How to design a trial to test the efficacy of Lutetium-PSMA products in metastatic castration-resistant prostate cancer?

This is the response to a comment from a drugdevletter.com follower.

Following a recent post on the SPLASH trial, which outlined various issues with its design, I received the following question.

My answer is simple

First, a disclosure: much of the following is inspired by working on research projects with Vinay Prasad and his team. You can gain a thorough and solid understanding of these concepts from reading his book Malignant: How Bad Policy and Bad Evidence Harm People with Cancer (available here).

• test the product in the most advanced settings first. In these cases, patients are at a higher risk of experiencing the event, enabling smaller sample sizes and faster results.

• overall survival as the primary endpoint: once again, in the most advanced settings, there is no justification for choosing an endpoint other than survival, as it saves little to no time (see this publication).

• no cross-over: until the product has demonstrated efficacy, and since overall survival is your primary endpoint, cross-over should not occur. It could hinder the interpretation of trial results (see what happened with sotorasib in CodeBreaK 200: here).

• a free investigator-choice control arm: the choice should be entirely unrestricted (see our paper here). In other words, investigators should have the freedom to treat the patient with the best treatment that would have been used if the patient were not part of the trial.

• enroll patients only in countries where the product will later be marketed, accessible, and sold (i.e., only in countries where care reflects what would be considered the best standard of care).

If this first trial is positive, what would be the next trials?

• if a trial has already demonstrated efficacy and you are testing the product in an earlier line, free access to the product upon progression should be guaranteed. The sponsor should provide access to this treatment if the investigator decides it is the best option.

• once again, the control arm should remain unrestricted, or reflect the best available care.

Concluding words

Running a trial is not an easy task. However, the core question is whether the trial will inform clinical practice: if the answer is clearly negative from the start, the trial should simply not be run. This proposal is based on fundamental principles of trial design. Trials should not be designed to simply bring drugs to market; they should aim to deliver better treatments for patients.