Abstract
Serum erythropoietin (sEPO) is an initial screening tool for distinguishing polycythemia vera (PV) from secondary erythrocytosis (SE), but defining 'subnormal' sEPO levels for PV diagnosis remains contentious, complicating its clinical utility. This study compares the diagnostic performance of sEPO across established subnormal limits, including reference interval (RI), clinical decision limit (CDL), and functional reference limit. sEPO levels were analyzed in 393 healthy donors (HDs) and 90 patients (41 PV and 49 SE), who underwent bone marrow biopsy and genetic tests due to erythrocytosis. The RI (2.5-97.5 percentile from HDs) of sEPO was 5.3-26.3 IU/L. A CDL of 3.1 IU/L, determined by ROC analysis in erythrocytosis patients, had a sensitivity of 80.5% and specificity of 87.8% for diagnosing PV. A functional reference limit of 7.0 IU/L, estimated based on the relationship between sEPO and hemoglobin, hematocrit, and WBC, increased sensitivity to 97.6% but decreased specificity to 46.7%. Using 5.3 IU/L as a 'subnormal' limit identified all three JAK2-negative PV cases, increasing the sensitivity and negative predictive value to 97.6% and 97.0%, respectively. Combining the RI, CDL, and functional reference limit may improve PV diagnostic accuracy.