Read the Full Video Transcript

Alicia Morgans: Hi, I'm so excited to be here today with Dr. Mack Roach, who is a professor of radiation oncology and urology at UC San Francisco. Thank you so much for being here with me today, Mack.

Mack Roach: I'm honored. As I was saying before we came on, when I sent you a copy of my paper, I didn't expect to get a response, but let me tell you about how I got into this.

So there was a gentleman, a doctor, Dr. Pilepich, who designed RTOG 75-06, 86-10, 85-31, these are all classic studies with hormones and radiation. And years ago, he got access to data from the RTOG looking at serum testosterone. And then he sort of retired, and he said, "Mack, how would you like to take this project over?" So I was like, "Okay, fine. Sure, I'll take it over." So I took it. So we had pre-treatment testosterone on 2,400 patients, treated all on randomized trials, and I took them and I sliced and diced them in different ways. And pre-treatment testosterone had no impact, which is sort of an interesting thing in itself because there's this myth, or there's a legacy or belief among many people, that a low serum testosterone is associated with a worse outcome. And it turns out most of that is surgical data.

But anyway, so we didn't find that. But what I did notice was that age, over 70, was associated with a better prostate cancer-specific outcome in terms of biochemical failure, in terms of metastasis-free survival. Now obviously, age over 70 is associated with worse overall survival, so older people die faster than younger people. And so, I was intrigued by this, and we subsequently published a paper formally addressing the issue of age from RTOG data. And then I also happened to occasionally stumble across a paper from the radical prostatectomy literature that suggested that age was an adverse prognostic factor. If you take two patients and you match them PSA for PSA, Gleason score for Gleason score, and you do a radical prostatectomy on them, the older patient has more advanced disease. And I was like, "Hmm, how does that work?" I mean, if age is favorable in radiation patients and is unfavorable in selected series, is this just a happenstance for this study and this one study, or is this a common pattern of some sort for some reason?

So I had a fellow from Egypt who wanted to come work with me, and he said, "Mack, Dr. Roach, I need a project." I said, "This is what I want you to do. Here's this software we use called EndNote. I want you to do a literature search. I want you to find all the papers you can find on radical prostatectomy, and all the papers you can find on radiation that look at outcome by age." And the age I picked for several different reasons. One is that the median age on all radiation papers in the RTOG is 70. It also turns out, I just turned 70. So I'm like, "Yeah, 70 is an important number that we need to understand." And then most of the surgical series were analyzed by age 65 and older.

So I said, "Okay, what we're going to do is, we're going to take age 70, plus or minus five years, so that'll get you 65 to 75, and look at whether or not it has an impact on outcome." And it turns out that as a general pattern, it appeared that age over 70 was a favorable prognostic factor on many radiation studies, and not an unfavorable prognostic factor on any radiation studies. And I said, "Okay, that's pretty consistent." And then it turned out that age was an adverse prognostic factor. Now the endpoint for the surgical series is a bit different because we have pathologic stage, we have other things that we don't have in radiation patients, but about it was sort of the opposite pattern. Okay, so let me show you figure two, which sort of summarizes the data. So it's sort of self-explanatory in summary.

So if you look at radiation studies, just over 50% of them show that age 70 plus or minus five years was associated with a better outcome. And there was another about 20% of the studies where it didn't reach statistical significance, but there was a trend in that direction. This could sometimes be explained by sample size and so forth. And in about 25%, there was no difference based on age.

With the surgical literature, about 50% of the studies showed that age was an adverse feature, either in pathology or biochemical failure, those sorts of cancer-specific outcomes. And then another 15 or 20% trended to be worse, but didn't reach statistical significance. And there were about a third of the studies in which the surgically treated patients' age did not have an effect. But there were no surgical papers that found that age was a favorable prognostic factor, and there were no radiation papers that found that age was an adverse prognostic factor.

So we tried to explain, or at least to discuss in the paper, some of their limitations, and also how to explain it. And we don't know exactly how to explain it. We do know that there are other cancer sites, for example, with glioblastoma multiforme, where age over 65 is an adverse feature in general. We do know that with breast cancer, premenopausal breast cancer is worse than postmenopausal breast cancer. So the notion that age could be a favorable or an unfavorable feature has been seen in other sites. What's somewhat unique about this is the fact that it could be modality specific. That it could be that age is important as a favorable prognostic factor with radiation, but unfavorable with surgery.

Now, we have other examples of differences between radiation and surgery in terms of prognostic factors, because for example, we have many randomized trials that show that when you add hormone therapy to radiation, the outcomes are better. But when you add hormone therapy to surgery, there are at least seven Phase III randomized trials with either radical prostatectomy versus radical prostatectomy and hormone therapy, where there's no effective hormone therapy in the surgical space, but it's favorable in the radiation space. So maybe it has something to do with hormone therapy use. Now, seven of the eight studies that used hormone therapy did show a favorable outcome. There was one study that did not use hormone therapy that did show a favorable outcome, even without hormone therapy. So I'm going to pause there, because I've said a lot, and see if you have any questions about our paper.

Alicia Morgans: Well, thank you so much. That's so interesting. And I wonder, these are studies that were pulled from the literature obviously, and one of the things that's important is none of these studies, at least as far as I know, were randomizing patients to radiation or surgery. Really, these were studies of radiation or studies of surgery. Do you think that it could be that the patients who end up getting radiation when they're younger may actually have more aggressive disease? Because in many cases, I think in clinical practice we have a younger patient, that patient often favors surgery because they do get to avoid hormonal therapy. And when a younger patient is getting radiation, in my clinic at least, it's often a patient who has maybe locally advanced disease, other poor prognostic signs that make the surgeon perhaps say, "Well, maybe it wouldn't be best if I operate on that patient." There's a high risk that that patient will need postoperative radiation anyway, and so maybe radiation would be the way to go. Do you think those factors might be?

Mack Roach: That's a good question, but here's the deal. So some of the radiation studies, for example, the study that had 2,400 patients, were randomized trials for radiation. So it showed that age. So they weren't, like some patients had more aggressive disease that were under the age of 70, because they were randomized, right? So they had similar amounts of disease.

And the surgical series that have been published that look at age actually find that young age is a favorable prognostic factor in some of the surgical series. Now, the age groups that they typically have are not usually in radiation studies. So some studies say age under 55 years is actually favorable in the prostatectomy literature. We don't treat as many patients in that age category. That's one reason why we focused on the median age, which is closer to 70.

Now, I think the thing that makes it a little bit complicated to try to figure out is that, if age is associated with more advanced pathologic stage in surgically treated patients, then that would imply that patients that are getting radiation that are older also have more advanced pathologic stage, and therefore, should be expected to do worse, but they do better. So that's why I think it's more likely biology than selection.

And if you look at the surgical literature, I think that if you look at older patients who undergo radical prostatectomy, they tend to be the healthier patients that undergo radical prostatectomy. So if anything, I mean, your question's a great question. I don't know the answer to the question. I'm not sure exactly how to apply the information. But let me tell you one problem when you go to analyze data, this is one of the take-home messages.

So some of the studies included both radiation patients and surgical patients in the same study. We only included those studies if they analyzed them separately. So if they analyzed them separately, then we could look at whether age was important in radiation patients and important in surgical patients. If they combined them, but they didn't analyze age separately, you really couldn't tell what the effect was, because in theory, you might have them cancel out, and they didn't look at age quite the same way.

So part of the take-home message is that, when you see a study, for example, my colleague Matt Cooperberg, published a paper in Cancer in 2010, and he used the CAPRA system to look at outcome in patients treated with radiation and surgery. About 75% of the patients that got surgery were under the age of 67, and about 75% of the patients who got radiation were over the age of 67. So when you analyze outcome and you have differences in the age distribution, how do you adjust for it? So you really have to be careful about doing comparative studies if you don't account for the fact that age can have a different biology. So I'm not quite sure how you adjust for it. One way is to analyze them separately.

We have a paper that's being submitted looking at artificial intelligence and race in prostate cancer. So this AI model is based on the standard features of Gleason score, PSA, and then the scan, and then the digital information that we get from pathology slides. I don't think the AI algorithm separately would break it down by... So if you combine it with surgical data, it'd be a little tricky. So we'll wait and see what AI does with that kind of information. But there are other problems that come out if this is a real thing.

Alicia Morgans: Absolutely. Well, the AI may be finding things in those tissues, in those slides, that we can't see, molecular changes that may be different between younger and older patients. And so that will be really, really interesting. And you make some great points about the surgical patients. I think the radiation population, I imagine, may be getting treatment because of their age, not because of the risk of their disease, making that older age population maybe more balanced among the risk groups. And that's really what I meant.

But to your point about the surgical patients, the older patients in the surgery group should actually be healthier and have better outcomes in most cases. So I think there are so many interesting things that have come out of this work. So many implications. I'd love if you could just give us a bottom line. I mean, how does this affect what you do when you see a patient in clinic? Because we often sort of have these tendencies, along the lines of what you described, but now there's data to kind of pool it all together to give us something to reference when we're talking to patients.

Mack Roach: That's a great question. In fact, it has implications. I saw a patient today who was 80, and the patient was asking me if he should have a radical prostatectomy. And I said, "Well, you could." I said, "I don't usually recommend radical prostatectomy for men that are 80, but let me just reassure you that if you're older, and you want to have a good chance for an outcome, you don't need to think, 'Oh, surgery is going to be better for me than radiation would be.'" I just use the information to encourage them to be optimistic about his results.

When you tell an older patient, "Hey, guess what? Your older age might be a favorable factor.", then I think, if nothing else, it makes the patient feel better. Because usually, old age is always bad. It is never, "I got good news for you. You're over the age of 70, and therefore you might actually do better than the patients under the age of 70 when it comes to cancer outcomes." So I think encouraging patients to be a little bit more optimistic about their prognosis, potentially, in radiation patients is, I think, one of the more straightforward concepts, ideas, or conclusions that I read from this data.

Alicia Morgans: Well, I absolutely love that, and I think to myself, when I see older patients, "Should we all be so lucky to reach where you are someday?" And I think that this is, as you say, really reassuring for those patients, as they're trying to make these decisions. And no one should have to make them, but when they have to make them, at least there is some information for them to draw from to help guide them in what can seem like an impossible choice.

So I thank you so much for the work that you do, and for your time and your expertise tonight.

Mack Roach: Honored to be able to talk to you about this topic. Thank you.