Development and Validation of a prediction model for severe PES during the treatment of unrelated UCBT in pediatric patients

Introduction

Umbilical cord blood transplantation (UCBT) is readily available and conveniently accessible, demonstrating low immunogenicity and high proliferative potential. Pre-engraftment syndrome (PES) is the most common immune response observed during the early stages following UCBT and occurs prior to neutrophil engraftment. Several studies suggested that PES may be attributed to unique cytokine storm during the immune reconstitution process post-transplantation. Few studies have described the potential risk factors and severity classification of PES, leading to uncertainty regarding the extent to which various parameters influence clinical outcomes in patients experiencing PES. There, we evaluated the combined effect of clinical characteristics and laboratory parameters on severe PES in pediatric patients undergoing UCBT.

Methods

From 2017 to 2024, a total of 123 pediatric patients with hematological malignant and nonmalignant disorders who experienced PES during the treatment of UCBT were included in this study. To establish a predictive model, patients were randomly divided into two groups: a training group and an internal validation group, with a ratio of 7:3, prior to feature selection and model development. Uni- and multivariable analysis was conducted to assess the potential parameters affecting PES severity, factors with a P value < 0.05 in univariable analysis were subjected to multivariable analysis. Finally, based on the multivariable Cox model, the optimal model's performance was illustrated using a nomogram, and evaluated through discrimination and calibration.

Results

Among the 123 patients, only 12 patients didn’t develop PES, while 54 experienced mild PES and 57 were diagnosed with severe PES. The mean monocyte counts in the graft (0.15×10⁹/L vs. 2.44×10⁹/L) and serum IL-6 level on the day of PES (6.79pg/ml vs. 45.70pg/ml) exhibited statistically significant differences between the mild and severe PES groups (P <0.001). Furthermore, patients with severe PES were more likely to experience PES within 7 days (P <0.001). No other elements demonstrated significant differences between the two groups (P >0.05).

Uni- and multivariable Cox analysis was applied to identify the potential risk factors affecting the severity of PES. Based on the above results, parameters with a P value of ≤ 0.05 were included in the multivariable Cox regression analysis. The multivariable analysis revealed that monocytes in the graft (HR: 1.320, 95%CI 1.052-1.656, P =0.017), donor chimerism on day 7 (HR: 1.018, 95%CI 1.007-1.028, P =0.008) and serum IL-6 level on the day of PES (HR: 1.006, 95%CI 1.002-1.025, P =0.016) were identified as independent risk factors for the severity of PES. Then, donor chimerism on day 7, monocytes in the graft and serum IL-6 level on the day of PES were selected to construct a visual nomogram to represent the predictive model. The predictive performance of this nomogram model was evaluated by C-index and area under the curve (AUC), which demonstrated outstanding discrimination with a C-index of 0.732 (95%CI 0.677-0.824) and an AUC of 0.881 (95%CI 0.811-0.950) respectively. In the internal verification, the corrected C-index was 0.692 (95%CI 0.678-0.894) and the AUC was 0.806 (95%CI 0.757-0.996), indicating good concordance between the predicted and actual incidence of severe PES.

The risk score for each patient was calculated based on the individual expression levels of the four factors, with an optimal risk score cutoff established at 91.9. Patients were subsequently divided into low-risk (score <91.9) and high-risk groups (score >91.9) according to the cutoff value. Compared to the low-risk group, patients in the high-risk group exhibited a significant higher 10-day cumulative incidence of severe PES [High-risk group: 92.2% (95%CI 76.7-97.5%) vs. Low-risk group: 24.7% (95%CI 7.25-47.5%), P <0.001]. However, the cumulative incidence of 100-day aGVHD was similar in the two groups [High-risk group: 80.7% (95% CI 67.4-89.0%) vs. Low-risk group: 75.8% (95% CI 63.2-84.5%), P >0.05].

Conclusion

This predictive model can be utilized to identify children at risk of severe post-engraftment syndrome (PES) following cord blood transplantation. It may facilitate preemptive interventions and treatments, ultimately reducing the risk of transplant-related mortality. Further validation with a larger sample size is necessary.