A common consequence of the evolutionary process in many species is extensive genetic diversity. As has become apparent in recent studies (Tennessen et al., Science 2012; Nelson et al., Science 2012), the growth of the human population coupled with what is believed to be reduced selective pressure, presumably in part due to the life-promoting and life-saving effects of modern medical interventions, has facilitated a truly impressive range of rare variants in human genomes. Some of these rare variants are expected to be associated with predispositions to various diseases and medical conditions. Hence, as a consequence of human evolution the need for personalized medicine arises.
It is in this context that a recent study (Chen et al. Cell 2012) by Michael Snyder, the chairman of the Stanford Genetics Department, and his colleagues is of special interest. It is a technological tour de force and tour de self applied to blood samples from Snyder himself and in which Snyder and his associates combine multiple systematic high-throughput types of analysis (genomics, transcriptomics, proteomics, metabolomics, and more) i.e., as the authors’ preferred jargon would have it, a multi-dimensional integrated personal omics profile (iPOP). This analysis revealed a genetic variant predisposing to type 2 diabetes mellitus that led Snyder to screen for type 2 diabetes which subsequently developed. The authors claim that therefore this approach is a guide to the personalized medicine of the future. In what follows, I evaluate these claims, informed in part by my personal experience with an infrequently-recognized form of personalized medicine that has been in operation for no less than 15 or so years and possibly much longer depending on how precisely the concept is defined. I emphasize at the start that I support a role for personalized medical care in several senses not limited to the purely genetic but note that this approach, the substantial hype notwithstanding, nevertheless faces challenges and limitations.
The specific type of personalized medicine with which I am associated is clinical histocompatibility testing. I direct a laboratory, one of about 150-200 in the United States, that performs genetic and immunological testing to guide the selection of pairs of donors and recipients for various types of transplants involving either solid organs (such as kidneys, pancreases, livers, hearts, and lungs) or blood or bone marrow containing hematopoietic stem cells as well as other more differentiated cell types. In this capacity, we provide to transplant surgeons or physicians information about potential transplant recipients and donors that is unique to the individual transplant participant (recipient or donor) so that the best pairings of donor and recipient can be selected. Thus, unique genetic and immunological attributes inform the treatment of transplant patients. Such personalized medicine has been in widespread operation since the 1960’s and with the input of DNA-based genotypic information since the mid-to-late 1990’s.
While our improved our ability to precisely define the genotypes at key loci, i.e. our ability to determine the identities of the most relevant alleles, definitely has improved average graft and patient survival and average quality of life over the past twenty-five years, it is also clear that the genetic relationship between recipient and donor is only one factor in optimizing clinical transplantation outcomes. The immunological histories of the recipients and donors can be important, as can a number of other factors such as inter-current infections or other medical problems or conditions, and uncontrollable factors such as age. Thus, genetics is not all-powerful in this setting and as some geneticists have put it, phenotype trumps genotype.
The sort of analysis performed on the blood samples from Michael Snyder and his mother are orders of magnitude more information-rich than the test results referred to in the preceding two paragraphs. Nevertheless, some of the limitations are similar. There is always risk of both false positive and false negative results. The likelihood of one or more false positives or negatives will of course rise with the number of independent tests performed. This caveat applies not only to DNA sequencing errors, which can be difficult to identify in the context of whole genome sequencing, but to every assay for every analyte assessed. Thus, even aside from financial considerations, there is reason to be cautious about pursuing clinical application for an approach that involves measurements on as many as ~100,000 molecular analytes in cells or blood or other body fluids. Quality control will be a major issue for any clinical application of these technologies.
Chen et al. conclude: “Our study indicates that disease risk can be assessed from a genome sequence and illustrates how traits associated with disease can be monitored to identify varying physiological stages.” But consider the identification of genetic variants in his genome that influence the likelihood of type 2 diabetes mellitus. Snyder takes his subsequent development of what was diagnosed as type 2 diabetes as confirmation of the value of the genomic information, but we do not know how likely it is that others with his combination of predisposing variants would actually develop the disease. After all, if Snyder, 54 years old at the initiation of the study, had known of these genomic variants in his twenties, assuming the technology had made it possible, it would have required waiting thirty years for the ‘prediction’ of disease to reach fruition. Would Snyder have become diabetic if he had not suffered the particular virus infection, with respiratory syncytial virus, which appeared to be associated with his initial loss of normal glucose control soon after the infection?
The variable penetrance of many disease-associated alleles makes these questions highly relevant. In fact, it is noted in the study that both Snyder and his mother have a telomerase gene variant associated with aplastic anemia and neither individual has experienced this condition after a combined 137 years of life. In an interview with Snyder in another journal (Genome Biology, 2012), he explained that as a result of the study, he changed his diet and other aspects of his lifestyle, but of course, just getting the information from routine testing of his blood for glucose concentration could have prompted the same alterations in personal habits. In fact, even without personal medical information indicating that he was diabetic, Snyder could have adopted the same practices based solely on the widely available current recommendations for diet, exercise, and other lifestyle characteristics.
The study of Snyder and colleagues, which is indeed the systematic study of Snyder, is extraordinarily impressive as a demonstration of the potential of the latest high-throughput technology to provide massive molecular detail, at a single point in time and longitudinally, on an individual patient. How precisely and completely it will serve as a guide for the future of personalized medicine remains to be seen. As Snyder notes in the interview, there were additional modalities of analysis that could be deployed, such as but not limited to methylomics (the totality of covalent modifications of genomic DNA and histones) and exposomics (the totality of chemical exposures). I remain skeptical that the wholesale application of these various types of assessment will be cost effective or affordable in most instances unless per test costs decrease rather dramatically, which of course remains possible. However, it is likely that more selective implementation of these methods prompted by particular clinical scenarios offers the prospect of usefully personalized medical care of enhanced effectiveness. Studies aimed at refining the uses of these methods for clinical purposes should occupy numerous investigators for some time to come.
Tennessen JA, Bigham AW, O’Connor TD, Fu W, Kenny EE, Gravel S, McGee S, Do R, Liu X, Jun G, Kang HM, Jordan D, Leal SM, Gabriel S, Rieder MJ,
Abecasis G, Altshuler D, Nickerson DA, Boerwinkle E, Sunyaev S, Bustamante CD, Bamshad MJ, Akey JM; Broad GO; Seattle GO; NHLBI Exome Sequencing
Project. Evolution and functional impact of rare coding variation from deep sequencing of human exomes. Science. 2012 Jul 6;337(6090):64-9.
Epub 2012 May 17. PubMed PMID: 22604720.
Nelson MR, Wegmann D, Ehm MG, Kessner D, St Jean P, Verzilli C, Shen J, Tang Z, Bacanu SA, Fraser D, Warren L, Aponte J, Zawistowski M, Liu X, Zhang H, Zhang Y, Li J, Li Y, Li L, Woollard P, Topp S, Hall MD, Nangle K, Wang J, Abecasis G,Cardon LR, Zöllner S, Whittaker JC, Chissoe SL, Novembre J, Mooser V. An abundance of rare functional variants in 202 drug target genes sequenced in 14,002 people. Science. 2012 Jul 6;337(6090):100-4. Epub 2012 May 17. PubMed PMID: 22604722.
Chen R, Mias GI, Li-Pook-Than J, Jiang L, Lam HY, Chen R, Miriami E, Karczewski KJ, Hariharan M, Dewey FE, Cheng Y, Clark MJ, Im H, Habegger L, Balasubramanian S, O’Huallachain M, Dudley JT, Hillenmeyer S, Haraksingh R, Sharon D, Euskirchen G, Lacroute P, Bettinger K, Boyle AP, Kasowski M, Grubert F, Seki S, Garcia M, Whirl-Carrillo M, Gallardo M, Blasco MA, Greenberg PL, Snyder P, Klein TE, Altman RB, Butte AJ, Ashley EA, Gerstein M, Nadeau KC, Tang H, Snyder M. Personal omics profiling reveals dynamic molecular and medical phenotypes. Cell. 2012 Mar 16;148(6):1293-307. PubMed PMID: 22424236; PubMed Central PMCID: PMC3341616.
Genome Biology. 2012, 13:147 doi:10.1186/gb-2012-13-3-147.