Hydroxyurea Treatment Is Associated with Elevated Serum Erythropoietin Concentration but Suppressed Global Hypoxic Transcriptional Responses in Sickle Cell Disease

Background The level of distorted erythrocytes due to polymerization of hemoglobin S in sickle cell disease (SCD) (Science 1949;110:543) is a major determinant of the severity of hemolysis and microvascular occlusion (Lancet 2010;376:2018). Erythropoietin (EPO) is elevated in SCD due to hemolytic anemia and a related increase in hypoxia-inducible factors (HIFs) (Eur J Haematol 2007;78:183). Hydroxyurea (HU) is widely used in the treatment of SCD. HU inhibits ribonucleotide reductase (Semin Oncol 1992;19(3 Suppl 9):1-10) and promotes γ globin synthesis thereby increasing HbF-containing erythrocytes (F cells) while suppressing sickle β hemoglobin production (J Clin Invest 1984;74:652 and 2003;111:231). Increased level of F cells reduces hemolysis and ameliorates clinical complications in SCD. We and others have observed an increase in serum EPO level with HU treatment in SCD despite an increase in the hemoglobin concentration, and we hypothesized that this may be due to the known increased affinity of hemoglobin F for oxygen and related tissue hypoxia (Blood 2009;114:4639). Methods Messenger RNA from peripheral blood mononuclear cells (PBMCs) was profiled using Affymetrix Human Exon 1.0 ST Array. Hypoxic transcriptional alteration was defined in 15 Chuvash polycythemia (CP) patients vs. 17 control individuals. CP leads to constitutive up-regulation of HIFs in the absence of anemia or hypoxia. Transcriptional alteration in SCD was determined in 13 HbSS subjects without HU treatment vs. 16 control individuals, and that induced by HU treatment was determined in 19 HbSS subjects with vs. 13 without HU treatment. For meta-analysis on serum EPO concentration, genomic DNA isolated from PBMCs was hybridized to the Illumina Human 610-Quad SNP array. Genotypes were imputed to 1000 genomes project phase 1 data. A linear regression model was applied adjusting for age, gender, hemoglobin concentration, and HU treatment. Results Gene expression changes by HbSS highly correlated with those associated with homozygous VHLR200W (Pearson's r=0.79, Figure 1A). At 5% false discovery rate (FDR), expression levels of 377 genes were altered in both VHLR200W homozygotes and HbSS by >1.2 fold. For these hypoxic genes, the correlation of expression changes between HbSS and homozygous VHLR200W reached r=0.97 (Figure 1B). In contrast to our hypothesis, HU treatment in general suppressed expression changes induced by HbSS (r=-0.85, Figure 1C), especially for the hypoxic genes (r=-0.95, Figure 1D). In VHLR200W homozygotes, 62 of the hypoxic genes correlated with plasma EPO levels (adjusted P<0.05, n=42). These EPO-correlated genes were the most strongly up-regulated hypoxic genes in HbSS (red points in Figure 1B) and also the most strongly suppressed by HU treatment (red points in Figure 1D). Consistent with previous observations, we found that EPO was elevated by HU treatment in two SCD cohorts, and this persisted after adjusting for covariates including hemoglobin concentration which reflects hypoxic as well as inflammatory and hemolytic responses: Walk-PHaSST (β=0.49, P=2.5×10-15, n=586) and PUSH children (β=0.34, P=2.5×10-7, n=387). This observation suggests that biological signals independent of hypoxic regulation may contribute to EPO production under HU treatment. In a meta-analysis for the Walk-PHaSST and PUSH children cohorts, SNP rs60684937, located within the first intron of MAP2K6, an upstream regulator of HIF signaling (Mole Cell Biol 2005; 25:4853), was significantly associated with EPO levels at genome-wide significance (combined P=3.5×10-8). The C allele of the SNP decreased EPO levels in both Walk-PHaSST (β=-0.30, n=388) and PUSH children (β=-0.24, n=249) cohorts. This association was validated in an additional 89 SCD patients from the Howard cohort (β=-0.39, P=0.011). Further investigations are needed to determine whether the causal polymorphism affects protein function or gene regulation of the nearby genes. Discussion Our study demonstrates a prominent release from hypoxic transcriptional responses by HU treatment in SCD despite an increase in serum EPO, a defining characteristic of an up-regulated hypoxic response. Our study hypothesizes that hypoxia-independent signals trigger EPO production in the setting of HU therapy and it identifies a potential genetic determinant in this alternative pathway. Figure 1. Figure 1. Disclosures No relevant conflicts of interest to declare.