Post-splenectomy early biochemical trajectories and predictors in 935 patients with Wilson disease: a group-based trajectory modeling study

Splenectomy is commonly used to alleviate portal hypertension-related hypersplenism in patients with Wilson’s disease (WD) cirrhosis. However, early post-splenectomy biochemical responses vary substantially among individuals, and objective tools to characterize these dynamic patterns remain limited. This study aimed to identify early postoperative biochemical trajectories after splenectomy in patients with WD and to determine preoperative factors associated with unfavorable trajectory patterns. We retrospectively analyzed 935 WD patients undergoing splenectomy (2001–2019). Serial ALT, AST, TBIL, and ALB were measured preoperatively and on postoperative days 1, 7, 14, and 21. Group-based trajectory models (GBTM) classified early postoperative biochemical patterns. Model selection was informed by comparison across multiple class solutions (K = 1–4), with K = 2 selected as the most parsimonious and clinically interpretable solution while avoiding very small subgroup proportions in higher-order models. Multivariable logistic regression identified predictors of unfavorable trajectories. GBTM identified two distinct trajectories for each biomarker. Unfavorable trajectory rates were 20.3% (ALT), 12.7% (AST), 13.6% (TBIL), and 46.1% (ALB). Model comparison supported the selection of K = 2 as the most parsimonious and clinically interpretable solution, while avoiding very small subgroup proportions in higher-order models. Higher Child–Pugh class strongly predicted unfavorable ALB and TBIL patterns. Sensitivity analyses excluding Child–Pugh yielded consistent results. Early post-splenectomy biochemical trajectories in WD exhibit heterogeneous patterns. These trajectories reflect short-term dynamic responses rather than definitive improvement in intrinsic hepatic function. The identified predictors may provide signals for early risk stratification but require validation against long-term outcomes.