TRAKJAK: a complete blood count-based prediction of polycythemia vera at initial erythrocytosis workup to reduce financial and ecological costs

Background Erythrocytosis is common in daily clinical practice, and polycythaemia vera (PV), a myeloproliferative neoplasm associated with JAK2 mutations, should be systematically considered in clinical reasoning process. Financial constraints and environmental burden of extensive testing prompted us to develop a predictive score for PV to guide decision to screen for JAK2 mutations. Methods We retrospectively collected clinical and biological data from patients with PV and non-PV erythrocytosis who had low erythropoietin levels in 2024 at our hospital (derivation cohort). Data were analysed using univariable and multivariable methods to develop the TRAKJAK score, which was then validated in a separate cohort of patients referred to our centre in 2025 for erythrocytosis. Results We included 134 patients in derivation cohort (96 [72%] with PV and 38 [28%] without) and 65 in validation cohort (33 [51%] with PV and 32 [49%] without). The TRAKJAK score required at least one positive parameter among RBC count ≥6.57 × 1012/L, basophil count ≥0.1 × 109/L, and platelet count ≥300 × 109/L. In validation cohort, negative predictive value (NPV) was 96%, sensitivity was 97%, positive predictive value (PPV) was 82%, and specificity was 78%. In global cohort, sensitivity was 98%, specificity was 89%, PPV was 94%, and NPV was 95%, with a false negative rate of 1.5% and a false positive rate of 4.0%. Conclusion Use of TRAKJAK could have prevented unnecessary JAK2 mutation screening in 62/70 patients (88%) who ultimately did not have PV. Like all clinical prediction tools, TRAKJAK is an aid to reduce over-investigation and must not replace clinical judgement.