Prof Nicholas James - (Institute of Cancer and Genomic Sciences, Queen Elizabeth Hospital, Birmingham, UK)
NJ: Obviously great to be presenting these things back to back and the results are strikingly similar. As you said, they’ve got some interesting differences as well. So these are the survival results from the abiraterone randomisation in STAMPEDE. STAMPEDE has a number of different randomisations within it, so highlighted in yellow here are the patients we’re looking at today. I won’t go over the rationale because Karim has just done it. STAMPEDE includes a broader range, so newly diagnosed patients were 95% of the total into the trial and they had to be either metastatic, pelvic node positive or high risk locally advanced. All of those patients, the last ones, the T3/4s, got radical radiotherapy as well and about 60% of the node positive ones got radical radiotherapy as part of their base care. We did permit relapsing patients who had had prior surgery or radiotherapy but without significant amounts of hormone therapy but actually they only made up 5% of the total. Pretty standard inclusion criteria in terms of performance status, consent and so on. Stampedetrial.org is the trial website; all the protocols and everything are on there if people want to look them up.
We had a single primary outcome measure of overall survival and we also showed data on three of our secondary outcomes: failure free survival, toxicity and skeletal related events. A simple one-to-one randomisation, as with LATITUDE. Our target was a 25% survival improvement and we had three interim analyses which we haven’t published but which met their pre-set criteria. So we had 90% power and 0.025 1-sided alpha to do this analysis.
We recruited just under 2,000 men extremely quickly this time, [??] is very efficient at harvesting patients, if you like, into trials, and randomised as indicated equally. Patient characteristics obviously differ from LATITUDE so PS2s were permitted but were only 1% of the total, 21% performance status 1, so just under 80% were PS0. Median age in the late 60s with a very big range, 39 up to 85. Just over half were metastatic and just under 90% of those had bony metastases so a slightly lower bony metastases proportion than LATITUDE. Then we have the other two populations, the node positive 20% and the node negative locally advanced 28% which obviously aren’t in LATITUDE at all. So there are some differences here. As I already mentioned, the ones to whom you could give radical radiotherapy in the main did have radical radiotherapy.
Our primary outcome was overall survival and we had 262 control arm deaths at the point of the data freeze versus 184 on abiraterone. That gives you a hazard ratio of 0.63 which is astonishingly similar to the LATITUDE one at 0.62. It means the data are really very convincing. So that’s a 37% improvement in survival. The curves you’re seeing here are overall survival for the entire population and then M0 and M1. For the M1 patients I didn’t show the curves but the median survival is about 40 months, so a little bit longer than LATITUDE, reflecting the fact that we’ve got all of the metastatic patients in the M1s, not just the high volume ones. So there are some differences in the metastatic patients. We’re projecting, the dotted lines you can see on here are the mathematical best fit curves so we can project those forwards. We’re thinking the median survival of the metastatic patients is going to come out somewhere in the region of 6½ to 7 years and given the hazard ratio is the same for LATITUDE I would imagine that will be similar, slightly shorter, reflecting the worse prognosis.
When we set STAMPEDE up we were anticipating the median survival of these patients was 2 to 3 years, so that’s a staggering improvement in a decade and a bit driven by obviously all the new agents that we have but in particular shifting them upstream. So the Forest plot, everything lines up, as with LATITUDE, over the point estimate for the trial. There’s no convincing evidence of heterogeneity by the various stratification factors we looked at. We did look at metastatic status in more detail and I’m sure this is going to generate a lot of discussion and controversy as to whether this is a dichotomous population and the two groups behaved differently. There’s not very many deaths here, which is obviously great news, but it means we can’t fully power that sub-analysis. But if you look at the way that the confidence intervals and the point estimates line up they overlie each other. There are more deaths in the M0 patients from non-prostate cancer causes which obviously dilutes your benefit because they’re not impacted by abiraterone. But the test for interaction by metastatic status is convincingly negative, it’s not borderline negative. So our conclusion, as already stated, was there was no evidence of heterogeneity so we feel these data apply to the M0s as much as the M1s. It was tested in the whole population, the whole population is positive and the test for heterogeneity suggests the effect is uniform.
Further evidence for the effect being uniform comes from the failure free survival data so we’ve got a lot more failure free events. A huge difference, as seen in LATITUDE, with a hazard ratio of 0.29, we have slightly different criteria. 61 leading zeros in the p-value of 71% improvement. The Forest plot here, again everything lines up over the point estimate, no good evidence of heterogeneity and in this case, if anything, the hazard ratio is more favourable for the M0s although we think there is no heterogeneity. So just skipping on to the skeletal related events, as with LATITUDE these are obviously extremely important and what we saw, this is focussing on the metastatic ones on this curve, is a 0.45 hazard ratio, a 55% reduction in skeletal related events, again mirroring LATITUDE. So the two trials are very consistent in their impact across all these key outcomes.
We also looked at new treatments and we looked at it in a slightly different way in terms of how we presented it. So from relapse time to life prolonging treatment is very similar but the pattern of treatment is different. Actually this is a point of difference between the trials because we are open label clinicians knew if their patients had abiraterone or not so we see far more use of AR targeting therapies in the control arm than we do in the abiraterone arm which is point of difference between the trials driven by the different design.
Safety outcomes, I won’t spend too much time on these. We saw almost exactly the same – increased cardiovascular hypertension, cardiac arrhythmias and ALTs. In principle these are grade 3-5 toxicities, in practice they’re not very hard to manage. Compliance, again this is a slight difference, we saw 21% stopping, apparently for excessive toxicity. We haven’t yet matched that to the reported toxicity in those patients. For the patients stopping at two years 70%, just about, got to completion of treatment.
So our conclusions are very, very similar to LATITUDE – abiraterone acetate and prednisolone improves survival by 37% across the population; failure free survival, on a slightly different definition, by 71%; SREs or symptomatic skeletal events, what we collected, go down by 55% which we think is clinically, and actually from the economic point of view, extremely important. It was well tolerated and the toxicity mirrors the CRPC setting. So, as with LATITUDE, we think this should be part of the standard of care for men starting on to androgen deprivation therapy, including the M0s. Now I appreciate that might be controversial but that was our conclusion.
