Background: Eosinophilia, typically defined as an absolute eosinophil count (AEC) exceeding 0.5 x 10 9/L, is a common laboratory finding with uncertain clinical significance. While individuals with mild eosinophilia are frequently referred for hematologic evaluation, the diagnostic yield and clinical utility of these referrals remains poorly characterized. This study aims to refine risk stratification for eosinophilia in individuals referred to hematology and to delineate the spectrum of associated diagnostic outcomes.

Methods: We conducted a retrospective study of patients referred to classical hematology at the Oregon Health & Science University between 2015-2025 with an eosinophilia-specific ICD code and an AEC ≥0.5 x 109/L. We collected data on patient demographics, absolute eosinophil counts, pertinent medical history, eosinophil-related diagnosis, and treatment.

Results: A total of 142 patients were referred and evaluated for eosinophilia with an absolute eosinophil count (AEC) ≥0.5 x 109/L over the 10-year study period. The mean age was 54 years. Sixty five percent (n=92) were female. Peak AECs ranged from 0.5 to 45.66 x 109/L. At the time of hematology referral 18.3% (n=26) had spontaneous resolution of eosinophilia (AEC of ≤ 0.5 x 109/L). Forty-five patients (31.7%) had mild eosinophilia (0.0-1.5 x 109/L), 71 patients (50%) had moderate eosinophilia (1.5-5.0 x 10 9/L), and 26 patients (18.3%) had severe eosinophilia (>5.0 x 109/L). The most common etiologies of eosinophilia were idiopathic (45.1%, n=64) and hypereosinophilic syndrome (HES, 20.4%, n=29). Other causes of eosinophilia were asthma (9.9%, n=14), parasitic infection (7.7%, n=11), autoimmune disease (4.9%, n=7), and drug-related reaction (4.9%, n=7). Among HES cases, 65.5% (n=19) were categorized as HES not otherwise specified (HES-NOS) and 6.9% (n=2) were malignancy-associated. No cases of idiopathic eosinophilia progressed to hematologic malignancy. Treatment was initiated in 42.2% (n=60) of patients and included systemic medications for asthma, parasitic infections, autoimmune disease, and HES. Eosinophilia-specific therapy with prednisone, mepolizumab, and/or hydroxyurea were administered in 22.5% (n=32) of patients. There were no significant differences in age, sex, race, or history of asthma, autoimmune disease, or malignancy between patients who did and did not receive eosinophilia-specific therapy.

Receiver operating characteristic (ROC) analysis demonstrated that peak AEC was a strong predictor of clinically meaningful eosinophilia, defined as a diagnosis of HES or the need for eosinophilia-specific treatment, with an area under the curve of 0.83. An AEC threshold of 1.5 x 109/L yielded a 97% sensitivity and 39% specificity for clinically meaningful eosinophilia, 97% sensitivity and 39% specificity for eosinophilia specific treatment, and a 92% sensitivity and 48% specificity for any systemic treatment. The optimal Youden's Index, which reflects the point on the ROC curve that best balances sensitivity and specificity to maximize overall test accuracy, occurred at an AEC of 2.11 x 109/L (sensitivity 85%, specificity of 61%, positive predictive value (PPV) of 39%, and negative predictive value (NPV) of 93%). Similar optimal thresholds were identified for eosinophilia specific treatment(2.11 x 109/L, sensitivity 82%, specificity 63%, PPV 32%, NPV 95%) and any systemic treatment (2.40 x 109L, sensitivity 78%, specificity 67%, PPV 46%, NPV 89%).

Conclusion: In patients referred for eosinophilia, an AEC threshold of 1.5 x 109/L has a high sensitivity for clinically meaningful eosinophilia and treatment need. The higher thresholds offer greater specificity, but may reduce sensitivity and risk missed cases. These findings support raising the threshold for hematology referral to 1.5 x 109/L or higher, especially in low risk and asymptomatic patients. Incorporating refined AEC cutoffs to guide hematologic evaluation may reduce unnecessary diagnostic workup in low-risk patients.

Keywords: eosinophilia, idiopathic, hypereosinophilic syndrome, absolute eosinophilia count

This content is only available as a PDF.
2025
Sign in via your Institution