“Apollothirteenize” the Trial that Fractures Broke

Late last November Bayer issued a press release notification of the premature halting of a study in which abiraterone and prednisone plus radium-223 was being compared to abiraterone and prednisone alone (ERA 223, NCT02043678). The announcement revealed that the study was being terminated early due to an imbalance in fractures and survival between the two arms – favoring the control arm

We cannot know the full story of this result until we have a chance to review the full data from this trial, preferably following submission to a peer-reviewed journal. Yet, while we wait for the final data to emerge, it may be useful to engage in a little reflection on how we can get it so wrong, how trials can be not only not positive, but positive for the control, and what this should mean about trial design in the future.

It may be even more fruitful, however, to consider this failure in the biology of the disease and how the two drugs may have negatively interacted.

I have no insider knowledge on this trial, and am writing solely as an interested observer. Yet I do have experience in some aspects of this work that may shed light on mechanisms, both biological and trial-conduct related.

So here are my hypotheses behind what might be going on.
  1. Patients on the combination arm were taken off bone-targeted therapy because it was felt that the radium ‘was’ a bone-targeted therapy.
  2. Patients on the combination arm suffered a greater degree of marrow failure.
  3. The addition of the radium on top of an abiraterone induced ‘flare’ paradoxically enhanced bone resorption thus affecting bone marrow and bone integrity, leading to worse outcome. 
Let’s consider these points individually.

Point one. Let's call this the “Unintended consequence” argument, for lack of a better term. In this scenario clinicians, perhaps aware that in the Alsympca study there was a reduction in skeletal-related events, in addition to an overall survival benefit, decided that they could forego a bone-targeted therapy such as zoledronic acid or denosumab. They may have figured that the risk of osteonecrosis or other complications from these therapies superceded the possible benefit, because their patient was already on a bone-targeted therapy. 

Thus, it was well-intentioned clinicians who may have, unintentionally, created vulnerability in their patients to fractures and other complications. This argument would likely explain the altered rate of fractures but, unless the skeletal events led to fatal downstream events, wouldn’t explain the survival differences. 

But the trump card here is that this is a placebo-controlled trial. Thus, it is unlikely that such changes in the treatment were made intentionally. 

Point two: Radium is known to induce bone marrow failure in some individuals. Perhaps it is higher than expected? This would perhaps explain the imbalance in deaths, but wouldn’t explain the imbalance in fractures.

And what about point three. Several years ago we observed a high rate of bone flare in patients being treated on the abiraterone phase II trials.1 This was prior to the design of the COU-302 study that made abiraterone a standard of care in this setting, and in fact was a pivotal observation to the design of that trial and outcomes analysis in mCRPC now codified in the “Prostate Cancer Working Group” recommendations that are the guidebook for the design of prostate cancer clinical trials.

In that study flare was defined as the appearance of brightened lesions on a standard bone scan that subsequently remained stable in the context of a ≥50% decline in PSA. Although not proven, it was, and remains to be speculated that the radiographic flare was the result of greater uptake of technetium, the bone scan tracer, in bone tissue that is healing and calcifying in response to the retreat of previously invading tumor. Now, if a patient is experiencing a PSA decline and tumor retreat and, instead of technetium, radium is integrated into the bone, it could affect bone healing and remodeling, potentially leading to a fracture and potentially a greater propensity for bone marrow failure and poor patient outcome.

Further, perhaps the concept of synergy versus additive therapies needs to be reconsidered. It was recently demonstrated that each cancer drug in a combination has variable activity in an individual and it is simply the law of averages that the addition of a second therapy will similarly result in variable responses, and that their co-administration will, in some result in very good responses. The reader is referred to the recent excellent paper by Adam Palmer and Peter Sorger on this topic in Cell.2 They conclude that the efficacy of combinations, in most cases, arise from dual independent activity of drugs ( the patient simply responds to both drugs) without synergy in which an exponentially beneficial effect may be observed as each drug makes the other MORE effective than it would e expected to be as a monotherapy.

Indeed, if this argument could be made about the lack of synergy in the in context of response, it is possible that similarly we could simply be observing this effect with respect to toxicity. In other words, perhaps there isn’t some sort of synergistic toxicity of abiraterone plus radium, maybe this just happened by chance.

Finally, a comment not on this trial in particular but on negative phase III trials in general. They are hard to watch. It’s a challenge to see a good idea not come to fruition and to see the potential for forward movement of the field lost.

But let's keep in mind that this is cancer, and by its nature unpredictable. So when we see a negative trial –a failure- we need to apply what I call the “Apollo 13 “ rule, can we make it ‘successful’?

In the failed NASA Apollo 13 moon mission the defect was simple: A heating wire in the liquid oxygen tank lost its insulation and sparked a small explosion. The resulting explosion, and loss of oxygen, resulted in the mission being cut short. It was only through a variety of inventive fixes could the astronauts be returned safely to earth. That inventiveness, and the ‘root cause’ investigations behind the failure, actually improved NASA and Space travel in general. It was a successful failure.

The same can happen in medical trials. This is one place where NIH sponsored trials come into play. In recent years I have been part of a team looking at genetic and hormonal data from CALGB 90401, for example, a randomized phase III study of Docetaxel vs Docetaxel plus Bevacizumab in CRPC. The study did not meet its primary endpoint of an improvement in overall survival, but we have managed to learn about the hormonal and genetic milieu associated with response to Docetaxel and peripheral neuropathy,3 for example. There are numerous other examples. A key to the making success of a failure, however, is for the data to be made available for investigators and peer reviewers. Hopefully, we’ll continue to talk about this trial for a long time, not because of its failure, but because of what this failure has taught us.

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: January 5th, 2018

Written By: Charles J. Ryan, MD


1. Ryan, C.J., et al., Phase II study of abiraterone acetate in chemotherapy-naive metastatic castration-resistant prostate cancer displaying bone flare discordant with serologic response. Clin Cancer Res, 2011. 17(14): p. 4854-61.

2. Palmer, A.C. and P.K. Sorger, Combination Cancer Therapy Can Confer Benefit via Patient-to-Patient Variability without Drug Additivity or Synergy. Cell, 2017. 171(7): p. 1678-1691 e13.

3. Hertz, D.L., et al., Pharmacogenetic Discovery in CALGB (Alliance) 90401 and Mechanistic Validation of a VAC14 Polymorphism that Increases Risk of Docetaxel-Induced Neuropathy. Clin Cancer Res, 2016. 22(19): p. 4890-4900.