Abstract
We aimed to investigate whether early clinical indicators were associated with eventual disease severity, and to develop a predictive model for severe asparaginase-associated pancreatitis (AAP). Seventy-five acute lymphoblastic leukaemia (ALL) cases with AAP admitted to Shanghai Children's Medical Center from March 2013 to August 2023 were divided into non-severe (n = 44) and severe (n = 31) groups based on Atlanta diagnostic and AAP grading criteria. We compared essential information, asparaginase(ASP) dosage form, cumulative dose, clinical characteristics and laboratory tests between the groups. Statistically significant indicators were analysed with multifactorial logistic regression to identify independent risk factors for severe AAP. Receiver operating characteristic (ROC) curves assessed the early predictive value of age, C-reactive protein (CRP) and fibrinogen (FIB) levels. In the early stages of AAP onset, significant differences in age, CRP, platelet count, red blood cell distribution width, albumin, calcium, FIB, and D-dimer levels were found between the non-severe and severe AAP groups (p < 0.05). Multifactorial logistic regression identified age (odds ratio [OR] = 1.204, p = 0.035), CRP (OR = 1.334, p = 0.003), and FIB (OR = 0.85, p = 0.008) as independent predictors of severe AAP. ROC analysis showed an area under the curves (AUC) for age was 0.681 (95% CI: 0.557-0.805), CRP was 0.766 (95% CI: 0.653-0.880), FIB was 0.735 (95% CI: 0.612-0.857). Optimal cut-off values for age, CRP, and FIB were 9.46 years, 48.5 mg/L and 1.265 g/L respectively. The combined AUC was 0.916 (95% CI: 0.845-0.986), with 0.903 sensitivity and 0.818 specificity, outperforming individual predictors (p < 0.05). Age, CRP, and FIB levels are good early predictors of severe AAP, and their combination significantly improves predictive accuracy.