Feasibility, reliability, and accuracy of point-of-care measurements of fetal hemoglobin

Abstract Background: Fetal hemoglobin (HbF) is a key determinant of disease severity and an important therapeutic biomarker in sickle cell disease (SCD), because HbF inhibits the polymerization of sickle Hb (HbS) and high levels of HbF mitigate nearly all clinical complications. Hydroxyurea, the main disease-modifying therapy for SCD, exerts its benefit largely by inducing HbF. Periodic quantitative monitoring of HbF is essential to evaluate treatment responses and to monitor adherence. However, standard laboratory techniques for HbF measurement such as capillary zone electrophoresis (CZE) and high-performance liquid chromatography (HPLC) require costly equipment and highly trained personnel, making these methods impractical for use in many low-resource settings where SCD is most prevalent. The GazelleTM Hb Variant (Hemex Health) is a portable, inexpensive, battery-operated device with single-use cartridges that can quantify HbF and other hemoglobin species by microchip electrophoresis with multispectral imaging at the point of care (POC), so it may have utility in this clinical context. Objective: To evaluate the performance of the Gazelle Hb Variant for quantifying HbF in children and adults with hemoglobinopathies, in comparison to both CZE and HPLC as reference standards. Methods: Individual blood specimens were split into aliquots for quantitative measurement of HbF using CZE (Capillarys Flex 2, Sebia), a portable HPLC system (SmartLife LC, PolyLC Inc), and the Gazelle Hb Variant (Hemex Health). Concordance was assessed by Pearson correlation, coefficient of determination (r2), root mean square error (RMSE), and Bland-Altman plots. The lower limits of detection of HbF by Gazelle were also evaluated. Results: We studied blood from 127 individuals, 1 month to 50 years of age, with a variety of hemoglobinopathies (Hb SS 90, Hb S trait 15, Hb C trait 5, sickle-β-thalassemia 5, Hb SC 4, and others 8). Of these, 123 (97%) specimens were successfully analyzed by Gazelle on the first completed attempt. Technical errors (leaked cartridges) occurred with 4 samples, but these were successfully analyzed with a second attempt. HbF values measured by Gazelle correlated strongly with CZE (r = 0.95, r2 = 0.89, RMSE = 5.7%) and HPLC (r = 0.95, r2 = 0.85, RMSE = 6.6%). The two reference standards were highly concordant with each other (r > 0.99 r2 = 0.99, RMSE = 2.0%). Bland-Altman analysis revealed a mean bias of +2.1% (95% limits of agreement: –8.3% to +12.5%) for Gazelle vs CZE, and +3.25% (–8.0% to +14.5%) for Gazelle vs HPLC, indicating that Gazelle consistently but sightly overestimated HbF. Gazelle's sensitivity to detect non-zero HbF was 17% for samples with HbF <5.0%, 70% for HbF 5.0-9.9%, and 89% for HbF 10.0-14.9%. Gazelle had 100% sensitivity for detecting HbF in all samples with HbF ≥15% by both CZE and HPLC, establishing a reliable detection threshold at that level. Below ~5% HbF, detection was inconsistent, suggesting a lower possible detection limit in that range. Conclusion: The Gazelle Hb Variant POC device may provide a feasible, reliable, and accurate cost-effective solution for quantitative monitoring of HbF in patients with SCD treated with hydroxyurea in low-resource settings. While its limited sensitivity at low HbF concentrations (<5%) warrants caution, its strong concordance with reference standards and predictable performance at moderate-to-high HbF levels (10-40%) support its use for periodic monitoring of hydroxyurea therapeutic responses. Indeed, levels of HbF with optimized hydroxyurea therapy are well above the reliable detection limit. These findings reinforce the potential of Gazelle to support the implementation of hydroxyurea treatment programs in low-resource regions by enabling decentralized, cost-effective, POC HbF testing without the need for advanced laboratory infrastructure. Hemex Health donated the Gazelle instrument and 200 single-use cartridges for this study but did not have study oversight, collect or analyze the data, or participate in the writing of the abstract.