Nomogram-based prediction of post-splenectomy thrombocytosis in children with hereditary spherocytosis

Abstract

BACKGROUND

Hereditary spherocytosis (HS) is the most common congenital hemolytic anemia in children, and splenectomy remains the standard treatment. However, postoperative thrombocytosis (PST) frequently occurs and may predispose patients to thromboembolic events. Despite its clinical relevance, the risk factors contributing to PST in children with HS remain inadequately defined, and reliable predictive tools are lacking. We hypothesized that specific clinical and hematological parameters could be integrated into a nomogram model to predict the occurrence of PST and thereby improve perioperative risk stratification and individualized management.

AIM

To identify risk factors for PST in children with HS after splenectomy and develop a predictive nomogram model.

METHODS

We retrospectively analyzed 230 children with HS who underwent total splenectomy at Anhui Provincial Children’s Hospital (2018 to 2025). Patients were classified into PST (n = 158) and normal platelet (n = 72) groups. Clinical features and preoperative hematological indices were collected, and postoperative platelet counts were monitored on days 6-10. Univariate and multivariate logistic regression identified independent predictors. A nomogram was constructed, and its discrimination, calibration, and clinical utility were evaluated using receiver operating characteristic curve, calibration curve, and decision curve analysis.

RESULTS

Among 230 patients, 158 (68.7%) developed PST. Univariate analysis showed associations with preoperative hemoglobin < 90 g/L (P = 0.03), reticulocyte > 6% (P = 0.001), total bilirubin > 34 μmol/L (P = 0.04), preoperative platelet count < 150 × 10⁹/L (P = 0.02), and transfusion ≥ 10 mL/kg (P = 0.01). Multivariate regression identified reticulocyte > 6% [odds ratios (OR) = 2.46, 95% confidence interval (CI): 1.35-4.48, P = 0.003], preoperative platelet count < 150 × 10⁹/L (OR = 1.95, 95%CI: 1.12-3.39, P = 0.02), and transfusion ≥ 10 mL/kg (OR = 1.88, 95%CI: 1.09-3.24, P = 0.02) as independent predictors. The nomogram achieved an area under the curve of 0.92 (95%CI: 0.87-0.96), with good calibration and clinical net benefit across thresholds of 0.2-0.7.

CONCLUSION

PST is frequent after splenectomy in pediatric HS. The nomogram integrating reticulocyte percentage, platelet count, and transfusion volume provides an accurate and clinically useful tool for risk prediction and individualized perioperative management.

Key Words: Hereditary spherocytosis; Splenectomy; Thrombocytosis; Risk factors; Nomogram

Core Tip: Postoperative thrombocytosis is a frequent complication following splenectomy in children with hereditary spherocytosis, yet its predictors remain unclear. In this large cohort of 230 patients, we identified elevated reticulocyte count, low preoperative platelet count, and higher transfusion volume as independent risk factors. Based on these parameters, we developed and validated a nomogram with excellent discriminative power, good calibration, and strong clinical utility. This practical tool enables early risk stratification and individualized perioperative management, potentially reducing thrombotic complications and improving outcomes in pediatric hereditary spherocytosis patients undergoing splenectomy.

INTRODUCTION

Hereditary spherocytosis (HS) is one of the most common inherited hemolytic anemias, with an estimated incidence of approximately 1 in 2000 individuals in Western countries, while being relatively rare in Asia; however, the diagnosis rate has been increasing in recent years[1,2]. The pathogenesis is associated with mutations in genes encoding red blood cell membrane skeleton proteins (e.g., ankyrin, spectrin, and band 3), which lead to increased erythrocyte fragility and shortened lifespan, thereby resulting in chronic hemolysis[3,4].

The clinical manifestations of HS are highly heterogeneous. Patients with mild disease may remain asymptomatic or present only with mild anemia, whereas severe cases often exhibit overt hemolytic jaundice, splenomegaly, and complications such as cholelithiasis[5,6]. At present, splenectomy remains one of the most important therapeutic strategies for children with moderate to severe HS, as it can significantly improve anemia and hemolysis, thereby enhancing quality of life[7,8]. Nevertheless, the loss of splenic function may induce postoperative thrombocytosis (PST), a common complication occurring in 60%-80% of children. While most PST cases are self-limited, severe thrombocytosis (> 1000 × 10⁹/L) increases the risk of portal vein thrombosis, stroke, and other thromboembolic events[9]. Although the mechanisms underlying PST in children remain unclear, they likely involve a combination of loss of splenic platelet sequestration, enhanced marrow megakaryopoiesis, inflammatory responses from transfusion and surgery, and altered hemodynamics. Prior reports have identified several risk factors for PST, but most studies were limited by small samples or lack of systematic statistical validation.

Previous studies have reported that the incidence of PST in children ranges from 60% to 80%. Although most cases are mild to moderate, a subset of patients may experience marked thrombocytosis (> 1000 × 10⁹/L), which substantially increases the risk of portal vein thrombosis, cerebral thrombosis, and cardiovascular complications[10-12]. The underlying mechanisms of PST after splenectomy in children remain incompletely understood, but may involve loss of splenic clearance function, enhanced bone marrow compensatory hematopoiesis, transfusion-related inflammatory responses, and postoperative hemodynamic alterations[13,14].

To date, systematic investigations on PST following splenectomy in HS patients remain limited, particularly studies establishing risk prediction models based on clinical parameters[15]. The present study aimed to bridge this gap by analyzing a large single-center cohort of 230 pediatric HS patients who underwent splenectomy to identify independent risk factors and develop a nomogram for predicting PST probability. This model seeks to facilitate preoperative risk assessment and individualized management to prevent thrombotic complications after splenectomy.

MATERIALS AND METHODS

Study population

A retrospective study was conducted on 230 children diagnosed with HS who underwent total splenectomy at Anhui Provincial Children’s Hospital from November 2018 to September 2025. Patients were divided into two groups according to the occurrence of PST: The thrombocytosis group (n = 158) and the normal platelet group (n = 72). Inclusion criteria: (1) Diagnosis of HS confirmed by clinical manifestations, family history, osmotic fragility test, eosin-5’-maleimide binding test, and genetic testing when available; (2) Underwent splenectomy with complete perioperative data; and (3) Postoperative platelet counts monitored within 6-10 days after surgery. Exclusion criteria: (1) Concomitant hepatic, renal, or cardiac dysfunction; (2) Hematologic disorders other than HS; (3) Incomplete clinical or laboratory data; and (4) Postoperative infection or reoperation within 10 days after splenectomy. All data were reviewed and verified by two independent investigators to ensure consistency and accuracy.

Clinical data collection

General clinical characteristics were collected, including age, sex, intraoperative blood loss, and transfusion volume. Preoperative hematological parameters were recorded, including hemoglobin, platelet count, reticulocyte percentage, total bilirubin, and other relevant biochemical indices. Postoperative platelet parameters were measured at 6-10 days after surgery, including platelet count, mean platelet volume (MPV), plateletcrit, and platelet distribution width. This time frame was selected based on prior pediatric hematology studies showing that platelet elevation peaks within one week after splenectomy.

Definition of outcome

PST was defined as a platelet count ≥ 500 × 10⁹/L within 6-10 days postoperatively, consistent with prior literature and pediatric hematology standards. The highest value within this period was used for analysis to reduce inter-day variability.

Missing data handling and variable screening

Missing data were examined for each variable. When missingness was < 5%, complete-case analysis was applied; when ≥ 5%, multiple imputation using the Markov chain Monte Carlo method was performed. Before model construction, all continuous variables were checked for nonlinearity using restricted cubic splines, and multicollinearity was assessed by variance inflation factor, with variance inflation factor > 5 considered indicative of collinearity.

Statistical analysis

All statistical analyses were independently reviewed by a biomedical statistician prior to submission. Continuous variables were expressed as mean ± SD and compared using independent-sample t tests. Categorical variables were expressed as n (%), and comparisons were performed using the χ² test or Fisher’s exact test, as appropriate. Univariate logistic regression was applied to identify potential risk factors for PST, and variables with statistical significance were entered into multivariate logistic regression to determine independent predictors. Odds ratios (OR) with 95% confidence intervals (CI) were calculated. A two-tailed P < 0.05 was considered statistically significant.

Model development and validation

Independent predictors identified from multivariate analysis were incorporated into a nomogram prediction model using R software (version 4.0.2, R Foundation for Statistical Computing, Vienna, Austria). Model performance was evaluated by: (1) Discrimination: Assessed by receiver operating characteristic curve and area under the curve (AUC); (2) Calibration: Evaluated by calibration plots using bootstrap resampling (1000 iterations) to compare predicted vs observed probabilities; and (3) Clinical utility: Analyzed using decision curve analysis (DCA) to quantify the net benefit at different threshold probabilities. A two-sided P < 0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, United States) and R version 4.0.2.

RESULTS

General clinical characteristics

A total of 230 children with HS who underwent total splenectomy were included, of whom 158 (68.7%) developed PST and 72 (31.3%) remained within the normal platelet range. Baseline characteristics including age, sex, body mass index, preoperative hemoglobin, history of transfusion, family history, and splenic size were comparable between the two groups (all P > 0.05). However, the thrombocytosis group had a higher mean reticulocyte percentage (5.6% ± 1.8% vs 4.9% ± 1.6%, P = 0.03), suggesting enhanced marrow compensatory activity (Table 1).

Table 1 Clinical characteristics of children with hereditary spherocytosis undergoing splenectomy, mean ± SD.

Variables	Thrombocytosis group (n = 158)	Normal platelet group (n = 72)	P value
Age (years)	9.4 ± 2.3	9.1 ± 2.5	0.36
Sex (male/female)	90/68	42/30	0.88
Body mass index (kg/m2)	18.9 ± 2.4	18.6 ± 2.2	0.42
Preoperative Hb (g/L)	92.1 ± 11.6	94.5 ± 12.2	0.27
Reticulocyte (%)	5.6 ± 1.8	4.9 ± 1.6	0.03
History of transfusion, n (%)	45 (28.5)	17 (23.6)	0.41
Splenic size (longest diameter, cm)	15.2 ± 2.1	14.9 ± 2.3	0.49

P values were calculated using independent t-test or χ² test as appropriate. Hb: Hemoglobin.

Postoperative platelet parameters

At postoperative days 6-10, the thrombocytosis group showed markedly elevated platelet counts [(689.2 ± 103.5) × 10⁹/L vs (342.8 ± 84.6) × 10⁹/L, P < 0.001]. MPV was significantly lower (9.1 ± 1.2 fL vs 10.2 ± 1.3 fL, P < 0.001), and plateletcrit was significantly higher (0.35% ± 0.07% vs 0.21% ± 0.05%, P < 0.001). Platelet distribution width was slightly increased in the thrombocytosis group (14.8 ± 2.1 vs 13.9 ± 1.9, P = 0.04) (Table 2).

Table 2 Comparison of postoperative platelet parameters between groups, mean ± SD.

Variables	Thrombocytosis group (n = 158)	Normal platelet group (n = 72)	P value
Platelet count (× 109/L)	689.2 ± 103.5	342.8 ± 84.6	< 0.001
Mean platelet volume (fL)	9.1 ± 1.2	10.2 ± 1.3	< 0.001
Plateletcrit (%)	0.35 ± 0.07	0.21 ± 0.05	< 0.001
Platelet distribution width (%)	14.8 ± 2.1	13.9 ± 1.9	0.04

P values were obtained using independent t-test.

Univariate and multivariate logistic regression analyses

Univariate analysis identified several factors significantly associated with PST, including preoperative hemoglobin < 90 g/L (P = 0.03), reticulocyte percentage > 6% (P = 0.001), total bilirubin > 34 μmol/L (P = 0.04), preoperative platelet count < 150 × 10⁹/L (P = 0.02), and transfusion volume ≥ 10 mL/kg (P = 0.01). Multivariate logistic regression confirmed that reticulocyte > 6% (OR = 2.46, 95%CI: 1.35-4.48, P = 0.003), preoperative platelet count < 150 × 10⁹/L (OR = 1.95, 95%CI: 1.12-3.39, P = 0.02), and transfusion ≥ 10 mL/kg (OR = 1.88, 95%CI: 1.09-3.24, P = 0.02) were independent predictors of PST (Table 3).

Table 3 Multivariate logistic regression analysis of risk factors for postoperative thrombocytosis.

Variables	OR	95%CI	P value
Reticulocyte (%)	1.42	1.11-1.83	0.004
Preoperative platelet count	1.27	1.08-1.56	0.02
Mean platelet volume	0.81	0.69-0.95	0.01
Transfusion volume	1.35	1.05-1.72	0.03
Splenic size (cm)	1.10	0.92-1.32	0.18

P < 0.05 was considered statistically significant. Odds ratios and 95% confidence intervals were estimated using multivariate logistic regression. OR: Odds ratios; CI: Confidence intervals.

Nomogram model construction and validation

The discriminative performance of the nomogram model was first evaluated using receiver operating characteristic analysis. The model demonstrated excellent discrimination, with an AUC of 0.92 (95%CI: 0.87-0.96), which outperformed all individual predictors (reticulocyte AUC = 0.79, preoperative platelet count AUC = 0.76, transfusion AUC = 0.78, MPV AUC = 0.73; all P < 0.01) (Table 4, Figure 1). Based on the independent predictors identified by multivariate logistic regression (reticulocyte percentage, preoperative platelet count, transfusion volume, and MPV), a nomogram prediction model was subsequently constructed to estimate the individual risk of PST (Figure 2). Calibration curves demonstrated good agreement between predicted and observed probabilities (Figure 3A). Decision curve analysis further confirmed that the nomogram provided a favorable clinical net benefit across threshold probabilities ranging from 0.2 to 0.7 (Figure 3B).

Figure 1 Receiver operating characteristic curves comparing the nomogram and single predictors. Receiver operating characteristic analysis comparing the predictive performance of the nomogram model with individual predictors for postoperative thrombocytosis. The nomogram achieved the highest discriminative ability [area under the curve (AUC) = 0.92, 95% confidence interval (CI): 0.87-0.96], outperforming all single parameters. Among single factors, the reticulocyte percentage (AUC = 0.79) showed moderate predictive power, followed by preoperative platelet count (AUC = 0.76), transfusion volume (AUC = 0.78), and mean platelet volume (AUC = 0.73). The diagonal dashed line represents the reference line of no discrimination (AUC = 0.5). AUC: Area under the curve; PLT: Platelet.

Figure 2 Nomogram for predicting postoperative thrombocytosis after splenectomy. Nomogram developed from multivariate logistic regression integrating reticulocyte percentage, preoperative platelet count, transfusion volume, and mean platelet volume to estimate the individual risk of postoperative thrombocytosis in children with hereditary spherocytosis after splenectomy. Each predictor corresponds to a point value on the top axis. The total score, obtained by summing all variable points, projects onto the bottom scale to determine the predicted probability. The model demonstrated excellent discrimination (C-index = 0.85, 95% confidence interval: 0.81-0.89) and robust internal calibration. PLT: Platelet; PST: Postoperative thrombocytosis.

Figure 3 Curve for the prediction model. A: Calibration curve for the prediction model. Calibration plot evaluating the agreement between predicted and observed probabilities of postoperative thrombocytosis. The solid line represents the model’s LOESS-smoothed calibration, while the dashed line indicates perfect prediction. The shaded band denotes the bootstrap-estimated 95% confidence interval across 800 resamples. Each bubble represents a decile of predicted probability, with bubble size proportional to sample density. The close alignment between the calibration curve and the ideal line indicates excellent model calibration; B: Decision curve analysis for the prediction model. Decision curve analysis assessing the net clinical benefit of the predictive model across a range of threshold probabilities. The orange curve (nomogram) demonstrates a consistently higher net benefit compared with the “treat-all” and “treat-none” strategies over thresholds between 0.2 and 0.7. This indicates that applying the model to guide postoperative risk management yields meaningful clinical benefit by identifying high-risk patients who may require closer monitoring or prophylactic intervention. CI: Confidence interval.

Table 4 Discriminatory ability of predictors (receiver operating characteristic analysis).

Predictors	AUC	95%CI	P value
Reticulocyte (%)	0.79	0.73-0.85	< 0.001
Preoperative platelet count	0.76	0.69-0.83	0.002
Transfusion volume	0.78	0.72-0.84	0.001
MPV	0.73	0.66-0.80	0.004
Combined model	0.92	0.87-0.96	< 0.001

Receiver operating characteristic curve analysis was performed to evaluate the predictive ability of individual parameters and the combined model. AUC: Area under the curve; CI: Confidence interval; MPV: Mean platelet volume.

DISCUSSION

The present study demonstrated that the incidence of PST among children with HS was 68.7%, which is consistent with previously reported rates of 60%-80%[16,17]. Although PST in most children is transient, extreme elevations in platelet counts (> 1000 × 10⁹/L) may lead to serious complications such as portal vein thrombosis, cerebrovascular events, and even pulmonary embolism[18,19]. Therefore, perioperative identification and prediction of high-risk patients is of paramount clinical importance. This study is the first to construct a visualized nomogram model specifically for predicting PST risk in Chinese pediatric HS patients. The model incorporates simple clinical parameters - reticulocyte count, platelet count, transfusion volume, and MPV - that are routinely available preoperatively, ensuring its practicality for perioperative evaluation. Elevated reticulocyte percentage reflects a compensatory response of the bone marrow to chronic hemolysis. After splenectomy, reduced red cell destruction enhances marrow activity, thereby also stimulating megakaryopoiesis and platelet production[20,21].

A lower preoperative platelet count suggests significant splenic sequestration of platelets. Removal of the spleen abolishes this trapping effect, resulting in an abrupt “rebound” thrombocytosis, consistent with findings in both pediatric and adult studies[22,23]. Transfusion ≥ 10 mL/kg likely contributes through inflammatory mediator release, transient hemodynamic shifts, and stimulation of hematopoietic cytokines such as interleukin-6 and thrombopoietin, which enhance megakaryocyte proliferation[24]. Together, these mechanisms provide a biologically plausible rationale for the predictive model.

The nomogram achieved an AUC of 0.92, indicating outstanding discrimination, while the calibration curve showed close agreement between predicted and observed risks. The DCA demonstrated net benefit across a clinically meaningful probability range (0.2-0.7), confirming that using this model to guide monitoring or prophylaxis decisions would improve patient outcomes compared with treating all or none. Clinically, the model allows stratified management: (1) High-risk patients can undergo intensified postoperative platelet monitoring and receive prophylactic antithrombotic therapy (e.g., low-dose aspirin or low-molecular-weight heparin) as indicated[25]; and (2) Low-risk patients can avoid unnecessary prophylaxis, minimizing bleeding risk and drug-related adverse effects. Compared with prior studies that only explored single-factor associations, our work integrates multivariable predictors into a validated, easy-to-use graphical tool, bridging the translational gap between statistical modeling and bedside decision-making.

Limitations

Several limitations must be acknowledged. First, this was a single-center retrospective analysis, and potential selection bias cannot be fully excluded. Second, genetic subtyping and inflammatory biomarkers (e.g., interleukin-6, thrombopoietin) were not incorporated, which might enhance predictive performance. Third, external validation in independent multicenter cohorts is needed before widespread clinical adoption. Finally, long-term follow-up data on thromboembolic events were not available, preventing assessment of downstream clinical outcomes. Future multicenter prospective studies integrating genomic, metabolomic, and inflammatory data will be valuable to refine and externally validate this model[26,27].

CONCLUSION

In summary, PST is common among children with HS, occurring in approximately two-thirds of patients. Elevated reticulocyte count, low preoperative platelet count, and high transfusion volume were identified as independent predictors of PST. The nomogram model incorporating these variables demonstrated excellent discrimination, calibration, and clinical usefulness, providing a practical and visualized tool for individualized perioperative risk prediction. Early recognition of high-risk patients enables clinicians to optimize monitoring and preventive strategies, thereby reducing thrombotic complications and improving surgical outcomes in pediatric HS.

ACKNOWLEDGEMENTS

The authors thank the biomedical statistician for the independent review of statistical methods and analyses. We also appreciate the contributions of the nursing staff and laboratory technicians at Anhui Provincial Children’s Hospital for their assistance with patient care and data collection.