2743 - Clinical Predictors of Post-Treatment Quality of Life Following Chemoradiation for Gastrointestinal Malignancies

Purpose/Objective(s): Radiation therapy (RT) is a key component in the treatment of gastrointestinal (GI) malignancies in curative and palliative settings. Functional Assessment of Cancer Therapy (FACT) is a validated tool for measurement of quality of life (QOL) in oncologic care, however FACT-G trajectories during RT and early recovery are not well defined across mixed GI populations. This retrospective study aimed to evaluate predictors of changes in QOL before, during, and after RT, after adjusting for baseline scores and relevant clinical and demographic factors. The independent association between RT modality and FACT-G score was also assessed.

Materials/Methods: 168 patients who underwent chemoRT for a GI malignancy at a single institution were included. FACT surveys were administered at baseline (Time 0), during RT (Time 1), and approximately 3 months post-treatment (Time 2). Demographic and clinical variables of interest included age, race, sex, ECOG, EQD2 10, clinical stage, tumor site (colorectal, hepatobiliary, pancreatic, esophageal, general GI), RT modality (3D-CRT, IMRT, PBT, SBRT), and concurrent chemotherapy. The primary analytic endpoint was FACT-G at Time 2, modeled using linear regression with adjustment for baseline FACT-G and select variables. Patients who completed surveys at all three timepoints were included in the regression. Longitudinal within-patient changes were assessed through Wilcoxon signed rank tests between timepoints. Nonparametric testing (Kruskal-Wallis) was used to assess FACT-G scores stratified by modality at each timepoint.

Results: The cohort of 168 patients had a mean age of 67 years (55% male, 45% female) and was clinically heterogenous. Site distribution was skewed towards pancreatic (n=65) and esophageal (n=39) cancers. In paired longitudinal testing, FACT-G decreased significantly from Time 0 to Time 1 (p<0.001) but was not significantly different between Time 0 and 2, suggesting at least partial recovery in early follow up. In adjusted multivariable analyses (n=47), during-treatment FACT-G was the dominant predictor of Time 2 QOL, with Time 1 FACT-G positively associated with Time 2 FACT-G (p<0.05). Compared with CRC, patients with pancreatic malignancy had a higher FACT-G (+11.30 points, p<0.05), with esophageal cancer demonstrating a similar but non-significant effect. RT modality was not an independent predictor of FACT-G at Time 2, however PBT demonstrated a non-significant trend toward a higher FACT-G at Time 2 relative to IMRT (+6.41, p=0.220).

Conclusion: During-treatment QOL and tumor site strongly predicted post-treatment FACT-G, underscoring the importance of early QOL assessment to identify patients at risk for persistently poor QOL. PBT demonstrated a non-significant, directional trend toward improved FACT-G that encourages larger, prospective studies. Limitations of this study included a low compliance at Time 2. Further research is needed to better understand patterns of missingness in QOL studies.