Glancing through a review of SLE genetics (
tandfonline.com/doi/full/10.…), I bumped into a familiar plot, which made me revisit the beautiful work by Kamitaki et al. from 2020 on the elegant dissection of the HLA GWAS signal (
nature.com/articles/s41586-0…), which is one of the fine demonstrations of the value of non-European populations, particularly Africans, in fine-mapping GWAS loci to pinpoint the causal variants.
One of the strongest GWAS signals of SLE erupts out of the HLA region in chromosome 6. Here is a Manhattan plot from a 2008 study (
nature.com/articles/ng.200).
The well-known complexity of the HLA region posed a significant challenge for scientists to decode the chr 6 signal. Given that SLE is an autoimmune disease there was a valid reason to believe that SLE, like other autoimmune diseases, might be bearing a HLA risk variant. The HLA-DRB1*03:01 allele was assumed to be that variant.
But at the same time, scientists couldn't stop looking at a nearby pair of genes--C4A and C4B--that codes for one of the strongest genetic risk factors of SLE known since ages: complement 4. The earliest documentations of complete C4 deficiency in SLE patients date back to 1974 (
pubmed.ncbi.nlm.nih.gov/4212…).
But then, scientists had their reasons for not betting on the C4 genes, despite its historical links to SLE. As opposed to complete C4 deficiency, partial C4 deficiency had conflicting reports of its links with SLE, which weakened the argument that commonly observed copy number variations of C4 genes (that could only partially reduce the C4 gene dosage) drive the SLE risk in the population.
Of course, one could ignore their biological intuitions and play the who's the causal variant game purely from a statistical angle. But there was a problem. The C4-B(S) allele (that lacked any C4A gene) with the highest SLE risk effect was in high LD with the HLA-DRB1*03:01 allele in the European populations. Is there a way to break this LD friendship? It turned out there is.
Unlike in the European populations, in the African Americans, the LD between C4-B(S) and HLA-DRB1*03:01 was very weak. Meaning, you can separate individuals who carry only C4-B(S) alleles and those who carry only HLA-DRB1*03:01 alleles and compare the disease risks. In doing so, the authors were clearly able to see that SLE risk was driven by C4-B(S) alleles but not by the HLA-DRB1*03:01 alleles.