Abiraterone acetate in chemotherapy-naive mCRPC — Phase III trial outcome

This morning, at the ASCO annual meeting, Dr. Charles Ryan will present the full interim results of COU-AA-302, the randomized, Phase III study of abiraterone acetate in chemotherapy-naive patients with metastatic, castration-resistant prostate cancer (mCRPC).

The abstract of this presentation was placed on line just after midnight Eastern time last night and shows that:

  • All patients enrolled had to have either asymptomatic or “mildly” symptomatic mCRPC and could not have received treatment with any form of chemotherapy.
  • The trial enrolled 1,088 patients at 151 centers in 123 different countries.
  • Half the patients were randomized to treatment with abiraterone acetate (1,000 mg once daily) + prednisone ( 5 mg twice a day); the other half to a placebo + prednisone (5 mg twice a day).

The detailed results of the trial, based primarily on time to radiographic progression (i.e., an increase in evidence of metastatic tumor load to bone and other organs) are as follows:

  • Average (median) follow-up was 22.2 months.
  • Median time to radiographic progression-free survival (rPFS) was
    • Not reached in the abiraterone acetate + prednisone arm of the trial
    • 8.3 months in the placebo + prednisone arm of the trial
  • Median overall survival was
    • Not reached in the abiraterone acetate + prednisone arm of the trial
    • 27. 2 months in the placebo + prednisone arm of the trial , which is the longest survival period reported to date in any randomized, Phase III, clinical trial of men with mCRPC.
  • Median time to onset of pain requiring opiate therapy was
    • Not reached in the abiraterone + prednisone arm of the trial
    • 23.7 months in the placebo + prednisone arm of the trial
  • Median time to initiation of chemotherapy was
    • 25.2 months in the abiraterone + prednisone arm of the trial
    • 16.8 months in the placebo + prednisone arm of the trial
  • Median time to PSA progression was
    • 11.1 months in the abiraterone + prednisone arm of the trial
    • 5.6 months in the placebo + prednisone arm of the trial
  • Grade 3/4 adverse effects reported in the abiraterone acetate + prednisone vs. the placebo + prednisone arms of the trial are:
    • Hypertension, 3.9 vs 3.0 percent
    • Hypokalemia, 2.4 vs 1.9 percent
    • Elevated levels of alanine aminotransferase (ALT), 5.4 vs 0.7 percent
    • Elevated levels of aspartate aminotransferase (AST), 3.0 vs 0.9 percent

As previously reported, the Independent Data Monitoring Committee concluded that overall survival,  rPFS, and other secondary endpoints all favored the abiraterone acetate + prednisone arm of the study and unanimously recommended unblinding the study and crossing patients over from placebo + prednisone to abiraterone acetate + prednisone.

Also as previously reported, we do not know (and we will now never know) the size of the true, overall survival benefit resulting from treatment with abiraterone acetate + prednisone compared to a placebo + prednisone in men with metastatic, chemotherapy-naive, castration-resistant prostate cancer.

Some additional information about the results of this study are available in a media release issued by a division of Johnson & Johnson, and some of the information above has been modified since being originally posted to reflect information in that media release.

According to the media release, patients treated with abiraterone acetate + prednisone obtained “an estimated 33 percent improvement in survival” compared to those treated with a placebo + prednisone. However, it is not clear to The “New” Prostate Cancer InfoLink exactly how the available data can be used to justify that claim, given the limited amount of data available on overall survival at this point in time. At the time of the recommended unblinding of the trial, only 150 progression events (deaths or radiographic progressions) had occurred in the abiraterone + prednisone arm of the study compared to 251 progression events in the placebo + prednisone arm.

The media release states, in addition, that Johnson & Jonson “plans to submit marketing applications with regulatory authorities to extend the use of [abiraterone acetate] in men with mCRPC who have not received chemotherapy, beginning in the second half of 2012.” This would seem to imply that abiraterone actetate is unlikely to be formally approved for use in the treatment of men with chemotherapy-naive mCRPC before the first few months of 2013.

12 Responses

  1. Reblogged this on GREG'S LEGACY.

  2. Thanks for the summary, Mike.

    Please can you clarify something for me. You state the median time to rPFS in the abiraterone arm was never reached, yet later indicate there were 150 progression events in the abiraterone + prednisone. (Here I suspect you have a typo sine you report abiraterone + placebo). Unless all 150 were deaths, how can this be?

    Similarly, with median overall survival, you indicate it was not reached, but surely there would have to be some number … or is this related to the fact that the study was unblinded. I am confused.

    Thanks very much for all your hard work at ASCO.


  3. Rick:

    You were correct about the placebo/prednisone typo. Thank you. That has been corrected.

    The fact that patients in Arm A of a trial haven’t reached median survival or time to progression or whatever when those in Arm B have does not imply that one can’t demonstrate a survival benefit or a time to progression benefit. However, it does imply that you will not be able to project the size of that benefit with the same degree of accuracy.

    I have to say that, from a statistical point of view, trying to understand exactly what these results are showing us was extremely complex, and I am no statistician. Dr. Halabi spent the best part of 30 minutes going through the statistical justification for the decision to unblind the trial and report the results, based on this interim analysis, but it was certainly a relatively “close call,” and I am not sure that most of the audience understood the decision any better than I did.

    I cannot explain your question about the apparent divergence between the median survival and the median time to progression data and the actual number of “progression events” reported in the media release. It is counter-intuitive. All that I can really tell you is that it was explicable to a highly sophisticated statistician. I do have concerns, however, about whether the FDA will see these data through the same lens as the trial’s Independent Monitoring Committee. Someone is going to look pretty dumb if the FDA decides that it cannot approve abiraterone for the treatment of chemotherapy-naive mCRPC based on these data because there is a difference in statistical interpretation. In that case the manufacturer would likely be faced with having to start this trial all over again, whereas if they had waited until the next scheduled interim analysis they would (almost) certainly have had utterly compelling data that didn’t require statistical torture to demonstrate effectiveness.

  4. Thanks for taking the time to respond in detail, Mike.

    As presented the results clearly need a little more clarification that I will seek next time I see Ryan — or more likely Cooperberg, who is a statistical whiz. It does seem strange that the impact of abiraterone + prednisone cannot be measured in a couple of key categories like survival and rPFS … I suspect this must be related to either the early analysis and/or the unblinding.

    The bottom line to take from all of this is that there is a very positive impact. Hopefully, as you say, the FDA will concur.

  5. Hello, do you know when and in which journal will be the original article of this study printed?
    Thank you very much

  6. Dear Mireaia:

    Regarding journal publication of these data … Sorry … Not a clue. It could be one of several journals, and publication could be as much as a year from now.

  7. Thank you very much for your answer.

  8. Please help me understand. … If this drug can be of positive use to a person whose system has deteriorated to the point of needing chemotherapy, then how could it possibly NOT be of an even MORE positive use to a person whose system has not been severely compromised by chemo?

  9. Dear Bruce:

    There are all sorts of possible reasons why drugs that work very late in a disease process may not work as well earlier in that disease process. There are whole books on this subject.

    However, in the specific case of abiraterone acetate, the issue may not be the abiraterone acetate at all. It may have to do with the fact that (at least at the moment) research suggests that this drug has to be given in combination with a corticosteroid like prednisone. There are two issues related to the use of prednisone.

    In the first place, long-term use of corticosteroids is a serious problem. Since men with even chemotherapy-naive mCRPC may be on therapy for an extended period of time, the side effects of corticosteroid therapy may become severe. In the second place, we know that corticosteroid therapy alone actually has some level of clinical impact in men with chemotherapy-naive mCRPC, so it may well be that there are limits to how much more effective abiraterone acetate + predisone is compared to a placebo + prednisone.

    However, you are correct in your assumption that, almost certainly, abiraterone + prednisone is highly effective in the treatment of men with mCRPC prior to chemotherapy. That is not the issue under discussion above. The issue is whether the trial might (that’s “might” not “has,” and it is only a small “might” at that) have been stopped too early for there to be reasonable certainty of this clinical effect based on the data now available. Regulatory agencies are required to make decisions about the approval or non-approval of drugs for specific uses based on actual data, and not just on what we think is the case. (What people think ought to happen has been proved to be a bad predictor of what actually does happen more times than I would care to start counting!)

  10. “Median overall survival not reached”

    Hi Rick and Sitemaster,

    I’m responding to your exchange in the early morning of June 4 with what I believe is an explanation. I’m looking at this with the enormous benefit of not having my mind loaded with a numbing amount of hugely important information from the previous day’s conference, and I hope I haven’t missed the point entirely.

    Rick: “Similarly, with median overall survival, you indicate it was not reached, but surely there would have to be some number … or is this related to the fact that the study was unblinded. I am confused.”

    Sitemaster: “I cannot explain your question about the apparent divergence between the median survival and the median time to progression data and the actual number of “progression events” reported in the media release. It is counter-intuitive.”


    This looks straightforward to me, but perhaps I am overlooking something. Here’s the way it looks to me:

    I’ll assume the 1,088 patients in the trial were evenly split between the intervention and control groups, though the same thinking would apply if the split were not even (say, for example, two to one). Such a split would give us 544 patient in the intervention group. If half, or 272 + 1 = 273 were still alive when the analysis cut-off point was reached, then the median for overall survival (half alive, half not) would not yet have been reached for the intervention group. At some point it would be reached, as enough patients in that group passed on to boost the total to 272.

    The press release reports 150 progression events in the intervention group. (We can confidently assume radiographic versus PSA since PSA median progression was reached but radiographic progression was not.) As 150 is far fewer than 272 — the median point, the median would not have been reached in the intervention group. On the other hand, there had been 251 progression events in the control group per the news release. Allowing for some patient analytical exclusions that normally occur in both groups, it is reasonable that the real median was somewhat below my conjectural value of 272, which is noted above, but at or lower than 251.

    As we would naturally follow, if the radiographic recurrence median had not been reached, neither would the overall survival median have been reached for the intervention group as you have to be alive to experience radiographic progression. However, as PSA progression typically occurs in trials well before clinically documented progression, it is understandable that the median for PSA progression in the intervention group could have been reached, as it was, despite the fact that the medians for the other end points for the intervention group were not reached.

    I hope this helps.

  11. Dear Jim:

    I am sorry … You are way off base. what you are addressing is not the issue. The only serious issue is whether or not the method of analysis used to evaluate the data really does clearly demonstrate that the study data met the study’s co-primary endpoint for progression-free survival based on radiographic progression. No one is suggesting or implying that this study reached or was anywhere near demonstrating overall survival (the other co-primary endpoint). It didn’t, and now it never will.

  12. I was just trying to address the technical point about not reaching the median for a couple of the endpoints. I certainly see your point that there is no useful evidence about ultimate overall survival.

    Thanks for your dedication in bringing these key conference findings to us.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: