The Vitamin D Receptor Gene and Disease Severity in Sickle Cell Anemia

Abstract 2122

Vitamin D is important in multiple aspects of health, including the cardiovascular, immune and skeletal systems and its effects are mediated through the vitamin D receptor (VDR). The systems affected by vitamin D are also perturbed by sickle cell disease (SCD). Vitamin D deficiency is common in SCD, but its contribution to disease manifestations is not yet known. In normal populations, vitamin D has been shown to be associated with hypertension and vascular pathology. That association may be particularly relevant to the inflammatory / endothelial damage seen in sickle cell disease. We have used clinical and laboratory data to create separate inflammatory and vaso-occlusive severity scores. Our hypothesis is that specific VDR polymorphisms are associated with disease severity in sickle cell disease.

DNA specimens from 1141 study participants in the NIH-funded Silent Infarct Transfusion (SIT) trial were used. In this multi-center international trial, the participants were children ages 4 to 13 years of age with SCD who were screened for the presence of silent cerebral infarction and had demographic and clinical data collected, as well as samples for a biologic repository for a self-renewing source of DNA. An initial 570 samples served as the discovery cohort. The subsequently enrolled 530 individuals formed our validation cohort.

We evaluated 79 single nucleotide polymorphisms (SNP) in the VDR gene and three associated genes, CYP27B1, VD binding protein, retinoid X receptor, and tagging SNPs from the African American population from Hapmap. The discovery cohort individuals had VDR haplotype information from a prior genome-wide association study (GWAS), and analysis for additional VDR-related SNPs was performed using a specifically designed Sequenom assay. The validation cohort was analyzed for SNPs that were significant in the discovery cohort.

Phenotype data was obtained from the demographic and clinical information of the participants, and was used to create the severity scores. The vaso-occlusive score includes: number of hospitalizations for pain, number of hospitalizations for acute chest, and avascular necrosis. The inflammatory severity score includes: priapism, transient ischemic attacks (TIA), silent cerebral infarct, systolic and diastolic blood pressure, transcranial doppler velocity, white count and baseline hemoglobin. The overall severity score includes all of the inflammatory and vaso-occlusive variables. To derive the scores, the variables were transformed into quartiles. Each individual subject was assigned values of 1, 2, 3, or 4 for each variable with 1 representing lowest severity and 4, the highest. In addition, in concert with prior analyses of the SIT data, the variable for number of hospitalizations for pain was used alone as a severity measure.

The severity scores were not normally distributed and were not totally continuous distributions, so the Kruskal-Wallis test was used in association analysis. To look for complex genetic models including potential gene-gene interactions for prediction of disease severity, the Multifactor Dimensionality Reduction (MDR) method was utilized, with repeat analyses performed for each severity score.

By univariate analysis, no associations were found between any of the VDR associated SNPs and the 3 severity scores. Using MDR in the discovery cohort, one SNP, rs7965281, was found to be associated with the inflammatory severity score. It remained significant after correcting for multiple comparisons with permutation analysis. In the validation cohort, rs7965281 was tested for association with each of the severity scores. There was no association with the inflammatory or the vaso-occlusive severity score but a trend towards association with the overall severity score (p= 0.08). All the SNPs were tested for association with the variable, number of hospitalizations for pain using regression analysis. Two additional SNPs, rs7855881 and rs34312136 were found to be nominally significant (p=0.01 and p=0.04 respectively). Rs7965281 shows a trend as well with p=0.06.

In the literature, rs7965281 is associated with reduced risk for cutaneous melanoma in a large population based study as well as with blood pressure in a British population. Further work in our validation cohort, including MDR analysis for gene-gene interaction using the 3 significant SNPs remains to be done and this may provide further direction in future research.