Predicting Mortality with the Frailty Index in Older Adults Treated with Radiation

Summary: Older cancer patients, particularly those with head and neck (HN) and lung cancers, face high risks of treatment-related toxicity and mortality. Traditional metrics like age, stage and clinician-assessed performance status (PS) using the Eastern Cooperative Oncology Group (ECOG) scale often inadequately predict risk. The Cancer and Aging Resilience Evaluation (CARE) registry enrolls adults over 60 years old, includes a geriatric assessment (GA) and assigns a frailty index (FI), which has shown promise in predicting early mortality in gastrointestinal malignancies. This study aimed to evaluate whether the FI could similarly predict early mortality in older adults undergoing radiation treatment for HN or primary lung cancer. Patients enrolled in a single-institution’s prospective CARE registry at consultation were retrospectively identified. Those who received non-stereotactic radiation for HN or primary lung cancer were included. Frailty was assessed using a 44-item index based on deficit accumulation, categorizing patients as robust, pre-frail or frail. ECOG PS was recorded as 0-1 or 2. The primary outcome was early mortality- death within 18 months of completing radiation. Descriptive statistics were performed using Chi Square and point biserial tests (SPSS version 29.0.2.0). Of the 74 identified geriatrics, 77% were male with a median age of 68 years, and the median cancer was Stage 3. Radiation was to the HN in 65% and to the lungs in 35% of patients. 38% received chemotherapy and 15% received immunotherapy. 31 (42%) of patients were robust, 19 (26%) were pre-frail and 24 (32%) were frail. 49 (67%) were ECOG 0-1 and 24 (33%) were ECOG 2+. Within 18 months of radiation, 23 (31%) experienced early mortality. Neither ECOG, FI nor age showed a significant correlation with early mortality (p=0.68, p=0.11, p=0.92). However, cancer stage was significantly correlated (r=0.24, p=0.04).The positive predictive values (PPVs) of “frail and pre-frail” and ECOG 2+ for early mortality were 0.40 and 0.33 respectively. The negative predictive values (NPVs) of “robust” and ECOG 0-1 were 0.81 and 0.71 respectively. The GA was designed to identify overlooked impairments in traditional oncologic assessments and better predict toxicity and mortality. While FI demonstrated a significant association with early mortality in gastrointestinal malignancies with better PPV and NPV than ECOG, our study did not find a significant correlation between FI or ECOG and early mortality in HN and lung cancers. Cancer stage alone showed a significant correlation with early mortality. FI numerically showed a slightly better PPV and NPV than ECOG in this study. These findings suggest that the heterogeneity of cancer types and their treatments do not allow for universal application of ECOG and FI. This underscores the need for further research to refine assessment tools and predictive models in this population and ensure treatment plans accommodate individuals. Katelyn Ragland, MD (Presenting Author) - University of Alabama at Birmingham; Andrew McDonald, MD (Co-Author) - University of Alabama at Birmingham