Take an example of failed target validation. Eli Lilly is just one company but this could be true of almost all the pharmaceutical companies.
Eli Lilly announced in August 2012, that they have stopped “ongoing clinical studies investigating pomaglumetad methionil, also known as mGlu2/3, for the treatment of patients suffering from schizophrenia” in Phase 3. (http://newsroom.lilly.com/releasedetail.cfm?releaseid=703018). On Feb 7, 2013, announced that they will “discontinue the Phase 3 rheumatoid arthritis (RA) program for tabalumab, an anti-BAFF (B cell activating factor) monoclonal antibody, due to lack of efficacy” also in Phase 3 (http://newsroom.lilly.com/releasedetail.cfm?releaseid=738769)
Phase 3 trials are expensive. If you consider a conservative estimate of $5000 per patient and then multiply that with 5000 patients then you get $25M for a trial. This does not account for all the resources, and money that have gone up to that point in developing the drug. The Pharma companies like to fail early so that they do not spend their resources and media goodwill in failed drugs.
However, phase 3 failures happen and these are not for safety. These are for validation that the molecule inside the human body for which this drug was developed is not really a good target for the drug. There are many causes for this failure but these may be the top three.
One target one disease hypothesis. Since drug development is expensive, most drug companies focus on one molecule to effect. The central hypothesis being: THAT single molecule if changed will modify the disease. However, cells inside the body are complex with different organs expressing different set of genes and thus THAT single molecule is in different environments. The animal’s trials and others are fine for safety but the environment between the animal and humans are so different that the results are unpredictable.
Published results not reproducible in every case for targets: Nature had an article “Reliability of ‘new drug target’ claims called into question” by Brian Owens that highlighted the problems of published results. The drug companies like Bayer know that much of the data that is published is not very reproducible for discovering new targets. They spend a small amount of dollars validating the published results with a CRO before they develop the target for High Throughput screening (HTS) or High Content Screening (HCS). The effect is not due to outright fraud but rather the pressure to publish results quickly, that encourages incomplete, not fully validated studies to be published. That is possible but more likely; the studies are done in isolated cell lines that may not be reproducible in other cell lines. Therefore, when these “targets” are then developed further they do not function as expected.
Negative results not published: Negative results are not typically published in Scientific literature. There may be a good reason for that: since there are many hypothetical causes for every effect the literature would be inundated with negative results. For example, consider the much discussed global warming issue: It has been established that Carbon-dioxide as a green-house gas is the major cause, but if you were publishing negative results on every possible cause such as water-vapor concentrations, volcano frequency, earthquake location, use of computers, use of fertilizers, use of plastics, or other esoteric causes such as ship motion disrupting ocean currents then you will be lost in the literature. However, not publishing the negative results means that the researchers do not necessarily know which experiments do not work. Let’s take a trivial example away from Pharma. Frisbee’s come in various shapes and sizes and they all function as advertised. However, one company called Quirky (http://www.quirky.com/) decided that it could make Frisbee’s stay up forever by putting a small fan in its center. Much product development and prototype’s followed. This led to the conclusion that the reason such a product does not exist in the market today is because the concept does not work. Similarity between publications and marketplace: the things and ideas that do not work, do not appear in the marketplace or in publications. Unfortunately, that means that much effort is wasted in trying to work on ideas and concepts that do not work.