Prognostic value of time-varying patient-reported symptoms and quality of life in cancer patients receiving chemotherapy.

Abstract

11044 Background: Prior research has demonstrated that pre-treatment patient-reported outcomes (PROs) predict post-treatment survival. Symptom burden, physical function and overall wellbeing are time-varying concepts, so making survival predictions using pre-treatment data may have limited clinical actionability. The prognostic value of time-varying patient-reported symptom burden, physical function (PF) and wellbeing (WB) are not well characterized. Methods: Six US-based cancer centers collaborated to develop and deploy an EHR-integrated ePRO monitoring system to facilitate active symptom management. Patients receiving chemotherapy for a gastrointestinal (GI), gynecologic (GYN), or thoracic cancer were asked to report on 12 common symptoms using PRO-CTCAE items and 2 quality of life items twice weekly for up to 180 days after starting chemotherapy. Herein, we employed Cox regression to model 180-day survival using time-varying overall symptom burden, calculated as the sum of 12 symptom scores, PF and WB. Multivariable models also included demographics, clinical variables, treatment goal, and comorbidities. Results: The cohort included 3,999 patients (45% GI, 23% GYN, 32% thoracic) who submitted 42,254 symptom reports and had 481 deaths within 180 days. Median age was 66 (IQR 15), 59% female, 86% White, 42% Medicare, 5% Medicaid, 50% retired, 10% disabled. The mean symptom burden score was 7.57 (SD 4.66, min 0, max 34); severe deficit in PF and WB were reported in 13.2% and 7.8% of questionnaires, respectively. In bivariate analyses, greater symptom burden, more deficits in PF and WB, increasing age, female sex, and palliative treatment goal predicted inferior survival. After adjusting for other covariates, symptom burden, moderate and severe PF deficits, and severe WB deficits predicted inferior 180-day survival (Table). Conclusions: There is a significant inverse linear relationship between time-varying symptom burden and survival for patients receiving chemotherapy. Among chemotherapy patients, assessments that elicit symptom burden as well as PF and WB may augment the prognostic value and usefulness of ePRO tracking systems. Clinical trial information: NCT03850912 . Unadjusted Multivariable Characteristic HR 95% CI HR 95% CI Symptom Burden (Linear) 1.13 1.11, 1.15 1.08 1.06, 1.10 Physical Function Deficit (ref = no deficit) Mild 1.65 1.13, 2.41 1.12 0.74, 1.69 Moderate 3.82 2.64, 5.53 1.89 1.23, 2.92 Severe 9.55 6.74, 13.5 3.61 2.35, 5.55 Overall Wellbeing Deficit (ref = no deficit) Mild 1.98 1.27, 3.07 1.30 0.81, 2.10 Moderate 4.18 2.74, 6.36 1.51 0.92, 2.48 Severe 9.97 6.49, 15.3 1.75 1.02, 3.00 Age (Linear) 1.02 1.01, 1.03 1.03 1.01, 1.04 Sex (Female, ref = male) 0.57 0.49, 0.68 0.61 0.51, 0.74 Treatment Goal (ref =curative) Palliative 4.27 3.22, 5.66 3.66 2.71, 4.95 Control/Other/Unknown 2.66 1.99, 3.55 2.52 1.85, 3.42