
NEW YORK (GenomeWeb) – The definition of the term "precision medicine" has changed over the years since the first human genome was sequenced in 2003. Researchers are increasingly finding that a simple overview of a person's genomic variants is not enough information to create medicines tailored to that individual's needs.
Indeed, the National Institutes of Health's All of Us Research Program — originally launched as the Precision Medicine Initiative — is not only aiming to collect data and samples from 1 million Americans, it is attempting to gather a variety of data from electronic medical records, surveys, wearables, and genetic testing that researchers can use to learn more about diseases and develop treatments.
As new technology becomes available, multi-omic research is becoming more prevalent. And in certain cases, different scientific disciplines are joining their skills and expertise together.
At Silicon Valley-based venture capital firm Andreessen Horowitz, General Partner Jorge Conde believes such inter-disciplinary companies are the future of precision medicine. Conde — who previously served as CFO, chief strategy officer, and chief product officer for Syros, and who cofounded human genome interpretation company Knome — oversees Andreessen Horowitz's biotech and healthcare BioFund, which totals $650 million.
He spoke to GenomeWeb about the next iteration of precision medicine, and how the combination of biology and engineering is bringing that future into being. Below is an edited transcript of the interview.
When a startup company or entrepreneur comes to you to ask for an investment, what kind of X-factor does it take for you to say yes?
We're essentially looking for three things. We're looking for technical founders that are the experts in their specific domain. Because we're working at these intersections of biology and computation around engineering, we tend to be most interested in founders that have a deep domain expertise in both disciplines. Our belief is that technical founders have traversed what we call the idea maze, that they've explored all the opportunities and run into all the blind alleys and dead ends that one could have with new technology, and when they emerge from the other side, they emerge with a unique idea or perspective on how to build up a technology.
The second thing is a technology that would enable a company to be built more like a technology company than a biotechnology company. In the case of a traditional biotech, they're trying to tackle the scientific problem. There's basic discovery work: we're going to try to find a target, and then we're going to develop a compound, and then we're going to optimize the compound, and then we're going to enter clinical trials. And that's proven to be a very effective model for discovering new therapies, but it also has binary risk — either it works, or it doesn't work. But at these intersections of biology and computation engineering, we look for platforms that are more modular and therefore more tunable and iterative. It's not as bespoke a model as traditional drug discovery where it's one molecule looking for a target. Here, you're building a capability and that capability can be improved and increased in sophistication over time, and therefore the risk is less binary and more iterative.
The third thing we look for are these ideas that can scale in the way that technology traditionally scales to hopefully very large companies and very large businesses that have an outsized impact. And we will, of course, want to wrap all of that around the right people and team, and a lot of that does come back to the technical founder. Can that person attract and retain the best people?
There are some specific technologies that people are talking about these days: AI, CRISPR, and synthetic biology are among those. Is there one that you're particularly excited about and why?
We have investments in, and we're interested in, all of those. In the case of AI, the real question is where will it be useful and how do you separate the hope and the hype? We think AI has a very powerful capability to help us deconvolute and better understand the black box of biology. AI is an uninterested participant in terms of trying to understand biology and doesn't require the same need to create mental frameworks for us to understand it. And so artificial intelligence can learn what drives biological systems in a much more comprehensive and non-hypothesis-driven way. And I think that can be a very powerful thing.
In fact, we have we have investments in AI applied to drug discovery and for diagnostics, and in the case of diagnostics, you can essentially train the algorithm on well annotated patient samples. And then over time, it will get better at doing that. It becomes a very powerful additive tool for us to essentially deconvolute the complexity of biology in a way that humans alone probably couldn't.
In the case of CRISPR, it's a foundational capability and tool. Most folks think about it for the applications of what we can do from a therapeutic standpoint. I think those are very important, but it also has diagnostic applications. And for me, one of the most eye-opening things was that it really transformed the way we can do basic R&D. Before CRISPR came along, the way you did target discovery and validation is very different than what you can do now with scalable CRISPR-based technologies.
And what we think is powerful about synthetic biology is that a lot of companies have been very successful with it in finding ways to automate and scale processes so they can try a lot of things. And I think that's a very important and powerful tool. But I think what's happening now is, as we get better at using all the other tools at our disposal, synthetic biology can become much more of a design-driven medium. Synthetic biology is increasingly being shown to not just touch what we can do in human biology but it's going to touch many other industries and transform them. And that can be as broad manufacturing, it could be therapeutics, it could be agriculture, food, textiles. So, we think that's just a really exciting area to be in right now.
You've talked about the future of precision medicine being in the programmability of living things. What do you mean by that?
We're developing medicines now that are [made from] redesigning living things or near-living things —genes and cells and microbes — and essentially repurposing them for therapeutic use, which for me is a very interesting departure from the past when chemicals or proteins and all the variations therein were targeting cells and genes and microbes. The second feature that I think is important is you start to have the ability to tune these things. If something doesn't work at first, you can go back, reengineer it, and optimize it.
But the related feature to that is that as you overcome challenges in the programming of the various cells, genes, or microbes, they become applicable to other therapeutic areas or other diseases, and they become a foundation on which you can build new applications or uses for them. In the bespoke nature of traditional drug development, you made a drug for a target and that's what it was good for — to make a different drug for a different target was essentially starting over. Here, because you have this foundational element, you can build upon success. As a result, if gene therapy is proven to be successful in one case, the challenges for using it in the second case or in the second disease hopefully are diminished. Every disease is going to have its own challenges — I don't want to minimize that. But there's a foundational element to programming these medicines.
Because of that, I think we're going to increasingly start to see the versioning of therapies kind of in the way we see versioning of software, where the first-generation CAR-T is less sophisticated than the second generation, and so on. And you'll start to see the medicines themselves improve over time as they get more sophisticated.
What's going to be required to do that? In the case of gene therapies, there you have a vehicle-and-cargo problem. You want to make sure that you have the right vehicle to deliver the gene therapies, and that has to be highly precise. Gene editing technology has to be exquisitely precise controlled. There is going to be an element of dosage when you think about these things.
The second area, and this is where I think AI becomes very important, is to be able to do biology at very large scale. I think one of the fascinating things about where we are and have come over the course of the last decade or so is both being able to read biology and write or modify biology at an incredible scale. The canonical example is sequencing. It's not only being able to sequence or read, but to write, you have tools like genome engineering, synthetic biology, organism design, organism engineering, the ability to do things at the cellular level, to have more control in terms of on switches, kill switches. All of those Lego components, as they come together, will allow us to essentially write or program biology at a scale that we've never been able to.
You've also spoken about combining biological science with the principles of engineering. Can you talk about what that would look like, and what you think the two disciplines can learn from each other?
When biology shifts to becoming more of an engineered discipline, you move away from the hypothesize-test-conclude-rehypothesize-retest scientific paradigm that's obviously very powerful and very important. When we see these more engineer-driven platforms, they tend to look more like the design-iterate-iterate-iterate-iterate type model where you can build on improvements from one generation to the next.
What we've learned from entrepreneurs that are working at these intersections is that when they build their platforms, and consequently when they build their companies, with that sort of engineering mindset, they're able to make that sort of iterative improvement over the short term that, when taken over a long period of time, take them a great distance. If they can get 2 percent improvement in regular cycles, over the course of the fullness of time, you have incredible shifts forward. I think that engineering iteration scale mindset is very valuable.
The flipside of that is what engineering can learn from biology. When pure engineers move into the biology space, they learn that uncertainty exists. And I think that becomes a very important thing for the engineering side to understand is to work within the constraints of the system as it exists, because you're not going to reengineer the entire system. I think when both sides learn from each other, it's proven to be a very powerful combination.
Ever since 2003 when the Human Genome Project was first declared complete, we've been promised precise medicines that would target specific genes and treat or cure diseases. Except for a few exceptions, that hasn't really happened so far. Why do you think that is?
I think there's been some great advancements in targeted therapies, and I think you can separate out the cases where the promise has been fulfilled and in cases where it hasn't. In my mind, it's a four-tier problem.
The first tier is you need to understand what you want the clinical benefit to be for the patient. That sounds very straight-forward but it isn't always. So, in the case of, for example, developing an Alzheimer's disease therapy, what is the benefit that you're trying to accomplish? Is it to improve memory recall, slow the progression of the disease or stop it, reverse the disease?
The second tier is whether the underlying biochemical pathways that would need to be modulated to have the desired patient effect are known. Again, to borrow the Alzheimer's example, it's unclear exactly what the biological pathways are that would need to be affected to have any beneficial effect for Alzheimer's disease. It could be tangles or plaque or any of the other things that have been floated as a hypothesis.
Then the third level below that is, assuming you know the pathways, do you actually have a specific intervention point that you think would have that cascading effect on the pathway that would have the desired patient impact?
The final layer of this four-tier process is can you make something essential that would hit that target, whether it's a small molecule, an antibody, whatever it is? And I think where it works well is when the three higher tiers are well understood and the fourth is really the problem you're trying to solve.
What's challenging is that in the vast majority of diseases, those aren't always very well understood or well known. I think that's what makes it particularly difficult, and I think Alzheimer's is a great example of that. In contrast, there's something like sickle-cell anemia. You're trying to get rid of the painful crises that sickle-cell patients have, among other things. The pathway for that is known — you've basically got to regain the function of hemoglobin, and you either repair the hemoglobin gene or you turn on fetal hemoglobin. So, if know to do that, the fourth tier is can we develop therapies that would do one or both of those things? And that's happening now, and so [we're] starting to see at least early evidence that would suggest that some people may be cured or at least have repression of disease.
I think what will help deliver on the promise of precision medicine is having capabilities and technologies that illuminate all four of those tiers in a way that we haven't had in the past, and when the promise falls flat is when you can't elucidate every single one of those tiers in a way that allows you to purposely move forward in developing the therapy.
So, would you say that CRISPR is the culmination of the promise of precision medicine — the kind of precise medicines that we were promised that would target specific genes or pathways?
It's a difficult question. We've been in this space enough to know that it's hard to call anything the culmination, and there's always more layers to the story of biology. But I think it's an incredibly promising technology. I think it will start in the ex-vivo therapies because that lowers the challenges in many ways and lowers the risks. And then I think we're going to see specific delivery in the body where it's, for lack of a better word, less challenging, whether it's the liver or the eye or any of the other organs that can be more easily targeted. Then I think the big challenge going forward is going to be the final domino for CRISPR: can you do the exquisite targeting systemically in the body and, if necessary, do re-dosing?
Finally, I wanted to ask you about what happened last November in Hong Kong with He Jiankui's germline genome editing experiment. The research community was angry about the ethics of it, but there was also some fear that it would translate into less funding for CRISPR research. Have you seen any corresponding backlash on CRISPR research or startup company funding because of that controversy?
No, I don't think I've seen any noticeable change in interest in the space or any funding associated with CRISPR. What we definitely have seen, appropriately, is that folks generally want to make sure that they're being cautious, and that things are being developed responsibly, and that things are being pursued and progressed with ethics in mind. But I haven't detected or seen any real drop off in terms of interest in the space.