Abstract

Aim: To identify the factors contributing to Hydroxychloroquine (HCQS)-mediated hemolytic anemia in G6PD deficient COVID-19 patients.

Materials and methods: We have developed Multiple Linear Regression (MLR) and Classification and Regression Tree (CART) models based on the published data (n=13) and evaluated the impact of G6PD variants in exerting hemolysis.

Results: African Ancestry subjects had significant HCQS-mediated drop in hemoglobin (pre- vs. post-therapy Hb, g/dl: 12.75 ± 1.57 vs. 6.78 ± 0.62, p=0.0008) than those from non-African ancestry (pre- vs. post-therapy Hb, g/dl: 13.27 ± 1.51 vs. 9.97 ± 2.34, p=0.04). Diabetics had significant HCQS-mediated drop in hemoglobin (pre- vs. post-therapy Hb, g/dl: 12.96 ± 0.98 vs. 6.82 ± 0.76, p=0.0007) than non-diabetics (pre- vs. post-therapy Hb, g/dl: 12.88 ± 2.13 vs. 9.13 ± 2.51, p=0.05). MLR model explained 53.75% variability in HQCS-mediated hemoglobin drop when African Ancestry, diabetes, hypertension and azithromycin used as input variables. CART model efficiently explained HQCS-mediated hemolytic anemia by projecting African Ancestry and diabetes as the key predictors (R²=1.00). Although, African Ancestry G6PD A (-) variant had lesser ligand propensity than G6PD Mediterranean, G6PD Y70H and G6PD Union, it is thermolabile similar to G6PD Orissa, G6PD Y70H and G6PD L235F.

Conclusion: African Ancestry diabetic patients are more prone for HQCS-mediated hemolytic anemia in G6PD deficient COVID-19 patients. MLR and CART models explain 53.75% and 100% variability in HQCS-mediated hemolytic anemia. In view of similar in silico profile of Indian G6PD variants to G6PD A (-), HCQS should be used with caution in diabetics during COVID-19 therapy.

Keywords: Glucose-6-phosphate dehydrogenase; Diabetes; African ancestry; COVID-19 ; Hydroxychloroquine, Hemolytic anemia