Rashid Sayyid: Hello, everyone, and thank you for joining us today in this UroToday Journal Club recording. I'm Rashid Sayyid, Robotic Urologic Oncology Fellow at USC in Los Angeles. And I'm glad to be joined today by Zach Klaassen, associate professor and program director at Wellstar MCG Health in Augusta, Georgia.

Today, we'll be discussing the recently published paper by Dr. Derya Tilki's group in JAMA Oncology that looks at persistent PSA following radical prostatectomy for prostate cancer and mortality risk. And this paper was published this month in JAMA Oncology, with Dr. Tilki as the lead author and Dr. Anthony D'Amico as the senior author.

So what about PSA clearance following radical prostatectomy? Historically, and this has been known for a while, the half-life of PSA is about two to three days. And so we would expect within half-lives for the PSA to nadir down to its lowest level. But we do expect undetectable PSA levels about one to two months post-radical prostatectomy. So we give ourselves more of a buffer.

But we know that patients who have persistent PSA, so after this one to two months after surgery, we do know that these patients have worse survival outcomes. And this has been previously shown very nicely by Dr. Tilki's group in their publication in European Urology in 2019, where they looked at patients who had PSA persistence, was seen here in the blue dashed line, versus those who have undetectable PSA. And as you would expect, patients who have a persistent PSA after surgery have worse cancer-specific survival, as seen on the left, and also worse overall survival, as seen on the right.

But also, importantly, in addition to its prognostic implications, we do know that there are very important treatment implications that come with PSA persistence. So salvage radiotherapy is one of the common forms of salvage therapy in this setting. And as we see, patients who do receive salvage radiotherapy, as seen in orange, do better in this setting compared to patients who do not receive salvage radiotherapy. And this is for overall survival on the left, as well as cancer specific survival.

And so the take-home message from these two Kaplan-Meier curves is that salvage radiotherapy in patients with persistent PSA is associated with improved cancer-specific survival and overall survival. So this really emphasizes the importance of detecting persistent PSA when it is present.

And so the next question, naturally, is what time point should we use to define PSA persistence? Now the EAU tells us that a PSA level is expected to be undetectable two months after radical prostatectomy. And the AUA roughly gives us a similar time frame, with six to eight weeks post-operatively.

But the question is, should the PSA be checked at the same time point in all patients? Or should we individualize this based on the patient's oncologic characteristics? And one question that's come up in the literature is, do patients with a higher PSA level before surgery need more time to clear their serum PSA?

And the thought process here is that these patients, who have their PSA checked one to two months, and have a high presurgical PSA, let's say, 20 to 30, do these patients just need more time to clear their PSA? And so if we check it at one to two months, is that too early for time frame, and these patients haven't reached a nadir yet.

And so the thought here is that this is a false positive finding, which leads to overtreatment, potentially, with salvage radiotherapy and ADT, and obviously can have a negative impact on a patient's quality of life. And so should we just be more patient in these group of men, and avoid all these negative sequelae?

And so the study hypotheses of this paper were as follows. So first of all, among patients with a persistent PSA two months after surgery, those who have a presurgical PSA greater than 20 actually have superior prostate cancer specific mortality and all-cause mortality compared to patients who have a lower presurgical PSA of 20 or less.

And this is a little counterintuitive. But if we think about this as we're all starting off at the same time point, where all patients have a persistent PSA, and then we work back from that, the theory here is that those who have a higher presurgical PSA are actually still along the path towards the nadir. And so these are patients who may potentially have undetectable levels with longer follow-up. And so it's a bit of a reverse thought process. But this is where they are trying to assess.

And then the second thing, they wanted to assess whether increasing persistent PSA level was associated with worse cancer specific and all-cause mortality. Well, the thought process that higher levels, obviously, would be associated with worse survival outcomes, suggesting a higher degree of residual disease.

And so to accomplish this, the study investigators used two cohorts-- the discovery cohort, which included men who underwent surgery for clinical T1-3, node negative M0 prostate cancer in the university hospital in Hamburg between 1992 and 2020. And staging in these patients was using conventional imaging, and done for those with a PSA of greater than 20, or a higher Gleason score of 8 to 10.

And then they used a validation cohort, which is a similar profile of radical prostatectomy patients from Johns Hopkins between 1990 and 2017. And so the time 0, so the start of follow-up in these patients with the date of radical prostatectomy, and the follow-up, as we would expect in clinical practice every two to three months for the first year, and then every six months for the next four years, and then annually thereafter.

In terms of the study endpoints, the co-primary endpoints, we touched upon them, were prostate cancer specific mortality and all cause mortality. And as is typical with these studies, survival analysis using Fine and Gray modeling as well as Cox proportional hazards modeling was used. The Fine and Gray is very similar to Cox in terms of its concept. But it does account for competing risks, for example, cardiovascular disease.

We know radical prostatectomy or prostate cancer patients in general are older. And so there also have a high likelihood of dying from other diseases. And so in terms of the statistical modeling, you account for the fact that, instead of getting censored, they're dying of these other causes. And so it really is just fine tuning these survival models, as opposed to Cox proportional models.

And so what they did was they put in interaction terms, and they correlated those with the outcomes. And there was a purpose behind these interaction terms. So they included in their model a persistent PSA versus undetectable PSA post-surgery, as well as the PSA level using a cutoff of greater than 20 or less.

And the point of this is to really evaluate that among patients with persistent PSA, if having a higher presurgical PSA or a lower presurgical PSA was associated with the cancer specific survival and the all-cause mortality. So basically, you're just adding this interaction term with the PSA to help figure out if there's a significant correlation between those two.

They also look to assess whether there was any interaction between the PSA level being persistent or undetectable, as well as time dependent radiotherapy or ADT use post-surgery. And the goal here was to evaluate whether use of radiotherapy or ADT in a time dependent fashion was associated with the survival outcome. So basically trying to understand what is the impact of salvage therapy in this setting on the survival outcomes.

They also correlated persistent PSA levels with all-cause mortality using Fine and Gray, as well as Cox proportional models, similar concept. When you look at this linearly, the higher the level you expect, the worse the survival outcomes. And basically, when they looked at these multivariable models, they adjusted them for many variables. And this is really with the goal of controlling potential confounders.

So they included those that were available, such as the age at surgery, the year was done, the Gleason score, stage, marginal status, nodal status, use of the salvage therapy, as well as prostate volume in a subset. So all with the goal of canceling all the noise that may influence erroneously these results.

And at this point, I'll turn it over to Zach, go over the results, discussion, and put these results in context for us.

Zachary Klaassen: Thanks so much, Rashid. Great introduction as always. This is the consort diagram from this study. And so we see at the top of this figure, 30,461 patients were included. This was then pared down to 1418 with a PSA after radical prostatectomy that was persistent. And we can see here, this was delineated as 446 with a PSA presurgical of greater than 20, and 972 with a PSA of less than 20 presurgically.

When we look at this Table 1 comparison of clinical factors among these 30,000 patients, we're going to just set this table up as follows. The clinical factors are on the left. There's then three groups-- undetectable PSA, 29,043, there's persistent, and PSA below the median, the median being 0.37. 706 patients are above the median. 712 patients.

There's a couple important factors here to highlight. So when we look at the pre-radical prostatectomy PSA level, the median, as expected, was lowest and detectable at just under 8, just over 11 for below the median of those with persistent PSA, and 16.5 with persistent PSA above the median.

When we look at the prostatectomy Gleason Score, as no surprise here, the Gleason 8 to 10 from left to right, from undetectable, 6.3%, below the median persistent PSA, 22.2%, and above the median persistent PSA, 35.7% Similar trend here for more higher risk T-staging, T3b and T4, 12.4% for undetectable, 42.8% for persistent PSA below the median, and 62.9% for persistent PSA above the median.

Similar trend also for margin status. Positive margins 18.4% Undetectable PSA, 44.6% for persistent PSA below the median, and 57% for persistent PSA above the median. Finally, here we see lymph node disease. Again, a similar trend. Positive lymph nodes only 8.3% for those with undetectable PSA, 31.4% for those with a PSA persistence below the median, and 53.2% above the median.

When we look at the all-cause deaths, just looking at the incidence, we see, not surprisingly, again, as we go from undetectable to persistent PSA above the median, we see increasing percentage here. And this is also a similar trend for all-cause deaths from prostate cancer, up to 59.3% for a persistent PSA above the median.

This table looks at the characterization of persistent PSA and time to undetectable PSA by pre-RP PSA levels. And so again, we have characteristics on the left. We have all patients with persistent PSA on the far right. And in the middle, we have this delineation between pre-RP PSA less than or greater than 20.

So I've highlighted a couple important points here. So time from radical prostatectomy to first assessment for persistent PSA level was lower in those who had a higher pre-RP PSA, two months, versus 2.23 months for pre-RP PSA less than 20 in this category here.

So what this tells us is that, as the hypothesis suggested, patients that had a higher pre-RP PSA were getting their PSA checked sooner after radical prostatectomy than those with lower PSA. This also held true for what was a time frame between first and second PSA assessments-- 2.25 for those with higher presurgical PSA, 2.3 months for those with a lower presurgical PSA.

When we look at the time from radical prostatectomy to undetectable PSA, in patients with a persistent PSA observed for six months after radical prostatectomy, we see that this is actually lower for those patients that had a PSA before surgery of greater than 22.96 months, and 3.37 for those with a PSA of less than 20. Not surprisingly, here we see that, in terms of this adjuvant therapy for both radiotherapy and ADT, both of these were less time from radical prostatectomy for those with a higher pre-RP PSA than those with a lower presurgical PSA.

This table looks at the multivariable interaction regression model, adjusted hazard ratios for all-cause mortality and prostate cancer specific mortality. This is assessed over a median follow-up of just over 6 years. Again, clinical factors on the left, we have all-cause mortality in the middle of this table, and then prostate cancer specific mortality to the right. There's a lot to unpack from this table. So I'll highlight several of the key points here.

When we look at this interaction term for persistent versus undetectable post radical prostatectomy PSA, as well as pre-RP PSA greater than or less than 20, this was a significant interaction term for both all-cause mortality, as well as prostate cancer specific mortality. And what this means in terms of the take-home messages from this analysis is that in patients with a post-RP persistent PSA, those with a PSA greater than 20 versus less than 20, had a 31% reduction in all-cause mortality, and a 59% reduction in prostate cancer specific mortality.

When we take those patients that had a post-RP undetectable PSA, similar delineation, greater than or less than 20 pre-radical prostatectomy PSA, we see that for those with a higher PSA pre-surgically, there was an 18% increased risk of mortality compared to those with lower PSA. And this does not play out in the prostate cancer specific mortality model.

Finally, from this part of the analysis, there's another interaction term looking at persistent versus undetectable PSA and radiotherapy in a time-dependent fashion. What we see here is that in the prostate cancer specific model, those with a persistent post-op PSA, radiotherapy led to a 45% decreased risk of prostate cancer specific mortality.

The bottom part of this model is different. The top was categorical PSA, below is continuous variant. And what this tells us here is that in patients who have an increasing PSA as a continuous covariate, there's a 14% increased risk of all-cause mortality, and a 27% increased risk of prostate cancer specific mortality.

The next two figures will look at cumulative incidence curves looking at, first, all-cause mortality among all 30,461 patients. When we look at this curve, we see that the highest eight-year point estimates of all-cause mortality was 15.84% for those with a post radical prostatectomy persistent PSA, and a pre-op PSA less than or equal to 20.

This decreases to 12.15% for those with a persistent PSA and a PSA greater than 20, and then down to 10.78% for those with a post-RP undetectable PSA, but a pre-surgical PSA greater than 20. And the lowest risk for all-cause mortality, 9.32% at eight years post-RP undetectable PSA, and a presurgical PSA of less than 20.

Similar figure, but also just looking at prostate cancer specific mortality. And we see the trend here, eight-year point estimates for prostate cancer mortality. Highest, again, 8.32% for post-RP persistent PSA with a presurgical PSA less than 24.33% for persistent PSA, and a pre-radical prostatectomy. PSA greater than 20, down to 2.63% for post-RP undetectable PSA in a presurgical PSA greater than 20, and 3.61% for post-radical prostatectomy undetectable PSA and a pre-RP PSA less than 20.

The last two figures here look at the effect of the level of persistent PSA after radical prostatectomy. This is all-cause mortality with the cut point at greater than or less than 1. And we see that, no surprise here, post-RP persistent PSA greater than 1, has an eight-year point estimate for all-cause mortality of 26.49%, compared to 18.41% for those with a persistent PSA less than 1.

Similar figure for prostate cancer specific mortality and similar trend here. Eight-year point estimates for persistent PSA greater than 1, 13.86%, and for less than 1, 8.17%. Both of these models were statistically significant.

So by way of discussion, as Rashid mentioned in the intro, there was a counterintuitive association observed between a significantly lower all-cause mortality risk and prostate cancer specific mortality risk in patients with a persistent PSA assessed at a median of 2.7 months, 17 months post-radical prostatectomy and pre-RP PSA greater than versus less than 20.

And the hypothesis here, again, was that more patients with a pre-RP PSA greater than 20 assessed for a persistent PSA at a conventional time frame of 1.5 to 2 months, post-surgical could have reached an undetectable PSA level if further PSA assessments were done before initiating post radical prostatectomy therapy for a presumed persistent PSA.

And the evidence from this paper is that patients with a pre-radical prostatectomy PSA greater than 20 had earlier PSA assessments following radical prostatectomy, which then prompted physicians to initiate post-radical prostatectomy therapy sooner. And this is substantiated by more frequent and shorter median time to post-surgical radiotherapy or ADT use during that first year following radical prostatectomy.

So the clinical significance of this is that we need to monitor PSA after radical prostatectomy for longer than 1.5 to 2 months before concluding that a patient has persistent PSA and initiating post-radical prostatectomy therapy. And this is ultimately to minimize overtreatment. And we know from several randomized controlled trials, as well as the artistic meta-analysis from several years ago, that there's no metastasis-free survival benefit for adjuvant versus early salvage radiotherapy.

So the take-home message from this study is simply that PSA level, assessed for at least three months after radical prostatectomy may minimize overtreatment. And in this study, a higher persistent PSA level was associated with a worse prognosis. We thank you very much for joining us for this UroToday Journal Club discussion. Thank you so much.