Making Licensing Pay

Does in-licensing improve attrition rates? It hasn't reversed the R&D productivity decline-in part because while in-licensed candidates are less risky than internally invented ones, they generate lower average revenues. But companies can do better: we show statistically that they need to place more emphasis on licensing products early. Moreover, to be successful at early-stage licensing, drug firms need focus, not on specific therapeutic areas, but on specific biological mechanisms.

Early-stage licensing can dramatically lower R&D attrition and increase returns, granted that companies focus their activities on specific mechanisms of action.

By Jeff Boschwitz, Behzad Aghazadeh, Catherine Arnold, Tong Zhang, and Charley Beever

Back in January, in our first article in this series on R&D attrition, we focused on the importance of managing the pipeline's mix and content relative to various classes of molecules and targets. (See "Know thy R&D Enemy: The Key to Fighting Attrition," IN VIVO, January 2005 (Also see "Know Thy R&D Enemy: The Key to Fighting Attrition" - In Vivo, 1 Jan, 2005.).) Drug companies, we noted, had already reduced their emphasis on novel targets, and thus, they assumed, their pipeline risk. But our analysis of a database of drugs developed between 1997 and 2004 showed that companies could leverage more important risk-reducers, including the class of molecules they developed (biologics targeting novel mechanisms fail less frequently than small molecules targeting precedented mechanisms) and the specificity of the mechanisms they worked on (targeted molecular therapies are less risky than broad). (For an in-depth discussion of the database, its components and the definitions of our categories, see sidebar, "Analyzing Pipelines: Our Methodology" in the January 2005 article.)

But in our previous discussion, we didn't focus on another dimension of the pipeline: do in-licensed drugs fail more than self-invented products? That's what we'll do here.

With good reason. The industry is doing a lot more licensing. Between 1997 and 2004, in-licensed drugs as a percentage of the late-stage pipelines of the top 14 pharmaceutical and top 5 biotechnology companies have risen some 26%. They now comprise 34% of the total development portfolio.

All this activity, however, hasn't reversed the R&D productivity decline—companies aren't producing the sales from new products required to replace and increase sales from patent-expiring drugs. The problem isn't that in-licensed candidates are any riskier than internally invented ones: in fact, they are less risky. In-licensed products succeed in late-stage development 35% more frequently than self-originated drugs. This advantage is, however, offset by lower average revenue generated in the market. (See Exhibit 1.) The net effect is that licensed drugs in total deliver as much downstream revenue as self-originated drugs, but not more.

But simple generalizations about in-licensing aren't enough to provide strategies for maximizing its effectiveness. In this article, we provide the outline of a risk-reducing, productivity-enhancing licensing strategy. In particular, companies need to place more emphasis than they do today on licensing products early, a more cost-effective approach. Indeed, they're already licensing more early-stage compounds (16% of their portfolios in July 2004, up from 11.9% in July 1997)—but they're also licensing more late-stage compounds (18.3% of their portfolios, up from 15%).

Moreover, to be successful at early-stage licensing, drug firms need focus, not on specific therapeutic areas, but on specific biological mechanisms. Opportunism works in late-stage licensing, but the universe of early-phase compounds is too large and too diverse. A mechanistic focus, and the expertise that drives the choice of mechanisms, not only yield higher success rates but help position a company to be the partner of choice—a term that makes sense only within relatively narrow categories, such as mechanisms.

Licensing Earlier

Some companies are hesitant to focus too many resources on licensing early-stage products—those in preclinical or Phase I development. They take the position that most of the available compounds have little value. Furthermore, internal scientists often resist embracing such projects because the external compounds don't meet their scientific standards or because they directly compete with internal research programs.

Our database showed, however, that these early-stage in-licensed compounds are just as successful in Phase II and Phase III as drugs licensed in those later stages. Just 14% of drugs licensed in Phase II make it to market; but 16% of drugs licensed in preclinical or Phase I and which make it to Phase II go on to approval.

More importantly, even though deal values for early-stage deals are rising while late-stage deal costs seem to have recently stabilized, the economics of Phase I and preclinical dealmaking are still at least as good as licensing during Phase II or Phase III, even factoring in the need to evaluate many low-value compounds before a suitable product is identified. (See Exhibit 2 and sidebar, "The Economic Model;" see also "Why Early-Stage Dealmaking is Hot," IN VIVO, December 2004 (Also see "Why Early-Stage Dealmaking is Hot" - In Vivo, 1 Dec, 2004.).)

This latter finding should not be surprising given the small supply of promising un-partnered compounds in late-stage development and the intense competition to acquire them from companies often pressured to fill pipeline gaps. Competition is particularly acute for drugs specific for unprecedented targets. In our prior article, we showed that 94% of Phase II projects of this type do not make it to launch, leading companies to scramble to fill the void left behind. However, as of July 2004, there were just 35 Phase III and 116 Phase II NCEs (new chemical entities) specific for unprecedented targets that were not already partnered with large pharmaceutical or biotechnology companies. Moreover, a large portion of these still unpartnered candidates are specific for niche markets and/or work on mechanisms relatively unattractive to the top drug companies.

Exhibit 2 also shows that early-stage in-licensing is far more economically attractive than developing self-discovered products. This finding does not mean that companies should cease early-stage research, though they should perhaps scale it back. Plenty of companies have argued, and we agree, that the knowledge and capabilities that come with robust internal research can drive licensing success for Big Pharma.

But this finding strongly suggests that licensing should be a source of ideas at least equal to in-house programs and thus a major contributor to the overall discovery research portfolio. In fact, some large biotech companies have implemented or are implementing portfolio management processes that compare self-originated products to external opportunities. A few Big Pharmas have also moved or are moving in the same direction, despite its management difficulties: the large number of internal early-stage programs creates a daunting challenge for the search and assessment of external comparator projects. But despite its expense, such a comparative program is probably a worthwhile anti-attrition program at least for the highest risk programs--NCEs for unprecedented targets, for example.

Licensing Focus

Still, companies need to do far more than simply emphasize early-stage licensing. They need to focus—narrowly.

Focus isn't universally popular. Some senior licensing executives believe that restricting their efforts within a rigidly defined framework will cause them to miss licensing gems. Indeed, such opportunism may be the best approach for late-stage licensing. There are just 850 unpartnered compounds in Phase II and III and an efficient organization can probably effectively assess most if not all of them.

But opportunism in early-stage licensing destroys value. In the first place, it's impossible to optimally separate the wheat from the chaff given the sheer number of early-stage projects: several thousand in preclinical and Phase I development at any given time, most of these in development by small biotech companies and academic or government agencies. And since not all mechanisms are created equal, companies need to place informed bets to reduce downstream attrition.

The key question is how narrowly in-licensers need to focus their attention and what criteria they should use. Our data suggests narrowing only as far as therapeutic areas is not sufficient. Therapeutic expertise, in fact, doesn't seem to be much of a predictor of licensing success—at least with mid-sized companies, where drugs licensed in new categories succeed as often (15% of the time from Phase II to launch) as in-licensed drugs in therapeutic areas well known to the developer (12% success rate).

On the other hand, a mechanism focus does increase licensing success. For example, as we showed in the previous article, success rates for non-specific (or broad-acting) cancer drugs are far lower than those for more targeted oncologics. It should be no surprise that Genentech Inc. , which has focused on both developing and licensing drugs with targeted cancer mechanisms, has been more successful in this therapeutic area than any of the Big Pharmas who for the most part have pursued both non-specific and targeted cancer therapies.

Further supporting the notion of a mechanistic focus is the fact that the first compound in a multi-product deal, or the only compound in a single-product arrangement, is far more likely to succeed than a follow-on product that's part of the same collaboration. (See Exhibit 3.) Such follow-on drugs often utilize different mechanisms of action than the first drug licensed in the alliance, which in general is the product that attracted the in-licenser in the beginning. It seems that in such instances licensors lose a degree of objectivity as the relationship between the companies evolves. Keeping with a mechanism focus can help guard against this unscientific subjectivity.

Becoming the Magnet

To be sure, companies need to focus their efforts on more than merely specific biological pathways. Drug companies which in-license biologics, and do so based on internal biologics expertise, will do better than the average in-licenser. In the first place, biologics get to market more reliably than small molecules. But in-licensed biologics have died in development more frequently than self-originated biologics—because, at least in part, the in-licensers lack the requisite skills and thus the decision-making capabilities for selecting and developing the right products. (See Exhibit 4.)

Moreover, focus—mechanistic or molecular—has another advantage: it attracts superior licensing candidates. Small companies are more likely to proactively approach such focused licensers early in their business development process, before reaching a stage where numerous competitors have a chance to assess the opportunity.

The importance of being a licensing magnet can be seen even when looking at the industry level. Mid-size pharmas and large biotechs have far lower success rates for in-licensed than self-originated drugs while the reverse is true for Big Pharma. (See Exhibit 5.) The reason seems to be that the best products go to big companies—they pay more and often are in a position to provide superior downstream development and marketing power. This creates an even greater imperative on mid-sized companies to create and execute a well-articulated licensing strategy that will allow them to build differentiating capabilities in the eyes of licensees.

Certainly, it is often difficult to resist the blandishments of opportunism. Focus requires choice—and companies can easily choose wrong. Better, argue the opportunists, to hedge one's bets, playing the broader market—a kind of index of available opportunities and mechanisms. But playing such an index will—like most indices--more than likely generate merely average returns, not better. And right now the average is poor. It takes 18 small-molecule Phase II NMEs specific for unprecedented targets to get one product launched—a 6% success rate (as opposed, for example, to a 17% rate for biologics or small-molecule NMEs specific for precedented targets). Developing a strategy to beat the average is critical. One approach is licensing early with a focus on specific unprecedented small-molecule targets.

In our next article based on this data, we will broaden our discussion to consider the profitability implications of structuring development portfolios with different mixes of licensed and in-house compounds.

The Economic Model: Licensing and Self-Discovered NPVs

Our analysis of licensing and self-discovered drug NPV was performed using published data on R&D costs (PhRMA) and product revenue (IMS, company filings), compound data derived from our pipeline database, transaction data from Windhover's Strategic Intelligence Systems database, and a 10% cost of capital. We discounted all cash flows back to the time of internal project inception (this could be 5-6 years before an early-stage licensing deal was signed). We assumed that drugs would reach their peak revenues on average 4-5 years after launch and that they would have an average of 11 years of patent life. Finally, based on average 2004 revenue of all drugs launched by Big Pharma and Big Biotech since 1997, we calculated that self-originated products on average generated $889 million in peak revenues while in-licensed drugs generated $653 million.

Behzad Aghazadeh, PhD, and Jeff Boschwitz, PhD, are senior associates and Charley Beever a partner, worldwide pharma business, for Booz Allen Hamilton. Catherine Arnold is director, and Tong Zhang tk, major pharmaceuticals equity research, for CS First Boston.