Read the Full Video Transcript

Christopher Wallis: Hello, and thank you for joining us for this UroToday Journal Club. Today we are discussing a recent publication entitled Prospective Evaluation of Clinical Outcomes Using a Multiplex Liquid Biopsy Targeting Diverse Resistance Mechanisms in Metastatic Prostate Cancer. I'm Chris Walls, an Assistant Professor in the Division of Urology at the University of Toronto. And with me today is my friend Zach Klaassen, Assistant Professor in the Division of Urology at the Medical College of Georgia.

This is the citation for this recent publication in the Journal of Clinical Oncology.

Advanced prostate cancer has changed dramatically in the last 10 or 15 years as we can see in this figure, FDA approvals in mCRPC, but a similar-looking figure could be put forth for approvals in mCSPC which has actually happened much sooner. As you can tell, I'm thinking back on history, targeting androgen access is a mainstay in the treatment of advanced prostate cancer. And so more novel androgen receptor signaling inhibitors have recently demonstrated OS benefits in both the mCSPC and mCRPC. However, despite these population-level benefits, 5% to 10% of patients will have primary resistance and not derive benefit from these agents. And the majority of those who initially derive benefit will acquire resistance over one to three years of therapy.

And these resistance mechanisms come in two flavors at least. The first is androgen-dependent pathways, and these typically serve to increase AR transcriptional activity and target gene expression while the androgen deprivation is still ongoing. And this may happen through genomic alterations in the androgen receptor itself, as well as the epigenetic alterations or expression of constitutively active receptor splice variants.

There are in addition, independent pathways. And the foremost of these is the development of neuroendocrine prostate cancer, which may occur as a result of the selective pressures forced by androgen deprivation. And so this transdifferentiation is highlighted in this figure here.

In terms of molecular profiling, there are two keys to help guide our prostate cancer treatment. Number one, we need to understand the molecular drivers of resistance. And number two, we need to monitor this evolution over time for individual patients to adapt the therapies to the changing tumor environment. And so historically this would require serial tumor biopsies. And there are a number of issues with the widespread implementation of these. The first of which is the morbidity of multiple bone biopsies. Additionally, bone biopsies are associated with inadequate tumor tissue sampling, and there are relatively complicated protocols required to obtain RNA from a bony biopsy.

An alternative, which is explored in this manuscript, is a liquid biopsy. And so this can include CTCs as well as circulating tumor DNA. It is in the past, already been proven to be prognostically relevant. And some, including the AR-V7 splice variants, have been demonstrated previously to be predictive of treatment response. Additionally, liquid biopsies have shown the ability to identify neuroendocrine prostate cancer.

That's the goal of this work, was to derive a CTC-based liquid biopsy transcriptional assay capable of measuring resistance mechanisms according to a variety of pathways, including splice variants, AR activity, and neuroendocrine prostate cancer.

To do this, they accrued a multi-institutional prospective cohort, which comprised the training set. And this was 99 patients with prostate cancer treated at the University of Wisconsin and Dana Farber, who had histologically confirmed metastatic prostate cancer. The authors used a validation set comprising two Phase II clinical trials between which 48 patients had a liquid biopsy that was prospectively collected and was able to be evaluated.

So in terms of the testing approaches, they collected whole blood from these patients and used the VERSA platform for CTC capture and extraction of messenger RNA. They then used quantitative RT-PCR to assess gene expression levels. Additionally, they isolated plasma from the whole blood to assess cell-free DNA using Qiagen's Circulating Nucleic Acids Kit.

So the primary clinical endpoint of this study was overall survival, and secondarily they assessed PSA progression-free survival and radiographic progression-free survival.

In terms of statistical analysis, they tried to make this relatively straightforward, but they used hierarchical clustering of gene expression data in the training sets and then nearest shrunken centroids classifiers using these identified genes. This classifier after derivation in this initial training set was then locked and applied to the validation set. For comparisons between clusters, the authors used the Fisher's exact test and the Mann-Whitney-U test for characterizing categorical and continuous data respectively and then compared survival outcomes using the Kaplan Meier method and Cox proportional hazards models.

At this point in time, I'm now going to hand it over to Zach to walk us through the results.

Zachary Klaassen: Thanks, Chris. So this first figure is a heat map that shows two distinct molecular clusters in circulating tumor cells from patients with metastatic prostate cancer. And you can see that on the left is cluster one in the blue, highlighting at the top of the figure. And cluster two is the orange to the right. And to sort of summarize this heat map, cluster one has low to the absence of AR-regulated genes and cluster two has high expression of transcriptional targets of the androgen receptor.

So based on this delineation of cohorts, the authors then looked at patient and clinical characteristics stratified by cluster one and cluster two. And you can see that the median age for these patients was 69 in cluster one and 72 in cluster two with a time since diagnosis of 3.8 years for cluster one and 1.8 years for cluster two. Importantly, there is a big difference in PSA between these clusters with a median PSA of 11.2 in cluster one and 216 in cluster two. The majority of patients presented with Gleason greater than or equal to nine disease in both clusters at 38%. 36% of patients had de novo metastatic disease in cluster one and 42% in cluster two. And the most common site of metastasis was bone metastasis in 100% of patients in cluster two and 81% of patients in cluster one, with lymph node metastasis being the second most common at 62% in cluster one and 54% in cluster two.

Looking at the type of disease, which is interesting, hormone-sensitive prostate cancer, 33% in cluster one and 15% in cluster two, CRPC in 59% of cluster one and 85% in cluster two, and neuroendocrine prostate cancer, 8% in cluster one and 0% cluster two.

In terms of prior treatment, broken down by cluster, current or prior use of ARSIs, 58% in cluster two and 45% in cluster one. And current or prior use of taxane chemotherapy, 46% in cluster two and 11% in cluster one.

This is the Kaplan-Meier plot of overall survival stratified by each of these clusters. Cluster one is in blue and cluster two is in orange. And the median overall survival for cluster one was 22.4 months compared to 8.6 months for cluster two with a hazard ratio of 3.45 and a 95% confidence interval of 1.91 to 6.21.

This is the Kaplan-Meier plot for overall survival stratified by AR-V status. AR-V negative in green and AR-V positive in red. The median overall survival for AR-V negative patients was 19.1 months compared to 8.6 months for AR-V positive with a hazard ratio of 2.49 and a 95% confidence interval of 1.39 to 4.47.

This is the multi-variable analysis for overall survival. A couple of points to highlight here. Even after adjusting, the cluster was still a significant factor for overall survival with a hazard ratio of 3.68 and a 95% confidence interval of 1.59 to 8.51. And you can see here that in the multi-variable analysis when we adjust for the other variables, visceral metastases and AR-V status are no longer predictors of overall survival.

This is the Kaplan-Meier curve of overall survival among phase two trial patients that received either abiraterone or enzalutamide. So this is the external validation of this cluster model. And you can see here, again, C1 cluster did not reach median overall survival compared to 15.2 months for C2 with a hazard ratio of 8.43 and a 95% confidence interval of 2.74 to 25.92.

Again, this looks at PSA progression-free survival in this phase two trial data with a C1 cluster PSA progression-free survival median of 12 months compared to 3.6 months for C2 and a hazard ratio of 4.64, and a 95% confidence interval of 1.53 to 14.11.

Looking at radiographic progression-free survival, cluster one at 40.6 months compared to a median of 2.69 months for cluster two with a hazard ratio of 5.63 and a 95% confidence interval of 1.83 to 17.41.

So to finish off the results section, the authors described a case of a patient that presented with a PSA of 175, extensive bone metastases, and liver metastasis. He then was placed on combination androgen blockade with docetaxel. And you can see here that these are the ctDNA fractions at multiple time points. So 11 months since diagnosis and then 14 and 17. So we see increasing ctDNA percentages over the course of time and over the course of treatment.

Delineating month 17, we can see the emergence of PTEN, RB1, and P-53. And so moving to the next slide, we also can see a concurrent copy number loss occurred for months 11 to 17 for RB1 and PTEN. So the conclusion from this sort of case example is that the timing of the CTC and ctDNA changes were concordant with both of these preceding clinical progressions at month 17. So predicting what was going to happen before the clinical imaging actually showed a progression of the disease.

This also highlights that treatment-emergent neuroendocrine prostate cancer can be detected before the onset of clinical progression, particularly with PTEN, RB1, and P-53 being highly expressed in neuroendocrine prostate cancer.

So several discussion points from this study. This is the first assay designed to detect three major mechanisms of resistance to ARSIs in prostate cancer, including AR-V status, AR pathway activation, and neuroendocrine prostate cancer. For the first time, this study identified a distinct cluster called C2, which had high expression of AR target genes, which was prognostic for overall survival, independent of other variables including AR-V status, CRPC, and neuroendocrine prostate cancer, which was also validated in the two Phase II trials that we showed previously.

Molecular classifiers represent an attractive way to follow prostate cancer disease status in addition to PSA and imaging. And from a hypothetical situation, you could envision blood samples being sent to the clinic for real-time, liquid biomarker studies.

So in conclusion, this is the first multiplex gene expression liquid biopsy assay that can assess multiple potential ARSI resistant mechanisms simultaneously in the treatment of metastatic prostate cancer.  Indeed, early identification of these molecular changes could help guide treatment decisions. And the heterogeneous nature of this cohort suggests that these findings may be generalizable to a broader patient population.

Finally, these findings are being further tested in multiple ongoing clinical trials for patients with hormone-sensitive prostate cancer, as well as castration-resistant prostate cancer.

Thank you very much. And we hope you enjoyed this UroToday Journal Club discussion.