新药技术转让的理想时段
转载自然杂志文章
From the analyst's couch:
The optimum time for drug licensing
James Kalamas1 & Gary Pinkus1
As the fortunes of 'big pharma' and the biotech industry become inextricably linked, effective drug licensing is becoming increasingly important. For the pharmaceutical industry, innovative biotech compounds have served to buttress lagging R&D productivity; whereas for biotech, partnerships with pharma have provided a stable source of much needed capital in the face of volatile public markets. In addition, pharma brings clinical development, portfolio management and commercialization skills that are lacking in many biotech companies. The optimal timing for drug licensing is an important strategic question, and the general view in both pharma and biotech is that the right time for these deals is during Phase II clinical development, although most deals take place sometime between preclinical and Phase II development. However, our analysis shows that pharma and biotech could create more value for themselves by doing deals earlier. To realize this value, both parties will need to change their approaches to drug licensing — both internally and in their dealings with potential partners.
The New Math
To understand optimal timing for deals, we developed a simulation model in which the major inputs included historical data by phase of clinical development (for example, probabilities of success, development costs, duration of each phase and historical licensing terms) and commercialization parameters for 350 biologicals on the market and in the clinic (for example, actual/projected revenue trajectory and margins). We then ran 10,000 simulations such that for each simulation a random set of input parameters was selected. The model determined the optimal timing (if there indeed was one) for a deal from the perspective of the biotech innovator and pharma partner, on the basis of the expected net present value of the compound at the beginning of preclinical development. Interestingly, the model showed a misalignment of reservation prices (the price below which a party in a negotiation is unwilling to strike a deal) between pharma and biotech. For deals that made economic sense, our simulation showed pharma should in-license more than 90% of the time at the start of preclinical development. For biotech, the model supported conventional wisdom, indicating that 55% of deals should occur during Phase II trials, with another 38% in Phase III.
This misalignment can be attributed to a mispricing of deals. When we sweetened deal terms offered by pharma to biotech (increases of 150% for preclinical, 100% for Phase I and 20% for Phase II deals), the model predicted optimal deal distributions for both pharma and biotech that favoured earlier deals. With more attractive licensing terms, pharma can still maximize value while encouraging biotech to partner compounds earlier in development.
Earlier is cheaper
On the basis of our modelling and industry experience, there is a compelling case for change in how drug licensing is transacted. In the case of pharma, early-stage licensing is a necessary component for creating robust long-term pipelines in a cost-effective manner. Competition — and, as a result, cost — for late-stage deals is increasing, and although late-stage deals come with higher probabilities of success, they also come with premium prices. Pharma is arguably best positioned to access less costly early deals while managing the increased risk associated with early-stage compounds.
Encouraging change
There are several prerequisites to effect this change. For pharma, a greater organizational focus on doing early deals will be required, appropriately balancing long-term investment in pipeline development with near-term earnings per share requirements. First, it will be necessary to recruit licensing staff as rigorously as R&D staff. This will mean hiring top licensing talent in a wider range of disciplines, including technology, finance, venture capital and law, and compensating them on par with research counterparts (who typically earn 20–40% more). Second, creating new managerial and staff incentives for earlier deals with more creative terms will be important. Minimum numbers of early deals should be set and meaningful bonuses offered for meeting those goals. Last, pharma must overcome the 'not invented here' syndrome, often fostered by research scientists, that can inhibit more aggressive in-licensing of drugs. To break down such barriers, executives must involve R&D thought leaders in the early identification and planning of biotech partnerships. Indeed, the head of R&D might have direct oversight of the licensing department, and managers in both departments should be assessed and compensated on the basis of the strategic importance and innovation of new compounds brought to market, regardless of whether they were developed internally or externally.
For biotech, senior managers will need to be pragmatic about the expected value of their compounds (probability adjusted) and take into account the non-cash benefits of a partnership with pharma (for example, validation of technology, development and commercialization skills and portfolio management discipline). According to the Wharton School of Business, drugs produced by pharma–biotech alliances are 30% more likely to succeed in gaining US FDA approval than those developed by a single company. This, combined with a thoughtful identification of the best pharma partners (in terms of complementary skills and cultural fit), will increase the probability that equitable early-stage deals can be struck.
Implications of the New Math
The model we propose is not a quick fix. The drug development and approval process will remain long, expensive and frequently disappointing. But done right, early licensing can become a productive and necessary addition to pharma's long-term strategy. Those that miss out will need to rely on their internal innovation to compensate, and many will find themselves with waning pipelines and likely take-over targets for stronger players in search of cost synergies and marketed products. Those biotechs who can convince partners to join them early on will become integral to pharma R&D; those who cannot will be branded as damaged goods, making it even more difficult for them to raise capital and remain afloat. Early adopters of the 'New Math', in both pharma and biotech, will be the winners and will probably serve as catalysts for the often discussed, but as yet unrealized, biotech industry rationalization.
Market indicators
On the basis of historical deal terms, the conventional wisdom that biotech should wait until Phase II to partner its compounds is supported by our simulation (Fig. 1). This behaviour is probably the result of mispricing deals. The theoretical distribution of deals expected if deal pricing is fair (that is, pricing reflects the risk-adjusted net present value of the compound at all phases, with both parties indifferent as to when licensure occurs) predicts a preference for earlier deals. This is not consistently supported by actual deal making (Fig. 2). With improved deal terms, big pharma and biotech preferences for earlier deals become aligned (Box 1, Fig. 3).
图一
a | Typical amounts of payments and royalties from pharma to biotech when deals are done at each phase of product development. b | Under these conditions, our simulation supports the biotech preference to wait, indicating that value is maximized when partnerships are consummated in Phase II and beyond. The data exclude 61% of compounds for which no deal terms could be agreed to.
注： 这是比较老的数据和行情，新的统计数据待更新
图二
Mispricing of deals is probably the culprit for biotech preference to wait until Phase II to license compounds. Actual deals shown represent 79 deals between biotech and top 12 pharma from 1991 to 2002, after screening out platform technology deals and some product deals for which information was incomplete. The theoretical deal distribution assumes that all deals are neutral — that is, pricing reflects risk-adjusted net present value of the compound at all phases, with both parties indifferent as to when licensure occurs, thereby reflecting the statistical fall-off of compounds in the pipeline due to attrition.
转载自然杂志文章
From the analyst's couch:
The optimum time for drug licensing
James Kalamas1 & Gary Pinkus1
As the fortunes of 'big pharma' and the biotech industry become inextricably linked, effective drug licensing is becoming increasingly important. For the pharmaceutical industry, innovative biotech compounds have served to buttress lagging R&D productivity; whereas for biotech, partnerships with pharma have provided a stable source of much needed capital in the face of volatile public markets. In addition, pharma brings clinical development, portfolio management and commercialization skills that are lacking in many biotech companies. The optimal timing for drug licensing is an important strategic question, and the general view in both pharma and biotech is that the right time for these deals is during Phase II clinical development, although most deals take place sometime between preclinical and Phase II development. However, our analysis shows that pharma and biotech could create more value for themselves by doing deals earlier. To realize this value, both parties will need to change their approaches to drug licensing — both internally and in their dealings with potential partners.
The New Math
To understand optimal timing for deals, we developed a simulation model in which the major inputs included historical data by phase of clinical development (for example, probabilities of success, development costs, duration of each phase and historical licensing terms) and commercialization parameters for 350 biologicals on the market and in the clinic (for example, actual/projected revenue trajectory and margins). We then ran 10,000 simulations such that for each simulation a random set of input parameters was selected. The model determined the optimal timing (if there indeed was one) for a deal from the perspective of the biotech innovator and pharma partner, on the basis of the expected net present value of the compound at the beginning of preclinical development. Interestingly, the model showed a misalignment of reservation prices (the price below which a party in a negotiation is unwilling to strike a deal) between pharma and biotech. For deals that made economic sense, our simulation showed pharma should in-license more than 90% of the time at the start of preclinical development. For biotech, the model supported conventional wisdom, indicating that 55% of deals should occur during Phase II trials, with another 38% in Phase III.
This misalignment can be attributed to a mispricing of deals. When we sweetened deal terms offered by pharma to biotech (increases of 150% for preclinical, 100% for Phase I and 20% for Phase II deals), the model predicted optimal deal distributions for both pharma and biotech that favoured earlier deals. With more attractive licensing terms, pharma can still maximize value while encouraging biotech to partner compounds earlier in development.
Earlier is cheaper
On the basis of our modelling and industry experience, there is a compelling case for change in how drug licensing is transacted. In the case of pharma, early-stage licensing is a necessary component for creating robust long-term pipelines in a cost-effective manner. Competition — and, as a result, cost — for late-stage deals is increasing, and although late-stage deals come with higher probabilities of success, they also come with premium prices. Pharma is arguably best positioned to access less costly early deals while managing the increased risk associated with early-stage compounds.
Encouraging change
There are several prerequisites to effect this change. For pharma, a greater organizational focus on doing early deals will be required, appropriately balancing long-term investment in pipeline development with near-term earnings per share requirements. First, it will be necessary to recruit licensing staff as rigorously as R&D staff. This will mean hiring top licensing talent in a wider range of disciplines, including technology, finance, venture capital and law, and compensating them on par with research counterparts (who typically earn 20–40% more). Second, creating new managerial and staff incentives for earlier deals with more creative terms will be important. Minimum numbers of early deals should be set and meaningful bonuses offered for meeting those goals. Last, pharma must overcome the 'not invented here' syndrome, often fostered by research scientists, that can inhibit more aggressive in-licensing of drugs. To break down such barriers, executives must involve R&D thought leaders in the early identification and planning of biotech partnerships. Indeed, the head of R&D might have direct oversight of the licensing department, and managers in both departments should be assessed and compensated on the basis of the strategic importance and innovation of new compounds brought to market, regardless of whether they were developed internally or externally.
For biotech, senior managers will need to be pragmatic about the expected value of their compounds (probability adjusted) and take into account the non-cash benefits of a partnership with pharma (for example, validation of technology, development and commercialization skills and portfolio management discipline). According to the Wharton School of Business, drugs produced by pharma–biotech alliances are 30% more likely to succeed in gaining US FDA approval than those developed by a single company. This, combined with a thoughtful identification of the best pharma partners (in terms of complementary skills and cultural fit), will increase the probability that equitable early-stage deals can be struck.
Implications of the New Math
The model we propose is not a quick fix. The drug development and approval process will remain long, expensive and frequently disappointing. But done right, early licensing can become a productive and necessary addition to pharma's long-term strategy. Those that miss out will need to rely on their internal innovation to compensate, and many will find themselves with waning pipelines and likely take-over targets for stronger players in search of cost synergies and marketed products. Those biotechs who can convince partners to join them early on will become integral to pharma R&D; those who cannot will be branded as damaged goods, making it even more difficult for them to raise capital and remain afloat. Early adopters of the 'New Math', in both pharma and biotech, will be the winners and will probably serve as catalysts for the often discussed, but as yet unrealized, biotech industry rationalization.
Market indicators
On the basis of historical deal terms, the conventional wisdom that biotech should wait until Phase II to partner its compounds is supported by our simulation (Fig. 1). This behaviour is probably the result of mispricing deals. The theoretical distribution of deals expected if deal pricing is fair (that is, pricing reflects the risk-adjusted net present value of the compound at all phases, with both parties indifferent as to when licensure occurs) predicts a preference for earlier deals. This is not consistently supported by actual deal making (Fig. 2). With improved deal terms, big pharma and biotech preferences for earlier deals become aligned (Box 1, Fig. 3).
图一
a | Typical amounts of payments and royalties from pharma to biotech when deals are done at each phase of product development. b | Under these conditions, our simulation supports the biotech preference to wait, indicating that value is maximized when partnerships are consummated in Phase II and beyond. The data exclude 61% of compounds for which no deal terms could be agreed to.
注： 这是比较老的数据和行情，新的统计数据待更新
图二
Mispricing of deals is probably the culprit for biotech preference to wait until Phase II to license compounds. Actual deals shown represent 79 deals between biotech and top 12 pharma from 1991 to 2002, after screening out platform technology deals and some product deals for which information was incomplete. The theoretical deal distribution assumes that all deals are neutral — that is, pricing reflects risk-adjusted net present value of the compound at all phases, with both parties indifferent as to when licensure occurs, thereby reflecting the statistical fall-off of compounds in the pipeline due to attrition.