Impact of baseline symptom burden as assessed by patient-reported outcomes (PROs) on overall survival (OS) of patients with metastatic cancer.

Abstract

12020 Background: Patients with metastatic cancer experience variable symptom burden, but serial symptom assessments using PROs may be challenging to implement in routine clinical practices. We aimed to determine if a single measurement of symptom burden at the time of metastatic diagnosis is associated with survival. Methods: We examined prospectively collected baseline PROs of patients newly diagnosed with metastatic breast, lung, colorectal, or prostate cancer using the revised Edmonton Symptom Assessment System (ESASr) questionnaire from a large province (Alberta, Canada) between 2016 and 2019. The ESASr was categorized into physical (PH), psychosocial (PS), and total symptom (TS) domains whereby scores were classified as mild (0-3), moderate (4-6), or severe (7-10). Multivariable Cox proportional hazards models were constructed to evaluate the effect of baseline symptom scores on OS. Results: We identified 1,315 patients, of whom 57% were men and median age was 66 (IQR, 27-93) years. There were 180, 601, 240, and 294 patients with breast, lung, colorectal, and prostate cancer, respectively. Approximately one-quarter of all patients reported moderate to severe PH, PS, and TS scores, with lung cancer patients experiencing the highest symptom intensity across all domains ( P<0.0001). While age did not affect symptom scores, women were more likely to report severe PH, PS, and TS scores as compared to men ( P=0.02, 0.002, and 0.007, respectively). On multivariable Cox regression analysis, older age (HR 1.02, 95% CI, 1.02-1.03, P<0.0001) and female sex (HR 1.67, 95% CI, 1.39-1.99, P<0.0001) were predictive of worse OS as were severe baseline PH and TS scores (see Table) . However, baseline PS scores were not related to OS. Conclusions: A single assessment of baseline symptom burden using the ESASr in patients with metastatic cancer has significant prognostic value. This may represent a feasible first step toward routine collection of PROs in real-world settings where serial symptom measurements can be challenging to implement. [Table: see text]