Christopher Wallis: Hello, and thank you for joining us for this UroToday Journal Club discussion. Today we're talking about a recent publication entitled, The Addition of Docetaxel to Androgen Receptor Axis-Targeted Therapy and ADT in Metastatic Hormone-Sensitive Prostate Cancer: A Network Meta-Analysis. I'm Chris Wallis, an Assistant Professor in the Division of Urology at the University of Toronto. With me today is Zach Klaassen, assistant professor in the Division of Urology at the Medical College of Georgia. We're discussing this publication to which we both contributed recently published and currently in press in the European Urology Oncology.

Most UroToday readers and viewers will know that the metastatic castration-sensitive prostate cancer space has evolved relatively rapidly over the last 5 to 7 years, with new indications for four treatment approaches up to the beginning of 2022 and more since then. And so, with this evolution of care, we've seen changes in study design. Docetaxel became a standard of care based on the data from CHAARTED and STAMPEDE. Following this, it was initially permitted and then mandated in part of the control arm for other trials. So in ARCHES and ENZAMET assessing enzalutamide, we saw the inclusion of docetaxel, both patients in the control arm and the intervention arm. However, this wasn't mandated nor strictly controlled, and so more recently a triplet approach comprising ADT and ARAT and docetaxel has been tested against the doublet of ADT plus docetaxel in the ARASENS trial, looking at darolutamide as the ARAT, and in the PEACE-1 trial, looking at abiraterone as the ARAT.

And so, the clinical conundrum we face is that while the triplet approaches in both PEACE-1 and ARASENS have proven benefit compared to the ADT-docetaxel based doublet, this is not the most commonly utilized doublet approach, and instead, most patients receive ADT plus ARAT. Triplet has not been formally compared to ADT plus ARAT as a doublet, thus, we are unclear on the additional benefit of adding docetaxel for patients who are receiving ADT and an ARAT.

To address this, we, among our author group, have performed a network meta-analysis including parallel design phase III, randomized controlled trials performed in mCSPC. We compared four different classes of interventions, ADT alone, ADT plus ARAT, ADT plus docetaxel, and the triplet approach of ADT plus ARAT plus docetaxel. Protocol for this was prospectively registered with PROSPERO. Search was performed with Medline and SCOPUS from inception until March 9th, 2022, and we further included reports from the proceedings of ASCO, ASCO GU, and ESMO between 2014 and 2022. All screening and extraction was performed in duplicate and the design intervention based on characteristics, follow up, and overall survival outcomes of each study were extracted.

ARATs were considered as a single treatment class, and we thus had these four treatment groups that could be compared. Publication bias was assessed with Egger's tests and funnel plots, and between-study heterogeneity was both assessed and quantified. Given between-study heterogeneity, we used random-effects meta-analysis and treatments were ranked according to their P score, a measure of the level of certainty that one treatment is better than another averaged across all the treatments. In addition to these frequentist network meta-analyses, we performed a Bayesian network meta-analysis using four parallel Markov chains with 100,000 samples and use SUCRA analysis to provide a treatment of preferred treatment approaches.

Now I'm going to hand it over to Zach to walk us through the results of this network meta-analysis.

Zachary Klaassen: Thanks for that great introduction, Chris. This is the PRISMA diagram from this study. There was two trials that were included from recognized databases. Ultimately, the search included 1,670 trials, and ultimately, an additional nine trials were included in this study. You can see these nine trials as adhere. These are well known to our viewers and listeners. This included ARASENSE, ARCHES, CHAARTED, ENZAMET, GETUG-AFU 15, LATITUDE, PEACE-1, STAMPEDE arms C and G, the SWOG1216 trial, as well as TITAN. Again, you can see the key inclusion criteria here, as well as a number of patients, the majority of these were big phase III trials with good randomization to these treatment arms, with a duration of follow up ranging from 2.8 to 7 years in follow up.

With regards to the network graph of the treatment comparison, this is the figure that Chris showed earlier, looking at the nodes and the delineation of the trials. To sort of summarize the breakdown here, we see that amongst patients that received ADT plus ARAT, there was 3,211 patients, for those that received ADT plus docetaxel, there was 2,366 patients, for those that received ADT plus ARAT plus docetaxel, there was 1,421 patients, and for those that received ADT alone, there was 4,548 patients. So a good representation of each of these groups among the 11 trials.

This is the forest plot for the frequentist NMA showing hazard ratios for overall survival for the treatment strategies. You can see that this is with comparison to the ADT plus ARAT group. What's interesting is that compared to ADT plus ARAT, there was no significant difference between that treatment regimen compared to ADT plus ARAT plus docetaxel or ADT plus docetaxel. However, not surprisingly, ADT plus ARAT was better than ADT alone. And so, looking at the P scores, the most likely first-line treatment would be ADT plus ARAT plus docetaxel, followed by ADT plus ARAT, followed by ADT plus docetaxel.

This is the forest plot for the relative effects of the treatment strategies on overall survival using the Bayesian network meta-analysis. Again, the control group is ADT plus ARAT, and again, we see no difference between ADT plus ARAT and the addition of docetaxel or ADT plus docetaxel alone. Again, we see that ADT plus ARAT is better than ADT using the Bayesian analysis. Similar to the P scores, we see with the SUCRA analysis that the most likely first-line treatment option would be ADT plus ARAT plus docetaxel, followed by ADT plus ARAT, and followed in third place by ADT plus docetaxel.

The third analysis looks at posterior ranking probabilities of the four treatment strategies for overall survival using the Bayesian random effects hierarchical model. This is the four options. ADT, ADT plus ARAT, ADT plus ARAT plus docetaxel, and ADT plus docetaxel alone. There was a 77% likelihood that ADT plus ARAT plus docetaxel would be the first choice in the firs- line setting among these patients and a 23% chance that ADT plus ARAT would be the first choice for these patients.

Several discussion points from this network meta-analysis. Based on three different approaches, including a random effects frequentist approach, the P score ordering, and the Bayesian network meta-analysis, for all three of these, the combination of ADT plus ARAT plus docetaxel was the most effective treatment for first-line metastatic hormone-sensitive prostate cancer.

Docetaxel was part of the protocol specified treatment approach and 8 of the 11 RCTs included in this network meta-analysis. In ARASENS, all patients received docetaxel. In CHAARETED, GETUG-AFU 15, and the STAMPEDE arms, docetaxel was determined by randomization. In ARCHES, ENZAMET, PEACE-1, and TITAN, docetaxel was a stratification variable. Not surprisingly, there are several limitations of this study. Network meta-analyses are based on aggregate data rather than individual patient data. That's the pooled estimates of treatment effect applied to the overall patient population study. Secondly, the use of aggregate data precluded a pooled analysis of toxicity and safety data.

In conclusion, this network meta-analysis compared the effect of four different treatment strategies on overall survival in metastatic hormone-sensitive prostate cancer. Although not reaching statistical significance, the triplet strategy of ADT plus ARAT and docetaxel had a modest OS benefit compared with ADT plus ARAT alone. Overall, the triplet strategy had the highest likelihood of being the most effective treatment strategy, being favored over the combination of ADT plus ARAT. However, this needs to be balanced against the potentially high risk of toxicity with triplet strategy, which was not assessed in this particular network meta-analysis.

An individual patient data meta-analysis of the existing trials or a large scale RCT directly comparing these two treatment strategies would provide the most nuanced and reliable assessment of the comparative effectiveness of doublet and triple therapies. However, in the absence of such analyses, this network meta-analysis provides the highest level comparative evidence for these treatment approaches in the initial management of metastatic hormone-sensitive prostate cancer. Thank you very much for your attention and we hope we enjoyed this UroToday Journal Club discussion.