Abstract
Objective
To evaluate serum soluble receptor for advanced glycation end products (sRAGE) and soluble vascular endothelial growth factor receptor 3 (sVEGFR-3) as biomarkers for hypercoagulable state in children with Mycoplasma pneumoniae pneumonia (MPP).
Methods
We retrospectively enrolled 352 children with MPP (hypercoagulable, n=98; non-hypercoagulable, n=254). Serum sRAGE and sVEGFR-3 were measured by enzyme-linked immunosorbent assay (ELISA). Spearman analysis, Box-Tidwell test, and restricted cubic splines (RCS) assessed correlations/nonlinear effects; least absolute shrinkage and selection operator (LASSO) regression selected variables; multivariable logistic regression analyzed risk factors (variance inflation factor [VIF] for multicollinearity); receiver operating characteristic (ROC) curve analysis evaluated predictive value (Youden index cut-off); Bootstrap (1000 iterations) validated internally; net reclassification improvement (NRI) and integrated discrimination improvement (IDI) assessed incremental value.
Results
Serum sRAGE and sVEGFR-3 were elevated in the hypercoagulable group (both P<0.001), with shortened prothrombin time (PT), activated partial thromboplastin time (APTT), and elevated fibrinogen (FIB) and D-dimer (all P<0.05); both correlated negatively with PT/APTT and positively with FIB/D-dimer (all P<0.001). LASSO selected age, duration of fever, lactate dehydrogenase (LDH), Mycoplasma pneumoniae (MP) antibody titer, sRAGE, and sVEGFR-3 as candidate factors; duration of fever, MP antibody titer, sRAGE, and sVEGFR-3 were independent risk factors by multivariable logistic regression (all P<0.05; VIF<2). Single-marker areas under the curve (AUCs) were 0.840 (sRAGE) and 0.709 (sVEGFR-3); combined AUC was 0.870 (95% confidence interval [CI]: 0.831–0.909; sensitivity 86.61%, specificity 71.43%), outperforming single indicators (DeLong test, P<0.05), with optimism-corrected AUC 0.871 (95% CI: 0.863–0.875). Versus sRAGE and sVEGFR-3 alone, the combined model yielded NRI/IDI of 0.497/0.048 and 0.9719/0.2835, respectively (all P<0.001).
Conclusion
Elevated serum sRAGE and sVEGFR-3 are associated with hypercoagulable state in children with MPP; combined detection may serve as an exploratory reference for early identification, but requires prospective longitudinal and independent multicenter validation for clinical translation.
Keywords: Mycoplasma pneumoniae pneumonia, soluble receptor for advanced glycation end products, soluble vascular endothelial growth factor receptor 3, hypercoagulability
Introduction
Mycoplasma pneumoniae pneumonia (MPP) is a leading cause of community-acquired pneumonia in children. Its pathogenesis encompasses direct microbial injury, dysregulated immune–inflammatory responses, and endothelial dysfunction. A frequently overlooked complication is a systemic hypercoagulable state that intensifies intra-pulmonary inflammation and predisposes to atelectasis and pulmonary embolism, as well as to life-threatening deep-vein thrombosis and cerebral embolism.1–5 Consequently, early recognition of hypercoagulability in pediatric MPP is essential to improve outcome and reduce mortality.
The clinical diagnosis of hypercoagulable state currently relies mainly on conventional coagulation parameters and inflammatory markers; however, these indicators have notable limitations. Conventional coagulation parameters show low sensitivity to early-stage and mild hypercoagulable states, and D-dimer (D-D) exhibits a lag when thrombosis has not yet formed.6–8 Meanwhile, C-reactive protein (CRP) lacks sufficient specificity for infections caused by atypical pathogens9,10 and fails to reflect the core pathological processes of endothelial injury and inflammation–coagulation interplay. Therefore, exploring novel biomarkers that are directly related to the pathological mechanism of hypercoagulability in MPP and can complement the shortcomings of conventional indicators is of great significance for improving early predictive performance.
The receptor for advanced glycation end products (RAGE) exists mainly in two forms: membrane-bound RAGE (mRAGE) and soluble RAGE (sRAGE). mRAGE is a multi-ligand pattern recognition receptor whose ligands include advanced glycation end products, high mobility group box 1, and S100 calcium-binding proteins, and it participates in the regulation of inflammatory responses, oxidative stress, and coagulation processes.11 Studies have shown that sRAGE can be stably detected in the human circulation.12 Recent research has identified prothrombin as a novel and direct binding partner of sRAGE, suggesting a direct interaction between sRAGE and the coagulation process.13 In addition, serum sRAGE is upregulated in various inflammatory and thrombotic diseases, and its levels are closely associated with disease severity and prognosis.14,15 In this study, serum sRAGE levels were measured by enzyme-linked immunosorbent assay (ELISA).
Vascular endothelial growth factor receptor 3 (VEGFR-3), also known as Fms-related tyrosine kinase 4 (FLT4), was previously considered to be expressed predominantly in lymphatic endothelial cells. Recent studies have confirmed that it is also expressed in vascular endothelial cells, where it negatively regulates the expression and activity of VEGFR-2, thereby inhibiting the vascular endothelial growth factor signaling pathway, maintaining the integrity of the vascular barrier, and preventing excessive vascular leakage.16 Soluble VEGFR-3 (sVEGFR-3) is the shed form of the extracellular domain of the membrane-bound receptor, generated by proteolytic cleavage, and can be stably detected in the human circulation.17 MPP infection-related inflammation can activate the coagulation cascade and cause vascular endothelial injury.6,18 Previous studies have found that thrombin can cleave VEGF-C/D to activate VEGFR-3 signaling,19 suggesting an association between the VEGFR-3 family and the coagulation process. As a shed product of the membrane-bound receptor, changes in serum sVEGFR-3 levels may reflect the degree of endothelial cell activation or injury. In this study, serum sVEGFR-3 levels were measured by ELISA.
Based on the above research background, this study proposes the following hypotheses: (1) serum sRAGE and sVEGFR-3 levels are significantly elevated in children with MPP complicated by hypercoagulable state, and both are correlated with conventional coagulation parameters at the time of confirmed hypercoagulable state; (2) serum sRAGE and sVEGFR-3 are independent risk factors for hypercoagulable state in children with MPP; and (3) the combined detection of sRAGE and sVEGFR-3 has superior predictive performance compared with either single indicator, and may serve as a complementary biomarker combination for early risk stratification of hypercoagulable status in children with MPP. To test these hypotheses, this study adopted a retrospective cohort design to systematically evaluate the expression characteristics, correlations, and combined predictive value of serum sRAGE and sVEGFR-3 in MPP with hypercoagulable state, thereby providing a theoretical basis for the early clinical identification of hypercoagulable state in children with MPP.
Subjects and Methods
Sample-Size Calculation
The sample size was estimated based on receiver operating characteristic (ROC) curve analysis for predictive tests. The null hypothesis area under the curve (AUC0) was set at 0.50 (indicating no predictive value) and the alternative hypothesis AUC (AUC1) at 0.70 (indicating moderate predictive performance), with a two-sided α of 0.05 and a statistical power of 80%. A previous study reported an incidence of hypercoagulable state of 30.59% in children with MPP.7 Estimation using the “Tests for One ROC Curve” module of PASS 15.0 software indicated that 23 cases were required in the positive group and 52 cases in the negative group, yielding a total sample size of 75 cases.
Given that this study also involved complex statistical modeling, including multivariate logistic regression analysis (with six independent variables expected to be included), least absolute shrinkage and selection operator (LASSO) regression for variable selection, restricted cubic spline (RCS) nonlinear fitting, and Bootstrap internal validation (1000 resampling iterations), the sample size was expanded following the empirical rule of “at least 10 outcome events per variable” (events per variable, EPV). Six independent variables require at least 60 hypercoagulable events, corresponding to a total sample size of approximately 196 cases (based on 30.59% incidence). However, because LASSO selection, RCS fitting, and Bootstrap validation (1000 resamplings) all demand greater sample stability, the sample size was further expanded. A total of 352 children with MPP were actually enrolled in this study (98 in the hypercoagulable group and 254 in the non-hypercoagulable group), providing sufficient statistical power to ensure the reliability of the study conclusions.
Study Population
Inclusion criteria: (1) Fulfillment of the pediatric MPP definition:20 respiratory symptoms (fever, cough, wheeze or dyspnea) plus radiographic pneumonia, together with at least one of the following: single serum MP-antibody titer ≥1:160 or a ≥4-fold rise in paired sera; positive nasopharyngeal MP-DNA or MP-RNA; (2) Age 5–12 years; (3) complete coagulation function test results available within 3 h of admission, including prothrombin time (PT), activated partial thromboplastin time (APTT), fibrinogen (FIB), and D-D levels; (4) at least one coagulation function retest result available during hospitalization.
Exclusion criteria: (1) Pneumonia attributable to other pathogens (bacterial, viral or fungal); (2) Concomitant respiratory disorders (bronchial asthma, bronchiectasis, chronic obstructive pulmonary disease); (3) Immunologic, hematologic or allergic diseases; (4) Cardiac, hepatic, cerebral or renal impairment; (5) Anticoagulants, NSAIDs or immunosuppressive agents within the preceding month; (6) Malignancy; (7) Congenital or genetic disorders; (8) Psychiatric or neurologic disease precluding cooperation; (9) Inability to complete treatment or follow-up; (10) Polymicrobial infection; (11) Prominent extrapulmonary manifestations or complications; (12) Coagulation factor or vitamin K deficiency; (13) Patients who have already been diagnosed with a hypercoagulable state upon admission.
Clinical data of 415 children with MPP treated in our department between January 2019 and January 2025 were reviewed; 352 met the above criteria and were enrolled. The flow of patient inclusion and group assignment is shown in Figure 1. This retrospective study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Xingtai People’s Hospital. Each child was assigned a unique numeric identifier; all data were anonymized before analysis to safeguard privacy.
Figure 1.
Patient inclusion and group assignment flow chart.
Clinical Data Collection
Clinical data were retrospectively extracted from the electronic medical record system and comprised: (1) baseline characteristics—sex, age, duration of fever, peak body temperature, length of stay, and extent of radiographic involvement; (2) first available laboratory results after admission—white blood cell count (WBC), neutrophil count, CRP, lactate dehydrogenase (LDH), MP-antibody titer, and coagulation tests (PT, APTT, FIB, D-D).
Diagnostic Criteria for Hypercoagulable State and Grouping
At present, there is no diagnostic standard specifically for hypercoagulable state in children with Mycoplasma pneumoniae infection, either domestically or internationally, nor is there a universal diagnostic standard for hypercoagulable state in pediatric infectious diseases. Based on the general pathophysiology of coagulation dysfunction (shortened PT and APTT indicating enhanced clotting factor activity, elevated fibrinogen indicating increased coagulation substrate, and elevated D-dimer indicating fibrinolysis), this study adopted the following operational definition: at the first coagulation retest during hospitalization, patients meeting two or more of the following criteria were determined to have a hypercoagulable state: (1) PT shortened by >3 s; (2) APTT shortened by >10 s; (3) FIB >4 g/L; and (4) D-D >0.5 mg/L. This definition was independently reviewed by two senior pediatricians in the research team, with good inter-rater agreement (Kappa = 0.814). Patients meeting these criteria at admission were excluded to ensure a non-hypercoagulable baseline; grouping was therefore based on the first coagulation retest during hospitalization.
Serum sRAGE and sVEGFR-3 Assays
All serum samples were collected simultaneously at admission (within 3 h of admission) and at the first coagulation function retest during hospitalization (in the hypercoagulable group, at the time when hypercoagulable state was confirmed at the first retest; in the non-hypercoagulable group, at routine retest during hospitalization). Serum samples were collected during routine clinical care and archived by our department (admission and follow-up samples paired by patient identifier), then stored at −80°C until analysis. All samples were thawed for the first time for the measurements in this study, with relevant ethical standards followed throughout their use. sRAGE and sVEGFR-3 concentrations were quantified in duplicate with commercial sandwich ELISA kits purchased from Wuhan Beilaiyin Biotechnology Co., Ltd. (catalogue nos. HM10778 and HM10802, China). After equilibration to room temperature (21–26 °C), calibrators and samples were pipetted into pre-coated micro-wells, followed sequentially by biotinylated detection antibody and horseradish-peroxidase–conjugated streptavidin. Plates were incubated at 37 °C for 30 min, washed, developed with 3,3′,5,5′-tetramethylbenzidine for 10 min at 37 °C, and the reaction was terminated with stop solution. Optical density was read at 450 nm with a fluorescence microplate reader, and analyte concentrations were interpolated from four-parameter logistic standard curves. Throughout the assay, technicians remained blinded to the patients’ group allocation (hypercoagulable vs non-hypercoagulable) to eliminate observer bias.
Statistical Analysis
Analyses were performed with SPSS 26.0; figures were generated in GraphPad Prism 10.0. Continuous variables were first examined for normality. Normally distributed data are presented as mean ± SD and were compared between groups with the independent-samples t test. Non-normally distributed variables are expressed as median (P25, P75) and were analyzed with the Mann–Whitney U-test. Categorical data are reported as n (%) and were compared using the χ2-test.
Spearman correlation analysis was used to evaluate the correlations between serum sRAGE and sVEGFR-3 levels and conventional coagulation parameters at the time of confirmed hypercoagulable state in children with MPP. The Box-Tidwell test was used to assess the linearity of continuous variables; for variables showing significant nonlinearity, RCS were applied to fit their nonlinear effects, with the optimal number of knots selected based on the Akaike information criterion (AIC). LASSO regression was used for variable selection, and the optimal penalty coefficient λ was determined by 10-fold cross-validation (note: this cross-validation was used only for LASSO variable selection, not for validation of diagnostic performance), thereby constructing a set of characteristic factors with optimal performance. The selected factors were further included in logistic regression analysis to identify risk factors for hypercoagulable state in children with MPP, and the variance inflation factor (VIF) was used to diagnose multicollinearity among the identified risk factors. ROC curve analysis was used to evaluate the predictive value of admission serum sRAGE and sVEGFR-3 levels for hypercoagulable state in children with MPP. For combined prediction, a multivariate logistic regression model was constructed, and the predicted probability calculated from the regression equation was used as the combined predictive indicator; ROC curves were plotted, and the optimal cut-off value was determined according to the maximum Youden index. Internal validation of the combined predictive model was performed using the Bootstrap method (1000 resampling iterations), and the optimism-corrected AUC and its 95% confidence interval (CI) were reported. To quantify the incremental predictive value of the combined model over single indicators, the net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were evaluated. A P value <0.05 was considered statistically significant.
Results
Baseline Clinical Characteristics
A total of 352 children with MPP were enrolled (hypercoagulable group, n = 98; non-hypercoagulable group, n = 254). Duration of fever, peak temperature, LDH, and MP-antibody titer differed significantly between the two groups (P < 0.05, Table 1).
Table 1.
Comparison of Clinical Characteristics Between the Two Groups
| Variable | Hypercoagulable Group (n=98) |
Non-Hypercoagulable Group (n=254) |
t/χ2/Z | P |
|---|---|---|---|---|
| Gender/n (%) | 0.127* | 0.722 | ||
| Male | 53 (54.08) | 132 (51.97) | ||
| Female | 45 (45.92) | 122 (48.03) | ||
| Age (year) | 7.00 (6.00,8.00) | 7.00 (6.00,8.00) | −0.307ˆ | 0.759 |
| Fever duration (d)/n (%) | 4.582* | 0.032 | ||
| ≥5d | 58 (59.18) | 118 (46.46) | ||
| <5d | 40 (40.82) | 136 (53.54) | ||
| Fever peak (°C)/n (%) | 4.536* | 0.033 | ||
| ≥38.5 | 82 (83.67) | 185 (72.83) | ||
| <38.5 | 16 (16.33) | 69 (27.17) | ||
| Length of stay (d) | 7.00 (5.00,9.00) | 7.00 (5.00,8.00) | −0.152 | 0.879 |
| Extent of lesion/n (%) | 1.895* | 0.388 | ||
| Single lobe | 15 (15.31) | 50 (19.69) | ||
| Two lobes | 27 (27.55) | 79 (31.10) | ||
| Multiple lobes | 56 (57.14) | 125 (49.21) | ||
| WBC (109/L) | 9.04±2.92 | 8.87±2.92 | 0.513# | 0.608 |
| Neutrophils (%) | 62.29±14.98 | 64.29±15.08 | −1.119# | 0.264 |
| CRP (mg/L) | 20.15±6.57 | 19.86±6.41 | 0.378# | 0.706 |
| LDH (U/L) | 186.73±21.72 | 181.34±19.36 | 2.261# | 0.024 |
| MP-antibody titre/n (%) | 5.495* | 0.019 | ||
| 1:80 | 28 (28.57) | 107 (42.13) | ||
| ≥1:160 | 70 (71.43) | 147 (57.87) |
Note: *χ2 value; ˆZ value; #t value.
Abbreviations: WBC, white blood cell; CRP, C-reactive protein; LDH, lactate dehydrogenase; MP, Mycoplasma pneumoniae.
Serum sRAGE and sVEGFR-3 Concentrations
Children in the hypercoagulable group exhibited markedly higher serum levels of both sRAGE and sVEGFR-3 than those in the non-hypercoagulable group (P < 0.05, Figure 2).
Figure 2.
Serum sRAGE and sVEGFR-3 levels in the two groups. (A) Serum sRAGE levels in the hypercoagulable group and non-hypercoagulable group; (B) Serum sVEGFR-3 levels in the hypercoagulable group and non-hypercoagulable group. Data are shown as box plots with medians and interquartile ranges; whiskers indicate the range. Between-group comparisons were performed using the Mann–Whitney U-test, and the Z and P values are indicated in each panel.
Comparison of Conventional Coagulation Parameters and Their Correlation with Serum sRAGE and sVEGFR-3 at the Time of Confirmed Hypercoagulable State in Children with MPP
Compared with the non-hypercoagulable group, PT and APTT were significantly shortened, while D-D and FIB levels were significantly elevated in the hypercoagulable group (all P<0.05). Spearman correlation analysis showed that serum sRAGE levels were negatively correlated with PT (r=−0.221, P<0.001) and APTT (r=−0.201, P<0.001), and positively correlated with FIB (r=0.266, P<0.001) and D-D (r=0.362, P<0.001). Serum sVEGFR-3 levels were negatively correlated with PT (r=−0.215, P<0.001) and APTT (r=−0.220, P<0.001), and positively correlated with FIB (r=0.228, P<0.001) and D-D (r=0.276, P<0.001) (Figures 3 and 4).
Figure 3.
Comparison of traditional coagulation parameters in MPP children diagnosed with hypercoagulable state.
Notes: Coagulation abnormalities in hypercoagulable status are manifested as shortened PT and APTT, and elevated FIB and D-dimer. (A) PT; (B) APTT; (C) D-D; (D) FIB. ***P < 0.001 and ****P < 0.0001.
Figure 4.
Correlation analysis of serum sRAGE and sVEGFR-3 with traditional coagulation parameters at the diagnosis of hypercoagulable state in children with MPP. (A) correlation between sRAGE and PT; (B) correlation between sRAGE and APTT; (C) correlation between sRAGE and FIB; (D) correlation between sRAGE and DD; (E) correlation between sVEGFR-3 and PT; (F) correlation between sVEGFR-3 and APTT; (G) correlation between sVEGFR-3 and FIB; (H) correlation between sVEGFR-3 and D-D.
Risk Factors for Hypercoagulability in MPP Children
Based on previous literature evidence, age, duration of fever, LDH level, and MP antibody titer were prespecified as core confounding factors, which, together with admission serum sRAGE and sVEGFR-3, constituted the candidate variable pool. The linearity assumption of the logit was tested for the continuous variables among the above factors. Box-Tidwell test results showed that the interaction coefficients of sRAGE (interaction term β=0.002, P=0.876) and age (interaction term β=−1.286, P=0.209) were not statistically significant (both P>0.05), indicating a linear relationship between these continuous variables and the logit-transformed values, which satisfied the linearity assumption of logistic regression; both were therefore included as linear terms in subsequent analyses. sVEGFR-3 (interaction term β=−5.637, P<0.001) and LDH (interaction term β=0.619, P<0.001) showed significant nonlinear relationships. RCS analysis determined that the optimal number of knots was three for both variables, and the effect curves exhibited a nonlinear, monotonically increasing pattern. Using the median knot as the cut-off point, sVEGFR-3 and LDH were converted into binary variables (sVEGFR-3: ≤7.45 vs >7.45 ng/mL; LDH: ≤183.56 vs >183.56 U/L) for subsequent analyses. This model was designed to identify independent risk factors rather than to maximize predictive performance.
Based on the above test results, age, duration of fever, LDH level, MP antibody titer, and admission serum sRAGE and sVEGFR-3 levels were used as independent variables, and whether children with MPP developed hypercoagulable state (yes=1, no=0) was used as the dependent variable to construct a LASSO model. When λ=0.0002, the 10-fold cross-validation error was minimized, and the selected influencing factors were age (−0.008), duration of fever (−0.058), LDH level (−0.022), MP antibody titer (0.060), serum sRAGE (0.199), and sVEGFR-3 level (−0.118) (Figure 5). The above variables were included in multivariate logistic regression analysis (enter method). The results showed that duration of fever, MP antibody titer, and serum sRAGE and sVEGFR-3 levels were all independent predictors for hypercoagulable state in children with MPP (all P<0.05), and the VIF values of all independent variables were below 2, indicating no significant multicollinearity (Table 2).
Figure 5.
Variable selection by LASSO model and cross-validation. (A) LASSO coefficient profiles of the candidate variables: age (blue), LDH level (grey), fever duration (Orange), sVEGFR-3 level (purple), MP antibody titer (brown), and serum sRAGE level (green); (B) Cross-validation curve for tuning parameter selection; the red dashed line indicates the optimal λ selected by minimum cross-validation error.
Table 2.
Risk Factors for Hypercoagulability in Children with MPP
| Variable | β | S.E | Walds | P | OR | 95% CI | VIF |
|---|---|---|---|---|---|---|---|
| Age | −0.094 | 0.113 | 0.694 | 0.405 | 0.910 | 0.730–1.135 | 1.008 |
| Duration of fever | 0.699 | 0.324 | 4.660 | 0.031 | 2.011 | 1.066–3.793 | 1.025 |
| Titer of MP antibody | 1.010 | 0.343 | 8.666 | 0.003 | 2.747 | 1.402–5.383 | 1.021 |
| Admission sRAGE | 0.008 | 0.001 | 65.431 | <0.001 | 1.008 | 1.006–1.010 | 1.046 |
| Admission sVEGFR-3 (>7.45 ng/mL) | 1.553 | 0.331 | 21.967 | <0.001 | 4.728 | 2.469–9.052 | 1.040 |
| LDH (>183.56U/L) | 0.269 | 0.318 | 0.714 | 0.398 | 1.308 | 0.702–2.440 | 1.012 |
Abbreviations: LDH, lactate dehydrogenase; MP, Mycoplasma pneumoniae; sRAGE, soluble receptor for advanced glycation end products; VEGFR-3, soluble vascular endothelial growth factor receptor 3; VIF, variance inflation factor.
Predictive Performance of Admission Serum sRAGE and sVEGFR-3 for Hypercoagulability in MPP Children
To make full use of the continuous information of sVEGFR-3 to optimize the prediction power, a joint logistic prediction model based on the original continuous variables was also constructed: Logit(P) = −16.709 + 0.008 × sRAGE (pg/mL) + 0.387 × sVEGFR-3 (ng/mL). This predictive model differs from the risk-factor model in Table 2, in which sVEGFR-3 was dichotomized to satisfy the linearity assumption of logistic regression. According to the maximum Youden index, the optimal cut-off value for combined prediction was determined to be a predicted probability of P=0.390, yielding a sensitivity of 86.61% (95% CI: 81.88%–90.26%) and a specificity of 71.43% (95% CI: 61.81%–79.43%). ROC curve analysis showed that the apparent AUC of admission serum sRAGE combined with sVEGFR-3 for predicting hypercoagulable state in children with MPP was 0.870 (95% CI: 0.831–0.909). After internal validation using the Bootstrap method (1000 resampling iterations), the optimism-corrected AUC was 0.871 (95% CI: 0.863–0.875). The optimism value was extremely small, indicating a very low degree of model overfitting and good stability. The DeLong test showed that the combined predictive performance was superior to that of sRAGE alone (Z=−2.907, P=0.004) and sVEGFR-3 alone (Z=−5.929, P<0.001), demonstrating higher predictive accuracy (Table 3 and Figure 6).
Table 3.
Predictive Value of Serum sRAGE and sVEGFR-3 at Admission for Hypercoagulable State in Children with MPP
| Variable | Cut-off | AUC | Sensitivity % | Specificity % | 95% CI | P |
|---|---|---|---|---|---|---|
| Admission sRAGE | 1523.00 pg/mL | 0.840 | 64.57 | 87.76 | 0.794–0.887 | <0.001 |
| Admission sVEGFR-3 | 6.905 ng/mL | 0.709 | 50.39 | 88.78 | 0.656–0.761 | <0.001 |
| Admission sRAGE+sVEGFR-3 | 0.390 | 0.870 | 86.61 | 71.43 | 0.831–0.909 | <0.001 |
Notes: The cut-off of 0.390 for the combined model refers to the predicted probability from the logistic regression equation, determined by the maximum Youden index.
Abbreviations: sRAGE, soluble receptor for advanced glycation end products; sVEGFR-3, soluble vascular endothelial growth factor receptor 3.
Figure 6.
ROC curves of admission serum sRAGE and sVEGFR-3 for predicting hypercoagulability in MPP children.
To evaluate the incremental predictive value of the combined model over single indicators, the NRI and IDI were further calculated. The results showed that, compared with the sRAGE-alone model, the combined model yielded a continuous NRI of 0.497 (P<0.001) and an IDI of 0.048 (95% CI: 0.028–0.068, P<0.001); compared with the sVEGFR-3-alone model, the continuous NRI was 0.972 (P<0.001) and the IDI was 0.283 (95% CI: 0.231–0.337, P<0.001), both demonstrating that the combined model was significantly superior to single indicators (all P<0.05).
Discussion
Currently, the diagnosis of hypercoagulability in children with MPP still relies on conventional coagulation indices such as D-D, fibrinogen, and APTT. These parameters exhibit limited sensitivity and specificity in the MPP setting, frequently yielding false-positive or false-negative results. Consequently, the identification of novel, more reliable biomarkers for the early detection of hypercoagulability in MPP children has become a major focus of current research.
Hypercoagulability denotes an increased propensity for thrombus formation due to enhanced clotting-factor activity, reduced anticoagulant capacity, or impaired fibrinolysis.21 In MPP, the pathogen triggers a systemic inflammatory response syndrome (SIRS) with abundant cytokine release that activates the coagulation cascade while simultaneously suppressing physiological anticoagulant and fibrinolytic pathways, thereby disrupting hemostatic balance. Moreover, MPP may directly injure the vascular endothelium, exposing subendothelial pro-coagulant constituents and further facilitating thrombus generation.6,8,22
Previous work has shown that ligation of mRAGE triggers intracellular signaling cascades that up-regulate inflammatory cytokines, adhesion molecules and pro-coagulant factors, thereby integrating inflammatory, oxidative and coagulation responses.23 Membrane-bound VEGFR-3 has likewise been implicated in modulating endothelial phenotype and in regulating the synthesis and release of coagulation proteins, positioning it as an additional contributor to hemostatic control.24 Notably, the sVEGFR-3 detected in this study is the circulating soluble form, a shed product of the extracellular domain of membrane-bound VEGFR-3, and its elevated levels may reflect the degree of endothelial cell activation or injury. In the present study we examined the serum expression of sRAGE and sVEGFR-3 in MPP children with hypercoagulability and evaluated their predictive utility. Compared with the non-hypercoagulable group, children in the hypercoagulable group displayed significantly higher serum concentrations of both sRAGE and sVEGFR-3 (P < 0.05), suggesting that sRAGE and sVEGFR-3 levels are associated with hypercoagulable state in children with MPP and that both may serve as serum markers reflecting endothelial injury and inflammation–coagulation co-activation.
D-D, a fibrin-degradation product, increases whenever intravascular coagulation and subsequent fibrinolysis occur, and is therefore regarded as a sensitive marker of both thrombotic burden and fibrinolytic activity.25 FIB constitutes the principal substrate for thrombin; elevated concentrations enhance blood clotting capacity.26 The APTT mirrors the integrity of the intrinsic pathway, and its shortening often reflects heightened clotting-factor activity.27 PT reflects the function of the extrinsic coagulation pathway, and its shortening suggests activation of the extrinsic coagulation system and a hypercoagulable state preceding coagulation factor consumption.28 Spearman correlation analysis showed that serum sRAGE and sVEGFR-3 were significantly negatively correlated with PT and APTT, and positively correlated with FIB and D-D levels (P < 0.05). These associations indicate that variations in sRAGE and sVEGFR-3 parallel changes in conventional coagulation indices, further supporting that both markers are associated with the hypercoagulable state observed in children with MPP. Although the correlations between sRAGE, sVEGFR-3, and conventional coagulation parameters were statistically significant, the absolute values of the Spearman correlation coefficients were all within the weak-to-moderate range of 0.20–0.36, suggesting that these two markers explain only a small proportion of the variation in coagulation parameters. This result is consistent with the general rule of biomarker research: a single molecular marker usually cannot fully reflect the complex coagulation cascade. Therefore, this study emphasizes the value of combined application of sRAGE and sVEGFR-3 rather than their use as standalone replacements for conventional coagulation parameters; their clinical significance lies in serving as a complementary tool for early risk stratification rather than as an independent basis for diagnosis.
Multivariable logistic regression identified duration of fever, MP-antibody titer, and serum levels of sRAGE and sVEGFR-3 as independent risk factors for hypercoagulability in children with MPP. Prolonged fever and higher peak temperature mirror the intensity of systemic inflammation,29 and elevated LDH suggests heightened tissue injury and inflammatory activity,30 however, LDH did not retain independent significance in the final model, and peak temperature was not among the pre-specified candidate variables, suggesting that their apparent associations with hypercoagulability were largely accounted for by the duration of fever and the admission biomarkers rather than providing independent predictive information. Rising MP-antibody titers reflect the severity of MPP infection itself.31 These associated factors suggest that sRAGE and sVEGFR-3 could be incorporated, together with conventional clinical indicators (such as duration of fever and MP antibody titer), into a risk stratification framework for hypercoagulability in children with MPP, providing a reference for early identification of high-risk populations. Causal inference cannot be drawn from this observational design, and these associations do not imply that sRAGE or sVEGFR-3 actively participate in the prothrombotic process; they may merely serve as markers reflective of endothelial injury or inflammation–coagulation co-activation. Furthermore, it should be emphasized that this is an observational study that did not evaluate the effect of any intervention, and the associations of the above factors with hypercoagulable state do not imply that treatments targeting these factors can directly reduce the risk of hypercoagulability. The feasibility of using sRAGE and sVEGFR-3 to guide clinical decision-making, and whether intervention strategies based on these markers can improve patient outcomes, require validation in prospective interventional studies.
ROC curve analysis showed that the AUC of combined serum sRAGE and sVEGFR-3 for predicting hypercoagulable state in children with MPP was 0.870, higher than that of either sRAGE or sVEGFR-3 alone, demonstrating relatively high discriminative performance. NRI and IDI analyses further confirmed that the combined model had significant incremental predictive value over single indicators. However, given the single-center retrospective design of this study, the above results only suggest that combined detection of sRAGE and sVEGFR-3 has the potential to serve as an exploratory biomarker combination for early identification of hypercoagulable state in children with MPP; its predictive thresholds, generalizability, and clinical utility still require prospective longitudinal studies and independent multicenter validation, and it is currently insufficient as a direct basis for clinical decision-making.
Although this study has achieved certain results, several limitations remain: (1) Selection bias: the enrolled subjects were children with moderate MPP hospitalized in the pediatric inpatient department, excluding mild outpatient cases and severe intensive care unit (ICU) patients, which limits the generalizability of the conclusions; the potential impact of long-term sample storage conditions on marker stability also cannot be completely excluded. (2) Lead-time bias: serum measurements were based on only two time points—admission and the first retest—failing to capture the dynamic evolution during the disease course, and it cannot be determined whether the elevation of the markers is a “precursor” or an “accompanying phenomenon” of hypercoagulable state. (3) Information bias: clinical data were retrospectively extracted from electronic medical records, the duration of fever relied on subjective reports, imaging interpretations were not blindly reviewed, and the diagnostic criteria for hypercoagulable state were based on an operational definition grounded in general coagulation pathophysiology rather than internationally recognized standards specific to pediatric MPP, all of which may introduce misclassification bias. (4) Limitations in causal inference: Given the retrospective observational design, no causal relationship can be inferred between sRAGE/sVEGFR-3 and hypercoagulable state; the observed associations may reflect epiphenomena, reverse causality, or unmeasured confounding. The causal direction between sRAGE, sVEGFR-3, and hypercoagulable state needs to be clarified by prospective cohort studies and in vitro experiments. (5) Detection and translational limitations: the ELISA method is time-consuming and costly, cellular and molecular mechanisms were not explored in depth, and the clinical translational value requires large-scale prospective validation.
In summary, this study found that elevated serum sRAGE and sVEGFR-3 levels are statistically associated with hypercoagulable state in children with MPP, and their combined detection demonstrated certain predictive discriminative performance (AUC=0.870), showing potential as exploratory auxiliary reference indicators for risk stratification of hypercoagulable state in children with MPP. However, these findings are hypothesis-generating results only, and the generalizability of the diagnostic thresholds, early warning value, and feasibility of guiding clinical decision-making require further validation in multicenter prospective studies and interventional trials.
Funding Statement
This work was supported by the Medical Science Research Project of Hebei (Grant No. 20261027).
Data Sharing Statement
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.
Ethical Statement
This study involving human participants was conducted in accordance with the Declaration of Helsinki and its subsequent revisions, with approval granted by the Xingtai People’s Hospital (Approval Number: 2025132). Given the retrospective nature of the research utilizing archived serum samples routinely collected during clinical care, informed consent was waived. All data collected were handled with strict confidentiality to ensure the anonymity of the participants.
Disclosure
No potential conflict of interest was reported by the authors.
References
- 1.Zhao X, Lv J, Wu M, Wu Q. Clinical characteristics and risk factors for Mycoplasma pneumoniae pneumonia in children. Front Pediatr. 2024;12:1438631. doi: 10.3389/fped.2024.1438631 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Li YT, Ma YM, Dai FL, et al. Comparison of thromboelastographic profiles and clinical characteristics in children with severe Mycoplasma pneumoniae pneumonia. Front Pediatr. 2025;13:1459455. doi: 10.3389/fped.2025.1459455 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Song S, Xu Y. A retrospective study of the clinical characteristics of 9 children with pulmonary embolism associated with Mycoplasma pneumoniae pneumonia. BMC Pediatr. 2023;23(1):370. doi: 10.1186/s12887-023-04188-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Gu H, Li B, Han Y, Yang S, Wang X. Risk factors for suspected pulmonary embolism in children: complication of Mycoplasma pneumoniae pneumonia. Eur J Radiol. 2024;176:111474. doi: 10.1016/j.ejrad.2024.111474 [DOI] [PubMed] [Google Scholar]
- 5.Liu J, He R, Wu R, et al. Mycoplasma pneumoniae pneumonia associated thrombosis at Beijing Children’s hospital. BMC Infect Dis. 2020;20(1):51. doi: 10.1186/s12879-020-4774-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wang J, Mao J, Chen G, et al. Evaluation on blood coagulation and C-reactive protein level among children with Mycoplasma pneumoniae pneumonia by different chest imaging findings. Medicine. 2021;100(3):e23926. doi: 10.1097/md.0000000000023926 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Li YT, Zhang J, Wang MZ, et al. Changes in coagulation markers in children with Mycoplasma pneumoniae pneumonia and their predictive value for Mycoplasma severity. Ital J Pediatr. 2023;49(1):143. doi: 10.1186/s13052-023-01545-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Zhou F, Yang S, Zhang W, Huang M. Assessment of serum parameters caused by the outbreak of Mycoplasma pneumoniae pneumonia in children after COVID-19. Sci Rep. 2025;15(1):28306. doi: 10.1038/s41598-025-13555-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Borgel D, Bianchini E, Lasne D, Pascreau T, Saller F. Inflammation in deep vein thrombosis: a therapeutic target? Hematology. 2019;24(1):742–13. doi: 10.1080/16078454.2019.1687144 [DOI] [PubMed] [Google Scholar]
- 10.Bautista-Becerril B, Campi-Caballero R, Sevilla-Fuentes S, et al. Immunothrombosis in COVID-19: implications of neutrophil extracellular traps. Biomolecules. 2021;11(5):694. doi: 10.3390/biom11050694 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Serveaux-Dancer M, Jabaudon M, Creveaux I, et al. Pathological implications of receptor for advanced glycation end-product (ager) gene polymorphism. Dis Markers. 2019;2019:2067353. doi: 10.1155/2019/2067353 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Wu F, Afanasyeva Y, Zeleniuch-Jacquotte A, Zhang J, Schmidt AM, Chen Y. Temporal reliability of serum soluble and endogenous secretory receptors for advanced glycation end-products (sRAGE and esRAGE) in healthy women. Cancer Causes Control. 2018;29(10):901–905. doi: 10.1007/s10552-018-1066-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Degani G, Altomare A, Digiovanni S, et al. Prothrombin is a binding partner of the human receptor of advanced glycation end products. J Biol Chem. 2020;295(35):12498–12511. doi: 10.1074/jbc.RA120.013692 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Fan H, Tang HB, Chen Z, et al. Inhibiting HMGB1-RAGE axis prevents pro-inflammatory macrophages/microglia polarization and affords neuroprotection after spinal cord injury. J Neuroinflammation. 2020;17(1):295. doi: 10.1186/s12974-020-01973-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Seizer P, von Ungern-Sternberg SNI, Haug V, et al. Cyclophilin A is a ligand for RAGE in thrombo-inflammation. Cardiovasc Res. 2024;120(4):385–402. doi: 10.1093/cvr/cvad189 [DOI] [PubMed] [Google Scholar]
- 16.Heinolainen K, Karaman S, D’Amico G, et al. VEGFR3 modulates vascular permeability by controlling VEGF/VEGFR2 signaling. Circ Res. 2017;120(9):1414–1425. doi: 10.1161/circresaha.116.310477 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Schmidt A, Hrupka B, van Bebber F, et al. The Alzheimer’s disease-linked protease BACE2 cleaves VEGFR3 and modulates its signaling. J Clin Invest. 2024;134(16):e170550. doi: 10.1172/jci170550 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Choi IS, Byeon JH, Yoo Y, Lee KC, Choung JT. Increased serum interleukin-5 and vascular endothelial growth factor in children with acute mycoplasma pneumonia and wheeze. Pediatr Pulmonol. 2009;44(5):423–428. doi: 10.1002/ppul.20961 [DOI] [PubMed] [Google Scholar]
- 19.Lim L, Bui H, Farrelly O, et al. Hemostasis stimulates lymphangiogenesis through release and activation of VEGFC. Blood. 2019;134(20):1764–1775. doi: 10.1182/blood.2019001736 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Evidence-based guideline for the diagnosis and treatment of Mycoplasma pneumoniae pneumonia in children (2023). Pediatr Investig. 2025;9(1):1–11. doi: 10.1002/ped4.12469 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Samuels JM, Coleman JR, Moore EE, et al. Alternative complement pathway activation provokes a hypercoagulable state with diminished fibrinolysis. Shock. 2020;53(5):560–565. doi: 10.1097/shk.0000000000001437 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Sikora JP, Karawani J, Sobczak J. Neutrophils and the Systemic Inflammatory Response Syndrome (SIRS). Int J Mol Sci. 2023;24(17):13469. doi: 10.3390/ijms241713469 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Fuhr JC, Ramos MEK, Piovesan F, Renner LO, Siqueira LO. Relationship of advanced glycation end-products in hypertension in diabetic patients: a systematic review. J Bras Nefrol. 2022;44(4):557–572. doi: 10.1590/2175-8239-JBN-2022-0006en [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Jerafi-Vider A, Bassi I, Moshe N, et al. VEGFC/FLT4-induced cell-cycle arrest mediates sprouting and differentiation of venous and lymphatic endothelial cells. Cell Rep. 2021;35(11):109255. doi: 10.1016/j.celrep.2021.109255 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Johnson ED, Schell JC, Rodgers GM. The D-dimer assay. Am J Hematol. 2019;94(7):833–839. doi: 10.1002/ajh.25482 [DOI] [PubMed] [Google Scholar]
- 26.Wolberg AS. Fibrinogen and fibrin: synthesis, structure, and function in health and disease. J Thromb Haemost. 2023;21(11):3005–3015. doi: 10.1016/j.jtha.2023.08.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Santoro RC, Molinari AC, Leotta M, Martini T. Isolated prolongation of activated partial thromboplastin time: not just bleeding risk! Medicina. 2023;59(6):1169. doi: 10.3390/medicina59061169 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Kumar A, GS R, Kuppusami BR, Ahmad S. Evaluation of Coagulation Profile Alterations in Patients Undergoing Laparoscopic Cholecystectomy With Carbon Dioxide (CO2) Pneumoperitoneum. Cureus. 2025;17(10):e94160. doi: 10.7759/cureus.94160 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Zhang C, Zhang Q, Du JL, et al. Correlation between the clinical severity, bacterial load, and inflammatory reaction in children with Mycoplasma pneumoniae pneumonia. Curr Med Sci. 2020;40(5):822–828. doi: 10.1007/s11596-020-2261-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wang LP, Hu ZH, Jiang JS, Jin J. Serum inflammatory markers in children with Mycoplasma pneumoniae pneumonia and their predictive value for mycoplasma severity. World J Clin Cases. 2024;12(22):4940–4946. doi: 10.12998/wjcc.v12.i22.4940 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Choo S, Kim SH, Lee E. Clinical significance of Mycoplasma pneumoniae specific IgM titer in children hospitalized with Mycoplasma pneumoniae pneumonia. BMC Infect Dis. 2022;22(1):470. doi: 10.1186/s12879-022-07456-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.