Unhealthy Prognosis for Venture-Backed Diagnostics

Posted April 26th, 2013 in Diagnostics, Personalized Medicine, VC-backed Biotech Returns

Personalized medicine and diagnostics are undoubtedly critical to the delivery of better healthcare in the future: getting the right drug to the right patient at the right time.  In theory, these therapy-guiding diagnostics should reduce the cost of healthcare while improving outcomes.   

But I’m far less sanguine about who is going to fund the new innovations required to bring these diagnostics to market.  As I noted in last week’s post-mortem regarding On-Q-ity, venture-backed diagnostics have been a painful sub-sector to invest in and the wreckage is adding up.  An SVB report from last year has it pegged as the most challenging subsector in the med tech space (here).  According to that report, during 2005-2011, over $1.6B was invested in 106 companies focused on diagnostics.  In 2012, according to MoneyTree, another $168M was invested.  I’m skeptical on the return profile of lots of these investments.

Here are the six tough realities of the diagnostic space today:

  1. Diagnostic science isn’t any less risky than therapeutics.  The underlying bioscience uncertainty around the relevance and robustness of an assay or test for impacting/predicting outcomes isn’t any different than the uncertainty about whether a drug is pharmacologically active.  Many initial observations don’t get validated.
  2. Capital intensity remains very high for many Dx companies, as the exit point typically requires commercialization.  It takes a lot of capital and time to validate a diagnostic and bring it to market.  Large validation sets and clinical trials are expensive.  OncotypeDx, Genomic Health’s and the personalized medicine industry’s flagship product, was developed with samples from 2600 breast cancer patients as well as ~500 clinical trial patients, according to their 2005 S1 filing.  The bar has only gotten higher since then.  There are a number of solid private companies with launched Dx products that have consumed >$100M in equity capital: CardioDx, XDx, Tethys, Crescendo, to name a few.  These will need to be very large exits to generate a venture return.  MDV, a big backer of these plays in the past, has decided to deemphasize diagnostics (here) – it’s a telling sign when one of the leading investors with deep expertise in the field has decided to leave it.
  3. Diagnostics have a hard time extracting value-based pricing.  The difficulty of communicating value and then sharing in that value (be it cost savings or better outcomes) is significant, making many diagnostic models more of a high volume, low cost commodity business where scale is essential.  Further, unlike chronic or routine therapeutic use, there are many single use tests on the genetics side, one-offs on acute settings (MRSA), etc that make for challenging price-to-value considerations.
  4. Regulatory uncertainly is far higher than in therapeutics.  The FDA may have a high bar for new drugs, but it’s a bar that in most areas is fairly certain.  There are plenty of guidance documents and new, innovative drugs often get very clear paths to approval (SPAs, breakthrough status, accelerated approvals, etc…).  There are occasions in the past where the goal posts have moved (like antibiotics, obesity), but even in those areas the regulatory certainty has recently improved.  In diagnostics, regulatory uncertainty is a systemic risk.  It stems in part from the complexity of CLIA-certified home brew testing vs FDA-approved pathways, and what the future status of those pathways will be.  The requirements for approval need to be much more clear if this concern is to be overcome.
  5. Reimbursement remains a mess.  When you have multiplexed tests (like many new personalized medicine approaches) where the diagnostic itself involves multiple activities, reagents, characterizations, etc…, it’s well accepted that the old model of CPT code-stacking just isn’t going to be viable.  But its very unclear what the new model will be.  The process of getting a “new code” is arduous and uncertain, and typically requires mountains of data.  It’s simply not clear how to get paid for expensive but high value tests.
  6. Lastly, the exit market for venture-backed diagnostic companies remains weak.  For reasons #1 and #2, it’s very rare to see a big “early stage” M&A transaction (the equivalent of a $300M Phase 1 exit).  But getting companies to revenue stage requires lots of capital (#3) before acquirers are interested.  The big national labs (Quest, LabCorp) prefer accretive deals so don’t make frequent buyers of innovation.  The public markets may open up for diagnostic and life science tools players, but that’s not been a great path for future appreciation in the field.

All the above is not to say that you can’t make money in diagnostics, and a number of smart investors have and will continue to do so.  For instance, HBM made 21.6x or more on the sale of BRAHMS to Thermo a few years ago.  And there are a lot of very interesting emerging diagnostic companies, like Foundation Medicine here in Cambridge, that may achieve the escape velocity required to scale.  But it’s a tough space, or as I wrote a few weeks ago, “not for the faint of heart”.

At Atlas, we’ve reflected on these realities and decided to stick to what works for us in the Life Sciences – which is early stage therapeutics investing.  Looking at our aggregate track record over past 20 years, therapeutics have outperformed diagnostics/tools by a huge margin (5-fold).  Over the past few years, four challenging experiences helped tee up the perspective: OnQity, Aureon, Molecular BioMetrics, and Helicos.

  • OnQity.  Our failed tissue biomarker and circulating tumor cell play; see my recent post for the summary of the story and lessons learned (here).
  • Aureon Biosciences.  After a decade in the “systems pathology” space, Aureon got two products onto the market via the LDT/CLIA pathway.  But payors were reluctant to reimburse fully for the tests in the absence of “more data” (despite several large trials), CPT code-stacking wasn’t working, and the timeline for getting clarity on reimbursement was uncertain.  Although a good number of urologists/surgeons wanted to use the test, keeping a commercial organization going was prohibitive and so in late 2011 the company was shut down (here) – even after getting two products launched.
  • Molecular Biometrics.  MBI had developed a diagnostic in the IVF space for identifying embryos more likely to be viable.  We joined the B-round to fund the launch of what we thought was a ready-to-go test.  But unfortunately the underlying data wasn’t nearly as robust as initially believed: in hindsight, it was fundamentally a failure of over-fitting data during the hypothesis testing and validation stage.  The initial diagnostic signature looked great but was an illusion without the outliers.  The product was voluntarily withdrawn and the company shutdown.
  • Helicos.  One of several companies in the Next Generation Sequencing space, a hot technology area plagued by capital intensity.  NGS isn’t necessary a “diagnostic” space, but it’s a new technology area that supports diagnostic applications as commodity solutions.  NGS technology and commercial development has just cost a ton and yet the space is commoditizing rapidly.  Pacific Biosciences has raised over $600M, Complete Genomics over $200M, and Helicos over $200M.  Lots of losses for investors here as these NGS companies struggled to compete with the established competitors, differentiate their offerings, etc…

These four ill-fated deals (spread out over several fund cycles) aren’t representative of all diagnostic companies.  I’m sure there are many good diagnostics companies that will succeed and deliver value to both patients and shareholders.  But for us, at least for the time being, they give us pause about diving into more diagnostics deals, as they frame up lots of the issues and challenges with the field (the six realities above).

It’s certainly true that investors can mitigate some of the above risks by focusing on near-commercial bets or funding through to the revenue-stage.  Roll-ups of smaller diagnostic plays can be had to accrue scale benefits faster (though more of a growth equity than classic venture play).  But these investment strategies preclude the funding of early stage innovative diagnostic opportunities.

Further, this approach raises the question of how does one generate a return from early stage diagnostic investing – the translation of academia-derived insights into new clinically-validated diagnostic tests.  Who is going to fund the next generation of new innovative Dx plays?  I’m hopeful that models of consortia-based funding or pooling of biomarker insights and capabilities will come together.  Qiagen’s CEO Peer Schatz speaks eloquently about these types of ideas, among others.  The Personalized Medicine Coalition has certainly been a positive force for pushing the field around policy changes and these more collaborative models.  Much like the Pharma industry has engaged actively in the early stage therapeutics ecosystem through novel partnership models and active corporate venture investing, the larger players in diagnostics need to do the same in personalized medicine. Much more of this collaborative activity is likely required to attract the much-needed private capital into the field.

This entry was posted in Diagnostics, Personalized Medicine, VC-backed Biotech Returns. Bookmark the permalink.
  • I made the comment that I disagreed with each and every one of your “hard realities” of diagnostics, but have yet to back my quick retort. I start my response with the acknowledgment that health care venture and innovation is hard in all aspects, so this is by no means a suggestion that one sector beats all. Venture is hard, and it is risky. At the same time, I find myself at odd with your (as always, well written) recent post.

    As I said in my original reply to your post on On-Q-ity, it is in the business model where most folks lose their footing. It takes technology. It takes a market opportunity. It takes a team. Those remain true for all investments. But it also takes a business model that makes sense, because ultimately they need to be a business in order to be successful. I think plenty of folks either ignore the last one or really get tripped up by it.

    As for the hard realities…

    1) Of course, at their cores, most health care innovation hinges upon a new and often unconfirmed aspect of biology. One of our companies, Epic Sciences, is exploring new understandings in the field of CTCs. So, in that regards, you may have a point. But you forget that with a therapeutic, you are not only looking for a bioscience understanding, but evaluating how an outside “agent” impacts that biology, through efficacy, through tox, through safety, etc. That sets the bar considerably higher. At the same time, technology is a huge enabler of leveraging new bioscience, in much the same way that raw processing power may allow new understandings of complex data sets, or that wireless technologies may give rise to new forms of communication. As a case in point, Applied Proteomics is using the latest advances in mass spectrometry and data visualization to develop completely new diagnostic tools. BioNano is developing cutting edge tools for genome mapping and structural variation analysis to elicit new understandings. Both would not be possible just a few years ago, given the complexity of the data and the limits of technology..

    2) Capital intensity is a relative term. I believe it an overgeneralization that just because a company needs to reach a time point further than its development that axiomatically it requires more capital. That is comparing apples to oranges. And, just as the differences in costs in starting a new Internet company has plummeted over the past 10 years, so have the costs and capabilities of bringing completely new diagnostics and tools to the field. Sera Prognostics is enrolling over 4,000 patients in their trial to evaluate a new tool in evaluating pre term birth risk. I can assure you the costs of that trial are a tiny, tiny, fraction of the costs of doing a similar level of clinical work in a drug or devices based trial. Also implied in your statements is somehow a sense that it represents a form of failure to take a technology and play the long ball to make it into an incredible–and sustainable–business. I see the folks at Astute Medical as just one of several examples, in their case building the next Biosite with a portfolio of hospital acute care diagnostics. A hallmark of great VC should be building incredible and long lived businesses.

    3) The best way to extract value based pricing? Simply put, the technology in question needs to very clearly, and very definitively drive explicit value, either through targeting a clinical outcome, improving a clinical outcome, driving lowers costs, or better still, combination of all of them. Simply confirming a biological understanding or developing great science is not enough. It needs to be an actual product and one that customers (ie payors, physicians) want to buy. In the words of Peter Drucker, “the purpose of a business is to create a customer.” Veracyte is an excellent case in point, where their Affirma test taking much of the ambiguity around thyroid fine needle aspirations and using that technology to explicitly drive lower costs in the health care system. I’m reminded of Luke Timmerman’s recent article highlighting a recent E&Y report that many biotech executives are ready to prove their product have value. To ignore answering that question upfront is naive and a fatal business flaw.

    4) I’m not sure how to answer this point, because I find that the FDA process is as clear as mud, in all aspects. And I’ve seen even the simplest of drug technologies get held up at the FDA for arbitrary, inconsistent, and yes, even political reasons. Putting CLIA based technologies to the side, the experience of our companies in diagnostics has been fairly straightforward. As with all things, data, the quality of that data, and the safety of the technology in question, are paramount. If the value is clear, it will get a approved. If we don’t believe that, it is time for all of us to hang up our gloves.

    5) Reimbursement is not a mess, it is a complex and lengthy undertaking that any product, irrespective of the sector (drug, devices, or diagnostics), need to respect and embrace and take head on. As with point #3, the easiest way to navigate the complexities of reimbursement: clearly and definitively demonstrate the value. Do some companies use stacking codes and the like? You bet. But if the underlying technology is simply taking advantage of a stacking arbitrage in the hopes of “staying under the radar” they are doomed to failure. Such strategies should only be used as a means to an end. If you have a $1000 product and you cannot prove explicitly $1000+ of value, you are toast.

    6) The last point is hardest to debate as it requires on some level a willingness to be forward looking rather than looking in the past. I make the observation that from the big deals such as Thermo’s very recently announced deal to buy LIfe or Inverness’ (now Alere) deal to buy Biosite to smaller transactions such as our own Helixis sale to Illumina and Life’s acquisition of Ion Torrent, there is a level of activity that has been ignored. In the global goals of both lowering the costs of health care and improving/expanding the access to health care, diagnostics innovations are critical underpinnings to achieving both. And if they are successful in doing so, they will get acquired, or even better, grow into the next Amgens, Genentechs, and Genzymes. Ariosa is only a start.

    Am I wearing rose colored glasses? That is a fair statement. As I said before, as a device investor, I can take no credit whatsoever for all the wonderful work and excitement going on in diagnostics and lab products. However, I do live in San Diego, which has quietly become, in my belief, the center of global innovation for all things diagnostics, sequencing, and lab products. Do we have the next Kendall Square or South San Francisco in the making? I believe we do–in fact, I think it is already here. From the efforts of the folks at Scripps, to the new J. Craig Ventor Institute and West Institute, to all the great science being done here at the various academic centers and research facilities, to the dynamic large, and growing companies like Illumina to the numerous small start-ups, many of which we can count in the Domain portfolio, I find myself in the center of a dynamic ecosystem that is doing really great things. Innovation does get rewarded, and in that regard, I remain an optimist.

  • LifeSciVC

    Thanks for the thoughtful note – you raise lots of points, and certainly Domain is one of the “smart investors” I refer to that are likely to do well in the field. Not meaning to get into a blog post/repost argument, but thought I’d comment on a few points you make. In your opening, totally agree that Business Models are key. Specific comments around your reflections on the six realities:
    1. On Risk profile, in addition to the “unconfirmed aspect of biology” – which is similar – I also think lots of early diagnostic studies fail because its easy to find signal when you are hunting for predictive variables – the multiple comparisons problem – in many datasets. Its just easy to get excited at an early Kaplan Meyer plot, but a lot harder to replicate them. 2. On Capital Intensity, no doubt clinical trials are less in the Dx space on a per patient basis but the aggregate capital required to launch Dx products in order to get to an inflection point are often (not always) high. And yes I believe “great VC should be building incredible and long lived businesses”. My point was that when you’ve raised $100M or more the exit value needs to be very large in order to generate an adequate return for the early stage investors. Simply a matter of math. And historic distributions suggest those valuation outcomes are very very rare, and far less common than in biotech. Raising $100M over 5-10 years just to get to $20M in revenues and sell for 10x revenues isn’t very attractive. 3. Totally agree with you on the need to articulate value in both drugs and diagnostics. Its key. Differentiated offerings are a must. But asking a payor to “share” their cost savings by accepting the higher pricing of a premium expensive diagnostic is often a hard thing to do in practice. 4. The regulatory complexity is my point – CLIA/LDT vs PMA/FDA adds complexity and its not clear if LDTs (or what type of LDT e.g., multi-analyte tests) will be regulated in the future by the FDA, etc…This clearly adds additional uncertainty to the space. This is of course why PMC and others have focused on it as an issue. 5. On weak exit environment – my assessment is certainly informed by the relative lack of big exits in the past; its true, this could change and great exits may be around the corner. But I’m not going to bet on it.
    Hopefully your optimism is warranted and great things happen in Dx – for patients and shareholders. More success is will foster greater interest in funding more innovation in the field – and that will be a good thing. And might even bring Atlas back to considering Dx investments. But for the time being, we can make far better returns sticking to our focus on high innovation quotient early stage therapeutics.