Christopher Wallis: Hello, thank you for joining us for this UroToday Journal Club. Today, we're discussing a recent publication entitled, Association of Treatment Modality, Functional Outcomes, and Baseline Characteristics With Men With Treatment-Related Regret Among Men With Localized Prostate Cancer. I'm Chris Wallis, an assistant professor in the Division of Urology at the University of Toronto. Joining me today is Zach Klaassen, an assistant professor in the Division of Urology at the Medical College of Georgia. This is the citation of this recent publication that we contributed to in JAMA Oncology.

Localized prostate cancer treatment is a complicated disease space. We have multiple guideline-recommended treatment options, including surveillance, surgery, and radiotherapy. While these may be informed by disease characteristics, for most patients, there are multiple appropriate treatment approaches, and therefore treatment decisions are somewhat difficult at times. Notably, adverse events or the side effects of therapy may differ between treatment modalities. However, the correlation between these symptoms and the impact on any given patient or their quality of life is very non-linear, so it's difficult to counsel a patient on how likely they are to be bothered from their side effects of therapy, even when we know how prevalent these are.

Treatment regret is an important metric that is really premised on this negative cognitive-based emotional feeling. And it's premised really when we consider it on a counterfactual comparison, i.e. what is the alternative result that someone may have received if they made a different decision in the past? And as a result, it is influenced by pre-decision expectations. In our view, treatment regret is a really good outcome measure, because it integrates the effects of treatment-related functional impairments, oncologic outcomes and associated anxiety, as well as behavior, emotional, and interpersonal changes that may be associated with the prostate cancer diagnosis and treatment. And all of these are considered through the lens of an individual patient's values and expectations.

And so our hypotheses in this study were that treatment modality would be independently associated with the likelihood of treatment regret, that this association would be mediated by the functional outcomes achieved following treatment, and that decision making style and expectations initially will influence the long-term risks of developing regret. And so to assess this, we relied upon the CEASAR cohort, which is a prospective population-based cohort of patients accrued from five population-based SEER registries, who were diagnosed with localized prostate cancer during 2011. Patients could be included if they were aged less than 80 years in clinically localized disease, PSA less than 50, and were enrolled within 6 months of their diagnoses. Patients completed surveys at baseline, 6 months, 12 months, 3 years, and 5 years after enrollment.

In this study, we focused on the regret measure as an outcome. This was developed and validated by Dr. Clark, and using his approach, we define significant regret as scores of 40 or greater and assess these at 3 and 5 years following enrollment. This highlights the conceptual framework through which we considered our analyses. Looking at treatment regret as the outcome on the far right, we can see that a baseline patient characteristics influence their pretreat and expectations, and these may be mediated through the decision-making style. And these characteristics, along with their tumor and disease characteristics, may influence the treatment modality selected. In turn, treatment modality influences functional outcomes. And together, both expectations, treatment modality, functional status, as well as the mediating effect of social support, may lead to treatment regret.

And so we considered a number of important exposures. In terms of primary treatment modality, we compared surgery to radiotherapy and surveillance. Now, we further considered patient-reported functional outcomes, as measured used in the EPIC-26 and the SF-36 to assess prostate cancer-specific and overall functional status. Finally, we added important demographic and baseline characteristics, including the decision-making style, measured using a PDM, or a participated decision-making, 7 tool, as well as the difference between expectation and outcomes, both in terms of oncologic efficacy and toxicity. And then we considered social supports, age, race, ethnicity, education level, and marital status that's potentially important in influencing a man's likelihood of developing treatment-related.

To analyze this, we built three successive logistic progression models. The first assessed treatment modality and adjusted for D'Amico risk group, age, education, comorbidity, race or ethnicity, ADT, use of pelvic radiotherapy, study site, and decision-making style. Model two added in treatment-related health problems, patient-reported functional outcomes, perceptions of treatment efficacy and toxicity, and social support. Model three added marital status, baseline social support, and baseline functional outcomes to the first model. As we know, D'Amico risk group may influence treatment decisions. We further stratified the models according to these criteria. I'm now going to hand it over to Zach to walk us through the results of this analysis.

Zachary Klaassen: Thanks, Chris. You can see this is the diagram of assembly of the CEASAR study court and final analytical cohort, and over a number of exclusion criteria. The analytical cohort, for this specific study looking at treatment regret, was 2072 patients, including 1136 that underwent surgery, 667 that underwent radiation, and 269 that underwent active surveillance.

This is the table 1 looking at baseline characteristics of these patients, and I've broken this down into two slides given the size of the table 1. We can see here on the far right as the p-value followed by all the patients together, and then sequentially active surveillance, radiation, and surgery. So looking at some of these highlights of the variables, we can see here that, not surprisingly, patients that underwent surgery were a little bit younger. We can see that the most common race was white patients at three-quarters in each of these groups. There's a pretty balanced division of patients based on educational level across these three groups. And we can see that nearly 80% of patients were married at the time of their enrollment.

In terms of D'Amico risk category, the majority of patients in each of these groups was low or intermediate risk, and we can see that the baseline PSA was 5 to 6 for each of these patients. Looking at the most common PSA level at the time of diagnosis, most common was 4 to 10 in 60-70% of patients, the most common clinical tumor stage was T1, ranging from 7485% of patients, and the majority of patients had a biopsy Gleason Score of 6 or 3+4. Subsequently, any ADT within 1 year was only seen in primarily the radiation cohort, at 32%. To conclude the baseline characteristics, we can see that this is the breakdown by site, and with a high proportion of patients either being in Los Angeles or in the Louisiana cohort.

This table looks at the pairwise association between treatment modality and patient regret at 5 years after diagnosis. And so what's important in this table is, at the left here, this is D'Amico risk category among all patients and then stratified by D'Amico risk individually, and to the right of this is the treatment comparison adjusted for basically patient demographics. And we can see, on the far right, this is adjusted for demographics as well as patient long-term functional outcomes.

Going back to the starred or the asterisks for the statistically significant differences, we see that surgery versus active surveillance was associated with treatment regret, with an odds ratio of 2.40. And this odds ratio decreased to 1.73, 95% confidence interval of 0.99 to 3.02 after adjusting for patient long-term outcomes. Looking at surgery versus radiotherapy, we see here a statistically significant odds ratio for regret of 1.57. However, this is nullified after adjusting for the patient long-term outcomes. In low-risk patients, we see significant regret for surgery versus active surveillance, with an odds ratio of 2.73, which was still significant after adjustment for long-term outcomes of the odds ratio of 2.08. Moving down to the high-risk D'Amico group, we see you that radiotherapy versus active surveillance was actually protective for regret, even after adjusting for long-term functional outcomes, with an odds ratio of 0.12. And we see that for high-risk patients surgery versus radiotherapy, odds ratio of 2.64. But when we adjust for patient long-term outcomes, this becomes non-significant.

This looks at the adjusted odds ratios for treatment-related regret, and you can see here that the ability of using a decision tool was associated with less regret. Sexual function was also associated with less regret. And we moved down to treatment here, surgery versus active surveil. This was associated with regret, with an odds ratio of 2.08. And then if we look at the increasing level of education level compared to lower levels of education, this was associated with less treatment-related regret. Not surprisingly, development of health problems due to prostate cancer treatment at 6 months, as well as perception of treatment effectiveness compared with expectations at 5 years, as well as perception of treatment of adverse events compared with expectations at 5 years, all were significantly associated with more treatment regret with significant odds ratios, as listed here.

This is the baseline characteristics associated with patient-reported regret at 5 years, we can see here that increasing age, the utilization of a decision tool, as well as social support were all protective for treatment related regret. As we see here in the D'Amico risk category, high-risk patients with an odds ratio of 1.5 was also associated with treatment-related regret.

This Bland-Altman plot looks at the scores at 3 and 5 years. The goal of this analysis was to see if we assessed outcomes of 3 years, if they differed from 5 years. And the summary from this figure is that, generally, the regret scores were consistent over time, regardless of when we analyzed the data.

Several discussion points from this study. This is the first study in men with localized prostate cancer to find higher rates of regret among those who were actively treated, either with radical prostatectomy or radiotherapy, compared to those on surveillance after adjusting for baseline differences. This was modified by D'Amico risk categories. We saw that low and intermediate risk patients that had active treatment were more associated with likelihood of regret compared to active surveillance, with the reverse effect noted in patients with high-risk disease. Comparisons between surgery and radiotherapy showed higher regret with surgery. However, this differed significantly only in the patients with high-risk disease. So a disconnect between patient expectations and treatment outcomes for both efficacy and toxicity contributes more substantially to regret than patient-reported outcomes, treatment modality, or clinical pathological features, and thus, treatment-related regret may be more modifiable than other contributors to the survivorship experience of these patients, given its link to pretreatment expectations.

So in conclusion, these findings suggest that more than 1 in 10 patients with localized prostate cancer experience treatment-related regret. There's a disconnect between patient expectations and outcomes that appears to drive treatment-related regret to a greater extent than disease, characteristics, treatment modality, or patient reported functional outcomes. And finally, improve counseling at the time of diagnosis and before treatment, including the identification of patient values and priorities, may decrease regret among these patients. We thank you for your attention. We hope we enjoy this UroToday Journal Club discussion.