In the discovery arena, the answer could be yes, if you cast a wide net on this fishing expedition.
We’ve been discussing second chances for drugs that have failed to achieve efficacy in clinical trials using phenotypic screening. This sounds like an interesting idea, but the real question here is “what are the chances of success using this approach?” After all, we started with the premise that drug discovery and development is inefficient. If phenotypic screening isn’t likely to lead to success, might this just be an academic exercise?
I contend that the use of phenotypic screening has the potential to be successful, to increase efficiency by shortening development timelines, and to drive development strategy. The real pearl in this oyster, however, is the prospect of discovering new biology.
So how can phenotypic screening increase efficiency in drug development? As I mentioned in the first article in this series, a lot of time and money has gone into a compound by the time it reaches Phase II clinical trials. I also mentioned that less than 20% of the drugs that enter Phase II actually succeed. About half of those fail for lack of efficacy or strategic reasons. If a second indication were found for one of those failures, development time would be accelerated significantly. The pharmacokinetic (PK) characteristics for the compound have been optimized (efficacy failures are distinct from PK failures). Depending upon the initial indication and the new therapeutic area, the safety hurdles are significantly reduced.
The subsequent blogs in this series (found and ) have focused on failed clinical candidates and approved medications. But this approach can be extended to developmental candidates as well. Why would anyone want to do this? Pharmaceutical companies usually want to focus on a single therapeutic area for a candidate in order to avoid a complicated and confusing development process. On the other hand, the development process could be driven efficiently if efficacy in multiple therapeutic areas could be identified early in the discovery process, obviating potential cross-area development issues.
For example, let’s say that a compound has been found to have activity in relevant animal models in oncology and inflammation. Let’s also assume that the test article is a targeted agent and not a general cytotoxic molecule. If the development path for that agent were selected purely on market potential, it’s obvious that the oncology path would be more lucrative. However, oncology clinical trials are notoriously lengthy, and any adverse events associated with those trials are now part and parcel of the candidate drug’s legacy. A significant number of adverse events in oncology trials are more likely patient- or disease-related as opposed to compound-related. Nonetheless, those adverse events are now associated with the candidate agent. So working the development from the inflammation angle might be a more prudent approach in this case.
But is there any evidence that phenotypic screening will yield new indications? This question really cuts to the chase for phenotypic screening. If we employ off-label use of currently approved medications as an indication of the chances for success, the answer would be an overwhelming “Yes!” About 20% of all drugs are prescribed for indications other than the approved indication or indications. For cardiovascular and anti-convulsants, the estimate for alternative use is something like 80%! Pregabalin was developed as an anti-convulsant and is used for neuropathic pain. The anti-inflammatory steroid dexamethasone is used in premature labor to enhance pulmonary development of the fetus.
One of the most intriguing stories related to phenotypic screening comes out of McMaster University in Canada. Researchers there were screening libraries of known compounds for agents that would normalize human acute myeloid leukemia (AML) stem cells without altering normal human stem cells. They identified thioridazine, a dopamine antagonist, as being quite active in their in vitro screen. That activity also was observed in mouse studies. The data begged the question, “What does dopamine have to do with cancer?” As it turns out, some AML cells have dopamine receptors, and the clinical prognosis is inversely associated with the density of dopamine receptors on the cancerous cells. Further, thioridazine was only active against cell lines with dopamine receptors. Additional studies show that some breast cancer cells also express dopamine receptors. A new chapter in cancer cell biology appears to be opened because of a chance discovery based on phenotypic screening.
When I began my career in the pharmaceutical industry, I asked a senior member of my therapeutic area what I needed to do to succeed. He glibly told me, “Get lucky,” as if this were something I could just walk down the hall and start doing. Selecting relevant model systems for phenotypic analysis might just be the key to engineering that luck!