Leaders in Pharmaceutical Business Intelligence (LPBI) Group

Observations on Finding the Genetic Links in Common Disease: Whole Genomic Sequencing Studies

Author: Larry H Bernstein, MD, FCAP

In this article I will address the following article by Dr. SJ Williams.

Finding the Genetic Links in Common Disease:  Caveats of Whole Genome Sequencing Studies

In the November 23, 2012 issue of Science, Jocelyn Kaiser reports (Genetic Influences On Disease  Remain Hidden in News and  Analysis) on the difficulties that many genomic studies are encountering correlating genetic variants to high risk of type 2 diabetes and heart disease. American Society of  Human Genetics annual 2012 meeting, results of DNA sequencing studies reporting on genetic variants and links to high risk type 2 diabetes and heart disease, part of an international effort to determine the genetic events contributing to complex, common diseases like diabetes.
The key point is that these disease links are challenged by the identification of genetic determinants that do not follow Mendelian Genetics.  There are many disease associated gene variants, and they have not been deleted as a result of natural selection.  In the case of diabetes (type 2), the genetic risk is a low as 26%.

Gene-wide-association studies (GAWS) have identified single nucleotide polymorphisms (SNPs) with associations for common diseases, most of these individually carry only only 20-40% of risk. This is not sufficient for prediction
and use in personalized  treatment.

What is the implication of this.  Researchers have gone to exome-sequencing and  to whole genome sequencing for answers. SNPs can be easily done  by microarray, and in a clinic setting. GWAS is difficult and has inherent complexity, and it has had high cost of use. But the cost of the technology has been dropping precipitously. Technology is being redesigned for more rapid diagnosis and use in clinical research and personalized medicine.  It appears that this is not  yet a game changer.

My own thinking is that the answer doesn’t  fully lie in the genome sequencing, but that it must turn on the very large weight of importance in the regulatory function in the genome, that which was once “considered” dark matter.  In the regulatory function you have a variety of interactions and adaptive changes to the proximate environment, and this is a key to the nascent study of metabolomics.

Three projects highlighted are:
1.  National Heart, Lung and Blood Institute Exome Sequencing Project (ESP)[2]: heart, lung, blood

• A majority of variants linked to any disease are rare
• Groups of variants in the same gene confirmed a link between
  APOC3 and risk for early-onset heart attack

2.  T2D-GENES Consortium
3.  GoT2D

• SNP and PAX4 gene association for type 2 diabetes in East Asians
• No new rare variants above 1.5% frequency for diabetes

http://www.phgfoundation.org/news/5164/

The unsupported conclusion from this has been

1. the common disease-common variant hypothesis, which predicts that common disease-causing genetic variants exist in all human populations, but   (common unexplained complexity?) each individual variant will necessarily only have a small effect on disease susceptibility (i.e. a low associated relative risk).

1. the common disease, many rare variants hypothesis, which postulates that disease is caused by multiple strong-effect variants, (an alternative complexity situation?) Dickson et al. (2010)  PLoS Biol 2010 8(1):e1000294

The reality is that it has been difficult to associate any variant with prediction of risk, but an alternative approach appears to be intron sequencing and missing information on gene-gene interactions.

Jocelyn Kaiser’s Science article notes this in a brief interview with Harry Dietz of Johns Hopkins University where he suspects that “much of the missing heritability lies in gene-gene interactions”.

Oliver Harismendy and Kelly Frazer and colleagues’ recent publication in Genome Biology  http://genomebiology.com/content/11/11/R118 support this notion.  The authors used targeted resequencing
of two endocannabinoid metabolic enzyme genes (fatty-acid-amide hydrolase (FAAH) and monoglyceride lipase (MGLL) in 147 normal weight and 142 extremely obese patients.

Related articles

• Synthetic Biology: On Advanced Genome Interpretation for Gene Variants and Pathways: What is the Genetic
  Base of Atherosclerosis and Loss of Arterial Elasticity with Aging (pharmaceuticalintelligence.com)
• Making It Personal: Geneticist Michael Snyder Puts a Face on Personalized Medicine (psmag.com)
• Genetic testing not useful for predicting Type 2 diabetes, study suggests (cbc.ca)
• Doors to New Type 2 Diabetes Treatments Opened By Discovery of New Hormone (medindia.net)
• Diabetes and the Obesity Paradox (iplanethealthnews.com)

English: The human genome, categorized by function of each gene product, given both as number of genes and as percentage of all genes. (Photo credit: Wikipedia)