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 Association of Dynamic Change in Patient-Reported Pain With Survival in Metastatic-Castration Sensitive Prostate Cancer: An Exploratory Analysis of the LATITUDE Study. I'm Chris Wallis, an assistant professor in the Division of Urology at the University of Toronto. With me is Zach Klaassen, an Assistant Professor in the Division of Urology at the Medical College of Georgia. You can see here, the citation for this recent publication in Prostate Cancer and Prostatic Diseases.

Metastatic castration-sensitive prostate cancer is relatively heterogeneous. Despite many changes in advances in the disease state, we need to center our understanding and thinking about it in the basis of the heterogeneity of clinical presentation, tumor biology, and prognosis. Between 2015 and today, there have been substantial advances in treatment approaches, with the treatment intensification showing improved overall survival. In addition to overall survival, we also need to consider other endpoints. So patient-reported effects and quality of life are critical, and among these, pain is among the most broadly applicable across the disease setting. In mCRPC, pain has been shown to be prognostic, but has not been well assessed in mCSPC.

When we consider how we assess the prognostic value of pain, a static baseline assessment can quickly become outdated and it can lose a predictive ability as disease progresses and over time. A dynamic assessment of pain, instead, may address this, as we can assess changes as they occur moving forward, and this may better reflect improvements we see with treatment intensification. And so, the authors of this study perform an exploratory analysis of the LATITUDE trial to assess for that dynamic change in patient-reported pain was associated the risk of progression or death in men with de novo metastatic castration-sensitive prostate cancer.

As you can see here, from the original publication of the LATITUDE trial, this was a global multi-centered double wide randomized phase III trial. The authors included men with high-risk de novo metastatic castration-sensitive prostate cancer, and this was defined as at least two out of the three following characteristics, three or more bony lesions, visceral metastases, or Gleason score 8 to 10. Patients were excluded if they had prior docetaxel or local therapy or if they had low risk disease not fulfilling the aforementioned high-risk criteria. Patients were randomized in a 1:1 fashion to receive either abiraterone and prednisone, in addition to ADT, or ADT with dual placebos. Treatment was given in 28-day cycles.

This highlights the study design we've just discussed and emphasizes that the co-primary endpoints are overall survival, radiographic progression-free survival, with numerous secondary endpoints. To emphasize this further, secondary endpoints included patient-reported outcomes, including pain, fatigue, prostate cancer symptoms, and health related quality of life. These were measured, respectively, with the BPI-SF, Brief Fatigue Inventory, FACT-P version 4, and EQ-5D-5L. Patient-reported outcomes were collected on the first day of the first three cycles, monthly during cycles four through 13, and then every 2 months thereafter until treatment ceased. In this exploratory analysis, the authors focused on both worst pain score and pain interference score. Each of these are rated on an 11 point scale, from 0 with no pain or no interference to 10 being the worst pain or complete interference.

Analytically, the authors sought to assess if a dynamic change in patient-report worst pain and pain interference was associated with overall survival and progression-free survival. They first used Cox proportional-hazards models to determine the association of baseline pain with these outcomes. Covariates included in these models included the treatment arm, either abiraterone or placebo, number of skeletal lesions, baseline PSA, presence of liver or lung metastases, nodal stage, Gleason score, ECOG performance status, and age.

They then used joint point modeling for longitudinal survival outcomes, and they first performed multivariable univariate joint point models to determine the association of dynamic changes in pain with overall survival and progression-free survival. They developed a time to event sub-model using Cox proportional-hazards models and longitudinal sub-models using linear mixed effects models. In these mixed effects models, they used in interaction term between treatment arm and time to evaluation. The time of assessment was included as a random slope, where patients themselves were included as random slope intercepts. The two models, this time to event sub-model and the longitudinal sub-model, were linked through a random effect. Finally, they used two independent multi-variable joint probability models that were assembled using Markov chain Monte Carlo algorithms, and these models were used to calculate baseline hazards modeled using penalized lines.

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

Zachary Klaassen: Thanks so much, Chris. This is the baseline characteristics from this study. You can see this is stratified by those who received abiraterone versus placebo. As you would expect from a large phase III RCT, these were well-balanced amongst some these groups. Just to highlight quickly, the median age for these patients was 67 years, the median PSA actually was higher in the abiraterone group at 47 compared to 36.7 in the placebo group. In terms of Gleason score, roughly half of these patients were less than or equal to Gleason 8, and the other half, Gleason 9 to 10. In terms of T stage, two-thirds of these patients were T3 to T4. In terms of skeletal lesions, roughly half had 0 to 10 lesions and greater than 10 lesions. In terms of liver or lung metastases, roughly 20% had liver or lung metastases. Roughly half of the patients, 54%, had N1 disease, and in this population, about 55% of patients were ECOG zero. In terms of ethnicity, roughly 84% were non-Hispanic, non-Latino men.

This next couple of figures highlight some of the outcomes in the overall cohort. This is PFS by treatment regimen. The abiraterone patients are in the dark gray line and the placebo patients are in the light gray line. The median rPFS for abiraterone was 44.8 months compared to 17.2 months for the placebo arm. With regards to overall survival, again, we see an early and consistent splitting of the curves, with a meeting OS for abiraterone at 45.5 months and for placebo at 32.4 months.

The next several tables will be similar to this one in terms of looking at the joint models for the univariate followed by the multivariate models. This first table is the univariate joint model for OS with worst pain score, which had an AUC of 0.74. You can see that the current value of worst pain score hazard ratio was 1.316, with a 97.5% confidence interval of 1.258 to 1.376. Additional significant factors in the integrated analysis included, not surprisingly, receiving placebo was associated with worse survival, more skeletal lesions, higher Gleason score, and worse ECOG performance status.

This is the univariate joint model for PFS with worst pain score. Again, we see a hazard ratio 1.319 for current value of worst pain score, which was statistically significant. Again, we see that placebo more than 10 skeletal lesions, as well as Gleason score, were all significantly associated with worse PFS, as well as younger age.

This is the univariate drug models for OS with pain interference score. So this is the other pain metric that they looked at. This had an AUC of 0.73. We see that the current value for pain interference score had a significant hazard ratio of 1.319, 97.5% confidence interval of 1.261 to 1.381. Similar to the other models, placebo, skeletal lesions more than 10, higher Gleason score, and worse performance that is also were associated with overall survival in this analysis.

This is the univariate joint model for PFS using the pain interference score, an AUC score for this model of 0.73. Again, pain interference score was significant, 1.282, which was statistically significant. And again, placebo skeletal lesions more than 10, Gleason score, N1 disease, and age were all statistically significant as well.

Moving onto the multivariable joint models for overall survival. There is some significant findings here. In this multivariable model, current value of worst pain score was significant, hazard ratio 1.265. However, current value of pain interfere score was not significant at 1.016. So we see that one of the pain metrics was significant. The other was not. Again, and not surprisingly, those that received placebo had more skeletal lesions, higher Gleason score, and worse ECOG performance status also were statistically significant for overall survival in this multivariable model.

Some similar findings for the PFS model. Again, we see that the current value of worst pain score was significant, hazard ratio 1.388, and again, the current value of pain interference score was not significant. Similarly to the overall survival model, those that received placebo had more skeletal lesions, were younger, and had worse Gleason score disease was also significant in the multivariable model for PFS.

This study does have several important discussion points. In the secondary analysis of the LATITUDE trial, both baseline pain and dynamic change in pain were associated with overall survival and progression-free survival in men with de novo metastatic castrate-sensitive prostate cancer. Compared to baseline pain, dynamic change in pain reported scores had a relatively stronger association with OS and PFS.

It's important to note that pain is a common symptom in patients with metastatic prostate cancer and these findings could in part be attributed to the underlying link between cancer progression and pathophysiology of pain, such that nerve growth factors secreted by the tumor inflammatory and immune cells trigger pain exacerbation by TrkA receptor-mediated actions, such as the release of neurotransmitters. Also, this is important as these findings are consistent with those found in the FIRSTANA study, which looked at docetaxel versus cabazitaxel in chemotherapy-naive mCRPC, whereas they found that pain progression at initiation of chemotherapy was also associated with worse overall survival.

In conclusion, dynamic change in patient-reported pain has a significant association with PFS and OS and men with de novo metastatic castrate-sensitive prostate cancer. Due to its independent and superior prognostic ability, dynamic changes in patient-reported pain have the potential to tailor therapy in mCSPC patient population based on response to initial therapy and could be considered a dynamic prognostic marker above and beyond its role as a very important tool for quality of life measurement. This work underscores the importance of patient-reported quality of life metrics, such as pain not solely to measure the patient experience, but also highlights a potential dynamic interplay of quality and quantity of life in this patient population. Finally, further studies in mCSPC patient cohorts are required to validate this association and to explore the underlying mechanistic biology.

Thank you very much for your attention and we hope you enjoyed this UroToday Journal Club discussion of a secondary analysis of the LATITUDE trial.