By Julia Karow
This article, originally published March 19, has been updated with comments from Eric Schadt at Mount Sinai School of Medicine.
A healthy individual's genome sequence can help uncover disease risks that are not obvious from family history or other risk factors, and high-throughput molecular profiling in an individual over time provides unprecedented insights into disease development, according to a new study by researchers at Stanford University School of Medicine.
For the study, which started about two years ago and is still ongoing, a team of researchers led by Mike Snyder, chair of Stanford's department of genetics, generated an "integrative Personal Omics Profile," or iPOP, of Snyder — also referred to as the "Snyderome" by the study's participants — that consists of his genome sequence as well as multiple transcriptomic, proteomic, metabolomic, and autoantibody profiles over a period of 14 months.
The results, published online in Cell last week, demonstrate that there is much to gain from studying individuals in great depth over time, rather than comparing groups of sick with healthy individuals at a single time point.
Analyzing Snyder's genome revealed that he is at increased risk for type 2 diabetes — something he did not expect based on his family history — prompting the researchers to monitor his glucose levels more closely. During the study, he was diagnosed with diabetes and was able to respond quickly with changes in diet and life style.
The study also revealed complex and dynamic changes in Snyder's omics profiles in response to two viral infections and suggested that the onset of diabetes was related to one of those infections.
"This is the essence of personalized medicine," said Snyder, adding that the approach, which allows scientists to identify changes relative to a person's healthy state, is "much more valuable than just taking one time point from this study and comparing it to 500 other people — that may actually be uninterpretable."
Generating similar longitudinal omics profiles for large numbers of individuals will be "extremely powerful," he said, enabling researchers to dissect complex diseases like type 2 diabetes much better than before.
The kind of study — generating many types of data over time on individuals to better understand disease, disease risk, and how systems respond to environmental stresses like viral infections — "has been one of the dreams of the systems biology community for a long time," said Eric Schadt, chair of genetics and genomic sciences at Mount Sinai School of Medicine.
"Of course, the true power of this type of study is realized when all of the dimensions of data are scored in populations of individuals, as opposed to a single individual, and then integrated to construct more accurate views of the molecular and cellular biology driving normal and pathophysiological states," he told CSN in an e-mail message.
The Snyderome
After Snyder joined Stanford in 2009, the study began as a proof of concept to apply genomic and proteomic analyses that his lab had been conducting to medically important problems. Overall, the scientists took more than 3 billion measurements, following at least 40,000 components at each time point.
"If you go to a doctor's office, the most he will measure is 20 things, and probably fewer," Snyder said. "If you think about the capabilities out there, there is no reason we should not be measuring thousands, if not tens of thousands or ultimately millions of thing. We should really be analyzing healthy and disease states and making this part of our medical care."
Making himself the subject of the study seemed an easy way to ensure the availability of sufficient study material and to obtain consent, he said.
Another purpose of the study was to test and evaluate different exome capture and sequencing technologies. Snyder's genome, for example, was sequenced using three different platforms — Illumina's HiSeq 2000, Complete Genomics' proprietary technology, and Life Technologies' SOLiD — though results from the SOLiD have "fallen by the wayside," Snyder said.
Last year, the researchers published a comparison of the Illumina and Complete Genomics platforms (IS 12/20/2011) and another study of three exome-capture methods (IS 9/27/2011). The results showed that no single sequencing platform is currently accurate and sensitive enough to provide a complete picture, so the scientists combined results from Illumina and Complete Genomics for their omics study.
They also sequenced the genome of Snyder's mother in order to phase his variants to the maternal and paternal chromosomes. This was important to be able to analyze compound heterozygous mutations, where two mutations are present in the same gene but reside on different alleles.
An analysis of Snyder's variants showed that 55 rare coding single-nucleotide variants or indels are present in genes with a known Mendelian disease phenotype.
Further analysis showed that Snyder has a "significantly elevated risk" for hypertriglyceridemia, basal cell carcinoma, and type 2 diabetes. The last two risks were not immediately apparent from his family history, he said, though he later learned that his grandfather, who lived to be 87, also had elevated glucose levels.
Snyder also had no other risk factors associated with diabetes, such as smoking or obesity, and his glucose levels were normal during the first part of the study.
Another mutation he and his mother carry in a telomerase gene has been associated with acquired aplastic anemia, but neither of them has any sign of the disease, suggesting that some disease associations depend on the genomic context and are "not always accurate in terms of a personal genome interpretation," he said.
Prompted by the genome results, the researchers monitored Snyder's glucose levels, which increased shortly after he had an infection with a respiratory syncytial virus and stayed elevated for several months until he made a "dramatic change" in diet and exercise and started taking aspirin on a regular basis.
Without the genomic risk information, Snyder's diabetes might have gone undetected much longer. "It's only because of this rather frequent careful monitoring that it was caught quite early and therefore managed quite early," he said.
Associating the onset of diabetes 2 diabetes with a viral infection is also a novel result. "Our interpretation, which one would like to see reproduced, is that the genome has me predisposed [to diabetes] and the RSV infection is a stress response" that triggered the disease, he said, noting that viral infections have been associated with type 1 diabetes in the past, but not type 2.
While following his glucose levels provided the "most dramatic" results from a medical standpoint, the other analyses conducted on components of his blood at various time points — RNA-seq, mass spec of 6,280 labeled proteins, up to 7,000 metabolites, and miRNAs — allowed the researchers to follow molecular and cellular pathways that are changing in "exquisite detail," he said.
Integrating several types of data was important for the analysis. "If you just did the transcriptome, you would miss what we think are some key things that are going on," he said. "We think whole omics profiling gives you a microscopic vision that has resolution beyond anything that's ever been looked at previously," providing a "much better detailed understanding of the processes going on."
For the study of complex diseases, "we believe it is a whole new paradigm," Snyder said. While complex diseases like diabetes are currently often studied as a single entity, "there are probably many ways to get to that state," he said. "Doing this kind of detailed [omics] analyses, we would hope to be able to dissect the different patterns leading to that state and then potentially manage them very differently."
Besides helping to better understand disease, the study also provided new biological insights. For example, the researchers discovered "extensive" heteroallele expression — meaning one allele is expressed in one state, for example in health, and the other during a disease state. In addition, they have seen considerable levels of RNA editing and discovered new editing mechanisms.
Snyder and his colleagues are now planning to expand their study to a larger cohort of initially 10 and eventually 250 subjects, focusing on individuals who are at risk for diabetes because of their family history, genomic risk variants, or high blood glucose levels.
The plan for the larger study, which is about to be launched, is to take omics measurements at regular time points during healthy states and to increase the sampling frequency during stress responses, such as viral infections.
The cost of omics studies like this is coming down quickly. Snyder's group currently pays about $3,500 for a human genome sequence, he said, so sequencing each sample with two technologies will cost about $7,000.
"Once you prove the principle and value, a lot of people will start to jump on this and make it cheaper, and it will become relatively inexpensive to the point where I don't think it should cost any more than what you're currently doing for individual tests" during a medical check-up, he said.
Like others, his group is working on genomic databases that integrate with electronic medical health records, which poses challenges in terms of data storage and informatics, he said.
"Right now, we want to collect as much information as possible because we don't know what's important," Snyder said. "But it does seem plausible that some day, you don't need all 40,000 data points, but maybe a few thousand data points are enough to give you the same information."
Snyder himself plans to remain a study subject "for the rest of my life." He has already undergone two additional viral infections and plans periodic updates of the publicly available database that contains his personal omics data.
Have topics you'd like to see covered in Clinical Sequencing News? Contact the editor at jkarow [at] genomeweb [.] com.