Early Risk Factors and Clinical Characteristics of Pediatric Hemophagocytic Lymphohistiocytosis: A Single-Center Retrospective Study Authors: Ziming Wang, Zhining Geng, Yongli Xuan, Xinhui Luo Published: 2026 ABSTRACT AbstractBackgroundHemophagocytic lymphohistiocytosis (HLH) is a severe hyperinflammatory syndrome in children with rapid disease progression and high mortality. Early identification of clinical characteristics and risk-related indicators is essential for timely diagnosis and intervention.ObjectiveTo investigate early clinical characteristics and potential early risk-related factors associated with pediatric HLH among children with suspected HLH.MethodsWe retrospectively collected clinical data from 205 children with suspected HLH who met the inclusion criteria at the Xinjiang Uygur Autonomous Region Children’s Hospital between July 2021 and February 2026. Univariate logistic regression, variance inflation factor (VIF) assessment, and LASSO–logistic regression were used to examine clinical variables associated with HLH and to retain potential early risk-related factors. The performance of the final LASSO–logistic regression model was further evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, bootstrap resampling for internal validation, and the Hosmer–Lemeshow test.ResultsOf the 205 patients, 56 (27.3%) were diagnosed with HLH. LASSO–logistic regression identified fever duration (OR = 1.551, 95% CI: 0.973–2.472), serum ferritin (SF) (OR = 1.019, 95% CI: 1.001–1.039), triglycerides (TG) (OR = 1.424, 95% CI: 1.003–2.022), fibrinogen (FIB) (OR = 0.832, 95% CI: 0.663–1.044), and D-dimer (D-D) (OR = 1.014, 95% CI: 0.990–1.038) as potential early risk-related factors associated with HLH. The combined model showed the highest discriminative ability, with an AUC of 0.748. The Hosmer–Lemeshow test indicated good calibration (P = 0.721), and bootstrap internal validation yielded a C-index of 0.722.ConclusionFever duration, SF, D-D, FIB, and TG were potential early risk-related factors associated with pediatric HLH and may provide clinically useful information for early risk assessment in children with suspected HLH.Keywords: hemophagocytic lymphohistiocytosis, children, risk factors, clinical characteristics, LASSO regression Introduction Hemophagocytic lymphohistiocytosis (HLH) is a severe and rare inflammatory disorder.1,2 It is primarily characterized by aberrant activation and reactive expansion of the monocyte–macrophage system, leading to multi-organ inflammatory injury. Without timely diagnosis and treatment, the disease can progress rapidly and may become life-threatening3. HLH is generally classified as primary HLH (pHLH) and secondary HLH (sHLH)4. Primary HLH is most commonly caused by genetic defects affecting cytotoxic function; in untreated patients, the median survival can be as short as approximately 2 months, reflecting a fulminant natural course.5 Secondary HLH may be triggered by infections, malignancies, connective tissue diseases, transplantation, and metabolic disorders.6,7 The main clinical manifestations of HLH include persistent fever, cytopenias, hepatosplenomegaly, liver dysfunction, and coagulation abnormalities.8 However, with the increasing use of chemotherapy and immunotherapy, the HLH-2004 protocol—comprising etoposide and dexamethasone, with the addition of immunosuppressive agents such as cyclosporine—has been widely adopted for HLH treatment.9 The estimated 5-year survival probability under the HLH-2004 regimen is approximately 61%.10 For primary/familial HLH, hematopoietic stem cell transplantation (HSCT) remains the preferred curative strategy, and some studies have reported a post-transplant 5-year survival rate of approximately 66%.11 Overall, the prognosis of HLH is closely associated with the underlying etiology, the extent of organ dysfunction/failure, and the timing of diagnosis and therapeutic intervention. Prior studies indicate that the overall mortality of HLH can still be as high as 40%, and mortality-related clinical conditions mainly include organ involvement and critical illness, bone marrow suppression with cytokine dysregulation, and coagulation disturbances.12–14 Li et al reported the use of machine-learning approaches to predict early mortality-related factors in HLH.15 In addition, Bethany et al developed a dynamic monitoring strategy to evaluate the association between immune parameters and prognosis during HLH chemotherapy.16 Guo et al established a prognostic scoring model based on the clinical characteristics of pediatric HLH.17 Nevertheless, studies focusing on early clinical characteristics and readily available laboratory indicators among children with suspected HLH remain limited. On the basis of the evidence above, the early manifestations of HLH substantially overlap with severe infection/sepsis and malignancy-associated inflammatory responses; meanwhile, the disease may evolve rapidly, leaving a narrow diagnostic and therapeutic window.18 The Chinese HLH-2022 guideline also emphasizes early recognition and risk-stratified management.19 Therefore, exploring potential early risk-related factors and clinical characteristics associated with pediatric HLH may provide clinically useful information for the early evaluation of children with suspected HLH. Materials and Methods Inclusion and Exclusion Criteria This was a single-center, retrospective, observational cohort study and is reported in accordance with the RECORD (Reporting of studies Conducted using Observational Routinely-collected health Data) guideline. Data were obtained from the electronic medical record system of Xinjiang Uygur Autonomous Region Children’s Hospital. We identified 205 children with suspected HLH admitted between July 2021 and February 2026. Suspected HLH cases were defined according to the HLH-2004 guideline9 and were subsequently verified through retrospective chart review. Case verification was performed by senior attending physicians with advanced professional titles at our hospital. Inclusion criteria were as follows: (1) age 1–18 years; (2) clinical suspicion of HLH after admission, defined as persistent or recurrent fever accompanied by one or more HLH-related clinical or laboratory abnormalities, such as cytopenia, hepatosplenomegaly, hyperferritinemia, hypertriglyceridemia, hypofibrinogenemia, abnormal liver function, or coagulation abnormalities, leading to initiation of an HLH diagnostic workup, including assessment of serum ferritin, triglycerides, fibrinogen, and/or bone marrow examination, as clinically appropriate. Exclusion criteria were as follows: (1) age >18 years; (2) a diagnosis of another disease at admission that was sufficient to explain all clinical manifestations; (3) multiple admissions for the same patient, in which case only the first eligible admission was included; and (4) missing key data. Definition Definite HLH was diagnosed according to the HLH-2004 guideline,9 based on either a consistent molecular diagnosis or fulfillment of at least five of the eight diagnostic criteria: fever, splenomegaly, cytopenias affecting ≥2 cell lineages, hypertriglyceridemia and/or hypofibrinogenemia, hemophagocytosis in tissues, decreased NK-cell activity, elevated ferritin, and increased soluble CD25 (sCD25). Suspected HLH was also defined in accordance with the HLH-2004 guideline. Patients were ultimately classified into the HLH group and the non-HLH group. Data Collection Data collected included: (1) demographic characteristics, including sex, age, and ethnicity; (2) clinical features, including fever duration, peak body temperature, and hepatosplenomegaly and/or lymphadenopathy; and (3) laboratory parameters, including neutrophil count (NEU), lymphocyte count (LYM), monocyte count (MON), hemoglobin (Hb), platelet count (PLT), C-reactive protein (CRP), serum ferritin (SF), triglycerides (TG), total bilirubin (TBIL), lactate dehydrogenase (LDH), blood urea nitrogen (BUN), total cholesterol (TC), aspartate aminotransferase (AST), alanine aminotransferase (ALT), albumin (ALB), fibrinogen (FIB), prothrombin time (PT), activated partial thromboplastin time (APTT), D-dimer (D-D), procalcitonin (PCT), and Epstein-Barr virus DNA (EBV-DNA). All laboratory data were obtained from the first measurements performed at admission. Statistical Methods All statistical analyses were performed using R software (version 4.3.0). Continuous variables with a normal distribution are presented as mean ± standard deviation (SD) and were compared between groups using the independent-samples t test. Non-normally distributed continuous variables are presented as median with interquartile range (IQR; P25, P75) and were compared using the Mann–Whitney U-test. Categorical variables are expressed as number (percentage) and were compared using the chi-square test. All tests were two-sided, and P < 0.05 was considered statistically significant. Variables were first screened through univariate between-group comparisons; those showing a potential association (P < 0.10) were selected as candidate predictors and entered into univariate logistic regression analyses. Pearson correlation and variance inflation factor (VIF) were used to assess multicollinearity among candidate variables. Finally, LASSO–logistic regression was applied to identify the final set of risk-related variables. The performance of the combined model was further evaluated using the area under the receiver operating characteristic curve (AUC), calibration curve, Hosmer–Lemeshow goodness-of-fit test, and bootstrap resampling for discrimination, calibration, goodness of fit, and internal validation, respectively. Results Baseline Characteristics A total of 205 children with suspected HLH were included in this study, of whom 56 (27.3%) were diagnosed with HLH and 149 (72.7%) were classified as Non-HLH (Figure 1). Within the Non-HLH group, the underlying diagnoses comprised infectious diseases in 87 patients (58.4%), hematologic malignancies in 36 (24.2%), and autoimmune diseases in 26 (17.4%) (Supplementary Figure 1). Age was presented as the median with interquartile range (IQR) in years. The median age was 8.00 years (IQR: 3.00–11.00) in the overall cohort, 8.00 years (IQR: 4.00–12.00) in the Non-HLH group, and 6.00 years (IQR: 1.69–9.00) in the HLH group. No statistically significant difference in age was observed between the HLH and non-HLH groups (P = 0.056). There were no statistically significant between-group differences in baseline characteristics, including sex, fever duration, hepatomegaly, lymphadenopathy, and splenomegaly (all P > 0.05) (Table 1). Notably, the median fever duration tended to be longer in the HLH group than in the non-HLH group (P = 0.097). Regarding laboratory findings, lymphocyte count, monocyte count, hemoglobin, and platelet count did not differ significantly between the two groups (all P > 0.05). However, the HLH group exhibited more pronounced markers of hyperinflammation and metabolic dysregulation: serum ferritin (SF) and triglycerides (TG) were significantly higher, whereas albumin (ALB) and fibrinogen (FIB) were significantly lower. In addition, D-dimer (D-D) and aspartate aminotransferase (AST) levels were higher in the HLH group. Other parameters, including C-reactive protein (CRP), EBV-DNA and procalcitonin (PCT), showed no significant differences between groups (Supplementary Table 1). Flowchart of patient selection and cohort classification. Table 1Baseline Clinical and Laboratory Characteristics of Children with Suspected HLH, Stratified by Final DiagnosisVariableTotal(n = 205)Non-HLH(n = 149)HLH(n = 56)PSex, n (%)0.347Female108 (53)75 (50)33 (59)Male97 (47)74 (50)23 (41)Age [M(P25,P75), years]8(3.00, 11.00)8 (4.00, 12.00)6 (1.69, 9.00)0.056Fever duration[M(P25,P75), days]8 (5, 14)8 (5, 14)9(7.5, 14)0.097Thermal spike[M(P25,P75), °C]39.6 (39, 40)39.5 (39, 40)40 (39.18, 40)0.013Hepatomegaly, n (%)0.711No94 (46)70 (47)24 (43)Yes111 (54)79 (53)32 (57)Lymphadenopathy, n (%)0.549No93 (45)70 (47)23 (41)Yes112 (55)79 (53)33 (59)MON [M(P25,P75), ×109/L]0.36 (0.19, 0.80)0.39 (0.22, 1.00)0.29 (0.16, 0.70)0.075SF [M(P25,P75), ng/mL]975.90 (607.70, 2000.00)814.80 (588.20, 1527.00)2000.00 (1029.72, 3543.50)< 0.001TG [M(P25,P75), mmol/L]1.63 (1.19, 2.33)1.52 (1.18, 2.19)2.14 (1.35, 2.83)< 0.001AST [M(P25,P75), U/L]66.00(40.00, 163.00)61.00 (39.00, 124.00)99.00 (43.00, 226.75)0.012ALB (, g/L)34.09±634.77±5.6132.26±6.650.014FIB [M(P25,P75), g/L]3.10 (2.14, 4.60)3.50 (2.30, 4.73)2.53 (1.49, 3.77)0.001PT [M(P25,P75), S]13.70 (12.70, 15.10)13.70 (12.60, 14.80)14.00 (12.85, 15.95)0.091D-D [M(P25,P75), mg/L]3.57 (1.80, 7.67)3.15 (1.70, 6.42)4.53 (1.92, 15.99)0.012LDH [M(P25,P75), U/L]533 (384, 793)516 (358, 788)626 (440.25, 843.25)0.061Abbreviations: MON, Monocyte Count; SF, Serum Ferritin; TG, Triglyceride; AST, Aspartate Aminotransferase; ALB, Albumin; FIB, Fibrinogen; PT, Prothrombin Time; D-D, D-Dimer; LDH, Lactate Dehydrogenase. Baseline Clinical and Laboratory Characteristics of Children with Suspected HLH, Stratified by Final Diagnosis Abbreviations: MON, Monocyte Count; SF, Serum Ferritin; TG, Triglyceride; AST, Aspartate Aminotransferase; ALB, Albumin; FIB, Fibrinogen; PT, Prothrombin Time; D-D, D-Dimer; LDH, Lactate Dehydrogenase. Univariate Logistic Regression Analysis Variables showing a between-group difference (P < 0.10) and/or deemed clinically relevant were entered into univariate logistic regression analyses. Variables with P < 0.05 in the regression screening were retained as candidate predictors, including fever duration, SF, TG, LDH, AST, ALB, FIB, PT, and D-D. In the univariate analyses, fever duration, SF, TG, and D-D were associated with higher odds of HLH, whereas ALB and FIB were associated with lower odds. These findings suggest that changes in these parameters may represent early risk-related clinical indicators in children with suspected HLH. The remaining variables did not reach statistical significance in the univariate analyses (Table 2). Table 2Univariate Logistic Regression Analysis Comparing the HLH and Non-HLH GroupsVariableβOR95% CIPMON−0.130.8820.747–1.0410.138Fever duration0.471.6011.063–2.4110.024SF0.031.0321.011–1.0530.003TG0.581.7801.293–2.450<0.001LDH0.011.0120.991–1.0340.261AST0.011.0140.989–1.0400.264ALB−0.070.9290.880–0.9810.008FIB−0.320.7300.594–0.8960.003PT0.111.1180.989–1.2630.075D-D0.031.0311.007–1.0560.012 Univariate Logistic Regression Analysis Comparing the HLH and Non-HLH Groups Collinearity Analysis To examine correlations among variables that were significant in the univariate logistic regression (P < 0.05) and to assess potential multicollinearity, pairwise associations were evaluated using Pearson’s correlation coefficients. Six variables were included: SF, TG, ALB, FIB, D-D, and fever duration. A Pearson correlation matrix was computed and visualized (Figure 2). The results indicated generally weak to moderate correlations among SF_100, TG, ALB, FIB, D-D, and fever duration. To further evaluate whether substantial multicollinearity was present, variance inflation factors (VIFs) were calculated for each variable. All included variables had VIF values < 10, and tolerances were > 0.9, suggesting no severe multicollinearity (Table 3). Pearson correlation coefficient matrix heatmap of variables (red indicates positive correlation, blue indicates negative correlation; color intensity represents the strength of correlation). Table 3Multicollinearity Diagnostics for Variables Included in the Risk-Factor AnalysisVariableVIFToleranceFever duration1.0570.946SF1.0760.930TG1.1990.834ALB1.1620.860FIB1.1230.890D-D1.1540.866 Multicollinearity Diagnostics for Variables Included in the Risk-Factor Analysis LASSO-Logistic Regression Analysis To further identify variables that were stably associated with HLH and to mitigate potential instability in regression estimates due to multicollinearity, six variables (fever duration, SF, TG, ALB, FIB, and D-D) were entered into a LASSO–logistic regression analysis. The penalty parameter (λ) was selected using 10-fold cross-validation. This approach controls model complexity and reduces overfitting while retaining the most stable predictors associated with confirmed HLH. Cross-validation indicated that a less stringent penalty at λ_min = 0.085 tended to retain more variables, whereas a more conservative penalty at λ_1se = 0.061 yielded a smaller but more parsimonious set of variables (Supplementary Figure 2). Considering cross-validation error, model parsimony, and clinical interpretability, five indicators were ultimately retained as potential early risk-related factors associated with pediatric HLH: fever duration, SF, TG, FIB, and D-D. In the LASSO–logistic regression analysis, five variables were retained in the combined set of early risk-related factors: fever duration, SF, TG, FIB, and D-D. The regression coefficients for fever duration, SF, TG, and D-D were positive, suggesting that higher values of these variables were associated with increased odds of HLH after adjusting for the other included variables. In contrast, FIB had a negative coefficient, indicating that higher fibrinogen levels were associated with lower odds of HLH. Specifically, each 1 mmol/L increase in TG was associated with an approximately 42.4% increase in the odds of HLH; SF (scaled per 100 ng/mL) was associated with an approximately 1.9% increase in the odds; each additional day of fever duration was associated with an approximately 55.1% increase in the odds; and each 1 mg/L increase in D-D was associated with an approximately 1.4% increase in the odds. Conversely, higher FIB was associated with lower odds of HLH; each 1 g/L increase in FIB was associated with an approximately 16.8% decrease in the odds of HLH (Table 4). Table 4LASSO-Logistic Regression Analysis of Early Risk Factors Associated with Pediatric HLHVariableβOR95% CIFever duration0.4391.5510.973–2.472SF0.0191.0191.000–1.039TG0.3541.4241.003–2.022FIB−0.1840.8320.663–1.044D-D0.0141.0140.990–1.038 LASSO-Logistic Regression Analysis of Early Risk Factors Associated with Pediatric HLH Model Discrimination, Calibration, and Internal Validation ROC curve analysis showed that the AUC values of fever duration, FIB, D-D, TG, SF, and the combined set of variables were 0.619, 0.629, 0.650, 0.652, 0.709, and 0.748, respectively (Figure 3). The combined model showed the highest discriminative ability. The Hosmer–Lemeshow goodness-of-fit test indicated good calibration of the combined model (P = 0.721), and calibration analysis demonstrated good agreement between the predicted and observed probabilities of HLH (Figure 4). After internal validation using bootstrap resampling, the optimism-corrected C-index of the combined set of variables was 0.722, indicating acceptable model stability and discrimination. ROC curves for early risk-related factors. Calibration curve for combined early risk-related factors. Discussion HLH in children is characterized by abrupt onset and rapid progression; the ensuing cytokine storm can precipitate multi-organ dysfunction within a short period, and clinical outcomes may be unfavorable in the absence of active treatment.20 With the wider implementation of standardized immunochemotherapy regimens based on etoposide in combination with glucocorticoids, overall survival in HLH has improved substantially compared with earlier eras. Under contemporary diagnostic and therapeutic conditions, children with primary HLH (pHLH) receiving first-line etoposide-based therapy have been reported to achieve a 3-year survival of approximately 77%, and peri-transplant outcomes have also improved relative to the HLH-2004 era.21 However, in the early clinical course, HLH often overlaps considerably with severe infection/sepsis, malignancy-associated inflammatory responses, and disease flares of autoimmune disorders.22 Moreover, disease-specific tests such as soluble CD25, NK-cell activity, and genetic testing are difficult to complete rapidly within the “critical window” in many healthcare settings, which may still lead to delays in diagnosis and initiation of therapy.23 Given that early initiation of standardized treatment and shortening the time to transplantation or intensified therapy are closely associated with improved survival,24 early recognition of potential risk-related factors and clinical characteristics associated with HLH is of substantial clinical importance. In the present study, female patients accounted for a higher proportion of the HLH group than male patients. Previous studies have reported variable sex distributions among pediatric HLH cohorts. Alsohime et al25 reported an approximately balanced sex distribution in children with familial HLH, whereas Akyol et al26 reported a slightly higher proportion of female patients in a primary HLH cohort. These findings suggest that sex may not be a consistent factor associated with HLH across different cohorts. The higher proportion of female patients observed in our cohort may be related to differences in cohort composition, underlying triggers, or disease subtypes. In the clinical evaluation of suspected HLH, prolonged fever duration has been identified as a clinical feature associated with HLH.27 Unlike the typically self-limited febrile course observed in common pediatric infections, children with HLH frequently present with persistent, refractory high-grade fever.28 In our cohort, the median duration of fever in children with HLH was 9 days, with an interquartile range of 7.5–14 days, and longer fever duration was associated with higher odds of confirmed HLH (OR = 1.551). This is consistent with findings reported by Luo and Ozturk et al,29,30 who noted that the median duration of fever exceeded 10 days in pediatric HLH cohorts and that the mean number of febrile days reached 1330 ± 14.05 days. These observations are generally consistent with our results, further supporting the potential clinical value of sustained high-grade fever in the early evaluation of children with suspected HLH. Fever in HLH reflects immune dysregulation, which is primarily driven by impaired function of natural killer (NK) cells and cytotoxic T lymphocytes, which may impair the clearance of infectious antigens or aberrant cells.31 Consequently, persistent antigen exposure continuously stimulates the immune system, establishing a self-amplifying loop involving reciprocal activation between T cells and macrophages.32 Excessive production of endogenous pyrogens—particularly interleukin (IL)-1, IL-6, and tumor necrosis factor (TNF)-α—can persistently cross the blood–brain barrier and act on the hypothalamic thermoregulatory center, thereby elevating the thermoregulatory set point and contributing to persistent fever.33 Therefore, prolonged fever duration may reflect sustained hyperinflammation and immune dysregulation, and may provide a clinically observable clue during the early evaluation of children with suspected HLH. Serum ferritin (SF) is an acute-phase reactant that typically shows mild-to-moderate elevation in infection or nonspecific inflammation.34 In HLH, however, SF often increases far more markedly, sometimes with a rapid and substantial increase; in some children, levels can exceed 10,000 ng/mL, a threshold that has strong diagnostic implication for HLH in pediatric populations.35 In our HLH cohort, the median SF level was 2,000 ng/mL, which is consistent with the median SF of 2,538 ng/mL reported by Francesco et al in a study of primary HLH.36 In the prediction model developed by Basu et al,37 an SF level >3,123 ng/mL was identified as an early risk-related indicator for HLH, which is generally consistent with our findings Hypertriglyceridemia was also associated with HLH in our analysis and has been reported as an early risk-related indicator in previous studies. In the study reported by Cai et al,38 each 1 mmol/L increase in triglycerides (TG) was associated with an approximately 154-fold increase in HLH risk, closely aligning with the 1.42-fold increase observed in our study. Elevated TG is a hallmark of metabolic dysregulation, and its underlying mechanism is largely attributed to profound suppression of lipoprotein lipase (LPL) activity.39 As a result, peripheral tissues are unable to adequately clear circulating TG, leading to impaired lipid clearance and subsequent rapid and pronounced hypertriglyceridemia. Clinically, elevated TG reflects inflammation-driven metabolic stress and lipid clearance dysfunction; its presence may indicate systemic metabolic disturbance and has been associated with organ dysfunction and adverse outcomes.40 In our study, coagulation abnormalities in children with HLH were characterized by decreased fibrinogen (FIB; OR = 0.832, associated with lower odds of HLH) and increased D-dimer (D-D; OR = 1.014, associated with higher odds of HLH), both of which were retained as potential early risk-related factors associated with pediatric HLH. During HLH progression, coagulation dysfunction is mainly driven by consumption of coagulation factors and insufficient hepatic synthesis; meanwhile, extensive fibrinogen degradation contributes to a hypercoagulable and hyperfibrinolytic state.41 In recent years, Luo et al29 in a study of pediatric sepsis complicated by HLH, reported a median FIB level of 2.43 g/L, and found that each unit decrease in FIB was associated with a 0.44-fold change in risk, which is consistent with our median FIB level of 2.53 g/L. Although FIB typically increases as an acute-phase reactant in common infections, it may decline rapidly in HLH, suggesting that the inflammatory response may be accompanied by substantial consumption within the coagulation–fibrinolysis system.42 Cheng and Zhang et al43,44 further reported in HLH-related clinical studies that each one-unit increase in D-D was associated with an approximately 1.08–1.12-fold increase in HLH risk, closely aligning with our findings (1.104-fold). Therefore, HLH-related coagulation–fibrinolysis imbalance may manifest as decreased fibrinogen and elevated D-D, indicating enhanced fibrinolysis and/or consumptive coagulopathy,45 and may provide pathophysiological context for considering these parameters during early clinical evaluation of children with suspected HLH. In summary, this single-center retrospective study found that fever duration, SF, TG, FIB, and D-D were potential early risk-related factors associated with pediatric HLH among children with suspected HLH. These indicators are derived from routine clinical tests, are readily accessible, and can be repeatedly monitored throughout the course of treatment, thereby supporting early clinical evaluation and further diagnostic assessment. Although our findings are based on associations rather than causal inference, they provide clinically relevant information that may help inform early assessment and management of children with suspected HLH. Several limitations should be acknowledged. First, as a single-center retrospective analysis with a relatively small sample size, selection bias cannot be fully excluded. Second, some potential confounders, such as differences in supportive care and the lack of HLH-related genetic testing and scCD25 testing, were not comprehensively quantified. Future multicenter, large-scale prospective studies are warranted further examine these associations and to incorporate additional variables for a more systematic evaluation of early risk-related clinical characteristics associated with pediatric HLH. Conclusions In conclusion, this single-center retrospective study described the early clinical characteristics and identified fever duration, SF, TG, FIB, and D-D as potential early risk-related factors associated with pediatric hemophagocytic lymphohistiocytosis. Our findings highlight the importance of early recognition of characteristic clinical manifestations and laboratory abnormalities in children with suspected HLH. Recognizing children with a higher likelihood of HLH may help clinicians initiate timely evaluation and management and may support clinical decision-making. Further multicenter prospective studies with larger sample sizes are warranted to confirm these findings and to further develop clinically applicable early risk assessment approaches for pediatric HLH. Acknowledgments We are also grateful to our peers for their suggestions and review of the article. All the authors have agreed to the article. Funding Statement This study was supported by the Science and Technology Department of Xinjiang Uygur Autonomous Region. The funded project was “Establishment of a Rapid Diagnostic Method and Implementation of Risk-Stratified Treatment for Pediatric Hemophagocytic Lymphohistiocytosis” (TSYC202301A046). Data Sharing Statement The data will be shared when asking for the corresponding author Xinhui Luo (mjiu511@gmail.com). Ethics Statements This study was reviewed and approved by the Ethics Committee of the Xinjiang Uygur Autonomous Region Children’s Hospital (approval number: KY2024112107). Given the retrospective design and the anonymized nature of the patient data, the requirement for written informed consent was waived in accordance with institutional and national regulations. The study was conducted in accordance with the ethical principles and standards of the Declaration of Helsinki. Author Contributions Ziming Wang: Conceptualization, Methodology, Data curation, Formal analysis, Investigation, Writing – original draft. Zhining Geng: Conceptualization, Methodology, Investigation, Data curation, Writing – review and editing. Yongli Xuan: Investigation, Data curation, Validation, Writing – review and editing. Xinhui Luo: Conceptualization, Supervision, Writing – review and editing. All authors took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Disclosure The authors declare that they have no conflict of interest. References Associated Data Data Availability Statement The data will be shared when asking for the corresponding author Xinhui Luo (mjiu511@gmail.com).