Why do 90% of drugs fail in clinical trials?

Professor Duxin Sun of the University of Michigan school of Pharmacy recently published a paper discussing why 90% of drugs fail in clinical trials. Despite advances in target validation, high-throughput screening, computation-assisted drug design, toxicity prediction, drug-formulation optimization, patient selection, and clinical trial design, the success rate of clinical drug development has been around 10-15% for the past several decades. The authors suggested that the structure -- tissue exposure/selectivity -- activity relationship(STAR) should be used as an optimization index. The problem and solution proposed by Professor Sun as an industry expert are very helpful for new drug discovery, but TODAY I want to analyze why the failure rate of new drug discovery has been maintained at about 90% from the perspective of payment.

In the West, there is a Sturgeon law that says "Ninety percent of everything is crap", so not only the research and development of new drugs, but also the best 10% in many fields are obviously different from other soy sauce products. Not only do only a handful of products make it to market, but only a handful of products make money for manufacturers, so the law is widely applied. Of course, everyone has different opinions on what crap means, and it's easy to get into the loop without a hard definition, such as defining the bottom 90% of CRAP standard. Of course, there are objective standards for the approval of new drugs in general, although there are also subjective elements of the department of Food and Drug Administration, so the success rate of new drugs hovering around 10% for many years may have a deeper reason.

The two checks and balances that determine human behavior are greed and fear, and drug discovery is no exception. Generally speaking, the higher the risk, the higher the reward, so the risk ceiling is the main factor. The return on the success of a new drug and the cost of a single drug determine how much manufacturers will tolerate the failure rate. MIT Professor Andrew Lo has calculated that if a single drug costs $200 million to develop and can be successfully marketed at $2 billion a year for 10 years (NPV at $12.3 billion at launch), then a 5% success rate is acceptable. At that cost, $30 billion would support 150 new product developments, with a 5% success rate per drug. At least 3 of those 150 projects would have a greater than 98% chance of success, resulting in a return of $36 billion. While research and development costs have soared over the past few decades, drug prices have risen in tandem, so the return on the market has been relatively constant and risk tolerance has stabilised at a relatively constant value of around 10%.

A success rate of 10% in different environments means different types of projects, and pharmaceutical companies will make adaptive adjustments according to the level of technology and payment environment to maintain this success rate. Although me-too is now a despised word, technology is lagging behind and even big pharmaceutical companies in Europe and the United States are using me-too to reduce risk. There are plenty of comments about me-Too from the likes of James Black and Paul Janssen. Do a search for them. If you go back to the 1980s and complain about the target validation data, people will ask you where the target is. If you go back to the 1970s, people will ask you what the target is. Programs were scarce at the time, with at least 20 structurally similar beta-blockers on the market, and even newer targets like statins with 10 highly similar products on the market. At the beginning of the 21st century, the regulatory environment changed radically. For several years, the pharmaceutical industry had to cut people and programs on a large scale to meet the requirements of the payment department, thus avoiding a sharp decline in the success rate. Periods of easy payments and technological breakthroughs (such as the last 7 or 8 years) have attracted a lot of capital to explore higher risk targets, so success rates have not improved significantly. In fact, the success rate of new drugs in different diseases is also different. In diseases with high willingness to pay, such as AD, the success rate of new drugs is far less than 10%.

So while STAR may significantly improve the success rate of drug discovery, the net result may be that it will drive manufacturers to chase higher returns and therefore higher risk targets, probably improving the overall level as other new technologies mentioned in the article, but not changing the reality that only 10% of products win in the inner roll. We always pursue the best programs that money can buy. Of course, the above calculation is based on the current parameters of the European and American markets, especially the American market. Risk tolerance could change if the US starts to restrict drug prices or if products cannot enter the US market. If you have to hit a 20% success rate to survive, you can either get better at it, or you can shop for a more mature project, or you can go home and sell sweet potatoes and avoid all risks completely.