Mean corpuscular hemoglobin modulates HbF distribution and sub-phenotypes of sickle cell disease

Abstract Background Fetal hemoglobin (HbF) prevents sickle hemoglobin (HbS) polymerization that injures RBCs leading to the phenotype of sickle cell disease (SCD). F-cells can be detected by flow cytometry if they contain at least 4-6 pg of HbF. A HbS/HbF ratio of ~70:30 is needed to “protect” fully the sickle RBC from polymer-induced injury. In a normocytic sickle cell with an MCH or mean corpuscular hemoglobin of ~30 pg, this is ~10 pg of HbF. Larger RBCs need more HbF to prevent HbS polymerization. The distribution of HbF/F-cell, in patients treated with gene therapy (GT, MCH ~30 pg), or in high HbF responders to hydroxyurea (HU, MCH ~42 pg), showed a mean of ~12 pg HbF/RBC after GT compared with HU where ~40% of RBCs had 10 pg of HbF. We therefore hypothesized that at any level of HbF, HbF/F-cell, which is the critical measure of the protection HbF provides against SCD complications, is determined by MCH. Accordingly, we studied complications of sickle vasoocclusion and the intensity of hemolysis in 2 cohorts of patients. Methods Patients were either HbS homozygotes or had HbS-β0 thalassemia, were not on chronic transfusion or apheresis, and were not transfused within 6 wks of data extraction. Cohort 1 included patients followed at our hematology clinic (BMC). Cohort 2 included patients who were participants in the Cooperative Study of Sickle Cell Disease (CSSCD) and not taking HU. HbF per cell (HbF/cell) was derived by MCH x (%HbF/100). Outcome variables included annualized vasoocclusive events and the hemolytic score, a linear combination of reticulocyte count, LDH, AST, and bilirubin calculated by principal component analysis with a mean of 0. A higher score indicates less hemolysis. Markov Chain Monte Carlo method in Open bugs was used to estimate the predicted values, and log-transformed median predicted values for each of the 4 hemolysis markers were used. A Poisson regression model was used with annualized vasoocclusive events and Hb.F, HbF/cell, MCH, MCV, age and sex as independent variables. Results Forty-six patients, 76% on HU, aged 40.7±8.0 yrs, HbF 11.6±8.8%, MCH 33.0±5.4 pg., with 265 admissions, comprised the BMC cohort. After adjusting for other independent variables, there was strong positive linear association between MCH and vasoocclusive events (p0.001) as well as admissions for acute vasoocclusive events (p0.0001). There was a negative linear association between HbF/cell and vasoocclusive events (p0.001). In the BMC cohort, males had an average number of total admissions higher than female patients by 1.964. Additionally, as a patient's age increased, the average number of total admissions per year increased by 0.172/yr. In the CSSCD cohort 1742 patients were analyzed, with a mean age of 22.8± 12.4, HbF 1.5±1.9% and MCH 28.9±3.2pg. The CSSCD cohort also showed a strong positive linear association between MCH and HbF/cell and vasoocclusive events (p0.001) after adjusting for other independent variables. Patients with higher HbF/cell had reduced pain episodes, and patients with higher MCH had more pain episodes after adjusting for age, sex and hemolysis score. Males and older patients also had more frequent vasoocclusive events. Additionally, HbF/cell was positively correlated with the hemolysis score, with higher HbF/cell being associated with higher hemolysis scores. Increased hemolysis score was also associated with increased pain episodes. Pearson correlation between MCH and HbF/cell was 0.5539. Conclusion Both cohorts were similar in important variables, and therefore comparable. Regardless of HbF level, individuals with lower MCH appeared to have fewer vasoocclusive episodes than individuals with higher MCH. The hemolysis score was negatively correlated to HbF concentration, showing as HbF increases, hemolysis decreases owning to the increased protection of HbF in sickle RBCs. Lower MCH was associated with higher HbF/RBC. HbF and MCH might determine the level of protection for a given F-cell and might be useful prognostically. HbF inducing therapeutics that do not promote stress erythropoiesis and increase erythrocyte heterogeneity are most likely to result in HbF levels more evenly distributed among erythrocytes with a commensurate benefit on both vasoocclusive and hemolytic complications of SCD.