PART II

Stephan: The worldwide health burden is caused by a handful of diseases. Heart disease, diabetes, obesity, prostate cancer, multiple sclerosis - all things we or our families are going to get - probably account for 90 percent of the disease burden in the human species. Less than 10 percent of diseases are very rare Mendelian mutations identified over the last 15 years. So from that perspective -- and if you believe those common diseases are predominantly caused by common variants floating around in the population -- and the association studies published to date support that assumption -- then you can get most of the information you want out of high-density SNP scans. We'll be doing one million polymorphisms, and using linkage disequilibrium and communicate on those common risk factors.

If you look at it that way, whole-genome sequencing might only provide [another] 10 percent. For example, we currently think autism is caused by private or rare mutations and we haven't found a whole-genome association signal to date. So autism might be teased out using a sequencing strategy rather than a genotyping strategy. On a practical level, the biggest bang for the buck is to use a one-million SNP array, capture the common variants, then slowly layer in more and more density.

KD: What have you learned from other direct-to-consumer genetic services, including the IBM Genographic Project?

Stephan: We've learned a lot from all those companies. For example, we've learned from the National [Geographic] Genographic Project that over 500,000 people have bought [the $99 kit], are interested in a genealogy strategy, and are not uncomfortable sending their DNA through the mail. It's also been valuable learning that the direct-to-consumer approach can work if privacy and anonymity are maintained. We've learned from DNA Direct that people are comfortable using a telephone [counseling] approach. That was valuable.

From some others, we've learned that if you can't maintain the highest-quality standards, people will quickly ferret that out and it's going to hurt the field. So from Day 1, we're making sure that every nuance of this operation is perfect. We've seen congressional hearings, FDA guidances, and we're following all those paths.

KD: Many people are interested in how Navigenics will compare to 23andMe.

Stephan: I've spent time talking to [23andMe co-founders] Anne [Wojcicki] and Linda [Avey] and, as I understand it, we're very complementary. Navigenics is focused squarely on medical risk assessment on actionable conditions. As I understand what they are doing, 23andMe is focused on ancestry genealogy, and that's going to be their space.

For example, we're calculating all ancestry in the background, so we can give ancestry-specific risk assessments -- but that won't be visible to the person -- just so we can refine risks. We don't want to dilute what we do best, by doing these offshoots.

KD: There's been a surge of whole-genome mapping papers in top journals this year. Are we close to really being able to gauge the risk factors for common diseases by surveying SNPs?

Stephan: What we've built in-house is a team of expert geneticists, epidemiologists, statisticians, computational biologists, and bioinformatics folks who vet every paper published in the world and try to understand whether that paper is real or not. Was the study appropriately powered? Was the case-control cohort correctly phenotyped? Or were they appropriately matched? Was the technology robust? Were the algorithms of the correct generation? Were the effects in the same direction with the same allele across multiple replication cohorts?

For example, and with no disrespect to the Wellcome Trust folks, whom we applaud for making their data available [2007 Nature paper identifying genes for seven common diseases], but their strongest association in bipolar disease did not make it through our quality control criteria. When we looked at the distribution of AAs, ABs and BBs of the associated SNP in their study population, we saw overlap in those clusters and [conclude] the association may have been a false positive. Because of that possibility, it won't be on the Navigenics menu. We really feel like a lot of this stuff is being pumped through peer review with little critical evaluation of how it was done.

KD: I presume you've volunteered your own DNA for genotyping to test the service?

Stephan: No one who's taken it is frightened or scared when they get their results back. It's all probabilistic -- just like a cholesterol test. People have an intuitive [understanding of] testing and they're not scared. One thing that was insightful to me was just how high the average lifetime risk for some diseases is. The average lifetime risk for Type 2 diabetes is 30 percent.
[In my case] for example, great news! I don't carry an ApoE4 allele. That has major implications on my stress levels and how I live my life going forward. I don't have to focus on implementing the promising prevention therapies coming online. I also learned that -- my mom died of breast cancer... for common [gene] variants, I'm loaded over the general population so that might at some point [encourage me] to get a checkup...

We have 20 conditions we're launching with -- I found I was average loaded for obesity... this is a validator to me, I'm in the average ballpark for body mass index. I'm below average for Type 2 diabetes.

KD: How will you communicate these genotyping results to the public?

Stephan: We'll have a very intuitive dashboard to provide you with an intuitive sense of what you should be focusing on in terms of managing your health proactively. It will encompass lots of information, supported in specific ways, taking into account average lifetime risk, how you're loaded relative to the general population, age of onset, a number of different nuances... Another aspect is that because we'll have the whole genome captured in a [high-density] way, we can update you with respect to predispositions to that disorder as they are published.

KD: How will you safeguard privacy of people's genetic data?

Stephan: Our position is that your genome belongs to you and you alone. We'll deliver the information directly to the person... and have strategies in place as to how the person should talk to their physician. There'll be printable materials to take into their physician. Clearly, GINA (the Genetic Information Nondiscrimination Act of 2007) plays a big role, and since it passed in the House, and the President said twice he'd sign it, we believe GINA is going to pass before the end of the summer... But our position is that GINA is critical, privacy is critical, and consumers who get this information need to be advised.

Subscribe to Bio-IT World magazine.