Response to Letter Regarding Article, “Genetic Architecture of Abdominal Aortic Aneurysm in the Million Veteran Program”

In Response:

We appreciate the comments from Drs. Zhang and Liu on our two recent publications that aimed to identify the genetic factors associated with abdominal aortic aneurysm (AAA)^1,2. In one paper, we performed a genome-wide association study (GWAS) on 180,000 study participants to interrogate common variants, and in the other, we developed the machine learning approach, HEAL, to analyze rare nonsynonymous variants in 401 whole genomes. We acknowledge the direct overlap between these two discoveries is relatively little: among the 24 genes identified from the common variants, only two (ERG, SMYD2) were found in the protein-protein interaction modules (“HEAL modules”) as an extension to the genes identified from the rare variants (“HEAL genes”). However, in a broader search, 26 additional genes in our HEAL modules were also reported in a large collection of AAA genes derived from ~100 studies³ . On the other hand, we replicated ten known AAA loci in our GWAS study. Therefore, there are numerous known AAA genes identified in either of our studies. So why was there little overlap?

One possible explanation is the two datasets focused on different aspects of the genome. The input for our rare variant study was deeply-sequenced whole genomes, in which half of the called variants were rare (AF4) in the low P value group (GWAS P < 0.001). This enrichment grew from 6- fold (Fisher’s exact P=0.03) to 17-fold (Fisher’s exact P=0.005) as we lowered the allele frequency cutoff from 0.001 to 0.0001, suggesting the HEAL gene signals were contained in the GWAS results, albeit these signals were relatively weak due to the limited statistical power of GWAS for rare events.

Note that we focused on rare variants with nonsynonymous effects. One known marker gene for AAA is Interleukin-6 (IL6), on which an intron variant was captured by our GWAS but missed entirely in our rare variant analysis. However, when examining regulatory regions in the same genomes where the rare variant study was performed, a strong IL6 signal was evident (manuscript in preparation).

Rare variants are often young in origin and exert larger effects on diseases and reproductive fitness ⁴ , whereas common variants can impact both essential and weakly-associated disease genes. A recent study systematically interrogated rare variants in complex diseases and discovered both known and novel genes⁵ , suggesting common and rare variants could converge on the same biological pathways or function distinctly. As for AAA, future studies will better reveal how common and rare variants work in tandem to affect this disease.