Changing definitions of clinically meaningful endpoints

We have gradually lowered our bar to define what constitute clinically meaningful endpoints for cancer patients. Originally, it was generally agreed that a new drug should improve overall survival (OS) and/or quality of life to justify its toxicities. When we decided we can’t wait for OS results, alternative intermediate outcomes arose that can be quickly measured and were assumed to predict overall survival. These are called surrogate measures. We accepted these as endpoints for clinical trials but with an important understanding that the surrogate should have been validated such that if a drug improves the surrogate measure, it also most likely improves survival. In recent years, we lowered our bar further to accept surrogate measure even if they are not validated but with a plan to test survival benefit in a later trial. Now, we are trying to convince ourselves and the patients that surrogates are clinically meaningful themselves. As a patient, you want to know if you are living longer or better as a trade-off of the drug toxicities but a committee of experts can now decide that those are not meaningful endpoints to you, you should care more about whether you developed metastasis or not based on special imaging technologies. And you should take this expensive drug because although it might take many years before you develop metastasis (or you may not develop metastasis at all before your natural death), this drug will help push the detection of metastases further away. You may protest that you need to know if you will live longer by taking this drug straightaway even before you have any evidence of disease. But because metastasis-free survival is now considered a clinically meaningful endpoint, you don’t need to know if you’ll live longer.

I am describing the story of this trial of enzalutamide in non-metastatic prostate cancer, in which patients up to 95 years of age were randomized to demonstrate a significant improvement in metastasis-free survival, a surrogate measure that the FDA has recognized as a clinically meaningful endpoint in itself. I encourage you to read this article in detail. In which the FDA makes it crystal clear that measuring overall survival is not important in this scenario and that if you did measure overall survival and found it negative, the outcome could still be considered positive if it does not show a detrimental effect in survival. As for me, I can’t understand how patients up to 95 years of age would benefit by prolonging metastases-free survival. There are more chances of facing a natural death before the prostate cancer has time to go metastatic and be lethal in this subgroup of patients.

In other news, palbociclib failed to improve overall survival versus placebo in PALOMA-3 trial. But the industry is still “encouraged by these reports”. Of course, because there was no detrimental effect on survival!

But what about drugs that do improve overall survival?

Enough with surrogates, here’s a drug called cabozantinib that actually improved survival in a disease as lethal as advanced hepatocellular cancer in a placebo-controlled phase 3 RCT as a second line therapy. Metastatic hepatocellular cancer has a median survival of around 6 months, so any improvement in overall survival is great, right? No. In this cabozantinib trial, even the control arm patients who were randomized to placebo lived for a median of 8 months, which means these are not average patients. These are patients with otherwise excellent health who do not reflect the majority of liver cancer patients we see in the clinic. Vinay Prasad has discussed the issues regarding the non-representativeness of this trial eloquently in one of his tweetorials.

Avoiding wisely

There have been a lot of discussions on Twitter about personalized medicine. Many people seem to think that personalized medicine is matching the name of a drug to the name of a mutation, but for me tailoring the care of cancer patients according to their goals and preferences is the true spirit of personalized medicine. There are many cancer patients who want to avoid surgery. For example, avoiding rectal surgery would also mean avoiding a stoma, which would be a huge impact on quality of life. In this study, which is a good example of how to apply observational data and the benefits of international collaboration, patients with rectal cancer who achieved a complete response after neoadjuvant chemotherapy and underwent a “wait and watch” strategy instead of immediate surgery were followed up for outcomes. The majority of patients didn’t have regrowth and they avoided surgery with this strategy. If there was a local recurrence, which usually happened within 2 years, that could also often be managed by salvage surgery. Thankfully, randomized trials are underway. Other practical concerns before this strategy can be implemented as routine care are discussed very well in the accompanying editorial.

Let me take some selfies

This time, I have two papers to share, both of which were done with my colleagues at Brigham and Women’s Hospital. In first work, we show that overall survival should be the endpoint for trials of PD-1 inhibitors not because benefit in progression-free survival might not translate to overall survival but because these drugs could improve overall survival without affecting progression-free survival. In any case, RECIST-based assessment of response or progression seemed less meaningful for these drugs. In another piece, we show that there is no association of affordability of modern cancer drugs with efficacy and safety data as assessed using the NCCN guidelines.

Dr. Gyawali is a research fellow at Program On Regulation, Therapeutics And Law (PORTAL) at Brigham and Women’s Hospital/Harvard Medical School. The opinions expressed herein are his own. You can read his previous blogs here.