Read the Full Video Transcript

Rashid Sayyid: Hello everyone, this is Rashid Sayyid, I'm a Urologic Oncology Fellow at the University of Toronto, and along with Zach Klaassen, Associate Professor and Program Director at Augusta University, we'll be discussing a study looking at a subgroup analysis from the SWOG S1216 trial that looks at Bone Biomarkers and Subsequent Survival in Men with Hormone-Sensitive Prostate Cancer: Results from the SWOG S1216 Phase III Trial of Androgen Deprivation Therapy with or without Orteronel. This study was recently published in European Urology with Dr. Lara as the first author.

We know that patients with metastatic prostate cancer commonly have skeletal or bone metastases, and these mets are predominantly osteoblastic in nature, manifesting as sclerotic bony disease. We know that circulating blood-based biomarkers of bone turnover are independent prognostic factors for overall survival in mCRPC patients, and highly elevated markers are predictive of better survival of bone targeted therapy. However, we don't exactly have these correlates in the mHSPC, hormone-sensitive, space, and so the prognostic and/or predictive value in this space remains unknown.

Orteronel, or TAK-700, is a non-steroidal CYP17A1 inhibitor, kind of similar to abiraterone in its mechanism of action. It inhibits testosterone biosynthesis similarly in the testes, the adrenals, and the prostate. This drug was evaluated in the SWOG S1216 trial that took patients with de novo mHSPC disease and randomized them in a 1:1 fashion to LHRH agonist plus the TAK-700 or the orteronel 300 milligrams given twice daily versus the LHRH agonist with an active control of bicalutamide 50 daily with a primary outcome of overall survival.

In the grand scheme of things, this was a negative trial, but there are some important considerations to take away from this. If we look at the overall survival here in the Kaplan-Meier curve, we see that the overall survival, the median OS, was 81 months, which was in line with other trials for the most part, so just under 7 years. If we look at the control arm, we see that the median overall survival was much better compared to the control arms from other trials at just under 6 years, or 70.2 months. So this is a sign first of using the active control as opposed to other trials which have used just placebo as opposed to ENZAMET, which, as well, used bicalutamide. Also, it's a sign of the ability to utilize rescue drugs in this space and how far we've come along in this disease space as well.

And so, the authors sought to conduct an ad hoc analysis of the SWOG S1216 trial, where they looked at the markers from baseline patients here using various immunoassays. They looked at bone formation biomarkers, specifically C-terminal collagen propeptide, and they also looked at bone alkaline phosphatase. They also looked at markers of bone resorption, such as the C-telopeptide, and pyridinoline as well.

Next, they used these markers in a way where they developed a model, and then they sought to evaluate the model. For this exercise, patients were split into a training set, one-third of the cohort, about 316 patients, and they also split the cohort into validation set as well, which was two-thirds of the total sample size. Starting off in the training set, split points were identified in each individual bone biomarker using survival tree algorithms, which utilized survival data, with the ideal split point being the value at which the log-rank test statistic for survival between the groups was maximized. Next, they looked at linear combinations of the biomarker splits that maximized the univariable differences in survival, and these were used to create categorical risk groups within the population.

Essentially, the first step, they looked at each bone biomarker alone, looked at the maximal survival with each one in different cutoffs, and then they took the four biomarkers together, and using the combination of the values for each, they created the cutoffs based on the maximal survival benefit with each combination.

Next, the predetermined split points from the training set were applied to the validation set. And so the elevation of each individual bone marker, the first step, and then the combination of the markers, the second step, were individually evaluated as potential overall survival prognostic factors using Cox proportional hazards models adjusted for various factors such as the treatment arm, whether you got orteronel versus not. They also looked at interaction term between the bone marker elevation in the treatment arm to see if there was a difference in the prognostic value based on which treatment you got. They looked at the disease extent as well, the performance status, race, Gleason score, age, the log-transformed PSA, and presence or absent of visceral mets as well.
They also applied the Bonferroni correction for the n = 5 models. So if we apply a p-value of 0.05 and we repeat the analysis more than one time, we're much more likely to find a significant value in the absence of a real significance. To correct for that, we perform the Bonferroni correction to minimize any inflation of the type one error rate or the false positive rate.

At this point, I'll turn it over to Zach to go over the results and the discussion of this study.

Zach Klaassen: Thanks so much, Rashid, for that introduction. This is the baseline characteristics of patients included in the pooled arms of the bone marker analysis. We can see here in the middle, this is all the patients in the S1216 primary analysis population. To the right of this is the patients, 949, that had baseline bone markers available. In terms of the percentage of patients in each arm, 50/50, which is as expected. In terms of minimal severity of disease percentage, again, roughly 50/50, the overwhelming majority of these patients had excellent performance status, well-balanced with regards to PSA, 30 in the primary analysis in 28 in the bone marker analysis. With regards to Gleason score, most commonly was Gleason greater than or equal to 8, at 59% in the primary analysis and 60% in the bone marker analysis. Visceral metastases, 14% and 13%. Age at study entry was just in the late sixties and the majority of patients were white, at 84% in the primary analysis and 86% in the baseline bone marker analysis. I will mention that there was 11% Black or African American in the primary analysis and at 9.4% in the bone marker analysis.

This table looks at the baseline serum bone marker concentration. This just gives everybody a reference for what was the medians for each of these markers in the study. In terms of bone alkaline phosphatase, median was 3, for C-terminal collagen propeptide median was 125, for C-telopeptide, median was 0.4, and for pyridinoline was a median of 1.6.

This table will take a minute to walk through. This is the hazard ratio estimates for elevated bone markers in the validation set, which was 633 patients. You can see on the top four lines here for bone alkaline phosphatase, CICP, CTx, and PYD, they split these into median survival by group with a high and a low, with a hazard ratio. We'll take a minute to focus on the bone marker combination. Low-risk patients were defined as CTx less than 0.6 and CICP less than 161, which had a median survival of 8.2. When we look at the intermediate risk, CTx less than 0.6 and CICP greater than or equal to 161 or CTx greater than 0.6 and CICP less than 286, with a median overall survival of 5.1 years. High-risk patients were delineated as greater than or equal to 0.6 CTx and a CICP greater than 286, with a median overall survival of 2.1 months. And so these were statistically significant when compared to the low-risk patients.

Graphically, this is how this looks. This is the Kaplan-Meier curve for the bone marker combination. Again, we see the low-risk patients in dark orange, the intermediate-risk patients in regular orange, and the high-risk patients in this orange-yellow color. We see early and consistent splitting of the curves between these three groups. To highlight, again, the median survival outcomes for low-risk patients was 8.2 years, for intermediate-risk patients was 5.1 years, and for poor-risk patients was 2.1 years.

These next several slides will look similar to this one. This is looking specifically at each bone marker. This is the survival probability for patients in the validation set for bone alkaline phosphatase. As we saw in the first table, the elevated bone alkalize phosphatase has a hazard ratio of 1.43 and 95% confidence interval of 1.02 to 2.01. Patients with low bone alkaline phosphatase had a median overall survival of 6.8 years compared to 3.3 years for high bone alkaline phosphatase. Similar curve looking at C-terminal collagen propeptide, also known as CICP. Elevated CICP had a hazard ratio of 1.92. Those patients with a low CICP had median survival of 7.6 years compared to those with a high CICP at 2.4 years.

For C-telopeptide, or Ctx, elevated CT X had a hazard ratio of 1.37 and those patients with a low CTx had a median overall survival of 7.7 years compared to 4.0 years for those with a high CTx level. Finally, for pyridinoline, or PYD, an elevated PYD had a hazard ratio of 1.77 and a median overall survival in low PYD patients of 8.2 years compared to 3.4 years for those with a high pyridinoline level.

By way of discussion, the results of this prospectively designed study demonstrated a strong association between biomarkers of bone turnover and overall survival in men with advanced or metastatic hormone-sensitive prostate cancer. Elevated levels of each of the four bone biomarkers showed statistically significant association with worse survival outcomes independent of traditional clinical risk factors. Importantly, using a regression tree analysis, unique subsets of men with different survival outcomes, such as delineated by low, intermediate, or high risk, patients were identified by employing combinations of bone biomarkers.

Secondly, bone turnover markers are poised for pragmatic clinical use. First, they can be conveniently obtained through phlebotomy and measured using commercially-available assays, and they can guide patient counseling and future research. Examples of this are future early-phase clinical trials can employ this information to identify high risk groups to screen for new drugs by identifying those with worse prognosis in order to increase pace of trial. Secondly, bone biomarkers could someday be a component of a biologically-informed multidimensional model that comprises complex molecular and clinical data to yield prognostic and predictive utility.

In conclusion, the subsequent survival of men with newly diagnosed metastatic prostate cancer following the initiation of ADT is strongly and statistically significantly associated with baseline serum levels of bone metabolism biomarkers. These results can be employed by clinicians in counseling patients and by researchers in the design and conduct of future trials.

We thank you very much for your attention. We hope you enjoyed the UroToday Journal Club of the SWOG study looking specifically at bone biomarkers.