So we know that 30& to 50% of patients, especially the high-risk patients, will experience biochemical failure within five years of radical prostatectomy. And current guidelines recommend early salvage radiation to the prosthetic bed with or without ADT depending on the risk factors. However, this paradigm is slowly shifting with PSMA-PET/CT being increasingly utilized to stage prostate cancer patients in this biochemical failure settings. And findings from such imaging tests have been shown effect treatment decisions. But to date, whether this change in management is associated with improved outcomes has yet to be determined.
Importantly, we currently have no predictive tools for biochemical relapse estimation among this patient population, meaning the population that has biochemical failure following a radical prostatectomy and then receives a salvage radiation. Emmett et al. have previously demonstrated that patients with negative or prostatic fossa-confined disease on the PSMA-PET have improved responses to salvage radiation therapy compared to those with nodal or distant metastatic disease. So that's the first hint that patients with localized or no evidence of any obvious disease respond better to salvage radiation therapy. But that's only a part of the puzzle. And can we do better by combining the findings from imaging with other clinical variables?
So the study objective was to build and validate a contemporary nomogram to predict biochemical relapse after PSMA-PET-based salvage radiotherapy post-radical prostatectomy for patients with evidence of recurrent or persistent prostate cancer.
And this was a retrospective cohort study of just over a thousand patients from 11 centers across five countries between 2013 and 2020. An inclusion criteria were having received open or lap radical prostatectomy. By lap, we mean, obviously, includes robotic. And patients received PSMA-PE-based salvage radiotherapy for PSA persistence or recurrence defined as a PSA greater than 0.1.
So the threshold's a little lower than the AUA definition is 0.2. So something to keep in mind. As for exclusion criteria, any evidence of distant metastases on PSMA-PET or CT scan. And if patient had received ADT prior to PSMA-PET or CT, then that's because we know that ADT can influence PSMA-PET/CT findings with initial flare and then subsequently it suppresses the PSMA expression, and thus any update on the scans.
For these patients, PSMA-PET scans or CT scans are performed prior to salvage radiotherapy. And so when we talk about PSMA-PET scans, could be either PET-CT or PET-MRI. And they used all known tracers, be it the Gallium-PSMA-11, which was the main one, about 70% of patients got that one, could have been a different tracer. 68Gallium-PSMA-I&T, 18F-PSMA-1007, or DCFyL, et cetera, et cetera, which for and above the remaining quarter of the patients. Importantly, scans were performed locally interpreted by two experienced readers. As for treatment and follow-up, essentially this was standard of care. Clinical decisions were performed in discretion the treating physician in accordance with the center of care recommendation of PSMA-PET findings. All the patients received intensity-modulated and image-guided salvage radiotherapy prostatic fossa with the possibility of a simultaneous integrated boost for local recurrence.
What about pelvic lymph nodes and ADT? Those who are administered according to individual patient risk factors for follow-up patients received regular serum PSA testing and keeping with standard care testing for these patients and patients will restate a time of biochemical relapse post-salvage radiotherapy. Now you may ask, while we already have finding PSMA-PET to help predict which patients do better versus not, "And so what variables were integrated in this nomogram?" So the variables were selected based on the available data, which is a practical approach and also investigator consensus. And they include the following variables in the nomogram, the pathologic grade group, which makes sense. The worst it is, the worst the outcomes you would expect. The pathologic T stage, the resection status, meaning the surgical margins. What was the PSA prior to the salvage radiotherapy? The failure is defined as 0.1, but patients could have 0.3, 0.4, two, three, et cetera, whether patients received ADT or not.
Dose to the prosthetic fossa, PSA persistence following RP, you would expect those who have PSA persistent do worse as opposed to those who nadir to undetectable levels and then recur. Whether there was evidence of pelvic lymph node involved in PSMA-PET, remember this in mets was an exclusion criteria, so it's going to be most part low volume metastases. And then PSMA-PET-detected local recurrence prior to salvage radiotherapy. Again, could have been have been negative PSMA-PET findings. The primary study endpoint was freedom from biochemical failure. This is from the time of completion of salvage to biochemical relapse, defined as PSA increased by greater than 0.2 following the salvage. And then the rates of this event were estimated using Kaplan Meier curves.
For secondary study endpoint, they looked at freedom from distant metastases detected on either a PSMA-PET/CT or bone skin. When they developed a nomogram, they used the initial dataset with deemed a learning data set about a thousand patients. This was split into training sets, so you used to build the model and then the internal validation set was about a quarter of the cohort. And that's to see how the nomogram performs in your set. Obviously, you would expect the nomogram to perform better in internal validation set because it's a similar cohort. And then the next step would be to do an external validation in a significant different cohort. So that helps with the generalizability or the external validity of the nomogram. And again, we mentioned that there were 11 centers. Out of those 11, the authors select the most dissimilar center and this one was used for external validation. The point being, if it does well in the most similar one, it's definitely likely to do better in the more similar one.
So this is a kind of a worst case scenario. The authors used a multivariable Cox proportional hazardous model with a stepwise variable selection method to build the model and essentially determine the weights of the different variables in the nomogram. And this is further elaborated on the next point, which says, "Coefficients from the Cox model were used to build the nomogram and predict the freedom from biochemical failure at one year, two years, and two and a half years." Next, when they perform the validation, the internal validation set was divided into low and high-risk groups based on the median number of points by the nomogram. So essentially the nomogram, based on the weights of the different variables, you get points added and the higher the points are, the more likely the outcome is to happen. So based on the results from the internal validation set, the cutoff was selected and patients were divided into low and high-risk groups.
And these groups, essentially, it's simplifying the analysis into a binary variable. And the two groups, the low and high-risk were compared for the two outcomes of biochemical failure distant mets using the log-rank test with survival analysis used with Kaplan Meier plots. Another important thing to consider in this setting is that a lot of the practices and the outcomes per se, are going to vary by the treatment center. So you need to account for the fact that some patients are clustered within one hospital and others within the other. And so to account for that clustering, the investigators employed a mixed effect modeling approach with the clustering variable being the center in which patients were treated. So again, it's you're trying to dilute any differences that are clustered by the hospital. And at this point, I'll turn it over to Zach to go over the results and discussion for this paper.
Zach Klaassen: Thanks so much, Rashid. So this is the patients and treatment characteristics for this study. And I've divided it into two slides just based on the size of the table. We can see here at the top, this is broken down by total cohort, the training set, 708 patients, the internal validation set, 271 patients, and the external outlier validation set, which was 50 patients. So most of the differences we're going to see are the external outlier validation set compared to the training and internal validation set. So age at first at the time of salvage radiotherapy was roughly 70 years of age in the training and internal validation set, 72.5 years. In the external set, the most common T stage amongst these patients was T2. Resection status of surgery was R zero in roughly two-thirds of the training and internal validation set and up to 88% in the external outlier validation set.
When we look at ISUP grade at surgery, most commonly was one plus two or three for these three sets. The PSA persistence at surgery was not present in roughly three quarters of patients in the training and internal validation set, and upwards of 84% in the external set. PSA before salvage radiotherapy most of the time was less than 0.5, and this is consistent with salvage radiotherapy. Local recurrence after PSMA-PET was seen in roughly 40% of patients in the training and internal set. And only 14% of patients in the external outlier set. Pelvic lymph nodes diagnosed at the time of PSMA-PET present in roughly 20% to 30% of the training and internal validation set and up to 94% in the external set.
Looking at the second half of this table, so this is again the patients and treatment characteristics. We look at the dose to the pelvic lymphatics, roughly less than 50 gray for the majority of patients in this study. Looking down to ADT use, roughly 30% of patients in the training and internal validation set and 20% of patients in the external set. And duration of ADT was most commonly six to 12 months. So this is the nomogram for freedom from biochemical failure after PSMA guided salvage radiotherapy. And we can see the point scale at the top. Total points is right here. And just highlighting the key variables, PT status was weighted quite heavily, particularly for T3B and T4. R zero status was roughly 55 points. ISUP grade was very heavily favored for additional points, particularly at ISUP grade group five 100 points. Pelvic nodes detected on PET scans, if they were, was roughly 45 points.
PSA prior to salvage radiotherapy, if greater than 0.5, was 70 points. And dose to the prosthetic fossa less than 66 gray, was associated with roughly 75 points. ADT, yes. No, no being 90 points. And we can see that this was then used to assess 12-month, 24-month, and 36-month freedom from biochemical failure. These are the calibration plots for the internal and external outlier validation sets. On the left is the internal validation cohort with a mean C index of 0.72 and a standard deviation of 0.06. On the right is the external outlier cohort with a mean C index of 0.67, standard deviation is 0.11. This is a Kaplan-Meier plot for freedom from biochemical failure. On the left is the internal validation cohort. On the right is the external outlier cohort. And basically what this is, the nomogram at the median split these into low and high-risk patients, low-risk being blue, high-risk being yellow.
And we can see that in the internal validation cohort, this split was statistically significant for predicting biochemical failure. Although this was not statistically significant in the external outlier cohort, we see this as trending towards significant P value of 0.09. Again, splitting between low and high-risk patients. Let's talk about freedom from distant metastases. Among 923 patients that had known metastatic status after salvage radiotherapy, 18.1% of these patients had distant mets. And the median time from the end of salvage radiotherapy to distant mets was 20 months. And you can see here the mean indices that they assessed. So internal validation cohort, C index of 0.7. External validation cord out of C index is 0.61. And then they did several analyses depending on the type of imaging. So for patients with PSMA-PET scan or CT scan for restaging after salvage radiotherapy, in the internal validation core, this was 0.75. And in the external validation core, this was 0.61.
So this is the Kaplan-Meier plot for freedom from distant metastases among all imaging modalities in internal validation cohort, this was statistically significant showing a difference between low and high-risk. And we see that this is also statistically significant in this cohort when PSMA alone or only was used in the internal validation cohort. So you can see a consistent early splitting of the curves between low and high-risk. However, this PSMA only in the external outliers cohort was not statistically significant, but this may be to sample size limitations as well.
So by way of discussion, this multicenter study is the first to develop a nomogram to predict freedom from biochemical failure for PSMA-PET-based salvage radiotherapy in recurrent prostate cancer. Besides clinical and therapeutic characteristics including pre-salvage radiotherapy PSA, ISP grade, pT stage, surgical margins, ADT use, and salvage radiotherapy dosed to the fossa, nodal recurrence on PSMA-PET was significantly associated with freedom from biochemical failure on multivariate analysis. The final nomogram achieved a C index of 0.72 in the internal validation hort and retained good calibration on the external validation set with a C index of 0.67.
Although the nomogram was created to predict biochemical relapse, which is not a surrogate endpoint for overall survival and recurrent prostate cancer, 16% to 22% of patients with prostate cancer do worry about PSA recurrence, which is associated with poorer quality of life.
However, the nomogram performed well for freedom from distant metastases prediction in the internal validation cohort, especially when considering only patients with PSMA-PET scans or CT scans for restaging with a C index of 0.75.
So in conclusion, this is the first nomogram to predict freedom from biochemical failure in a contemporary multicenter patient cohort receiving PSMA-PET-based salvage radiotherapy after radical prostatectomy that achieved competitive prognostic value.
The presence of pelvic nodal recurrence detected on PSMA-PET scan was introduced as a significant variable.
And the nomogram achieved stable prediction even within an external validation set that had significantly different patient and treatment characteristics.
Well, thank you very much for your attention, and we hope you enjoyed this UroToday Journal Club discussion.