All Articles

Drug Development Should Be Getting More Expensive, Not Less

The most expensive part of drug development by far (in both money and time cost) is clinical trials. As such, it often doesn’t make sense to me why when confronted with the problem of falling R&D productivity, the instinct is to speed up, cut costs, or even forego preclinical development. The way to cut clinical trial costs is to improve the effect size of a medicine. You can enroll less patients and achieve a significant result more quickly. The way to get around falling R&D productivity is not to run more poorly run, large clinical trials. We should be increasing the standard for initiating clinical trials, increasing efficiency by decreasing the denominator of dubious trials.

If the speed of clinical trials is the real issue with drug development, give me a drug currently in clinical trials that is not getting to patients fast enough? IND requirements are there for a reason and no one thinks we should be giving patients more garbage in clinical trials. Even if you can eliminate pen & paper inefficiencies, cutting out inefficient administrators, how much can you really hope to decrease costs? Definitely not a 10x. Clinical trial services is crowded and has an incredible track record of burning cash without providing lasting value. Even if you have cut costs, there are a limited number of accessible patients to enroll in trials so scale is difficult to acheive without reaching moral bankrupcy.

The goal should be improving the predictive validity of preclinical assays and experiments. Investors need to stop falling for contrived assays that aren’t rigorous and poorly represent the underlying biology. For the most part, there also lies responsibility on the founders as well to be thorough and not lie during fundraising, being fully transparent about missing preclinical evidence.

Poor in vitro and in vivo models are not an excuse to jump straight to human. When we evaluate such models, we need to understand what metric they perform poorly on. Do they have low specificity or low sensitivity? More often than not, the deficiencies with model systems are that they lack specificity, not sensitivity. If a drug target is not shrinking tumors in a mouse, why do we think it will shrink tumors in human? The reality is that preclinical models do in fact have sizable utility.

Recently, a team published a case study describing the development of IGFR1 inhibitors for cancer therapy. Of course there are no approved IGFR1 inhibitors, but drug developers have spent upwards of $1.6 billion studying these drugs in clinical trials (16 different IGFR1 inhibitors across 183 clinical trials). Surely, this is not an issue of clinical trials not being fast or cheap enough. As the authors write, this is an example of poor preclinical dissection of biology, and rushed clinical development. The preclinical data never looked that good anyways.

As noted, mouse models do have adequate sensitivity. If you can cure Tyler Jack’s KPC mouse model, you likely have a drug for PDAC. Investigators described the IGFR1 preclinical package as compelling, without noting long term cures in tumor bearing mice. Investigators need to be more honest with themselves and investors need to have a role in shutting these companies down when the data doesn’t look good.

The added benefit of doing another preclinical study to refocus or calibrate expectations more than makes up for the potential losses from a failed clinical trial. Experiments are orders of magnitude cheaper in animals, not to mention the ethical issues with giving patients garbage.

Is the FDA requiring more data in drug approvals? Is the standard for initiating a clinical trial increasing or decreasing? When we set precedent for future centuries, we should be conservative with what we tolerate. Not only will it save money in the short term, but we are still in the early innings for drug development and future generations will benefit from high quality, interpretable studies.

For companies like Creyon or Battery Bio who claim to want to do drug development without preclinical studies, we should be dismissing these outright. As we have learned from EQRx, the FDA is not to be trifled with and your dreams of fast follower revenues deserve the bulk of public derision of the biopharma industry.

Published Aug 24, 2023

Harvard-MIT PhD Student