Pancreatic stone protein versus conventional biomarkers for prognostic discrimination in sepsis: a meta-analysis of paired AUC comparisons

Background Pancreatic stone protein (PSP) has been proposed as a prognostic biomarker in sepsis, but its comparative performance versus C-reactive protein (CRP) and procalcitonin (PCT) remains uncertain. Methods We performed a meta-analysis of direct within-study paired AUC comparisons between PSP and CRP and/or PCT in critically ill adults with sepsis, septic shock, or acute infection-related conditions in whom sepsis development or clinically relevant adverse outcomes were prospectively evaluated. Differences in AUC (ΔAUC) were pooled using random-effects models. The principal conservative analysis included one comparison per independent cohort and comparator. Results Seven studies comprising 803 unique participants contributed 21 paired comparisons. In the principal conservative analysis, PSP showed a higher prognostic AUC than CRP, with a pooled ΔAUC of 0.179 (95% CI 0.095–0.262; p < 0.001), although between-cohort heterogeneity was substantial (I² = 81.65%). Compared with PCT, the pooled ΔAUC was 0.153 (95% CI 0.114–0.193; p < 0.001), with no observed heterogeneity (I² = 0.00%). The mortality-only sensitivity analysis showed a similar pooled effect for PSP versus CRP (ΔAUC 0.17, 95% CI 0.08–0.26), with substantial heterogeneity persisting, and for PSP versus PCT (ΔAUC 0.153, 95% CI 0.114–0.193), with no observed heterogeneity. Exclusion of the clinically distinct intra-abdominal infection cohort (Michailides et al.) likewise did not materially change the PSP-versus CRP estimate (ΔAUC 0.166, 95% CI 0.077–0.256). The direction and approximate magnitude of the pooled effects were also preserved across alternative assumptions for the within-study correlation. Conclusions Available comparative evidence suggests that PSP may have greater prognostic discrimination than CRP and PCT. The findings were broadly maintained across sensitivity analyses, but the limited number of independent cohorts and substantial heterogeneity in the CRP comparison warrant caution. Further prospective studies are required to confirm these findings and establish their clinical relevance. Trial registration This systematic review was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO; CRD420261340301).