No magic markers
Another challenge to consumer genetic testing has been consistency. Although most sequencing processes are similar, and the DNA itself is a unique code, reports from various groups for a single person can have striking differences. "People interpret the information that comes out of the sequence differently," notes John Boyce, co-founder of Delphi Bio, a biotech consulting firm, who also founded the annual Consumer Genetics Conference.

The main problem, as explained by Boyce and others, is that solid, predictive biomarkers have not been agreed on for most conditions. And because one company might select a different set of biomarkers for a disease than another, "the estimates of odds conferred may vary widely," Pearson says.

Genome-wide association studies, in which genetic variants are matched with different health outcomes across a study population, promised to give researchers a tidy map of diseases and related genetic variations. But, particularly in studies that use short-cut sequencing methods in which common variants are sampled rather than full genomes, the results so far have been disappointing to many population geneticists. As Pearson points out, "a spelling variant does not get to be common in the population if it's really harmful," which means that most harmful variants are rare—and confidently predicting risk for an individual based on a relatively small sample population can be problematic.

"The sample populations are not large enough to truly find the predictive biomarkers," Boyce says. These relatively small sample sizes could hamper companies' ability to make exacting correlations between genetic variations and health or disease risk.

Another company Boyce is involved with, GnuBio, aims to drop the cost of sequencing to $30 per genome in the next two years. With fast and cheap sequencing, he expects the number of logged genomes will skyrocket. "Once the numbers [of genomes] are high enough," Boyce says, "we should have the power to see deeper into the genome and find the predictive biomarkers."

Showing the science
Most of the companies offering personal genome scans use their own algorithms for calculating a person's disease risk or other health information, and the validity of the results has remained relatively unexamined.

Since the Pathway Genomics consumer test controversy this spring the FDA has sent public letters to five companies requesting meetings to discuss the regulatory future of the latter's genetic products. Four of the firms—23andME, deCODE, Knome and Navigenics—offer direct-to-consumer genome scans (Personal Genome Service; Complete Scan; KnomeCOMPETE; and Health Compass, respectively), which the FDA has labeled as medical devices and, as such, they would be subject to closer inspection and possibly premarket approval.

"It is important that [the scans] be analytically and clinically accurate so that individuals are not misled by incorrect test results or unsupported clinical interpretations," Alberto Gutierrez, head of the Office of In Vitro Diagnostics at the FDA, wrote in letters to the companies.

Rather than pulling all of the tests off the market immediately, however, the FDA plans to evaluate the claims each company is making and the data they have to back them up.

"By law, they're medical devices," Gutierrez explains of the tests. Depending on how vital or potentially risky the information the tests claim to provide is, services might be evaluated differently. If one test is "going to give people a risk assessment that a kid is going to be bald when they grow up, I think that's a very low risk," he says. However, if a test is offering to predict how a person's body is going to metabolize a drug, "the risk of that information being misused is something that we would be concerned with."

Some working in the industry have voiced support for more government guidance. "We welcome the interest," Pearson says of the FDA's inquiries. He adds that his group shares the agency's concern that, "that new technologies are used reliably and accurately."

But the FDA—already spread thin as it deals with other regulatory challenges, including food safety and drug approval—might need to rally additional resources to provide premarket approval if the consumer genetic testing market expands as rapidly as many are predicting. "It would be impossible for them to sustain the burden of all the new tests," says Paul Kim, a partner at the law firm Foley Hoag, who has studied the regulation of the industry.

Boyce agrees, noting that if interest in personal genome sequencing is "going to become such a tidal wave at some point, it's going to be hard to regulate." He also worries that premarket approval could stifle innovation and new company start-ups.

Until the FDA meets with companies to assess the data backing up their claims and conclusions, the fate of these tests remains unknown. Gutierrez notes that the agency might simply seek to verify that product labeling is truthful and complete enough for consumers to know what kind of answers and information they might—or might not—get from a genetic test as well as whether they will be able to act on that information. The agency might also take a broader approach to ensure support for consumers who choose to dive into their genome. "It wouldn't be out of the question for us to require [that there be] genetic counseling on hand," Gutierrez says.

Back in the lab
Most commercial tests are not entire genome scans, but rather analyses of common genetic alterations that have been linked to disease risk and other health issues. This shorthand version, although more economical and efficient, can leave much of the genome unexamined. Such details might not concern someone specifically looking for their family's risk of age-related macular degeneration, whereas for purposes of future discovery leaving out so much of the genome can be problematic, much as a partial population census would likely fail to bring back crucial details about a community.

2 Comments

1. 1. nexstepone 05:05 PM 6/29/10

   I agree with the general content of the article, however I will suggest that attempting to define the "success of gene testing" in concrete terms is by it's very nature inherently flawed. We all want our healthcare professional to provide us with definitives. Give my problem a name and an ICD9 code. However human kind represents a dynamic, multi-factorical complexity such that the best we can provide is an increase or decrease in the probability of an outcome. Being able to take the complexity of analysis and reduce it to a reasonable analysis is the challenge. Any analysis needs to be in the hands of the TRAINED healthcare professional, not the consumer.
   
   I will suggest that the area of drug metabolism, gene analysis has the probability to improve the status quo. A study last year by MEDCO reported that $145 Billion of the $292 Billion spent on prescription drugs DID NOT WORK. The MEDCO findings indicate that the introduction of a drug into a patient protocol is at best, a toss of the coin in terms of it's probability of meeting the drug's intended target or goal.
   
   Prescription drugs that utilize hepatic pathways have been defined. I will further suggest that KNOWING the degree that a drug pathway is compromised provides for an INFORMED decision by the healthcare professional. The group of genes called CYPs (Cytochrome P450s) represent the intersection in the pathway for many drugs. CYP450s serve a dual purpose, they are part of the pathway of elimination (metabolizing) of drugs, both prescription and OTC however in some instances these genes are needed to activate the drug (Pro-drug). Some foods, nutrients, OTC drugs, exotoxins and other prescription drugs can and do impact how these genes express (produce the enzyme/protein). In effect these other substances have a probability of compromising the elimination or activation of drugs that use CYP pathways, even if the gene is "normal" in function (EM - extensive metabolizer).
   
   The diversity of function in drug metabolizing genes is an extremely important variable to consider in optimizing a patients drug protocol. CYP genes vary in function from poor metabolizers (PM - NO FUNCTION) to ultra-rapid metabolizers (URM - up to 13 duplicates) of the same gene. Knowing the function of a CYP is an important factor to consider. In my experience in analyzing this group of genes for patients, most drugs use more than one pathway for metabolism. So complexity is the norm and the possible combinations of variables exceeds even the brightest of minds.
   
   The pharmacologic and toxic effects of certain drugs are exaggerated in a significant percentage of the population due to a heritable deficiency in CYP 450 enzymes (Tucker, 1994; Meyer, 1994; Smith et al., 1998). The two major polymorphically (many forms) expressed CYP450 enzymes are CYP2D6 and CYP2C19. The incidence of the poor-metabolizer genotype varies among different ethnic groups. Five to 10 percent of individuals in Caucasian populations are of the PM (no function) CYP2D6 genotype and have deficient metabolism of debrisoquine (an antihypertentive drug metabolized by CYP2D6) and over 25 other drugs. However, less than 1 percent of Japanese subjects are defective in CYP2D6 activity. In contrast, 20 percent of Japanese subjects are poor metabolizers of S-mephenytoin (an anticonvulsant metabolized by CYP2C19), whereas less than 5 percent of Caucasians are so affected (Kaneko et al, 1999).
   
   Significant improvements in patient outcomes and sharp reductions in the cost of ineffective medications and the effects of ADRs and ADEs can be achieved by considering the effects of genetic variation and of other drugs, supplements and foods in a patients protocol.
   
   Several examples follow.
   
   Drug-gene interaction: Warfarin therapy
   
   Warfarin is an anticoagulant commonly used to prevent and control blood clots, but it is complicated to use because the optimal dose varies greatly among patients. If the dose is too strong, the risk of serious bleeding increases and if the dose is too weak, the risk of stroke increases. In a 2006 study, the AEA-Brookings Joint Center for Regulatory Studies estimated that formally integrating genetic testing into routine Warfarin therapy could allow American users to avoid 85,000 serious bleeding events and 17,000 strokes annually. They also estimated integrating genetic testing into Warfarin therapy could save $1.1 billion annually.
   
   Drug-gene and drug-drug example: Plavix and Protein Pump Inhibitors
   
   Plavix (clopidogrel) is the second largest selling drug in the world. It is commonly prescribed to reduce the risk of blood clots and potentially fatal heart attacks in patients who have received stents to open clogged coronary arteries. Researchers have found that patients with variations in the gene cytochrome CYP2C19 have a 3.6 times greater risk for major adverse cardiovascular events such as death, heart attack, and stroke; the risk was greatest in CYP2C19 poor metabolizers.
   
   Recent studies have also created confusion about the appropriate use of Plavix and Proton Pump Inhibitors (PPIs) a group of drugs including Prilosec, Prevacid and Nexium, that produce a pronounced and long-lasting reduction of gastric acid production. The Medco Outcomes Study, a retrospective analysis of 16,690 patients taking clopidogrel for a full year following coronary stenting revealed that patients who also took a PPI for an average of nine months experienced a 70% increase in the risk of heart attack or unstable angina, a 48% increase in the risk of stroke or stroke-like symptoms, and a 35% increase in the need for a repeat coronary procedure. The proposed mechanism is that PPIs inhibit CYP2C19-mediated metabolic bioactivation of clopidogrel into its active ingredient. Clopidogrel is a pro-drug that requires the enzyme to take it's active form.
   
   Drug-food example: grapefruit and common statins
   
   Statins are the most widely prescribed cholesterol-lowering medications. Grapefruit contains the compound bergamottin, which interacts with the cytochrome P-450 and P-glycoprotein enzyme systems. These enzyme systems are responsible for breaking down statins, as well as other drugs, into more usable chemicals and transporting them in the body. When a person consumes grapefruit or grapefruit juice at or around the time of their statin dose, bergamottin prevents the enzyme systems from breaking down the drug, causing high amounts of the statin drug to accumulate in the body. This can cause a variety of very dangerous health problems, such as liver damage or severe muscle and kidney damage.
   
   So the bottom line is, there is NO simple solution. The only way for the healthcare professional to determine the viability of a drug working as intended is to KNOW how this group of CYP genes function and what other substances have influence on them. By having an understanding of how the pieces of the puzzle interact, and how our own unique puzzle is laid out, can the healthcare professional have a probability of decreasing a negative effect (side effect, ADR or ADE). Unfortunately most physicians were not trained in pharmacogenomics and the analysis is time consuming. The result is that the probability of a physicians "guess" being correct is not better than a toss of a coin. Ever heard the statement by your physician "let's try this, and if it doesn't work we will try something else?" The future in drug protocol effectiveness is based on Personalized Medicine, and in this case, Pharmacogenomics is the key to that reaching that end.

   Reply | Report Abuse | Link to this

2. 2. inbetweener 01:20 PM 7/3/10

   Scratch that, I'm blind and didn't spot that the link was in the text, but still citing it at the end clearly would be nice.

   Reply | Report Abuse | Link to this