoni AP, et al. Cytokine and Endothelial Activation Patterns Related to Severe and Non-Severe Respiratory Viral Infections. J Med Virol 2026;98:e70784. [Crossref] [PubMed]
10. Yan Z, Wang Y, Sun W, et al. Th1/Th2 cytokine expression and clinical significance in children with respiratory syncytial virus pneumonia. China Med Guide 2023;20:99-101,105.
11. Li T, Huang X, Yang J, et al. TNF-α and IL-1β levels and TLR4 signaling pathway in children with respiratory syncytial virus pneumonia. Chin J Hosp Infect 2020;30:919-22.
12. Xu S, Tian X, Lu H. Clinical features of children with respiratory syncytial virus–associated pneumonia and analysis of peripheral helper T cells and Th17/Treg-related cytokine levels. J Clin Pulm Med 2025;53:262-4,269.
13. Zeng Z. Correlation between serum IgE, IL-4, IL-6 and IFN-γ levels and disease severity in children with respiratory syncytial virus pneumonia. Chin J Child Health Care 2025;37:136-8.
14. Xu C, Zheng L. Association of adenosine deaminase, interleukin-6, lactate dehydrogenase, serum amyloid A and cytokine levels with severity of viral pneumonia in children. Infect Dis Inf 2025;38:230-4.
15. Cheng Q, Chen S, Zhou H, et al. Diagnostic and prognostic value of serum ICAM-1, sTREM-1 and IL-6 in children with viral pneumonia. Henan Med Res 2025;34:1252-6.
16. Qu C, Zhang F, Yuan X, et al. Serum miR-146a and miR-145 levels in children with respiratory syncytial virus pneumonia and their clinical significance. Chin J Prev Med 2024;58:1733-8. [Crossref] [PubMed]
17. Qi W, Sun C, Li Y. Expression of IL-1β, IL-6 and IFN-γ in pediatric patients with influenza A viral pneumonia and their relationship with disease prognosis. Chin J Pract Med 2024;19:5-8.
18. Ma X, Yang Y, Li Z, et al. Diagnostic value of adenosine deaminase, procalcitonin and neutrophil-to-lymphocyte ratio in pediatric pneumonia and their relationship with disease severity. Adv Mod Biomed Sci 2024;24:4513-5.
19. Ma B, Guan H, Wang H. Early diagnostic value of combined detection of serum hs-CRP, PCT, sB7-H3 and LTB4 in pediatric pneumonia. J Clin Exp Med 2021;20:2435-9.
20. Xu D, Chen Q, Wang R, et al. Expression of miR-146a and miR-155 in peripheral blood mononuclear cells and their relationship with inflammatory response in infants with respiratory syncytial virus pneumonia. Chin J Child Health Care 2020;28:316-9.
21. Wang K, Liu L, Li Z, et al. Application value of serum LTB4 and PCT detection in pathogen diagnosis and disease severity assessment of pediatric pneumonia. J Matern Child Health Res 2020;31:1054-8.
22. Fan Z, Liu Y, Fan F, et al. Correlation between PCT and CRP levels and disease severity in pediatric viral pneumonia. J PLA Med Sci 2020;32:49-53.
23. Zhou Y, Yang J, Huang J, et al. Clinical significance of hs-CRP, WBC, PCT and T-cell immune markers in pediatric infectious pneumonia. Lab Immunoassay Clin Appl 2019;26:1701-5.
24. Li Y, He X. Clinical value of procalcitonin, C-reactive protein and D-dimer in pediatric pneumonia. World Latest Med Inf Abstr 2019;19:1-2,5.
25. Xue H, Su Z. Early differential diagnosis of pediatric pneumonia using cellular immunity markers and serum hs-CRP and PCT. Shaanxi Med J 2018;47:530-2.
26. Xiao J, Liu L, Zhang L, et al. Serum Cytokines as Predictive Biomarkers for Disease Severity in Pediatric Adenovirus Pneumonia. J Inflamm Res 2025;18:17077-88. [Crossref] [PubMed]
27. Wang G, Liu J, Yu X, et al. The value of combined examination of serum serum heparin binding protein, tumor necrosis factor alpha, interleukin-6, platelet count, and c-reactive protein in evaluating the condition and prognosis of children with adenovirus pneumonia. Front Pediatr 2025;13:1620206. [Crossref] [PubMed]
28. Liu X, Zhang X, Li J, et al. Diagnostic value of heparin-binding protein in severe adenovirus pneumonia in children. Chin J Contemp Pediatr 2022;24:1014-9. [Crossref] [PubMed]
29. Tan L, He Z, Liang Y, et al. Correlation analysis between the severity of respiratory syncytial virus pneumonia and the expression levels of inflammatory cytokines in bronchoalveolar lavage fluid among infants and young children. Front Pediatr 2025;13:1482029. [Crossref]
30. Duan Y, Ou Y, Li J, et al. Serum cytokine levels in children with community-acquired pneumonia caused by different respiratory pathogens. Ital J Pediatr 2025;51:147. [Crossref] [PubMed]
31. Lin J, Chen J, Yuan S, et al. Local and systemic cytokine profiles in children with pneumonia-associated lung consolidation. Front Immunol 2025;16:1546730. [Crossref] [PubMed]
32. Duray AM, Lee B, Abood RN, et al. Predicting infectious etiology and severity in hospitalized pediatric pneumonia using blood cytokine biomarkers. Front Pediatr 2025;13:1693879. [Crossref] [PubMed]
33. Ganguly M, Nandi A, Banerjee P, et al. A comparative study of IL-6, CRP and NT-proBNP levels in post-COVID multisystem inflammatory syndrome in children (MISC) and Kawasaki disease patients. Int J Rheum Dis 2022;25:27-31. [Crossref] [PubMed]
34. Lampidi S, Maritsi D, Charakida M, et al. Multisystem inflammatory syndrome in children (MIS-C): A nationwide collaborative study in the Greek population. Eur J Pediatr 2024;183:1693-702. [Crossref] [PubMed]
35. Lubell TR, Gorelik M, Abel D, et al. Development of a Model to Identify Febrile Children at Low Risk for Multisystem Inflammatory Syndrome. Pediatr Emerg Care 2023;39:476-81. [Crossref] [PubMed]
36. Santos-Rebouças CB, Piergiorge RM, Dos Santos Ferreira C, et al. Host genetic susceptibility underlying SARS-CoV-2-associated Multisystem Inflammatory Syndrome in Brazilian Children. Mol Med 2022;28:153. [Crossref] [PubMed]
37. Levine DA, Uy V, Krief W, et al. Predicting Delayed Shock in Multisystem Inflammatory Disease in Children: A Multicenter Analysis From the New York City Tri-State Region. Pediatr Emerg Care 2023;39:555-61. [Crossref] [PubMed]
38. Vansevicienė I, Krunkaitytė U, Dekerytė I, et al. Does Multisystem Inflammatory Syndrome Only Mimic Acute Appendicitis in Children or Can It Coexist: When Should We Suspect MIS-C? Medicina (Kaunas) 2022;58:1101. [Crossref] [PubMed]
39. Tran DM, Pham DV, Cao TV, et al. Severity predictors for multisystemic inflammatory syndrome in children after SARS-CoV-2 infection in Vietnam. Sci Rep 2024;14:15810. [Crossref] [PubMed]
40. Akcay N, Kihtir HS, Durak C, et al. Mortality Risk Factors Among Critically Ill Children With Acute COVID-19 in PICUs: A Multicenter Study From Turkish Pediatric Critical COVID-19 and MIS-C Study Group. Pediatr Infect Dis J 2022;41:742-50. [Crossref] [PubMed]
41. Martin B, DeWitt PE, Russell S, et al. Characteristics, Outcomes, and Severity Risk Factors Associated With SARS-CoV-2 Infection Among Children in the US National COVID Cohort Collaborative. JAMA Netw Open 2022;5:e2143151. [Crossref] [PubMed]
42. Dashti M, Alkandari H, Malik MZ, et al. Genetic insights into MIS-C Post-COVID-19 in Kuwaiti children: investigating monogenic factors. Front Cell Infect Microbiol 2025;14:1444216. [Crossref] [PubMed]
43. Wurm J, Uka A, Buettcher M, et al. Clinical and Laboratory Biomarkers as Predictors of Severity in Pediatric Inflammatory Multisystem Syndrome-temporally Associated With SARS-CoV-2: Data From a Prospective Nationwide Surveillance Study in Switzerland. Pediatr Infect Dis J 2024;43:675-81. [Crossref] [PubMed]
44. Çakmak F, Demirbuga A, Demirkol D, et al. Nailfold capillaroscopy: A sensitive method for evaluating microvascular involvement in children with SARS-CoV-2 infection. Microvasc Res 2021;138:104196. [Crossref] [PubMed]
45. Yu R, Yan T, He J, et al. Clinical efficacy of combined detection of serum procalcitonin and C-reactive protein in early differential diagnosis of bacterial and viral pneumonia and analysis of related inflammatory response mechanisms. Biomark Med 2025;19:657-66. [Crossref] [PubMed]
46. Pan H, Wang G, Guan E, et al. Treatment outcomes and prognostic factors for non- malignancy associated secondary hemophagocytic lymphohistiocytosis in children. BMC Pediatr 2020;20:288. [Crossref] [PubMed]

(English Language Editor: L. Huleatt)