Identification of Novel Variants Associated with Fetal Hemoglobin Levels in Healthy Donors (the INTERVAL study)

• Published in: Blood 2019-11, Vol.134 (Supplement_1), p.2243-2243
• Main Authors: Kawabata, Emily, Lessard, Samuel, Paul, Dirk, Bronson, Paola G, Peters, Robert, Krishnamoorthy, Sriram, Astle, William, Butterworth, Adam
• Format: Article
• Language: English
• ISSN: 0006-4971; 1528-0020
• DOI: 10.1182/blood-2019-123977

Sickle cell disease (SCD) affects more than five million people worldwide, predominantly in sub-Saharan Africa. Hereditary persistence of fetal hemoglobin (HbF) is an uncommon genetic condition in which production of HbF in early life is not suppressed. SCD symptoms are reduced in patients carrying this condition, suggesting that increased HbF levels may be a promising therapeutic strategy to ameliorate the symptoms of SCD. This is exemplified by the HbF-raising compound hydroxyurea, currently the most commonly used US Food and Drug Administration approved drug treatment for SCD. However, hydroxyurea is only effective in ~70% of patients and carries a black box warning for carcinogenicity, hence there is a need for additional SCD treatments. Previous genome-wide association studies (GWAS) have identified four loci robustly associated with HbF levels, including variants in the BCL11A region. These initial genetic discoveries have led to promising ex vivo gene-editing approaches to silence or reduce levels of BCL11A, which are currently being tested in clinical trials. To identify novel potential therapeutic targets to raise HbF levels, we conducted the largest GWAS of HbF levels in ~11,000 healthy blood donors from the INTERVAL study among whom HbF was measured in whole blood using a mass spectrometry approach. We ran linear mixed models accounting for age, sex, blood group, technical effects and 10 principal components of ancestry for 14,910,742 variants either directly genotyped using the Affymetrix UK Biobank array or imputed from a combined 1000 Genomes/UK10K reference panel. In addition to confirming previously reported signals at the BCL11A, HBS1L-MYB and HBB loci, stepwise conditional analysis identified six novel genomic regions at genome-wide significance (p0.5), including a rare (frequency=0.2%) variant near GRIK2. To identify likely causal genes and mechanisms, we integrated our results with relevant transcriptomic and epigenomic datasets. Preliminary results suggest that a common (frequency=26%) 29 base-pair indel upstream of CHAC2 is highly likely (posterior probability=0.77) to be the causal variant at this locus. The variant maps to a site of active chromatin and GATA1 transcription factor binding specifically