Data from a small, Phase II, French trial known as the TAXOMET trial were also presented yesterday at the ASCO meeting.
The TAXOMET trial was a prospective, randomized, mulri-center trial in which non-diabetic men with metastatic castration-resistant prostate cancer (mCRPC) were randomized to treatment in either
- Arm A: with docetaxel (75 mg/m2 every 21 days + prednisone (5 mg twice a day) + metformin (850 mg twice a day) or
- Arm B: with docetaxel (75 mg/m2 every 21 days + prednisone (5 mg twice a day) + a placebo (twice a day)
The trial enrolled 99 patients between January 2013 and December 2015 at a total of 10 different centers in France.
Here are the basic results of the study, as presented yesterday by Martin et al. (see also this report on the MedPage Today web site):
- 50 patients were initially randomized to Arm A.
- 49 patients were initially randomized to Arm B.
- 95/99 patients were evaluable at the end of the trial.
- No significant differences were observed between the patients in Arm A as opposed to Arm B in terms of
- PSA response rates (72 percent of patients in each arm showed a PSA decrease of ≥50 percent)
- Overall response rate (ORR), which was 28% in both arms
- Clinical or biological progression-free survival (PFS)
- 7.3 months in Arm A
- 5.8 months in Arm B
- Overall survival (OS)
- 24.2 months in Arm A
- 19.7 months in Arm B
- There was just one difference between the two arms in terms of adverse events.
- Diarrhea was more common in patients treated with metformin
- 70 percent in patients in Arm A
- 50 percent in patients in Arm B
- There were few Grade 3/4 adverse events.
- Diarrhea was more common in patients treated with metformin
- There was no difference in qualirty of life scores between the two arms during the treatment period.
Martin et al. concluded that
This is the first prospective randomized controlled trial to evaluate the combination of [metformin] with [docetaxel + prednisone] in mCRPC. The addition of [metformin] has no meaningful clinical benefit in this setting.
Now some readers may feel uncomfortable with this conclusion. This was a small trial, and there was a clear numerical (albeit not statistically significant) benefit in terms of overall survival (4.5 months).
The key question that would seem to your sitemaster to be posed by these data is, “Do they really show no meaningful clinical benefit?” Might it have been more appropriate to conclude that this trial wasn’t large enough to show a statistical significant overall survival benefit, despite the presence of a numerical benefit?
On the other hand, it does has to be said that the small numerical improvement on PFS (just 1.5 months) doesn’t seem to correlate well with the numerical improvement in OS. There is a strong possibility that using metformin in the treatment of mCRPC is really not clinically beneficial, But it is also possible that thsi is simply because of the late stage of the prostate cancer in these patients, and that if metformin is to have therapeutic benefit it needs to be used much earlier in treatment.
Filed under: Living with Prostate Cancer, Management, Treatment |
THE "NEW" PROSTATE CANCER INFOLINK 
Thanks for your report and comments.
This amounts to an encouraging pilot study.
Hmmm …. Maybe …. I would rather take the view that it is “interesting” scientifically, but provides little reason for any degree of excitement.
What catches my interest is that 4.5 month advantage in overall survival. If this were a large, Phase III clinical trial aimed at approval for metformin for this class of patients and with statistically significant results, 4.5 months would most likely be enough to merit FDA approval. It is often hard to establish statistical significance with such a small trial. There is a huge gap in the “confidence interval” for both the A and B groups, which clearly indicates that the small size of the trial is having a large impact on the outcome for OS. The statistics suggest that the true impact of metformin could be as good as 33.7 months versus 14.8 months, or as bad as 17.2 months for metformin versus 36.8 months. The final statement in the announcement is not at all justified, in my view.
BIG Difference in Baseline PSA Between Groups A (Metformin, PSA 80.3) AND B (No Metformin, PSA 54.5)
I just learned this morning of another glaring problem in this small trial, which is now being touted almost as a guideline in some quarters, though thankfully not in Sitemaster’s report. (Thanks as always for your thoughtful reporting.) Here’s the source. And here’s the statement about PSA, which was not in the conference abstract:
“the median baseline prostate-specific antigen (PSA) level was 80.3 ng/mL in the metformin group, and 54.5 ng/mL in the placebo group.”
Am I going nuts, or does this suggest a rather large difference in risk in the two groups, with the metformin group likely having substantially higher risk at the outset based on PSA, and yet the metformin group still did a bit better judging by the statistics, which should be seen as shaky due to the large spans of the confidence intervals. Obviously the research team did not stratify their groups for PSA; I’m wondering what other keys they failed to use for stratification.
Isn’t anyone putting on their critical thinking caps? (My professors in experimental design and statistics would have had my hide if I had not observed errors like this!)
In the last paragraph you say, “On the other hand, it does has to be said that the small numerical improvement on PFS (just 1.5 months) doesn’t seem to correlate well with the numerical improvement in OS.” Actually, both PFS and OS appear to have the same percentage change (26% for PFS and 23% for OS), so the hazard ratios are probably pretty close as well.