How Do We Know When What We Are Doing Isn't Working?

I quip sometimes when lecturing that a clinician will make their first decision to use a new therapy based on the data - but the second time they make the decision to use that therapy it will be based on their experience. Although obviously a gross oversimplification,  I think it does reflect the fact that we obtain biases as we treat patients. 

But also, when we scrutinize the success or failure of a therapy in an individual patient we are likely to do so based on three parameters:  Did the treatment induce a response? What was the toxicity? and how long did the clinical benefits of the treatment last?

It is the third of these three points that deserves more attention than it typically gets. All of our approved drugs in CRPC received their regulatory approval because of their ability to improve survival. That may drive the decision of when to start a therapy, but it doesn’t do so much to inform us as to when to stop a therapy. 

So, how do we know when what we are doing isn’t working? It may seem obvious, but let’s consider a few points:
  1. In most studies in mCRPC the depth of the PSA response ( e.g. 50% or 90% decline) is roughly proportional to the duration of PSA response, or time to PSA progression.
  2. In mCRPC, the PSA may rise for months before radiographic progression occurs.
  3. Radiographic progression can occur without symptoms
  4. Symptoms can worsen without radiographic progression.
  5. A therapy might continue to benefit a patient even while the PSA is rising.
This latter point is not really proven yet. In order to do so,  we would need to do take patients who are on a given therapy but experiencing a PSA rise and randomize them to either continuing the original therapy or switching to a control (e.g. a placebo) and see which is associated with a longer time to disease progression. We did a small study like this at UCSF where we used GMCSF in the ‘off phase’ of an intermittent chemotherapy protocol. it didn’t delay progression too much compared to just observing. 

Another point worth raising is how a rising PSA in the context of an ongoing therapy can mean different things based on the depth of response that the patient experiences. Imagine two patients with a PSA of 100. If one has a 95% decline on a treatment to 5ng/mL, then he doubles to 10, he is still at a 90% decline and it is likely that his disease remains under good control. A second patient whose PSA declines from 100 to 49 ng/mL still has ‘responded’ but when his PSA doubles at the time of resistance he is pretty much back to baseline – at 98 ng/mL. One would guess that this latter patient is more like to have experienced radiographic progression at this point. 

So what are the data that might help us link a PSA rise to a radiographic progression event? We can consider data from some completed clinical trials. For example, I broke these data down from the COU-302 study and found the following:

Time to PSA progression – abiraterone arm                                      = 11.1 months

Time to PSA progression – placebo/prednisone arm                         = 5.6 months

Time to radiographic progression- abiraterone arm                          = 16.5 months

Time to radiographic progression – placebo/prednisone arm          = 8.2 months

If you do the math on this, consider the following. In a typical scenario the time from PSA progression (PSAP) to radiographic progression is about 5 months (16.5 for radiographic minus 11.5 for PSA). Is there any benefit to continuing the abiraterone during that time? we can’t be sure but we typically do continue it in practice. 

Looking closely at the above you see that the time from PSAP to Radiographic progression is 5.4 months in the abiraterone arm and only 2.6 in the prednisone arm. Although not a very long time in either arm, it is in fact double in the abiraterone arm than in the prednisone alone arm. And again keep in mind that we are looking at the median of an entire population of over a thousand patients and this incorporates the 15% or so who have no benefit in terms of PSA ( and the outliers on the other end who have extraordinarily long responses). I have had patients who have steadily rising PSA’s on abiraterone and enzalutamide for as long as a year before developing symptoms or radiographic progression.

Finally consider a theoretical clinical trial and see if it conforms with your philosophy of treating patients.  In this trial we are testing a new drug that is relatively well tolerated. In Arm 1 the clinician changes therapies upon the first rise in PSA. In Arm 2 the clinician continues the therapy on the original drug even after PSA progression occurs until such time as radiographic or clinical progression occurs. 

What is likely to happen? 

I would imagine that the patients in Arm A would cycle through all the available drugs relatively quickly whereas the typical patient in Arm B may not get all the available CRPC drugs. He probably would experience fewer side effects ( fewer drugs equals fewer potential side effects), but we can’t really predict which arm will live longer. I would bet that patients in Arm 2 might have less anxiety due to less switching treatments all the time. 

My approach is to not only look at PSA and scans but to look at the other factors that are associated with prognosis – and many have been published. First, I set the stage that a change in therapy is likely coming. Two, I reassure them that it is standard practice to ‘treat through’ a PSA rise for a while and three, I look at Alk Phos, LDH, Hgb and of course symptoms, to determine if what I am doing is not working. 

It’s not an exact science. It’s an art, as we all know. But when the vectors of all of the above start moving in the same direction, it’s time to start thinking about what to do next.

Written by: Charles Ryan, MD, B.J. Kennedy Chair in Clinical Medical Oncology, Director and Professor of Medicine, Division of Hematology, Oncology and Transplantation, University of Minnesota, Minneapolis, Minnesota

Published Date: August 13th, 2018