Personal DNA Testing: Expert Q&A
On July 11, 2008, Rudy Tanzi answered questions triggered by a NOVA scienceNOW segment on commercial DNA testing, offering caveats about such testing.
Q: Dr. Tanzi,
Isn't it a bit premature and unethical for Navigenics to have genetic tests like this?
Q: Can you tell us more about the calculation of personal risks for chronic diseases that are given by Navigenics? Is this credible data? (We don't think so.) Anonymous
Q: The genetics associated with common diseases such as cancer, heart disease, and diabetes are very complex. How do companies quantify an individual's risk of getting one of these diseases? Which genetic markers are being used? Have they been extensively studied? Is the basic knowledge about the association between genes and disease really at the point where we can use that information to start predicting disease risk? Don't you think the general population, with its lack of genetic literacy and understanding, will misinterpret the significance of their results?
Anonymous
Rudy Tanzi: The fact is that we are still in the very early and foundational days of being able to reliably predict one's lifetime risk for common age-related diseases, e.g. cancer, diabetes, cardiovascular disease, Alzheimer's disease. Personally, I have serious concerns about the validity of the numbers being provided by Navigenics regarding one's lifetime percent chance for getting common age-related diseases (as shown on the NOVA segment). This is because for the various diseases they are testing, there are many different risk-conferring and protective gene variants, and it is the combination of all of these risk-conferring and protective gene variants that influence one's lifetime risk for late-onset common diseases. For most of these diseases, only one or two established genetic risk factors are known; the majority of risk/protective genetic factors involved have yet to be established.
For example, in the genetic test for Alzheimer's disease, Navigenics appeared to be reading the results of a single variant (SNP) that acts as a marker for the late-onset Alzheimer's genetic risk factor, APOE. Based on that single marker, they tell a client their percent chance of getting Alzheimer's disease. The problem is one cannot reliably do this with only the APOE data.
If we consider all of the combined genetic risk and protective factors contributing to one's lifetime risk for Alzheimer's disease, it has been estimated that APOE is only 30% of the pie. Many other unknown gene variants, both protective and risk-conferring, combine with APOE to influence one's genetic risk; we have not yet confirmed the identity of the other genetic variants, which make up 70% of the equation. So, to inform a client of their lifetime risk for Alzheimer's based solely on one gene variant, as Navigenics appeared to have done on the NOVA segment is, in my opinion, not a very good idea, especially when one considers the potential psychological consequences. This is also the case for most other common diseases that Navigenics (and similar companies) are attempting to test for.
While the scientific intentions of Navigenics and similar companies are worthy and embody the future of personalized medicine, I and many other geneticists strongly feel that it is much too early to be doing genetic testing for these common age-related diseases because we currently do not possess sufficient genetic data to do so in a reliable and responsible manner. The identification of the full set of genes for each disease being tested will be necessary to reliably and accurately carry out the types of analyses that Navigenics is attempting. As a geneticist, I would be very interested in hearing from Navigenics as to how they calculate and justify the lifetime risk percentages for Alzheimer's and other common diseases in the absence of the complete genetic data required to reliably do so. In my opinion, for a consumer to seek out these genetic tests now is premature and ill-advised.
[Editor's note: Navigenics was given the opportunity to respond to this issue, as well as address the q&a in general. The Navigenics response appears at the end of the q&a.]
Another interesting point is that, to my knowledge, the Affymetrix SNP chips being used by Navigenics for genetic testing were originally intended for research purposes to find disease genes, and were not specifically designed for clinical testing. The chips do not concentrate on risk-conferring SNPs for disease, but instead include SNPs that tag sections of the genome that are inherited as "blocks." These genomic "blocks" were determined by the second phase of the human genome project called the "HapMap." The chips were originally intended to test as many blocks of the genome as possible to find genetic associations between diseases and specific genomic blocks (via SNPs that tag each block). The chips were not specifically designed for clinical testing for disease risk. Navigenics has adapted their analyses to use these chips for this purpose. In my own lab, we are using these chips in an effort to find the remaining 70% of the genetic factors (beyond APOE) that influence risk for Alzheimer's disease (as part of our Alzheimer's Genome Project being funded by the Cure Alzheimer's Fund).
Finally, I would like to note that in my interview with NOVA I repeatedly emphasized that I believe it is premature and ill-advised for consumers to seek out genetic testing for common diseases by Navigenics and similar companies, simply because we need much more genetic data to carry out such tests with a reasonable level of scientific reliability and accuracy. Most human genetic experts believe we are minimally 5-10 years away from being able to predict one's lifetime genetic risks for common diseases with reasonable accuracy. While I believe that folks have the right to seek out information about their genetic risk factors for disease, the testing must be reliable and accurate, it must be delivered with proper genetic and psychological counseling, and it must be protected from potential discrimination. Currently, the Genetics Information Non-discrimination Act (GINA) protects against genetic-based discrimination by employers and health insurance companies, but not by life insurance and long-term care insurance providers. The worst-case scenario is being denied life insurance or long-term care insurance, or having higher premiums based on one's genetic test results, and especially on unreliable ones.
Q: How does the genetic testing with Navigenics compare to the one offered by 23andme for about $1000? When you are determining disease risk, does this approach give any hint to the problems of, say, chromosome deletions as opposed to translocations that could have been caused by environmental exposure regardless of genetic makeup? Myrna James, Winchester, California
Tanzi: Dear Myrna,
I do not believe that the tests being carried out by these companies detect chromosomal deletions or other structural variations caused by environmental factors. As mentioned in my above answer, the genetic tests being offered by all of these companies are in my opinion, premature and unreliable. This is because we do not yet know all of the genetic risk and protective factors contributing to one's lifetime risk for these diseases.
Q: Dr. Tanzi, thank you for taking my question.
I am a customer of 23andme.com and am very happy that I had my genetic code mapped.
Recently, I found out that I have a "substantially higher risk (4.2X greater) of diffuse-type stomach cancer" because of my TT genotype. This result is based on studies carried out in Japan and Korea. Apparently, my risk is even greater because I am male.
My question is, if you had received these results, what would you do with them? I am more than willing to change behavior because of my results, but I'm wondering what expert would be able to give me the best advice on this (and frankly, on other findings from my genetic test). Do you have any suggestions?
Thank you. Anonymous
Tanzi: This is a classic example of why seeking out the tests being offered by Navigenics, 23andme, and similar companies is, in my opinion, ill-advised. It is not scientifically reasonable to use a single study of one particular population and ethnicity regarding a potential disease risk factor and then extrapolate the results for all individuals who take the test. We must first wait for testing of that particular variant in many independent study samples of various ethnicities, and then we must carry out meta-analyses (collective analyses) of all the published data. Only in this way can we ultimately determine first, whether this is indeed a genuine risk factor and second, if it is, in which populations and ethnicities it functions as a risk factor. For most genetic variants associated with disease, we do not yet have this information. Our laboratory and particularly, Dr. Lars Bertram, are currently carrying out the required meta-analyses for a few diseases, including Alzheimer's on http://alzgene.org, schizophrenia on http://szgene.org, and Parkinson's disease on http://pdgene.org. While there are rare, familial, early-onset (<60) mutations that can guarantee some forms of cancer, for the late-onset common forms of most cancers, in which prevalence increases with advanced age, the genetic landscape is very complex, involving many different gene variants that collectively influence one's lifetime risk. And, of course, the environment also plays a large role. For most cancers and common age-related diseases, labs are still carrying out the research necessary to establish the full sets of genetic risk factors involved. Until we accomplish this goal, we cannot carry out reliable and accurate genetic testing for most common age-related diseases, including most forms of cancer.
Q: I would like to know if there is a way to find out what ethnic/nationality I am. I know who my mother is, but she was gang-raped when I was conceived. Is there a test for that, and how much is it? Anonymous
Q: I'm adopted. What kind of testing would help me learn about my birth family? Do companies doing this kind of testing have a large enough DNA database to find my family information? Anonymous
The SNP chips being used could theoretically provide information about ethnicity, but I do not believe this service is provided as part of Navigenics' genetic testing.
Q: There is a bio-science company (Genelink) launching a subsidiary company (GeneWize) August 1, 2008. Their claim is that no other company currently has the capabilities to collect, test, analyze, and then customize and mass-produce a personal nutritional supplement. A DNA collection kit is sent to your home, and you follow the instructions and return it. Within a couple weeks you are supplied a detailed analysis and a supply of personalized and customized vitamins (all in one pill). My question is this: Do you believe this product is viable and beneficial, and would you take it yourself? Steve Cosentino, West Palm Beach, Florida
Q: It is now possible to obtain DNA testing and personalized nutritional supplements formulated and based on the results. The goal would be to provide nutritional support to boost enzyme performance where scientific studies indicate such support can prevent disease. Do you feel this is a good idea whose time has come? Judi Spillers, Melbourne, Florida
Tanzi: Dear Steve and Judi,
In my opinion, we are scientifically and medically not at the point of using genetic tests to tell you exactly all the supplements you should be taking. However, based on simple blood tests (non-DNA-based), some interesting data can be obtained. For example, if a blood test shows that your homocysteine levels are too high, potentially placing you at risk for heart disease and Alzheimer's, you could take extra folic acid to attempt to get those levels back down. Otherwise, I would not recommend companies selling supplements based on genetic testing.
Q: What will happen when we do personal genetic profiling for insurance risk stratification? Maybe we will be able to do this with a national health insurance program, but with private payers any hint of risk and your premium is sky-high. Should we make a three-year-old pay a high health insurance premium for a high rate of breast cancer risk? Greg Sharon MD, Chicago, IL
Tanzi: Dear Greg,
While we are currently unable to carry out reliable genetic profiling for most common diseases, in 5-10 years we will likely have enough data to do so. By that time, we will need comprehensive genetic privacy laws, which are not yet fully in place. Currently, the recently passed Genetic Information Non-discrimination Act (GINA) protects from genetic-based discrimination by employers and health insurance companies. The problem is that life and long-term care insurance are not yet protected. So, GINA will need to be expanded by the time reliable and accurate genetic testing for lifetime risk for these diseases becomes possible, most likely 5-10 years from now.
Q: We lost a 34-year-old son to colon cancer. He had colon polyps as an infant and eventually lost all the colon except the bottom five or six inches by resections. Was this predicable decades ago by DNA testing? Dr. Tom Mote, San Antonio, TX
Tanzi: Dear Dr. Mote,
So sorry to hear this. No, it was not possible to predict this when your son was an infant.
Q: I have the hepatitis C virus. Host response plays a role in the natural progression and response to treatment of the disease. Do you think it will be helpful in the future to have genetic sequencing to help predict the probable disease progression or responses to treatment in this and other diseases based on "host response" factors? Vic Cabot, Grand Rapids, Michigan
Tanzi: Dear Vic,
Yes, this will be possible with more research, but we are not there yet. Genetic labs are currently working on this for hepatitis and other pathogens. I would give it 5-10 years.
Q: My grandmother has Parkinson's, diabetes, and a heart condition. How can I be tested for genetic DNA? What is the personal cost of this? Also, my grandma, mom, aunt, and I all required a hysterectomy by age 30. Now I have a 16-year-old girl. I am concerned for her! Will this DNA test tell why all of us have needed this surgery? Leeann, Palo Alto, California
Tanzi: Dear Leeann,
Unless these conditions involved early onset (under 60 years old) and showed a familial pattern of inheritance, we could not reliably test you for risk. We are still in the process of collecting the full set of genetic variants (both risk-conferring and protective) that contribute to inheritance of susceptibility for the more common, late-onset (>60 years) forms of these disorders. Until then, one's lifetime risk for these and other diseases cannot be reliably and accurately determined. It may take 5-10 years before we can reliably inform someone of their lifetime risk percentages for these and other common age-related diseases.
Q: My father and his father passed of Huntington's disease. My physician suggests I join a study. Would that really help anyone? CJ, Chicago
Tanzi: Dear CJ,
Yes, groups such as the Hereditary Disease Foundation, High-Q Foundation, and the Huntington's Disease Society of America can provide valuable genetic information about the disease as well as up to date information regarding ongoing clinical trials of new therapies. As an aside, the Huntington's disease gene was the first disease gene to be localized using genetic linkage analysis. I was fortunate enough as a student to work on that pioneering study headed by Dr. Jim Gusella between 1980 and 1983.
Q: I have prostate cancer and my family has a history of having this disease. Knowing this and having access to genetic testing, would it be possible for someone like me to have his prostate removed before the cancer starts? Can my two boys take advantage of this knowledge? Yes, it's a major decision, but my grandfather died from prostate / bone cancer, and that's what I have as well. David Starek, Seneca, South Carolina
Q: I hear all the time about breast and ovarian cancer but little to nothing about testicular and prostate cancer. Are there any tests available for those two diseases, and do either have a genetic component to them? Anonymous
Q: Breast cancer and prostate cancer have many mirror similarities. In considering this, are there any similar genetic changes that might predict prostate cancer? Gerald Wery, Rochester Hills, Michigan
Tanzi: The breast cancer genes BRCA1 and 2 have been linked to other cancers including prostate cancer in some studies. Over a dozen gene variants have been associated with prostate cancer. However, it is not completely clear as to which variants best predict the disease. Several groups are currently planning to offer genetic testing for prostate cancer. As with Alzheimer's, diabetes, etc, I believe that meta-analyses of all the available genetic data will be necessary to determine which variants most reliably predict risk. Until then, I would have concerns about the validity of such tests.
Q: My mother has had Hashimoto (an auto-immune disease) for 40 years. I was diagnosed with it about three years ago. Is there a test to see if my child will get it? Will we be able to create a vaccine for genetic disorders once we have detected them?
Thanks for considering my question. Sophie, Oviedo, Florida
Tanzi: Dear Sophie,
Variants in several genes have been associated with Hashimoto disease, e.g. interferon-gamma and interleukin 4. However, to my knowledge, a reliable test for risk has not yet been established. For a vaccine to work, an antibody must target and immobilize a particular protein leading to clinical benefit. Some genetic diseases may be amenable to this mode of prevention/treatment, e.g., Alzheimer's, some cancers, multiple sclerosis.
Q: Dear Dr. Tanzi
I am a second-generation Parkinson's patient with more non-motor than motor-specific symptoms-more to do with the prefrontal functions as has been studied by the team at Bethesda (NIH, etc.).
I am fearful for my sons and daughters, but have more inclination and concern for our sons. What recent work has been done in this area? I once had an opportunity to speak to Dr. Howard Federoff before going to the first World Parkinson Congress, and he had stated that there was work being done in a method by which (I can't remember if the method was genetic) markers of some sort were under development. Is there any further work being done in this area? (I do applaud Rep. Slaughter's legislative work in this area.)
My symptoms have progressed to the wearing-off state where a 2 hour and 30 minute meds interval (Comtan limit of 1600 mg) requires additional dosages at the 1 hour and 15 minute midpoint. My ability to filter visual stimuli forced me today to stare at the floor tiles as my wife led me through the shopping mall due to being sort of "engulfed" by my visual surroundings, about 20 to 30 minutes before my next meds.
I don't see STN DBS as a possbility, perhaps a multiplexed-probe system that addresses prefrontal issues, but that seems far down the road, as in the case of the work being done by Dr. Bohen in relationship between the research on Alzheimer's and PD at Ann Arbor, but his work has little to do with sensory stimuli as I understand, and is only a study at this point. TMI? Not a cure either. The Colorado work I believe may help with my off periods, but there are no trials in Rochester. That would leave Gene Transfer, pending a HESC breakthrough. But the best improvement seen is about 40%. My prayers are with you and your wonderful and devoted colleagues.
I thank you for your valuable time. Gary Hilburger, Rochester, New York
Tanzi: Dear Mr. Hilburger,
Thanks, and I wish you the best of luck with your treatments. Dr. Federoff is an outstanding clinician, who could provide you with the best information on this question.
Q: Discovery of genetic contributions related to diseases and behavioral traits such as mental illness and aggressiveness that sometimes leads to criminal actions is very close. Do you think doctors or parents should make a decision to terminate a pregnancy based on the research available? What will this do to the gene pool? Could this potentially eliminate crucial traits to the survival of the human race? Glenn, Lakeland, Florida
Tanzi: Dear Glen,
It is likely that at some point, perhaps 10-15 years from now, it may be theoretically possible to predict susceptibility to certain criminal behaviors with reasonable accuracy. Even now, there are certain chromosomal patterns that can point to this. But, I personally believe that to screen for such susceptibility in advance and then weed out the carriers, would be a grave mistake with potentially disastrous consequences for the human race. I also believe that strong support from family and friends can go a long way even on the backdrop of such genetic susceptibility. The strength of our genepool depends on genetic variation. I personally do not believe we will ever have the wisdom to determine who should be born or not; the arrogance inherent in such a process would be mind-boggling.
Q: Has a gene been discovered for alcoholism? Is there a test to prove that a person is alcoholic?
Thank you. Anonymous
Tanzi: Yes. Genes affecting the neurotransmitter GABA have been associated with alcoholism. However, I do not believe they can be reliably used to test for susceptibility in the absence of the knowledge of other risk-conferring and protective gene variants. Many labs are currently trying to find the full set of genetic risk factors that confer susceptibility to alcoholism. I do not believe there is a gene variant known that absolutely guarantees alcoholism. So, it would not appear to be possible to "prove" such a condition.
Q: Is there currently a test for the delta 32 mutation against HIV infection, and where could one get the test? Anonymous
Tanzi: I believe a company called HIVmirror tests for this. I would certainly hope folks who seek such a test do not do so with the belief that a positive finding means they do not have to be as careful with unprotected sex.
Q: Diabetes is prevalent in my mother's side of the family but not my father's. Both my siblings (one boy, one girl) now have diabetes, but I do not. Is there a genetic test to tell whether I will get it, and is there any way to avoid getting it? Anonymous
Tanzi: Except for some rare mutations known to directly cause diabetes, usually early in life, the genetic risk factors that work together to influence risk for late-onset, common forms of diabetes are not yet firmly established. Reliable and accurate genetic testing will require knowledge of that full set of genes, likely requiring 5-10 more years.
Q: How much impact does epigenetics have upon the health of an individual? Liz Whelan, England
Tanzi: Epigenetics plays a huge role in how our genetics influences our health. We are in the earliest possible days of determining how epigenetic factors work together with genetic risk variants to influence one's susceptibility for disease. This is a very important and blossoming field in genetics.
Navigenics Response
Navigenics is appreciative of the opportunity to respond to this Q&A, which brings up relevant issues but overlooks several critically important points that reinforce the need for the availability of personal genetic information.
The majority of the questions submitted in this post are from people seeking more information about genetics and, in particular, about their own genetic makeup, either because they don't know their own family history or because they or someone in their family has been diagnosed with a concerning health condition. Posts like these demonstrate the strong societal interest to use the information that has been unlocked by the genomics research of the last decade. It is time for these discoveries to move into the hands of physicians and consumers, who can use them to improve their health. Indeed, we believe that future generations will look back at these years and mark them as the beginning of the preventive genomic medicine era.
Navigenics was founded to understand the nuances of this new field and deliver this information in a responsible, accurate, and transparent fashion. This is clearly a challenging task. Most of the scientific literature in genomics is not of sufficiently high quality to use in the care of patients. Indeed, when some geneticists argue that the science is too early for use in patients, they are correct to a large extent. However, there are clearly a handful of complex genetic diseases with well-established independent risk alleles that have been reproducibly shown to raise the risk of disease. The criteria that Navigenics uses to judge the quality of a scientific study are extremely conservative and consequently less than 5% of the genomic literature meets our standards. We therefore agree with geneticists who feel that genomic testing companies that have less rigorous standards do a disservice by either giving a false sense of reassurance or by provoking unnecessary anxiety.
Any effort toward a major paradigm shift, particularly when it relates to medicine and public health, should be met with healthy skepticism, and we respect and understand the opinions of some geneticists in the field. However, there is a puzzling opinion sometimes expressed by geneticists that physicians or consumers should not use genomic information because "it is too early." The typical argument put forth is that before genomic testing is used in patient care, medical science should wait decades to have all the genetic markers fully discovered and have all the extremely complex models in place between multiple genetic risk factors and environmental risk factors. In summary, they urge waiting until genomic medicine is completely and perfectly predictive. Why is this opinion puzzling? Because physicians have used probabilistic risk factors for decades and have consequently helped countless patients despite having this "imperfect" information.
Readers will no doubt find the following scenarios rather familiar: Your doctor tells you to stop smoking or you may get lung cancer. A patient presents to an emergency department with chest pain and is admitted to the hospital because their risk factors of diabetes and high cholesterol put them at elevated risk for heart disease. These sorts of clinical encounters happen thousands of times a day and demonstrate physician acceptance for the use of environmental risk factors in patient care. The study of environmental risk factors over the last century has resulted in known exposures that raise an individual's risk of disease. These include smoking as a risk factor for lung cancer, obesity as a risk factor for type 2 diabetes, and a lack of exercise as a risk factor for heart attack. Inherent in these studies is the understanding that an exposure to an environmental risk factor is by no means completely predictive of developing disease. For example, an individual can be obese and never get diabetes, be a heavy smoker and never get lung cancer, or have elevated cholesterol and never get a heart attack. Yet we as a community commonly tell an individual that these exposures will raise their probability of disease, based on these population-based studies. Modern medicine has not waited for the exhaustive discovery and analysis of every environmental risk factor that exists before using this information. Rather, physicians have used what incomplete probabilistic information they have for the good of their patients.
Genetic risk factors that increase an individual's risk of disease are identified in the same fashion as environmental risk factors. Common risk alleles that have been studied in thousands to tens of thousands of individuals are associated with an increased probability of disease. Like environmental risk factors, these genetic markers simply indicate a probability of disease and not a diagnostic certainty. Furthermore, it is important to understand that genetic and environmental risk factors have similar probabilistic values. It is commonly thought that the genes Navigenics identifies through whole-genome association studies have risk effect sizes that are so small as to be negligible. This is not the case. For example, the effect size, or relative risk of having a heart attack if you carry two copies of an allele on chromosome 9 is the about the same as if you are a smoker or if you have an LDL cholesterol level of greater than 160. What this means is that genetic risk markers give physicians an additional tool to assess patients for risk of disease that equals or surpasses the usefulness of environmental risk factors. Given the looming health-care crisis that will be caused by common and chronic diseases like diabetes, common genetic markers give health-care providers an additional tool to work toward practical solutions aimed at prevention now in our generation, rather than pushing the responsibility off for another decade.
In addition to the importance of stringently selecting the highest-quality genetic markers, we also believe that the laboratory facilities in which such tests are conducted must be of the highest quality. Many companies use research facilities to conduct these tests. We agree with those geneticists that feel that these facilities were never intended to be used for tests that would be used in patient care. The testing performed by Navigenics is done in a government-regulated, CLIA-certified molecular diagnostic laboratory, with stringent quality metrics that are intended to insure accurate results for patient care. Additionally, every individual who takes the Navigenics test is paired with a certified genetic counselor and can avail themselves of both pre- and post-test counseling so they understand that this information is to be used in combination with other risk information so that the individual and their physician have another tool in their tool kit to facilitate prevention.
In summary, many of the concerns regarding genomic testing are valid, and consumers and physicians should be extremely discriminating about which service they choose for their personal genomic information. It is imperative that genomic testing services use only highly validated and robust genetic markers and that they convey this information in a responsible fashion. Furthermore, it is important to understand that these genetic markers are important tools to arm physicians to help their patients. Every diagnosis of breast cancer, and every death from heart disease, serves as a frequent reminder that modern medicine can not be complacent with its current methods and should not wait for a hypothetical perfect predictiveness that may be decades away. It is not too early to start responsibly communicating robust and validated probabilistic genetic risk information to those who are interested.
Dietrich A. Stephan, Ph.D.
Cofounder, Navigenics
Chairman, NIH Neuroscience Microarray Consortium
Director, Neurogenomics Division, TGen
Vance K. Vanier, M.D.
Chief Medical Officer, Navigenics