Industry Insights: Five Key Figures From The Atlas Annual Review

I’ve always been captivated by and drawn to the intersection of raw emerging science, ambitious determined talent, aggressive capital, and savvy strategy that come together in an often-combustible mix to generate novel therapeutics. 

At the earliest stage, it’s critical to figure out what you’re going to aim at (the molecular target) and what type of therapeutic you’re going to use to hit that target.  Also essential: conviction that if you are technically successful and your product runs the gauntlet through regulatory approval, there will there not only be patients interested in using it (presumably your original inspiration for the medicine), but also, physicians interested in prescribing it, and payors willing to reimburse for it.

(I note in passing the strategy outlined above is deliberately target-centric, representing the approach many drug developers take. For a broader perspective, see this characteristically insightful discussion by Derek Lowe of an exceptionally comprehensive review [plus rich supporting materials] by Arash Sadri arguing target-based approaches haven’t been as productive as we tend to imagine.  Meanwhile, the utility of phenotypic-based approaches [see this 2011 piece I wrote for Forbes] may be gaining traction in the industry.  As noted in this recent publication by thoughtful scientists at Pfizer, phenotypic-based approaches were critical for the discoveries of ivacaftor for cystic fibrosis, daclatasvir for hepatitis C, and risdiplam for spinal muscular atrophy.)

The complex web of considerations involved in coming up with new medicines is not only what makes the process so challenging and exciting but is also why I look forward each year to Bruce Booth’s annual review of the industry.

Booth, as most readers know, is a partner at Atlas Venture. The Cambridge, Mass.-based firm invests in early-stage therapeutics companies. Booth also writes about the industry at LifeSciVC, and curates the always worthwhile “From the Trenches” blog. 

Atlas’s yearly examination of the state of the industry examines many of the key factors that influence medicine creation.  The last year, in particular, has been especially challenging for young and emerging biotech companies.  As Booth acknowledges, “2023 couldn’t end fast enough for most of us in biopharma; it’s been a tough year in the capital markets, and the industry is facing it’s fair share of headwinds.”  Even so, he recognizes that “despite all that, great science and medicine continues to advance.”

The entire 54-minute video, available on YouTube, is must-see for anyone in the industry (I’ve already watched the whole thing twice).  Five key figures – associated with important themes – stand out.

$80 Billion. This is the approximate projected size of the market for obesity medicines, a category, we learn, that is now considered “likely the biggest drug class in history.”  On the other hand, Booth notes, there are also very high healthcare costs associated with obesity which could presumably be ameliorated by effective treatment; certainly, this is the argument manufactures will continue to make.

Commentary: The newfound success of obesity medicines is fascinating. Up until very recently, obesity was considered a drug development graveyard and a space most companies wanted to avoid. The rationale: the amount of weight loss that seemed to be attainable through medicines had been fairly minimal, while the risks (remember fen-phen?) were well-documented. In particular, the concern was that regulators, aware of how widely new medicines in this category would be used, were likely to look at candidate medicines extremely critically – and to demand costly, and lengthy, cardiovascular outcome studies to document safety. (It was also often assumed that medicines in this category would be regarded as mere “lifestyle drugs,” like sildenafil for erectile dysfunction, rather than more essential medicines, like statins for reducing risk of cardiovascular disease.)

The success of the new Eli Lilly and Novo Nordisk medicines – delivering not only a remarkable degree of weight loss, but also documented evidence of cardiovascular benefit (as Booth highlights) – is what has changed the benefit-risk calculus so dramatically. 

In short, the drugs seem to work for many patients, and will give doctors something more powerful in their armamentarium than to helplessly advise “diet and exercise,” with minimal expectation of actual benefit (an experience that I vividly recall from my time in clinic).

One lesson: better medicines can completely reframe a therapeutic category; credit to Lilly and Novo for persisting.

Digital/data/technology question: will the arrival of effective weight loss medicines (with more in the offing, including oral formulations) create an opportunity for digital health companies to support patients on these medicines, as well as patients who can’t or won’t take these medicines, or will the arrival of effective pharmacological solutions largely obviate the need for a tech-supported behavioral fix?

21%.  This is the fraction of big pharma drug launches — roughly one of out of five new FDA approvals — that end up becoming “blockbusters.” These products are defined as generating annual sales of more than $1 billion. Another 18% are “high sellers” – $0.5B-1B a year; the rest of launches attain sales lower than that. But when accounting for the cost of R&D for each of these drugs, it turns out that the vast majority of revenue is derived from the blockbusters; high sellers cumulatively bring in slightly more than they cost, while the rest are estimated to contribute negatively to R&D returns, according to Booth. 

To bring the point home, Booth notes that just seven medicines – representing just 4% of total approvals over the last decade or so – generated approximately 28% of the total industry revenue.

Commentary: It is not news – or should not be news – that investment returns in pharma, like returns in many other industries including book publishing, music, and venture – follow a power law, rather than a bell-curve distribution (a distinction I discuss in my 2007 WSJ review of Nassim Taleb’s Black Swan, entitled “Shattering the Bell Curve,” here). 

But surprise or not, the continued dependence on blockbusters is a source of fragility for pharma companies who urgently need to find the next one before revenues dry up. It’s also a concern for patients with conditions associated with smaller markets. This can include not only rare genetic diseases, but also patients with cancers that might be exquisitely characterized and amenable to a precise molecular therapy. As Booth points out, not only can such populations be challenging to identify in the context of clinical trials, but even if a medicine for such a narrow indication is successfully approved, “if you’re a larger company, building a commercially sustainable long-term franchise based on such tiny market opportunities is incredibly challenging.”

Ideally, big pharmas would be able to support themselves through the development of a portfolio of products that might be exquisitely effective for relatively small populations. This was explicitly the ambition of precision medicine. Yet in practice, developing medicines for small, molecularly defined populations hasn’t generally been as efficient as was hoped and expected at the beginning of the genomics era.

Gene therapy provides another example of what Booth describes as “a real disconnect between clinical impact and value for patients” on the one hand, and the view of “the market and the street,” on the other. 

Despite gene therapy offering exceptional promise for patients, as well as a number of recent approvals, Booth points to data showing most companies in this space have been shellacked by the market; presumably this reflects investor concerns about the commercial promise.

In contrast, what seems to occur far more commonly is that the industry uses molecular characterization of disease to identify promising targets that can then be targeted in a range of conditions, often with little if any molecular characterization of the patient. This approach seems especially common in the very hot therapeutic area of inflammation and immunology, where the goal seems to be to replicate the success of the TNF-alpha inhibitors adalimumab (Humira) and the etanercept (Enbrel), each an example of the so-called “pipeline in a pill” (or injection in this case). 

A single medicine that can effectively treat a range of patients is of course a positive development. But the concern is that if the only medicines that big pharma can develop with favorable economics are those for huge markets, many patients with potentially treatable conditions are likely to be left behind.

12%. This is the fraction of new drug approvals over the last six years that are (a) from big pharma and (b) sourced internally. More than half of all new drugs approvals — 57 percent — are from “emerging pharmas” (i.e. not “top 20”). Many of these new drugs (perhaps not surprisingly) are for smaller market indications. 

But even within top 20 pharma, according to Booth’s data, only about a quarter of new approvals originated in-house; the rest relied upon the broader biotech ecosystem, and ultimately were in-licensed or obtained through the acquisition of a smaller company.

Commentary: In 2010, savvy industry analyst Andrew Baum, then at Morgan Stanley, famously suggested that pharmas (especially those not particularly good at internal research) consider evolving their model from R&D to S&D (search and develop), as I discussed here.  In today’s world in which internal discovery groups account for such a small fraction of all launched products, it seems reasonable to revisit Baum’s question, particularly through the lens of what qualities and capabilities are most critical for today’s big pharma discovery organizations?

6%. This, astonishingly, is the cumulative success rate for clinical trials in 2022 – down from 16% in 2017.  In other words, a product that enters a phase 1 clinical trial has only a 6% of emerging with regulatory approval, according to data Booth presented. 

Commentary: Why have things gone from bad to worse over the last five years?  Contributing factors suggested by Booth include challenges associated with the many new modalities that are coming forward (itself a good problem to have); tougher regulatory review; and unexpected late-stage blow-ups (admittedly a rather tautological “explanation”).  I suspect that another factor could be an increased interest in oncology, since the success rate for oncology drug development is notoriously low – I have heard “5%” thrown about.

Interestingly – and despite what seems to me like an all-out effort by every big pharma I know, and involving all the top management consulting firms – the composite development time associated with taking new products from preclinical studies through registration has actually increased slightly over the last five years, with most of the uptick occurring in phase 3. 

I suspect that as result of concerted efforts on the digital and data front, we will start to see some meaningful improvements in the efficiency of clinical trials – I don’t know any big pharma that’s not working on this. It’s possible technology can also help the industry achieve meaningfully reduced clinical development costs (which apparently haven’t changed much over the last five years either, according to Booth, and remain staggeringly high). But as I’ve discussed, a critical need for the industry right now is to improve the probably of success for trials, especially late-stage trials. Meaningfully increasing this value would be incredibly important and (as Andreas Bender has reviewed) also exceptionally difficult (extravagant claims to the contrary).

1. This is the number of times that Booth mentioned AI or ML in his talk. Specifically, it was in the context of drug failures (there were many), and included in this list was BEN-2293, which he described as the “lead program” of BenevolentAI. Booth’s conclusion: “Even AI and machine learning can’t help you escape the challenges of clinical trial attrition that faces so many companies.”

Commentary: Atlas has decided over the years to stick close to its knitting and focus on what might be called traditional early stage drug development; they are extremely excited about novel biology, but aren’t focused on emerging digital and data technologies.  Other early-stage biotech venture firms – Flagship comes to mind – are aggressively leaning into such technologies. 

Booth’s presentation captures a view of drug development that, in my experience, largely mirrors the perspective of most senior biopharma drug developers. It’s not so much anti-digital/data/technology as it is an anti-digital/data/technology exceptionalism. 

To hear and read Booth is to come away with an appreciation for the magnitude of our collective challenge. It’s exceedingly difficult to come up with new medicines, especially given the complexity of biology and disease, and our still relatively shallow understanding (see here).  While digital/data tools like Alphafold can be selectively helpful, they don’t seem (yet) to offer magic answers to the many challenges the industry faces.

Yet even with these caveats, we also must remember that, as I’ve discussed and as Booth highlights, we live in an incredibly exciting time for biological technologies. There are, as Booth emphasizes, a huge number of promising emerging modalities (even if most, as he also points out, are not yet associated with significant commercial value).  

“Fundamentally,” Booth observes, “we are awash in great science and great ideas.”

While I am not optimistic that AI will offer great stand-alone oracular value in the near-to-medium term (i.e. “develop this, not that”), I believe there are exceptional uses of digital and data technology to support and enhance emerging biological tools – in particular, to automate processes, capture data more consistently, and accelerate learning cycles. Harvard’s Marco Iansiti and Karim Lakhani have pointed to Moderna as a leading success story here.

Particularly in situations where the process is the product (cell therapies come to mind), the ability to rapidly iterate and optimize your product will likely depend upon how effectively you leverage the powerful digital and data technologies now at hand.

Ultimately, I suspect, the question won’t be — and shouldn’t be — whether to prioritize biological or digital technologies. It will be how to most effectively integrate both categories of technologies and pragmatically leverage their combined power to get better at developing the impactful new medicines we all hope to deliver to the patients awaiting them.