Digging into STAMPEDE data fir SBRT

[Javelin18]

The topic of whether to irradiate the prostate when there are distant metastases cones up regularly on the forum. Here’s an interview with Dr Koontz, Duke University, that describes the reasoning behind irradiating with metastatic disease.

hematologyandoncology.net/a...

I agree with Dr Koontz, except for her interpretation of STAMPEDE results, Take a look at the last two lines of table 3 in this paper.

thelancet.com/journals/lanc...

What the data show is the majority if patients in the trial didn’t have docetaxel treatment. The only systemic treatment was ADT. There is a strong benefit to radiation, as shown on the last line of the table, when patients were treated with docetaxel.

It seems reasonable to me that patients with higher metastatic burden would be more likely to have castration resistant disease, since they would have had the disease, and been under ADT treatment, for a longer time.

Based on this possible secondary correlation, I don’t think it is valid to conclude that radiation has no positive effect on outcome. I believe it likely that patients in the study didn’t benefit from radiation because they didn’t have effective systemic treatment.

Dr Koontz points out that distant metastases can create metastases. However I don’t think it is proper to conclude that the prostate shouldn’t be irradiated if disease has progressed beyond oligometastatic. To get rid of cancer, you have to get tid of ALL the cancer.

[Tall_Allen]

STAMPEDE selected de novo metastatic patients - no one in the trial was treated for a long time before entry.

[Javelin18 in reply to Tall_Allen]

Thanks for clarifying that. Do you see the large difference in the effect of SBRT for patients treated with or without docetaxel, at the bottom of table 3? What do you make of the difference?

[Tall_Allen in reply to Javelin18]

What difference? Perhaps you are misinterpreting that? All I see is that with or without prior docetaxel, debulking made no difference in overall survival (p interaction = 0.63).

[Javelin18 in reply to Tall_Allen]

Control without docetaxel i357 deaths in 844 patients. SBRT without docetaxel 342 deaths out of 847 patients. Control with docetaxel 34 deaths in 184 patients. SBRT with docetaxel 28 deaths in 183 patients.

[Javelin18]

This tells me that SBRT combined with docetaxel has a significant benefit. SBRT without docetaxel has no benefit.

[Tall_Allen in reply to Javelin18]

You have to understand statistics.

There was no difference in survival among men who had no doc, whether they had SBRT or not - 95% confidence interval of the hazard ratio was 0.80-1.08. If it crosses 1.0, we cannot reject the null hypothesis that there was no statistically significant effect discernable. In other words, there is strong evidence that the average survival was no difference.

There was also no difference in survival among men who had doc, whether they had SBRT or not - 95% confidence interval of the hazard ratio was 0.49-1.34. If it crosses 1.0, we cannot reject the null hypothesis that there was no statistically significant effect discernable. In other words, there is strong evidence that the average survival was no difference.

The interaction p value was 0.63, meaning we cannot reject the null hypothesis that there was no interaction effect between taking doc and SBRT. So, there is strong evidence of no interaction effect.

[Javelin18 in reply to Tall_Allen]

When I said I understand statistics, I meant that I’ve spent most of my career doing probabilistic risk analysis for rocket launches. That primarily entails Monte Carlo analysis of flight parameters to develop probability density functions. It also includes Baysian inference of probability of individual events. P value of the interaction of two input variables is new to me.

I found this article that discusses the use of probability for clinical diagnosis and treatment.

ncbi.nlm.nih.gov/pmc/articl...

Since the concept is new to me, and I have a bit of brain fog from treatment, could you explain your interpretation of the P value being .63

[Tall_Allen in reply to Javelin18]

To repeat: The interaction p value was 0.63, meaning we cannot reject the null hypothesis that there was no interaction effect between taking doc and SBRT. So, there is strong evidence of no interaction effect.

The null hypothesis is always that there is no effect. The p value helps us decide if we can reject the null hypothesis. By convention, we cannot reject the null hypothesis if p is greater than 0.05.

In this case, they tested the hypothesis that there is no interaction between taking docetaxel and using SBRT on survival. If P had been less than 0.05, it would have meant that we can reject that hypothesis, and we could have concluded that there was an interaction effect. But since P was greater than 0.05, we cannot reject the hypothesis, so we conclude that there probably was no interaction between taking chemo and SBRT on survival. Ergo, the effectiveness of SBRT did not depend on whether the patient had taken docetaxel.

I know it is difficult to wrap one's head around double negatives when one has chemo brain, but you can trust me on this.

Another way of thinking about this is with multivariate logistic regression. If overall survival (OS) at the last follow-up is modeled as:

OS= a + b x S + c x D + d x S x D + ...

where S is whether they had SBRT, and D is whether they had docetaxel

The high p values for the coefficients of S and SxD show that we cannot rule out that b and d are equal to zero. Only the coefficient of D (whether they had docetaxel) is reliably non-zero.

[Javelin18 in reply to Tall_Allen]

I understand the double negative, but I also believe this warning; from the article on probability in clinical diagnosis:

Limitations: Neither hypothesis testing nor statistical tests lead to proof. It merely indicates whether the hypothesis is supported or not supported by the available data. When we reject a null hypothesis, we do not mean it is not true but that it may be true. By default when we do not reject the null hypothesis, we should have this limitation in mind and should not convey the impression that this implies proof

Biology isn’t physics. There are unknown cofactors that are not modeled. I don’t think it is correct to make the absolute statement that SBRT with grater than 4 metastases is of no benefit. I think a more accurate way to say it , is studies haven’t shown a benefit.

[Tall_Allen in reply to Javelin18]

Proof is created by randomized clinical trials. We use statistics to determine whether the difference between the treatment and the control could have been a chance occurrence. But it is the power of randomization and control that allows us to go further than mere association to causation.

The proof is stronger than you think. It is not only true that studies haven't shown a benefit, It is true that studies have shown that there is no benefit across the population of men with de novo metastatic PCa. There were two major, well-done randomized clinical trials (STAMPEDE and HORRAD) that showed exactly the same thing. That's why it reaches the highest level of evidence (1a).

It is true, however, that RCTs predict population differences, and the individual's results may vary, or as you suggest, there may be a subgroup for whom it doesn't apply. One such subgroup is the luminal/basal distinction. I know that the STAMPEDE investigators are currently looking into that. Until we have that data, patients are ill advised to ignore the evidence.

[Javelin18]

I do understand statistics. However, I’ve just returned home from my Lu-PSMA treatment, so I’ll need some rest before responding

[EdBar]

Well this is purely anecdotal, I was dx with stage 4 Gleason 9 back in March of 2014. Outlook was not very good, 3-5 years since I had extensive Mets throughout my skeleton and in several nodes. A local RO, really brilliant guy, Top Doc award winner in ATL, recommended radiating my prostate, he said he has had patients who seem to get more mileage out of ADT by doing this, reducing the tumor burden. It also stops it from spreading into nearby organs which are difficult to treat. After a lot of research and after reading Snuffy Myers book it made sense to have my prostate debulked.I underwent IMRT to my prostate and several nodes in the fall of 2014. Later I became a patient of Snuffy and I recall him saying that it was one of the best things I could have done, and likely extended survival by several years. Well here I am nearly 8 years after dx, I had a 6 year run of my PCa being undetectable, ya undetectable, and only recently became resistant although PSA is still extremely low. I have undergone other treatments too of course, I believe in aggressive treatment but I feel fortunate to be where I’m at given where I started. I’d do it again in a heartbeat.

Ed

[Javelin18 in reply to EdBar]

Thanks for sharing your experience. I’d like to read that book. Do you have a title or link to where I can find it n

[Javelin18 in reply to EdBar]

I typically ask a question, then think I should have tried to find the answer before asking. I found this book on Amazon by searching Snuffy Myers.

amazon.com/Beating-Prostate...

Is that the book you’re referring to?

I found it funny that the next hit in the search was

amazon.com/GUND-Sesame-Stre...

I always liked the Snuffleupagus on Sesame Street.

[EdBar in reply to Javelin18]

That’s the one, it’s a bit dated due to when it was published. Read the section about research done at MD Anderson if memory serves me, about where the majority of mets are generated from even after becoming metastatic - no surprise it’s the prostate. That’s what convinced me to have it radiated, to prevent further mets from being generated by it. Snuffy once referred to it as the “mother ship” during a visit.