Founder Stories: Jeff Huber, Co-Founder of GRAIL

I'm Jeff Huber, I'm CEO of GRAIL. GRAIL's mission is to detect cancer early when it can be cured. The premise behind that is relatively straightforward. Cancer that is discovered at early stage, stage 1 or stage 2, depending on cancer type, the outcomes are quite positive, 70%, 80%, 90% positive. Cancer that's discovered at late stage, stage 3 or stage 4, is roughly the inverse. It's 80% to 90% negative where people die. So our goal is instead of detecting cancer at late stage, which is the dominant case today, with the exception of breast cancer and a couple of other screen cancers, most cancers on the order of 80% are discovered at late stage, stage 3 or stage 4. So instead of detecting here, where the odds are against you, we want to detect here, where the odds are in your favor. And if we can make that happen, we think that that's the fastest and best path to be able to save millions of lives. The underlying mechanism for doing that is what we call ultra-intense genome sequencing. It's based on a team and technology that was spun out of Illumina. I'll share a little bit more about that in a moment. But with ultra-intense genome sequencing, what we ultimately want to do is create a blood test that could be integrated deeply in the medical system so that if you go in for an annual physical exam, They'll typically do a blood draw anyway, tell you your glucose and cholesterol. There should be one more thing, the GRAIL test, that gives you the opportunity to save your life. As was mentioned, I am at GRAIL now, but I spent the prior dozen years at Google. Maybe a little bit of background or the story of how GRAIL got to where it was today might be useful. So I was a dozen years at Google. The first 10 of those were developing products— Google Ads, which is the economic engine behind Google, the thing that has enabled Google to do all of the other crazy things it has done, Google Apps, Gmail, Calendar, Docs, that set of communication collaboration products, and then Google Maps. I was involved in the early days and then ran Google Maps for several years, most notably in the period where Google made the commitment to effectively map the world and generate the data itself that's enabled the very high-quality experience it has. Collectively, those were products that are now each used by more than a billion people and are each billion-dollar products for Google. As I was coming up on the 10-year anniversary at Google, though, I used that as an opportunity to pause and reflect a little bit, thinking back over that decade of what did I accomplish, what were the places where I had the biggest impact on Google, but also if I were going to contemplate another decade, at Google, what were the things that were both going to have an impact for Google but that would be energizing for me? As I reflected back, it was really the early days of Ads, the early days of Apps, the early days of Maps, building the team and foundation, setting the strategy, but also each of those was opportunities to learn new things. So it felt like as I was contemplating a second decade, rather than turning the crank again and doing a fourth big system at Google, that this would be an opportunity to take Google in a new direction, and also to have me continue learning. And as I was looking across the set of opportunities and looking broadly across the technology landscape and science landscape, the thing that really resonated for me was the shift, the phase shift that was happening in life sciences. Specifically enabled by new technologies like next-generation sequencing, where you're effectively able to digitize biology. And I've always looked at innovation as coming at intersections. I was coming from a background in computer science, and it felt like with this new direction or this phase shift in life science, that that was a really interesting opportunity to drive, to visit that intersection, to learn at that intersection, and to be able to bring the power and tools and technology that I was familiar with from building very large-scale billion-user applications at Google in computer science to this wave of data that was going to be coming in life sciences with next-generation sequencing. So I became one of the founding members of what became the Google Life Sciences team and now is called Verily within Google, or Google itself became Alphabet, and started exploring exactly that. Of what was the power and potential in building a technology team to take on those challenges. Serendipity happens because right in that window I was then contacted by Illumina, who is the leader in genome sequencing, who was looking at a similar phenomenon, was seeing all of this data being generated by their sequencing machines at— on a trajectory of being unprecedented scale, and they decided they wanted a big data person and I was the big data person they wanted. So I ended up joining the board of Illumina. That then starts getting back to the GRAIL story. As I foreshadowed earlier, the team and technology that ultimately became GRAIL was incubated, started in R&D in Illumina. So when I was there attending the first board meetings at Illumina and we were doing the R&D reviews, one of the projects that they were very excited about and it was clear that there was enormous potential was this concept of being able to use next-generation sequencing taken to its logical extreme or maybe even a little bit beyond to be able to detect cancer at its earliest stages. And where did that come from at Illumina was they had acquired a company called Veranada that was working in non-invasive prenatal testing. Non-invasive prenatal testing has been— has now become over the last several years one of the most successful molecular diagnostics and the concept behind it is that you could do a blood draw of the mother's blood and find in the mother's blood DNA from the infant that would enable you to detect fetal abnormalities, things like Down syndrome. As Veranata was developing and had commercially launched that product, they had gotten to the point of having done about 100,000 of those tests,. And they found in roughly 20 cases an abnormal signal, or a signal that they couldn't explain. And they were racking their brain trying to figure out, is this some new signal, signature of a fetal abnormality? What's going on? But as they were doing it, then those children, those babies were delivered perfectly fine and healthy. And then they were looking for what else could be going on. And they coincidentally just hired the former director of the National Cancer Institute. Rick Klausner, who came in and looked at the data that they were presenting and promptly said, oh, that's cancer, which was a big surprise to the team. But as they dug into it, that's exactly what was happening. They were finding cases with perfect specificity, literally 100% accuracy, where this was diagnosing cases, unfortunately, of late-stage cancer in the mother, that hadn't yet been diagnosed. And that led to the light bulb within Illumina of saying, okay, if this test that was developed for this other purpose that doesn't have the kind of sensitivity that it would need to have, but is still detecting the signal, what would it take to be able to go from the late-stage diagnoses that are happening here to be able to detect cancer at its earliest stages? And that was really the genesis of GRAIL. And as we were talking, the team was coming and they were— it was an A+ team that had been put together within Illumina. There was an early collaboration with Memorial Sloan Kettering that has continued to this day that has driven great clinical work of confirming the science and biology behind it. But the kind of discussions that we had at the Illumina board meeting was, wow, this is beyond just being a— should be a good business opportunity to know this is really a moral and ethical imperative to make happen. Because if we can do this, if we can get this right, if we can make this scale, this is the best path to be able to save millions of lives. In parallel to all of this happening, there's a personal dimension of it as well, which is my wife Laura, who was 46 years old, perfectly healthy, did everything right, ate right, exercised, no family history of cancer, was diagnosed with late-stage colon cancer. Hardly any symptoms leading up to it whatsoever, no symptoms until just a month or two before diagnosis. And even then, the diagnosis was still an odyssey of trying to figure out exactly what was going on, but ultimately culminated in the spring of 2014 in a diagnosis of stage 4 colon cancer. Colon cancer that had metastasized to her liver and through her lymph system. That started an 18-month period of intensive treatment. She vowed to fight. I vowed to fight with her. But ultimately, that was a losing fight because of where and when it started. She passed away in November of 2015. I was approached by the Illumina board in December, the next month, about becoming the CEO of GRAIL, given my known passion for this topic. And I'm very confident that if GRAIL had existed 3 or 4 years earlier, that Laura could have had a very different outcome. Colon cancer that's detected at an early stage has quite good prognosis. The surgical resections have a very high success rate. She could be alive today instead of where we are. So I was approached by the Illumina board in December of 2015. GRAIL was announced in January of 2016. My role was announced in February, the next month. GRAIL had its first employees, which was that team that came over from Illumina. So we had 40 employees on day one in March of 2016. A year ago this week, and we've been on a rocket ship ride since then. We started with 40 people on day 1. We've roughly quadrupled. We're about 160 people now. We've built a sequencing lab in our facilities in Menlo Park in Northern California that we believe is the highest capacity sequencing lab in the world. We've launched our first clinical study that we call CCGA, the Circulating Cell-Free Genome Atlas. It's a 10,000-subject study. That really for us is building the reference library of what is this underlying analyte circulating nucleic acids fragmentary DNA that's shed by the cancer from its very earliest stages. What does that look like across cancer type and stage in 7,000 subjects? And then what does it look like in the asymptomatic population? So we're also doing 3,000 subjects that are age and demography matched to those to be able to build out that reference library. We're also in the planning stages and early launch stages of the next phases of our clinical studies that will ultimately be hundreds of thousands of subjects that we'll be doing on the path to demonstrating clinical scientific rigor, clinical rigor, clinical evidence, clinical utility, impact on patient outcomes, impact on health system economics, and then ultimately, the commercial products that we'll be introducing. Across all of those, so the clinical studies are of unprecedented scale. But to really make this work, we have to do 3 things. We have to have the best science in the world to be able to identify this signal and be able to do it at incredible precision. So if you think about what we're trying to accomplish, we're looking for fragmentary DNA in the blood. But cancer at its earliest stages, the concentrations of that are very low. So we have to have assays that have the precision to be able to detect literally a handful of molecules in a full tube of blood. So we have to have the best science around measurement and generating that signal. In our clinical plans, we are executing these clinical studies of unprecedented size and scale. So there's an operational challenge around that, but then generating all of that data, bringing it back, is a tremendous challenge. And then finally, from a technology perspective, I mentioned with the ultra-intense genome sequencing, we're sequencing an order of magnitude broader than typical liquid biopsies are doing, where the focus there is on kind of mapping a mutation to a druggable therapy. Instead, we're casting our headlights very wide of looking for cancer wherever it may be, and cancer is spectacularly heterogeneous. So we're looking at over 500 genes and other regions to be able to detect cancer wherever it's coming from at its earliest stages. And then going to depth, depth is really the measure of the oversampling rate of being able to detect that scant signal, that handful of molecules in a full tube of blood. The net effect of that is that we're generating on the order of a terabyte of data for every test that we do. So take that then to, with the kind of scale that we aspire to get to, if we can get to a mid-range milestone of, if we could be integrated deeply in the medical system, if we were doing screening for the adult population in just the US, that would be 100 million people a year times a terabyte of data roughly each. That makes GRAIL the first zettabyte-scale application. Which is roughly the size and scale of Amazon and Google and Facebook and YouTube today. So we're building the capabilities from the beginning to be able to do that, to be able to handle that scale of data, but then also be able to generate the machine learning, the learning models to be able to generate the insights that we need. So there's lots more, and I'm sure there'll be some interesting questions in Q&A, but why don't we save them for that? And we'll continue with the show. Thank you.