Reminiscing

Today marks an historic day for 23andMe (or, should I say, 23andWe?). A paper out today in PLoS Genetics represents a potential sea change in the way genetic studies can be conducted. At least when it comes to 'consumer genetics', a new research model that directly involves the 'human subject' element of research is making its scientific debut. (And in a journal with open access so anyone can read it!)

A seminal component of genetics research is, and has always been, the so-called human subject, a real, live person who contributes his or her highly valuable information (typically in the form of a biological sample in combination with some level of phenotypic data) to a study protocol. By examining the molecular components of these samples, scientists have been attempting to correlate genetics to biology in a measured, highly-controlled environment, in keeping with the traditional approach to academic research. By establishing, and sometimes vociferously defending, the firewall between the subjects and themselves--with an intervening IRB-approved consent form typically guaranteeing no personal benefit back to the participants--researchers have been absolved of, or disallowed, as the case may be, any responsibility or ability to communicate directly with their subjects (sounds almost feudal, doesn't it?), as if post-study phenotypic details are irrelevant to the task at hand. Resulting data are often 'de-identified' to further remove any connection back to said patient. And, heaven forbid, data from these studies rarely get back to the person from whose precious sample these discoveries were derived. (Augie Nieto's ALS story in the Wall Street Journal points out this conundrum. Another gripping read about patient sample use is Rebecca Skloot's "The Immortal Life of Henrietta Lacks".)

The 23andMe research model is precisely what motivated the company's formation. It arose from a frustration (not just mine) that scientific progress wasn't happening fast enough. Funding, often in the form of federal grants, was, and still is, hard to come by and is a painfully slow process. Obtaining large enough sample cohorts, in order to achieve statistically relevant findings, is another huge hurdle and can take years of painstaking labor. Even so, the information so carefully gathered during this process can be quite limited. Recontacting subjects has to be consented up front, and is expensive and time consuming to a point where it's often scrapped.

My time spent at Perlegen brought these issues into high focus. Indeed, they were what drove our CEO at the time, Brad Margus, to run an ad in Science, inviting any researcher who had assembled a significantly large sample cohort to work with us on a joint study (we'd find the money to get it done). We got one response. One! As luck would have it, that ad led a collaboration with Mayo Clinic and one of the earliest 'genome-wide association studies' (GWAS) of Parkinson's disease. We were funded by the Michael J. Fox Foundation, and through them I met Sergey Brin from Google (aka Anne Wojcicki's husband!), whose interest in Parkinson's has since been well publicized and is covered in a new Wired article by Thomas Goetz. The interaction with Sergey sparked a connection that would form a bond with my partner-in-crime, Anne, who was just as frustrated and bent on change. And the rest, as they say, is history.

What we envisioned four years ago--an online community where individuals get to interact directly with their own genetic data, share it however and with whomever they choose and answer whatever research surveys suit their fancy--is alive and well. Today's PLoS article is just the beginning. As highlighted in this 23andMe blog post, there are currently 650+ genome-wide association studies being conducted in 23andMe's digital, virtual lab, with over 29,000 people (not subjects!) actively participating. No doubt there will be snarky comments about some of the initial phenotypes that are producing these new discoveries (do YOU smell it?)--but that's not really the point. What matters is that innovation in this field IS possible, all because of the very people who are opting in to a whole new way of knowing themselves.

Good News on the Patent Front

It may be a bit premature to celebrate just yet, but I'm thrilled to hear news that the BRCA patents have apparently been invalidated by Judge Sweet. As reported in BusinessWeek:

U.S. District Judge Robert Sweet in New York ruled the patents invalid today, saying they “are directed to a law of nature and were therefore improperly granted.” The judge sided with the American Civil Liberties Union, medical groups and the Public Patent Foundation who sued to invalidate the patents.

As I'd stated in a previous post, Myriad's patent position was illogical and troubling, especially as full genome sequencing is now amazingly within our grasp. Patents on every gene would be like needing to have 1000 people show up to paint your house, just because they claimed a right to that small piece of wall. The concept of gene patents may not have been challenged partly because people weren't contemplating the genome as a whole, but rather in small, discreet chunks.

There will certainly be fall-out from Judge Sweet's ruling but I don't think I'm the only person who'll have a spring in her step based on this early news. Congratulations to the ACLU and all the people they were representing.

Read All About It Round 2

Update:

Thought I'd post a quick update to this topic. Last Friday I had a conversation with New York Times writer Andrew Pollack (he called me to discuss my original blog post below). Based on that exchange, and regardless of any articles that may ensue, there are a few things that could use clarification.

First of all, in my opinion, the business of personal genetics shouldn't be compared to the solar industry (which Mr. Pollack was postulating). Unlike alternative energy technologies, of which there are many exciting options that could supplant solar altogether (we'll see!), human DNA is what it is...there is no alternative (i.e. it's not a solar versus wind versus Bloom Box debate). At this point, most people seem to agree that having access to our genetic data will become mainstream, especially as next gen sequencing hits its stride. The path to having this concept become fully realized will certainly continue to hit the occasional regulatory bumps (for good reason) but information wants to be free and I can't see how any regulatory body can completely block our desire and/or need to access it. So I would bet that the field of personal genetics becomes such an every-day concept that future generations will be amused that we even questioned its viability.

Secondly, not all personal genetics companies are created equal. Comparing 23andMe or Navigenics to deCODE is like comparing apples to appaloosas. Neither of the two true start-ups were saddled with over $200 million in convertible debt (or are based in Iceland). As I've said before, I have great respect for Dr. Stefansson and the ground-breaking work he and his team have accomplished, but their unfortunately bankrupted model is a very different animal. Expecting deCODEme, their DTC--direct-to-consumer--play, to rescue its parent unfortunately wasn't in the cards (not that this was what they originally intended).

Sticking with the horsey theme (appaloosas are horses in case anyone wasn't sure), it's a bit nonsensical to view all entrants in an equestrian race as equals. Some stumble right out of the gates, others get an early lead only to lose it, and others stay with the pack and grab the lead toward the finish line (if there is such a thing). Whatever the case, this race isn't a sprint. It's an endurance run that will require patience, commitment and focus. May the best company, or companies, win.

I think I'll go read my New York Times now...one of the last great newspapers.

*****************************************************

Original post:

In what looks like a desperate attempt to make some headlines, I've gotten wind of the fact that Andrew Pollack, a writer for the New York Times, has been digging around, looking for dirt and hoping to stir up controversy for a story he's apparently writing about the personal genetics industry. It's a sorry state of journalism that we live with on a daily basis, but this hits too close to home for me not to comment.

Put simply, genetic data is just another source of information that can empower individuals to start taking control of their own health (and learn some fascinating ancestry insights in the process...just check out Brad Templeton's recent post). The objective behind 23andMe (a company I co-founded) is to remove the barriers of access to these data and begin to explain it to people who have a right to know. Yes, it's a controversial notion to some, but to others--I'd say a good majority of free-minded thinkers--it's a no-brainer that we should have knowledge of our own DNA. We still have a lot to learn about what our particular string of As, Ts, Gs and Cs codes for in terms of our predispositions to diseases and other traits, but that's also a goal of 23andMe--to open the door to the research process and let anyone who's so-inclined take a more active role in the wonderment of discovery, something that can benefit us all.

So if Andrew Pollack decides to take a swipe at this fledgling industry, he won't be the first, or the last. It's an easy bandwagon to jump on, along with the other nay-sayers, but there's a growing wave of data-empowered people who won't bat an eye, and who may just help lead us into a new age of personalized health that Andrew himself may find quite beneficial. Now that's something to write about.

Thoughts from JPM10

Last week I attended the 28th annual JP Morgan Healthcare conference here in San Francisco. As usual, the crush of black-suited bankers and biotechies in the far too tight halls of the Westin St. Francis gives one pause...what if there was an earthquake in the midst of this yearly frenzy?

Based on hallway conversations, Twitter tracks and general buzz, the overriding theme was one of optimism and, hell, giddiness, particularly around the ever more real promise of truly low-cost, accessible full genome sequencing. Illumina and Complete Genomics were the belles of the ball (Check out ILMN's stock performance this past week.) Daniel MacArthur goes into more detail on this topic, with links to even more technical background if you're so inclined.

A dead spot for me during the conference, however, came during the Myriad Genetics presentation. Here was their CEO, Peter Meldrum, boasting about his impressive margins (87%!), market share and IP estate (over 200 patents!), details that must have pleased the bankers. For me, however, the thought that a company is profiting so handsomely from intellectual property that corners the market on information simply residing in the human genome is unfathomable. To show such exceptional profits, primarily from their BRACAnalysis test (assessing a woman's risk for breast and ovarian cancer via BRCA gene mutations), especially with the on-going ACLU lawsuit, made my skin crawl. Was I the only person in the room thoroughly appalled?

The juxtaposition of these two extremes, full genome sequencing versus gene patents, is worth pondering. As individuals are able to access their entire genetic profile, and all the increasingly pertinent health information therein, the math just won't work if gene variants, along with the inherent knowledge regarding disease or pharmacogenetic associations, are patented. How many licenses would be required in order to provide comprehensive information to an individual about his or her own genome? Will licensing fees eventually surpass the cost of sequencing itself?

As Dan Vorhaus and John Conley wrote in The Genomics Law Report last August:

At least for the moment, whole-genome sequencing and gene patents coexist, despite considerable uncertainty. But while it’s impossible to accurately predict even a few months into the future in such a turbulent field, there is a basic economic truth that cannot be ignored: high-cost licenses for the rights to specific gene patents are fundamentally incompatible with low-cost whole-genome sequencing.

Don't get me wrong...I'm all for the protections offered through the US Patent and Trademark program when applied appropriately (and when there's something novel that has actually been invented). Unfortunately, though, the system is imperfect and there is precedence for it to be righted--albeit expensively and painstakingly--as need be. In the case of ACLU v. Myriad Genetics, though, will it simply come down to lawyerly skill, rather than fundamental truths, that wins the day? Myriad clearly has a lot of money in its coffers to pursue a victory in the courts. If only a lawyer of Andrew Hamilton's stature would come forward and do for the ACLU what Mr. Hamilton did for John Peter Zenger back in 1735. Now those were the days.

It's the Phenotype!

James Carville's irreverent phrase-- [It's] the Economy, Stupid--which rather unintentionally became a slogan for Clinton's successful '92 campaign, often comes to mind when considering all the opinions and perspectives being doled out these days regarding genome-wide association studies (GWAS) and their contribution and utility in the march toward a more personalized approach to healthcare. In this context, and when considering the complexity of genetics research, the importance of phenotype information gets short shrift. Just take out "economy" in Carville's adage, insert "phenotype" and let's talk turkey.

Clearly the technological approach used to date in collecting genetic data in GWAS is not the be-all, end-all solution on the path toward our understanding of the human genome as it relates to disease and drug response. In particular, using a common disease-common variant strategy to uncover the genetic basis of these human conditions is not without its flaws but was at least a good starting point, given the technological means available as the HapMap project enabled selection of commonly occurring variants in our ancestrally chunky genomes. (Thanks to David Cox for the much-appreciated education I received on this subject during our Perlegen days...the company's contribution to mapping human haplotype structure was, for its time, ground-breaking and will hopefully not be forgotten.)

During the last decade, even with the dramatic drop in genotyping costs being offered by Affymetrix and Illumina, scientists were still limited in their ability to survey the genome (mostly because of cost...still one of the key rate-limiters of genetic discovery despite the advances). These companies leveraged HapMap data to construct "genome-wide" SNP arrays (Affy's 500K and Illumina's 550K designs being the notable designs) that provided, at the time, the best coverage. These arrays became the heart and soul, if you will, of the majority of studies that were conducted (grant reviewers' eyes must have glazed over at the number of proposals suggesting the use of these gene/bead chip designs).

Even then, everyone knew the holy grail for true GWAS would be sequencing the entire genome but it just wasn't feasible. So, what with the limitations of the HapMap data and the resulting-- one could say inherently flawed--genome-wide arrays, scientists sallied forth and left hypothesis-driven candidate gene studies in the dust (thank goodness!). And, despite what some detractors are eager to flippantly malign, there are quite a few success stories as a result of these efforts. The role of Complement factor H in age-related macular degeneration is one shining example. (R.J. Klein, et al., "Complement factor H polymorphism in age-related macular degeneration," Science, 308(5720): 385-9, 2005)

So, what now? From everything we're hearing, the $1000 fully sequenced genome is right around the corner (yeah, I'll believe it when I see it, too). However long it takes for this to truly be the case, the game-changing nature of this mind-blowing technological advancement will, undoubtedly, expose other research limitations that haven't been the over-riding issue at hand.

Which brings me to the adulterated Carvillism. As the ability to more extensively probe the genome comes to pass with 'next-next' gen sequencing tools, we will bump up against the other, likely even hairier, problem of obtaining comprehensive phenotypic information from the human subjects enrolled in large genetic studies. Without this critical information, it will still be very difficult to make heads or tails of all the data flying off the sequencing platforms. How will we make correlations between genetic mutations and disease manifestations if our ability to collect that equally important phenotypic data doesn't improve as well?

One could argue that a primary downside of the GWAS conducted over the past decade had as much to do with the quality, or lack thereof, of the phenotypic information being used to recruit and include 'human subjects' in these case/control study designs. Using often-times 19th century-derived, antiquated disease definitions as a means of establishing a 'case' leaves a lot to be desired.

A case in point: Parkinson's disease. As scientists learn more about the genetics that underly this condition, subtypes are emerging, all still under the label of Parkinson's, since the outward disease symptoms appear quite similar. But the molecular basis of these subtypes can be agonizingly slow to tease out since the broader definition allows for inclusion of everything together in monolithic studies. It's a chicken and egg problem (how can you know what you don't know when designing a study?). And it may be one of the confounding issues that dilutes statistical power because each form of the disease is under-represented in a pool of cases all thrown into one pot. Garbage in, garbage out, as the old saying goes.

The frenzied autism debate, not that I want to get into it here, is another good case for disease label/phenotype inadequacy, IMHO. It seems there's a pretty big divide between the scientific community (that appears to liberally apply the autism label in defining cases for GWAS studies) and parents (who, obviously, have the singular perspective of their child's experience with the condition, or children's experiences, as the case may be, in those families who unfortunately have more than one child affected). Maybe I'm missing something here, but I've yet to see any rigorous, adequately powered studies (thousands of cases/controls) that home in on well-defined sub-phenotypes of autism (yes, using information from very observant parents who've watched their children's symptoms progress first-hand, on a day-t0-day basis).

Wouldn't it be worth gathering a cohort of, say, only those kids whose parents saw a remarkable decline in development (for instance, very verbal, developmentally on-track kids who abruptly took a backward turn and lost their verbal/social skills), or kids, diagnosed with autism, who suffer from severe digestive ailments? By enrolling only those children who meet these specific, tightly defined inclusion criteria, is it possible that we might make scientific progress? What if there is a small subset of autistic kids who have a unique genetic signature across their ADME genes that adversely affects their ability to cope with chemical exposures, including vaccines? Have we even looked at this thoroughly and carefully? If so, please send me links to the papers! But if those studies included a gallimaufry of phenotypes, don't bother.

I get the public health angle on vaccinations (and all three of my kids were vaccinated according to schedule) so I'm not espousing that parents shouldn't get their kids immunized. But, just like we talk about personalized medicine, isn't it possible that one day we could have a more personalized approach to vaccine administration? Why has the medical community agreed that one-size-fits-all medicine isn't working and yet we don't hear this same sentiment when discussing immunization schedules? Bottom line...as always...we need WAY more data, including much more highly developed phenotypic definitions. And until we've exhaustively pursued these types of studies, it's irresponsible of the medical and scientific communities to say that it's game over.

My prediction, and I'm sure I'm not alone, for this new decade is that phenotype will surface as a key issue in genetics research, and we'll see more and more companies emerging to address this very thorny problem. And, mark my words, it will include self-reported data coming via web-based entry programs, along the lines of PatientsLikeMe and CureTogether. Just might make sequencing look pretty darn easy.

ClubMed

I'm just back from my annual November trek to Boston, where for the past four (or is it five?) years I've attended the Partners Personalized Medicine Conference, run by Dr. Raju Kucherlapati of PCPGM. This year over 600 people attended (!), and there was a definite shift in the wind...what was more of a theoretical discussion back in the early days is becoming a reality, slowly but surely. Instead of talking hypotheticals, pharmacogenetic case studies are now possible, and having them led by the likes of Richard Hamermesh of HBS makes for a lively debate. The discussion around colorectal cancer drugs Erbitux and Vectibix, in particular, elicited strong opinions, which is what you hope to hear at meetings like these. Raju always joins right in the mix, too--whether you agree with him or not, his invaluable and articulate perspective never ceases to amaze.

It was an honor to be on the organizing committee (my contribution was minimal, to say the least) and, for the first time, I had the opportunity to moderate a panel. It consisted of physicians asked to ponder the question 'who will benefit?' from personalized medicine. Of course it's assumed that the patient stands most to benefit from targeted medicines (that's the whole point!). What I had hoped to explore was how a more personalized approach to therapeutic intervention is beneficial (or not) to other stake-holders in the medical ecosystem. It's a complicated and controversial perspective; the players--doctors, pharmas, medical systems, pharmacies--may be loathe to admit to (financial) incentives that run counter to what's best for the patient. What truly are the drivers that will propel this mission forward, and can they be aligned with the most efficient patient care?

Other than the altruistic (do no harm...don't be evil?) directive of--and passion, for most--doctors to provide the best care to their patients, what else might motivate them to push for a more personalized strategy? Let's be honest...doctors are human, like the rest of us, and they have other drivers that elicit certain behaviors. The veil has certainly been pulled back on the cozy relationship that had insidiously cropped up over the years between pharmas and docs, resulting in policies to crack down on drug-related adverse side effects of an entirely different sort, purely financial. It's at least a start in removing incentives that may be contrary to patient care. But what other drivers could be built into the ecosystem that more positively and proactively drive patient-centric behavior? And does it have to be tied to money? (My guess is, probably.)

Dr. Troyen Brennan of CVS/Caremark was on the panel, representing his slice of the healthcare pie. He claimed that a personalized approach to drug prescriptions WAS in the best interest of his industry. But I still don't get it. From the perspective of the retail pharmacies (as with the drug companies), they clearly stand to make more money in our existing (albeit shifting) paradigm of 'shotgun' therapy. How does a pharmacy receive any benefit from its customers getting fewer drugs that actually work for them?

This takes us to the broader issue of our very broken healthcare system. Sheila Walcoff (wish we had more Republicans like her!) had a slide depicting a DC view of personalized medicine that might well have been the Pentagon's plumbing blueprint. Yes, it's a complicated jumble, but without appropriate motivations even the most streamlined process won't flow.

We've seen (and personally experienced) systems where doctors are wrongly incentivized: HMOs. We can't perpetuate patients being denied care because it dings a doctor financially (been there, done that). Any other ideas out there?

Hopefully next year we'll have even better case studies to debate. And even further down the road, maybe the name of the meeting will change to Personalized Health, incorporating not only the therapeutic component but also, the best medicine of all, a healthy dose of prevention.

Friday Lite - Ancestry Anecdotes

One of my hobbies over the past few years has been building my family tree on www.ancestry.com. I posted a blog about it a while back, http://spittoon.23andme.com/whats-in-a-name/.

Since that post, 23andMe has launched a new feature called Relative Finder. What this feature performs is a comparison of your genetic profile (if you're a full 23andMe customer) to everyone else's in the company database and then lines up those individuals who look like they're genetically related to you, sorted by parents, grandparents/grandchildren...all the way down to 10th or even more distant cousins. Now that the size of the 23andMe database has been revealed--30,000 and growing--it's a pretty sizable number of people to be compared to. And it's turning up some VERY fascinating things.

For instance, it appears that my husband's (yes, Greg Mendel) aunt and first cousin are also my 10th cousins (yes, my in-laws!). This aunt also shows up as my mother's 7th cousin. How real is this? We'll have to trace our genealogical (or "paper") trees to see if we can find a connection. A fifth cousin on "Greg's" side has turned up as my father's 7th cousin. And one of the 23andMe Scientific Advisory Board members is also showing up as my mom's 7th cousin. Needless to say, these revelations need further investigation, but it's pretty mind-blowing!

The reason this is so revolutionary is that this new tool represents, really for the first time in human history, the ability to find relatives through a broad swatch of the genome and through comparisons to thousands of other genetically-armed individuals. Genetic genealogy services up until now have primarily focused on the Y chromosome and the mitochondrial DNA, but those tests offer only small components of your genetic past. Having data across all 23 pairs of chromosomes--as the 23andMe service does--opens up a whole new means of discovery. Doing genealogy research takes on a whole new angle with the advent of this tool. Comparing "DNA cousins"--those who look related through the genome--to "paper cousins"--those who are related through historical records--makes this hobby all the more interesting and accurate.

I'll continue to update my ancestry findings as I learn more...Friday's a good day for it!

Enjoy your weekend!