By Turna Ray
Recognizing that not everyone might wish to know their genetic predisposition to Alzheimer's disease, direct-to-consumer genomics firm 23andMe last week gave customers the option to learn whether they have variants in the APOE gene linked to the disease.
Under this model, the company will provide more than 60,000 of its customers with the opportunity to learn their APOE status with the click of a few buttons indicating they understand the risks and limitations of learning this information. This process ensures that customers are aware of the risk of this potentially sensitive data, but some are questioning whether the direct-to-consumer model is the appropriate mechanism for delivering information on APOE genotypes, even with additional precautions.
While variants in the APOE gene have been shown to increase the genetic risk for Alzheimer's, critics of DTC genomics firms often highlight the complex and evolving genetic underpinnings of this serious and irreversible illness as an example of the kind of gene-disease risk information that companies should not be selling directly to the general public. Some have even charged that reporting genetic risk data on Alzheimer's without the involvement of a doctor, and without any real preventative treatments, is irresponsible.
23andMe is assessing customers' risk of Alzheimer's disease based on the three main APOE variants: ε2, ε3, and ε4. The company is not providing information on variants in other genes that have been linked to the disease or to age-of-onset risk.
Allen Roses, director of Duke University's Deane Drug Discovery Institute, believes that 23andMe should not report APOE status through DTC channels. Researchers at Roses' lab at Duke first discovered the link between the APOE ε4 allele and increased risk of late-onset Alzheimer's disease. Subsequent research has well established that having one or two copies of APOE ε4 increases people's risk of developing the disease. However, APOE ε4's role in the overall development of the disease is still unsettled.
According to Shirley Wu, science content manager at 23andMe, for customers of European descent, the risk estimate provided by the company will inform them of their chances of developing Alzheimer's between the ages of 50 and 79.
Late-onset Alzheimer's disease manifests in people usually after age 65, with most diagnoses occurring after the age of 80. 23andMe told PGx Reporter that there is little population data about what specific APOE genotypes mean in terms of Alzheimer's risk for normal individuals under 50.
"The presence of an ε4 allele does not predict the age-of-onset risk, just a general average spread over approximately 45 years [55-100]," Roses told PGx Reporter via e-mail. "This is not accurate for the individual without the age-specific positive and negative predictive values."
Furthermore, he noted, "The absence of an ε4 allele also does not predict risk, just a delayed age of onset distribution."
After learning their APOE status, there is little 23andMe customers can do with this information without viable strategies to prevent Alzheimer's, observed Roses, who is currently investigating whether Takeda Pharmaceutical's diabetes drug Actos can delay Alzheimer's onset in people with a high risk of the disease based on age and genetic factors.
In Roses' view, APOE genotype data can only help in differentially diagnosing people who are cognitively impaired. "APOE ε4 homozygotes have a more well-defined predictable onset curve, but in normal ε4/ε4 individuals there is still a 30-year guess as to when cognitive impairment might begin," he explained.
From a public health perspective, Muin Khoury, director of the Centers for Disease Control and Prevention′s Office of Public Health Genomics, has urged for caution when selling genomic risk information with unproven clinical utility broadly to consumers. "My take on this has always been more conservative than others," Khoury said. "For tests that don't have the necessary clinical utility, I believe they should be offered in a medical setting, with counseling. The DTC aspect of all this makes it even more urgent that these tests are validated from a clinical utility perspective."
Earlier this week, the National Institute on Aging and the Alzheimer's Association released new diagnostic guidelines for Alzheimer's disease. In that document, the first update for the guidelines in 27 years, the NIA and AA state that while the presence of APOE ε4 can provide information about people's genetic risk for Alzheimer's, tests based on APOE genotype are currently only useful for research purposes.
An outspoken proponent of providing unfettered access to genomic data, 23andMe believes that people with the money to purchase its gene risk reports should be able to access them without permission from a doctor, even if the data are imperfect, the research is evolving, and the information lacks utility.
The company decided to begin reporting APOE variants to customers after upgrading in November from Illumina's HumanHap 550v3 chip, which tested around 600,000 SNPs, to Illumina's HumanOmniExpress, which gauges around 1 million SNPs.
"23andMe attempted to test for the APOE variants on a previous version of our genotyping platform with the intention of reporting on Alzheimer’s disease risk to customers who wished to view it," a 23andMe spokesperson told PGx Reporter. "However, our quality control process deemed that the data necessary to determine APOE status was not of high enough quality to do so on that version of the platform."
[ pagebreak ]
Using the new Illumina chip, "the data for the APOE variants have passed our quality control criteria and we are pleased to now be able to offer a report on Alzheimer’s disease to customers," the spokesperson said.
The chip upgrade has not only allowed the company to improve its genotype imputation power and report new associations, such as Alzheimer's, but the company is also charging existing customers more to have their samples run on the new platform. In order to get genetic risk reports on new diseases, 23andMe customers whose samples were analyzed on the old chip must either pay $49 plus a subscription fee of $9 per month, or they can pay $299 one-time for the life of the current platform.
Until 23andMe began providing information on APOE variants last week, it was the only company among the big three consumer genetics firms that didn't report on the genetic risk of Alzheimer's disease.
Decode Genetics' DecodeMe service reports on variants in the APOE and CLU genes to gauge the risk of developing the illness in customers of European descent. Navigenics, which previously used to sell genomic risk information under a DTC model but now markets its service through doctors and employer wellness plans, also reports information on APOE status and the risk of Alzheimer's disease.
Data Limitations
Before receiving their APOE genotypes, 23andMe customers must read and understand a note explaining that "estimates of Alzheimer's risk due to the APOE ε4 variant presented in [their] report are established in European populations only," and that "more research is needed to establish the variant's effect in non-European populations."
People of non-European ethnicity can further learn from information provided by 23andMe that the APOE ε4 variant's association with the risk of Alzheimer's is less consistent in African American and Hispanic populations, and more research is needed to confirm the "slightly stronger effect" of the variant in Asian populations.
Given these caveats, if customers still wish to learn their APOE status, they can click the "learn more" button to do so. In addition to Alzheimer's disease, 23andMe customers need to similarly "opt in" for reports on Parkinson's disease and hereditary breast cancer as determined by BRCA mutations.
The company notes that family history, environmental factors, and non-genetic factors such as high blood pressure and obesity, also contribute to Alzheimer's risk. However, these factors aren't considered in the risk score provided by 23andMe.
So, given the limitations of the genetic data associated with the risk of a debilitating and incurable disease, why report this information at all, especially for individuals of non-European ethnicities?
"We don't make any assumptions about what the customers may or may not want to see," 23andMe's Wu said. According to Wu, although 23andMe customers can self-report their ancestry, not all customers have chosen to report this information. "So, we don't want to make any assumptions about what they would want to see, but we do make clear what [group] the results are applicable to."
Similarly, Wu reasoned that when it comes to age-specific risk estimates, 23andMe is providing an individual's lifetime risk of developing Alzheimer's between 50 and 79 years, but "it's not an assumption of whether or not the person falls in that age range."
Research Advances
Since testing for APOE variants can only provide a risk for developing Alzheimer's and cannot say definitively who will develop the disease, the NIA recommends that such genetic testing should be used in a research setting to identify participants for studies.
"Most researchers believe that the APOE test is useful for studying AD risk in large groups of people but not for determining any one person’s specific risk," NIA states on its website. "Someday, perhaps, screening in otherwise healthy people may be useful if an accurate and reliable test is developed and effective ways to treat or prevent AD become available."
23andMe informs customers that the data provided in its reports are for research and educational use, not for diagnostic purposes.
While reporting APOE ε4 status may not be informative for many 23andMe customers who aren't of European ancestry, the company may be able to use customers' data to conduct research on the variant's prevalence in minority populations and whether the marker increases their risk of Alzheimer's.
"We're definitely looking at doing our own research on Alzheimer's disease and genetics," Wu said, adding that 23andMe will likely need to get more information from customers with incomplete self-reported ethnicity data to advance such research.
"We'll be conducting surveys to see what their experiences with aging, dementia, and Alzheimer's [are], specifically," she added.
23andMe has published research on genes linked to customer traits, such as hair curl and freckling. The company claims its consumer-funded research model allows access to a large study cohort and can speed the pace of genetic discoveries. 23andMe plans to eventually publish studies on genetic risk of diseases, gleaning data from its customer base.
But many of 23andMe's critics don't agree that research is the company's primary motivation for selling gene risk data. This information is not being provided by 23andMe as part of "qualified research … but has a purely commercial intent," Roses said.
In a more traditional research setting, Roses is working on refining the genetic risk prediction capabilities for Alzheimer's disease. His team's research suggests that varying lengths of the TOMM40 rs10524523 poly-T polymorphism — located at intron 6 of the TOMM40 gene and linked to APOE ε3 and APOE ε4 polymorphisms — can be used to craft a three-allele risk prediction system for gauging the age of onset for Alzheimer's.
[ pagebreak ]
Earlier this year, Takeda began collaborating with Zinfandel Pharmaceutical, a company started by Roses, to conduct a study in which normal subjects between ages 60 and 87 will be randomized to receive either Takeda's type 2 diabetes drug Actos or placebo. The aim of this trial is to assess whether Actos can delay Alzheimer's onset.
Study participants will be randomly assigned to receive placebo or Actos only if they are deemed to be at high risk for developing Alzheimer's within five years based on their age and a genetic test assessing their TOMM40 status. Study participants who fall in the low-risk category will be randomized to one of two placebo arms. (PGx Reporter 01/12/2011).
23andMe mentions research on TOMM40 and other gene variants in its educational materials to customers about ongoing genetic research in Alzheimer's, but the company notes that these newer genetic findings require further validation.
Meanwhile, although APOE ε4's association with increased Alzheimer's risk has been replicated in numerous clinical studies, there's still room for improvement. Currently, there isn't conclusive data explaining why some people expressing the variant don't develop the disease.
Between 20 percent and 25 percent of people in the general population carry at least one copy of the APOE4 variant. In late-onset Alzheimer's patients, 40 percent have at least one copy of the ε4 variant, but a large proportion of Alzheimer's patients have no copies of APOE ε4. Furthermore, the contribution of APOE and other genes, environmental factors, family risk, and lifestyle to a patient's overall risk of developing the disease is not fully understood.
Anxiety? No Worries.
Customers concerned about their risk of Alzheimer's disease are advised by 23andMe to learn their family history and talk to their healthcare provider. The company also provides genetic counseling through Informed Medical Decisions, an independent genetic counseling service that can answer customers' questions about their 23andMe test results over the phone.
While DTC genomics companies provide resources to mitigate customer anxiety over test results, these companies maintain that the majority of customers understand the limitations of the data and there is little evidence to suggest that having imperfect knowledge about one's risk of diseases is causing psychological harm.
Proponents of the consumer genomics model often counter criticism in this regard by citing the REVEAL study conducted by Robert Green of Brigham and Women's Hospital, which examined the effect on anxiety, depression, and test-related distress when adult children of Alzheimer's patients learned their APOE status. 23andMe includes a video featuring Green in an effort to educate customers about the APOE gene and Alzheimer's disease.
In the REVEAL study, researchers found no significant differences in psychological measures between the group of patients who learned their APOE genotype and the group that didn't. However, when discussing this study, consumer genomics proponents rarely point out that those who tested positive for the APOE ε4 variant were more concerned about their results than those who tested negative.
In the group where participants were told their APOE status, those who were ε4-negative "had a significantly lower level of test-related distress" than those who were ε4-positive, the researchers wrote in a paper published in the New England Journal of Medicine in 2009.
[ pagebreak ]
"Additional questions about risk perception and the effect of testing six weeks after disclosure suggested that subjects understood that their risk was higher or lower according to their genotype, and they had the expected negative or positive feelings about this news," the researchers reported. "Thus, subjects were not immune to the negative implications of learning that they had an increased risk, but these feelings were not associated with clinically significant psychological distress."
Green et al. suggest in the paper that larger studies that follow subjects for longer than one year, as was done in REVEAL, are necessary in order to conclude whether people experience long-term psychological effects after learning their APOE genotype.
Furthermore, the outcome of this study, which was crafted to test psychological effects of genotype data in subjects with a family history of Alzheimer's and who received genetic counseling as part of the study, is hardly a justification for the DTC genomics model, since consumer genomics firms are providing gene disease risk information to the general population. After all, 23andMe customers can get genetic counseling if they choose, but it's not a prerequisite for access to the Alzheimer's report.
"Most of the subjects were of fairly homogeneous and similar ancestry," the REVEAL study authors acknowledge. "If APOE genotyping had been provided without genetic counseling or to subjects who had no family history of Alzheimer’s disease, the results might have been different."
Nowhere in the paper do the REVEAL study authors dismiss the possibility that people may experience psychological harm from misunderstanding their APOE genotype outside of the context of family history and environmental risks. In fact, the authors point out that "these concerns are amplified by the recent emergence of companies offering direct-to-consumer genetic testing, with most of them evaluating single-nucleotide polymorphisms that indicate the APOE genotype, in the absence of guidelines for deciding which associations between genes and disease have sufficient clinical validity and usefulness to justify disclosure and with no gauge of the effect of such disclosure."
In 23andMe's educational video featuring Green, he states that he didn't receive compensation from 23andMe, and that he is not endorsing the DTC genomics testing industry. Green further informs 23andMe customers that some medical specialists advise against learning one's APOE results and even when given the opportunity to learn the results, many people decide not to.
Referencing the REVEAL study, CDC's Khoury observed that "there have been a number of [behavioral] studies lately that haven't documented [consumer] anxiety in relation to getting information about their risk to diseases. So, people say, 'What's the big deal then?'" In Khoury's view, "the big deal is that you don't know how consumers are going to behave when offering tests on a population basis that … don't have the clinical utility and validity for improving outcomes."
Studies showing that genetic risk information doesn't cause anxiety are of little help in advancing the field of genomics, Khoury said. "What is needed are large studies showing that applying genomics to healthcare improves outcomes," he said.
Roses also acknowledged that the REVEAL study indicated a lack of anxiety over APOE status, but noted that this "is not a real reason for buying these tests" because APOE genotype "is not particularly useful in normal individuals for an accurate assessment of their proximal risk."
If there were a therapy available for Alzheimer's disease prevention, he said, "things would change dramatically, but not based on APOE alleles."
Have topics you'd like to see covered in Pharmacogenomics Reporter? Contact the editor at tray [at] genomeweb [.] com.