The transition to personalized medicine is gathering momentum, but the high cost of clinical development remains a barrier between the pharma industry and a secure path to a profitable new era. Development costs will have to shrink dramatically as blockbusters dwindle and products for small or in many cases very small markets take center stage. Recent news suggests the industry is still living on blockbuster development budgets while many approved products are likely to generate boutique revenues.
Not surprisingly, the industry is still trying to produce new blockbusters for indications like Alzheimer’s disease that affect millions of people. Sometimes the industry will succeed. However, like most broad indications, Alzheimer’s has proved a high-risk, high-cost target for drug development. Approved products may slow the progression of dementia for some patients for 6 – 12 months. The potential market is vast and amyloid deposition remains an attractive target despite several failed studies of amyloid inhibitors in mild-to-moderate AD. The industry has advanced several blockbuster candidates, including Lilly’s solanezumab, Pfizer/Johnson & Johnson’s bapineuzumab and Genentech’s crenezumab. The cost of J&J’s program exceeded $700 M. Genentech’s Colombia study of crenezumab will cost $100 M.
Against a backdrop of staggering development costs and constant clinical failures, FDA published a new guidance document in February, Alzheimer’s Disease: Developing Drugs for the Treatment of Early Stage Disease. The guidance acknowledges some of the challenges of demonstrating efficacy in Alzheimer’s and Mild Cognitive Impairment (MCI) studies and suggests a variety of outcome measures. These include the Clinical Dementia Rating – Sum of Boxes (CDR-SB) score as a single primary efficacy measure that combines both cognition and function; certain isolated cognitive measures; and time to a dementia diagnosis. FDA also expressed willingness to consider additional outcomes measures.
As with every indication, operational issues also figure in Alzheimer’s development costs and failures. Among the challenges is variability associated with subjective outcome measures, which translates to large sample sizes and still higher costs. You can keep variability within bounds provided you can promptly detect spikes in variability data. In one Alzheimer’s study, a Health Decisions study team observed a spike in ADAS-Cog (Alzheimer’s Disease Assessment Scale-Cognitive Subscale) scores at a site. Streaming data enabled us to note the spike promptly. We found that the trained interviewer was on sick leave and a nurse without appropriate training was filling in. Similarly, we detected data changes because of rotation of specific test elements. The ability to quickly detect such problems enabled us to keep variability, and costs, within bounds. Unfortunately, like other Alzheimer’s candidates to date, the drug failed.
$100 M Study Costs vs. $50 M Revenue Projections
Dreams of new blockbusters look less realistic in March 2013 in light of the types of drugs approved in 2012. With 39 new FDA approvals, 2012 at first looked like a banner year for drug development. Some industry reactions were almost euphoric. Drug Baron has provided a dash of realism. For starters, four of the 39 approvals were not drugs. The Drug Baron analysis finds one likely blockbuster among the 35 approved drugs with a couple of other possibilities. Many of the approved drugs have small markets. According to the analysis, at least 15 of 2012’s new FDA approvals will struggle to generate $50 M in revenues.
Genentech’s Colombia Alzheimer’s study will cost $100 M. More than 40% of the drugs approved in 2012 will generate revenues on the order of $50 M. Something doesn’t compute.
Returns from the class of 2012 will not be sufficient to fund a robust new generation of development on the current development model. Profitably developing drugs for smaller markets will require a much higher success rate, substantially lower development costs or both. Indeed, a development model for the new era must be far, far more efficient. We need adaptive design but adaptive design alone will not be enough. We also need a leap in operational efficiency through techniques such as adaptive monitoring and early detection of problems like spikes in variability, as described above. The time has come for traditional clinical development to make way for personalized medicine.