Abstract

Hydroxyurea (HU) is an effective treatment for most patients with symptomatic sickle cell anemia, yet the fetal hemoglobin (HbF) response to treatment is variable. A capacity to predict an individual’s HbF response to HU would aid the selection of patients for treatment and reduce toxicity from unfruitful dose escalation. Unfortunately, this is presently not possible. We hypothesized that HbF levels and the HbF response to HU are regulated as complex genetic traits and previously showed that 12 single nucleotide polymorphisms (SNPs), associated with a 20% to 30% difference in baseline HbF concentrations, were found in the introns of 4 genes, PDE7B, MAP7, MAP3K5 and PEX7, spanning the genomic region from 136.1 Mb to 137.5 Mb on chromosome 6q (Cell Mol Biol 50:23, 2004). To begin to define the genetic predictors of the HbF response to (HU), we examined SNPs in candidate genes and genetic loci in 214 patients with sickle cell anemia whose HbF levels were available before HU treatment was started and after these patients reached a stable dose of this drug. Forty-six candidate genes were chosen because of their possible role in HbF regulation and HU metabolism and 226 SNPs in these genes were examined by mass spectrometry. A computer application developed in STATA was used to carry out multiple linear regression analysis with simultaneous adjustment for age, sex and the α- and β-globin gene cluster haplotypes for each SNP and combinations of nearby SNPs. Dominant, codominant and recessive models for modulating HbF expression were tested. In this QTL analysis, SNPs in a member of the cytochrome P450 family (CYP2C9), in aquaporin 9 (AQP9) and in the chromosome 6q qtl described above were significantly associated with the HbF response to HU. The effect of genotype on the magnitude of HbF response to HU was examined for selected SNPs in AQP9 and CYP2C9. In AQP9, AA was associated with an average increase of 6% in HbF compared with GG (rs1867380; OR 6.6, p<0.001). In CYP2C9, AG was associated with an average increase of 3% and GG with an average increase of 11% (rs2209331; OR 1.6 and 7.5, p=0.05, 0.000). An effect was also noted for the 6q qtl. We also treated the increase in HbF as a discretized variable, comparing individuals in the lower two quartiles of HbF response with individuals in the top quartile of HbF response to HU (Blood 89:1078, 1997). These same genotypes were more common in good HU responders than in poor responders (p<0.05). CYP2C9 (10q24) encodes a member of the cytochrome P450 superfamily of enzymes, monooxygenases catalyzing many reactions involved in drug metabolism, plays some role in the metabolism of HU derivatives. AQP9 (15q22.1–22.1), belongs to a family of water-selective membrane channels and stimulates urea transport, permitting passage of many uncharged solutes. These results begin to define the pattern of genetic heterogeneity that may be used ultimately to predict a patient’s HbF response to HU. As multiple genes are very likely to play roles in this response, the interactions and predictive value of their polymorphisms will need to be modeled with methods that account for simultaneous associations.

Author notes

Corresponding author