Prospective evaluation of comprehensive geriatric assessments in multidisciplinary bladder cancer care and implications for personalized vulnerability phenotyping

Abstract

Purpose: Frailty predicts adverse outcomes in bladder cancer (BC). Current guidelines endorse completion of Comprehensive Geriatric Assessments (CGAs) in older adults prior to treatment election to objectively measure frailty, however, these are rarely performed in urologic practice due to inadequate resources. We hypothesized CGA implementation would be feasible and identify multifaceted vulnerabilities beyond standard risk assessments in a multidisciplinary BC clinic and developed a novel method to visualize "vulnerability phenotypes" to guide supportive interventions.

Methods: Adults with BC were prospectively enrolled (June, 2020-July, 2021). Initially, patients underwent standard of care (SOC) risk assessment (N = 27). Subsequently, patients completed CGAs augmented with body composition assessments (N = 67). CGA completion time, rates, and patient-reported burden were assessed. Interdependence of CGA domains were quantified using Spearman correlation coefficients and compared decisional conflict and regret between arms. Vulnerability phenotypes were visualized using Spider Plots, generated in R. Clinical and survival associations with CGAs were evaluated using Cox proportional hazards models.

Results: 94 patients were enrolled with a median age of 72 years. Instrument completion in the CGA cohort was 79% to 100%. 91% of patients reported CGA completion was at most minimally burdensome. CGAs identified vulnerabilities including 31% vulnerable-to-moderately frail, 21% with mild-to-severe depression, 3% with mild-moderate dementia, and 40% at risk for malnutrition-malnourished. Frailty measures across instruments were weakly correlated (rho <0.4). In this heterogeneous cohort, vulnerability domains were not significantly associated with decisional conflict/regret, survival, nor complication rates after treatment. A novel Spider Plot tool is proposed to facilitate communication of the dominant vulnerability-driving individual risks.

Conclusions: CGAs can be successfully incorporated into uro-oncology practice with low perceived burden, identifying key vulnerabilities with implications for clinical care. Weak correlations across instruments support the value of gathering information across discrete domains. We present a novel approach to visually characterize personalized vulnerability phenotypes.