Drug Discovery Prologue (Dunning Kruger Saga Part 2)

Drug discovery is awfully slow, expensive and prone to high failure rates.

About 96% of drugs being developed will fail. Some researchers have estimated that the current probability of success is not much better than a random approach.  

The average cost of bringing a drug to commercialization is estimated to be between $1b to $2.6b. This cost figure includes cost expended on the many failures. Over the last 7 years, the average costs have risen by over 100%. At current trends, within 20 years, the average cost per drug to commercialization will balloon to over $20b.

The average time to bring a drug from discovery to market ranges between 10 to 15 years.

In 2019, the pharmaceutical industry spent $83 billion dollars on R&D. This figure is likely to be much higher in the wake of COVID19. Adjusted for inflation, this is about 10 times what the industry spent per year in the 1980s. For the industry as a whole, over ten years from 2010 to 2019, Deloitte’s estimated the returns on R & D has fallen from 10% to 1.9%. In 2020, the return increased slightly to 2.5%.

The major causes of failure in drug discovery come from a combination of efficacy and side effects. Other significant factors include commercial factors and avoidable errors such as bad trial designs and non-compliance with regulatory requirements.

This is a problem that is in desperate need of a solution. The size of the pharmaceutical industry is estimated to be about USD$1.5t in 2020, and this is projected to grow at about 7% CAGR for the next 8 years.

The next blog post will examine the fundamental reasons for the dismal outcomes in drug discovery. Then we will take a stab at the possible solution providers.

Yours One Legged

Biotech Investing- A Failed Approach (Dunning Kruger Saga- Part 1)

A lesson from my recent pivot into the lifesciences and healthcare field is an old familiar one. 

The first area I explored is the field of drug discovery. More particularly, single product narrow focus drug discoverers. It is the equivalent of looking at mineral explorers listed on the ASX. The main rationale for this approach is that drugs are important drivers of  the entire industry. I need to first understand the drugs and treatments in order to understand the ecosystem.

Being a numbers person, naturally I started looking at the various published studies on the historical success rates of various drugs in different areas eg oncology, vaccines, infectious diseases, etc. The supposition is that if I know the probabilities of success, I may be able to put bets on expected positive outcomes. I could then size the positions based on the Kelly Criterion and run a portfolio on this basis.

For example, say the base historical success rate of oncology is 3%. If the total addressable market for a particular drug in development is over $1b, we can estimate that if the drug achieves commercial success, the market cap of the company should be well over $5b. Given a 3% success rate, the entry price for a positive expectancy must be below $150m (assuming no dilution).

Let’s say we can get an entry at $100m. At 3% probability of success based on being undiluted (an unrealistic assumption already), but with a potential 50 bagger, we have a positive expected outcome. The Kelly Criterion based on this number says we should risk 1% of the portfolio. A more practical half Kelly stance puts us on 50 basis points.

Let’s now say we can find 200 positions at 50 basis points to fill out our entire portfolio. What can possibly go wrong?

We have just made an assumption of independence of outcomes between all the bets. In other words, we are assuming all of them are separate throws of dice with each outcome not influencing one another. Biology does not work that way, so drugs seldom work (or don’t work) that way too. Developments with one drug candidate can have significant influence on many other drug candidates in adjacent areas. When the debris has settled, I might just find that 20% of my portfolio has been wiped out because they are in essence all connected in some way. It could be a particular biological pathway, it could be a mechanism of action, it could be a flawed delivery mechanism, it could be a previously undiscovered genetic problem, it could be a downstream side effect, it could really be anything if I just did not know where to look.

More importantly, my competitors will usually have an army of 15 or more PhDs in various areas of medicine and science, and they collectively have over 300 years of experience. They have expert networks and resources from which to draw upon. They are paid very well to monitor a much larger pool of drug candidates in development, way bigger than my pool of 200 that I will struggle to just keep up with.  And I bet you they knew where to look well in advance.

This is one contest I want to avoid.

Let me know your thoughts in comments below.

Yours One-Legged.