“Research Productivity”; these two words have haunted our industry for the last decade. They have served as a rallying cry, a portent of doom and now may actually be driving some positive change. But before we jump to any conclusions on the productivity of specific segments of the business, let us remember that when industry analysts quote numbers for productivity, they are comparing investment in research against drugs launched and that the timeline for this process is still 12-15 years. That means that any improvement made today will not have an impact until about 2020. So while the fact remains that productivity in the pharmaceutical industry is not meeting anyone’s expectations; there is a great deal of innovation occurring in the early part of the pipeline and it is having an impact.
As with most productivity problems, the solutions that are working in drug discovery are complex and involve technology, technique and, perhaps most importantly, process changes. Simply improving productivity in one part of a linear process will only shift the bottleneck downstream. The racecar analogy is often used to illustrate this; if you had a formula one car would you get to work any faster in a heavy traffic area? The image of a vehicle capable of traversing your commute in mere minutes trapped in a traffic jam for hours brings this home.
So to truly improve productivity, we must examine the process and be willing to make changes that will increase end result productivity. This does not mean that the capacity at all points in the pipeline has to be increased. Because most of the cost comes from projects that fail and the direct costs escalate exponentially as projects progress, shifting failure to an earlier point in the pipeline or modifying preclinical processes to decrease the probability of clinical stage dropout will result in drastic changes in overall costs. It has been estimated (Dimasi; 2002) that a 10% change in clinical success rate will have an impact of >$200 MM per drug!
One approach that is becoming widely applied in drug discovery and preclinical development is broad profiling. Lead compounds are tested against a battery of assays to determine specificity as well as pharmaceutical suitability. This has already had a positive impact in both guiding medicinal chemistry projects and in identifying the projects that will never become a good drug. But the cost of profiling has meant that it is used sparingly and often only applied after significant cost and effort has been expended. This results in projects that are about to achieve significant milestones suddenly requiring months of additional effort, which may result in project termination. The question we must ask ourselves is whether having that information at the outset would have resulted in a more efficient process. Could the scientists involved have used broad perspective data to bring this project to a successful conclusion? Perhaps more importantly, would they have chosen a different series if they had this perspective?
In my experience, the answer to all of these questions is “yes”. Not only that, but the underlying cost structure is very positive. That is, the cost of generating “decision critical” data is an order of magnitude less than the loss from failure or restarting a program.
There are many ways to efficiently gather and utilize this data. At Amphora Discovery, we have applied this approach internally and for customers at many points in the drug discovery process. The most extreme process change is to gather the data at the outset by testing full chemical diversity libraries against both therapeutic targets and counter-screens at the HTS stage. If done correctly, this allows the omission of several steps in the process including confirmation, profiling and potency testing. More importantly, it allows the Chemist to begin with a series that already has the desired selectivity profile so they can focus on driving potency and pharmaceutical suitability rather than tackling the more difficult multidimensional problem of selectivity. Broad profiling of active compounds immediately after HTS or computational selection is second in its impact on process efficiency savings. This allows the combination of the confirmation (and possibly potency determination) steps and still provides the critical selectivity information for medicinal chemistry. The impact of perspective data on efficiency decreases as the project progresses but the importance of having the information does not. You will not realize the same costs savings by adding profiling at the lead stage but you will still need the data throughout this process to ensure that your project’s profile is not changing during optimization.
So the benefits of broad profiling are clear. Not only that, but the impact is at least somewhat validated by its increasing use in the industry. The big winners in this game will be the companies that are organizationally nimble and can take advantage of process and budgetary flexibility to realize increased productivity.