External validation of the AAA+ survival model in a large Chinese cohort with essential thrombocythaemia.

Abstract

Triple A Plus (AAA+) is a recently proposed survival prediction model for essential thrombocythaemia (ET) intended for settings in which next-generation sequencing is unavailable. However, independent validation in non-Western populations remains limited. A retrospective cohort of 995 consecutive patients diagnosed with ET at a single centre between 2001 and 2022 was screened. Primary external validation was performed in 850 patients who had complete baseline variables required for the original AAA+ calculation. In the external-validation cohort (n = 850; deaths = 116), AAA+ demonstrated strong discrimination for overall survival (C-index 0.892, 95% confidence interval 0.866-0.917). Time-dependent analyses indicated superior discrimination across follow-up, with 5- and 10-year area under the curve values of 0.902 and 0.912, respectively, exceeding those of AAA (0.873 and 0.894) and International Prognostic Score for Essential Thrombocythemia (0.759 and 0.857). For 10-year all-cause mortality, decision-curve analysis based on locally recalibrated absolute risks showed that AAA+ provided favourable net benefit across most clinically relevant thresholds. Diagnostic period-adjusted analyses showed higher hazards of death in intermediate- and high-risk patients than in low-risk patients, with no clear evidence of temporal heterogeneity or monotonic attenuation of the AAA+ risk gradient over time. AAA+ exhibited robust risk stratification and good discrimination for overall survival in this large Chinese ET cohort.