Personalized Genetic Scans: With gifts like these…

It was Christmas season in the Bay Area.  A strapping and entirely healthy 27-year-old received from his parents, in lieu of a traditional gift, a personalized DNA analysis from one of the several companies providing such a service directly to the consumer.  The results revealed that he was a carrier of genetic variants that increased his risk for amyotrophic lateral sclerosis (ALS).  The young man was frightened and the family devastated.

The analysis did not detect the rare and highly penetrant ALS mutations, such as those for Cu/Zn SOD1 or alsin genes, in its ALS screen, but rather looked at a common variant identified in a single, not yet replicated, case control study.1  In that study, the variant increased ALS risk by about 30%, typical of a low risk disease-associated single nucleotide polymorphism (SNP) marker.  The test is less “predictive” of ALS risk than by simply determining if the individual is an athlete, or a smoker.2

The company, 23andMe, clearly stated in its report and on its website that this result reflects only a research finding that is of little clinical utility at the present time.3  Of 105 diseases, traits, and conditions for which information is provided to consumers by this service, 27 are currently listed as “clinical reports” which describe associations that “have a substantial influence on a person’s chances of developing the disease or having the trait,” whereas 78 are “research reports” based on “limited scientific evidence” including information that “does not have a dramatic effect on a person’s risk for a disease.”  Although the ALS test fell into the latter category of a “research report,” this fine distinction can of course be lost on the recipient of the information.

And even the value of a “clinical report” is a matter of debate.  What does it take for a test to be clinically useful?  Several biostatisticians, most recently Dan Weeks and colleagues, have revisited this question for the relatively weak genetic associations conferred by SNPs for common diseases.4,5  Such associations are usually derived from case-control studies searching for genetic variants associated with disease risk.  In adequately powered studies, very high P values can be found for individual markers, yet each marker – even if causal – usually increases risk by no more than 30%.  Furthermore, all SNP markers used in these screens represent – by definition – variants that are common in the general population.  To classify an individual person, however, one needs a test that has a very high positive predictive value (high likelihood of disease in those with the variant), and a high negative predictive value (low risk of having the disease when the variant is absent).  For nearly all complex disorders studied to date, the discriminative ability of the SNP associations fall far short of reaching any reasonable threshold for clinical utility.

Another real-life example brought to our attention recently was a perfectly healthy young woman whose report indicated that she carried two alleles indicating a predisposition to multiple sclerosis (MS); her fears were compounded by the knowledge that a second cousin had severe progressive MS.  The most important MS variant, rs6897932, encodes a functionally active SNP in the interleukin-7 receptor alpha chain.6  The MS-associated C allele is present in 73% of healthy individuals compared with 79% of individuals with MS.  A test in which the “disease-associated” marker is present in nearly three quarters of the healthy population is obviously nonspecific and on its own useless as a discriminator.  The strongest susceptibility gene for MS genomewide, HLA-DRB1*1501, is not tested in the current direct-to-consumer (DTC) testing platforms.  HLA-DRB1*1501 increases MS risk by 320% (or 3.2-fold), thus it is a moderately strong risk allele.  However, because the allele is common in the general population, only one in 250 individuals who carry DRB1*1501 will ever develop MS.  In aggregate, it is likely that more than 100 variants contribute in some way to MS risk, but even when all of these variants are discovered they are expected together to account for only a minority (less than one-third) of total MS risk.7  As for many other complex diseases, environment plays a greater role than genes in determining risk for MS.

The DTC genetic testing services (see Table) operate in a narrow space that appears to have largely escaped regulation, including federal and state oversight.8-11  Some observers have likened some of the information received as akin to obtaining a horoscope.10  The companies tout the value of the tests to determine your ancestry; to find which of your genes are shared with relatives, famous historical figures or celebrities; and even to share your genetic information with friends (e.g.  a genetic Facebook).  One dating service, ScientificMatch, even promises to use DNA to select for compatibility (“a much lower chance of cheating”).12

The potential negative fallout from DTC genetic testing, even when intended for recreational use, can be substantial.  It can induce unnecessary concern (as in the present examples), or conversely lead to a false sense of security.  There is the significant potential for a loss of privacy and even control of one’s DNA samples, a concern that has become more realistic as several of the companies move into the arena of “genetic research.”  Finally, there is the possibility that the information could even be incorrect, due to a sample handling error or communication of a disease association that might not apply to individuals of all ancestries.  The companies mitigate these concerns by advising consumers to “seek the advice of health professionals if you have questions or concerns arising from your genetic information.”  The problem here, of course, is that most physicians feel quite unprepared to discuss the significance of a genome-wide SNP blueprint with a patient, especially in a hurried clinical setting.13

Viewed in a wider context, these anecdotes highlight the challenge of how to promote scientific and medical literacy in a world in which the sheer volume of available data far outstrips the capacity of any lay person, or professional, to absorb new information, ignore misinformation, and separate the essential from the nonessential.  For many physicians, there is probably no more anxiety-inducing situation than the patient who arrives for a consultation accompanied by volumes of material downloaded from Google, PubMed, and elsewhere.  The potential for information overload needs to be managed of course – in medicine as in other areas of endeavor.  More importantly, we should not lose sight of the fact that personalized genetic testing has significant potential to contribute to health care and also to help promote scientific literacy, a worthwhile goal for a nation like the United States, in which only 4 in 10 citizens believe in evolution.14

The whole genome association scans currently offered by DTC providers test 500,000 to 1 million SNP markers, or approximately 10% of the common variations known to exist in the human genome; rare variants including deleterious mutations are not captured in these tests, nor are copy number variations, deletions, and other architectural changes in DNA that are increasingly viewed as important in determining overall health and disease.  As the technology advances over the next couple of years, complete whole genome sequences are expected to become cost-effective as screening tools, and these should provide information not just about the linear sequence of our genes but also will reveal insertions, deletions, transpositions, and even full epigenetic profiles.  With these advances in the depth of interrogation of individual genomes, the resultant clinical utility of the information should dramatically increase.  Perhaps the most important achievement of the current DTC ventures is to set the stage for a new generation of personalized genetic tests that may actually be diagnostically useful for presymptomatic assessment of risk for many common problems, especially when most or all of the causative variants and their relationship with environmental triggers are revealed.  But we are not yet there.

Time Magazine recently named the DNA testing service developed by 23andMe as its 2008 Invention of the Year.15  Perhaps, but for some users it is certainly not the gift of the year – at least not yet.  Our two young people concerned about ALS and MS can be reassured, but lingering anxiety will remain.  For their families, a New Years’ resolution: give video games to the kids next year.

Stephen L. Hauser MD and S. Claiborne Johnston MD, PhD

Editors

References:

1. van Es MA, van Vught PW, Blauw HM, et al. Genetic variation in DPP6 is associated with susceptibility to amyotrophic lateral sclerosis.  Nat Genet 2008;40:29-31.

2. Gallo V, Bueno-de-MesquitaHB, Vermeulen R, et al: Smoking and risk of amyotrophic lateral sclerosis: analysis of the EPIC cohort.  Ann Neurol in press 2009.

3. 23andMe web site. https://www.23andme.com. Accessed March 16, 2009.

4. Jakobsdottir J, Gorin MB, Conley YP, et al. Interpretation of genetic association studies: markers with replicated highly significant odds ratios may be poor classifiers.  PLoS Genet 2009;5(2): e1000337. doi:10.1371/journal.pgen.1000337

5. Janssens AC, Gwinn M, Bradley LA, et al. A critical appraisal of the scientific basis of commercial genomic profiles used to assess health risks and personalize health interventions.  Am J Hum Genet 2008;82:593-599.

6. Hafler DA on behalf of the International Multiple Sclerosis Genetics Consoritum.  Risk alleles for multiple sclerosis identified by a genome wide study.  New Eng J Med 2007;357:851-62.

7. Oksenberg JR  et al: The genetics of multiple sclerosis: SNPs to pathways to pathogenesis.  Nat Rev Genet 9:516, 2008.

8. NerveCenter: Genome scans get personal with online consumer services.  Ann Neurol 2008(2):A15-A17.

9. Hogarth S, Javitt G and Melzer D. The current landscape for direct-to-consumer genetic testing: legal, ethical, and policy issues.  Ann Rev Genomics Hum Genet.  2008;9:161-82.

10. Patch C, Sequeiros J, and Cornel MC. Genetic horoscopes: is it all in the genes? Points for regulatory control of direct-to-consumer genetic testing. Eur J Hum Genet  2009;doi:10.1038/ejhg.2008.246.

11. Kaye J. The regulation of direct-to-consumer genetic tests.  Hum Mol Genet 2008;17:R180-183.

12. ScientificMatch.com website: http://www.scientificmatch.com. Accessed March 16, 2009.

13. Calefato JM, Nippert I, Harris HJ, et al. Assessing educational priorities in genetics for general practitioners and specialists in five countries: factor structure of the Genetic-Educational Priorities (Gen-EP) scale.  Genet Med 2008;10:99-106.

14. Gallop Survey Feb.  11, 2009 quoted by Fox News Feb.  12, 2009; Harris Poll quoted in Reuters Life, Nov.  29, 2007

15. Hamilton A. The Retail DNA Test. Time Magazine Oct.  27, 2008.