J. J. Thomas, C. Claunch, A. Arenas, P. M. Jhaveri, D. A. Hamstra, H. Mekdash, C. H. Chapman, and S. Sharma; Department of Radiation Oncology, Dan L. Duncan Comprehensive Cancer Center, Baylor College of Medicine, Houston, TX
Purpose/Objective(s): Our study aims to identify patient subgroups with metastatic oropharynx cancers who may derive the most benefit from escalated therapy, using demographics, disease characteristics, and treatment patterns. Materials/
Methods: The National Cancer Database (NCDB) was queried for patients diagnosed with metastatic squamous oropharynx cancers between 2004 and 2019. Variables including age, gender, race, Charlson-Deyo comorbidity index (CCI), high-risk human papillomavirus (HPV) status, location of metastases, number of metastatic sites, and type of treatment received were evaluated. Patients with unknown HPV status or missing data were removed from analysis. Descriptive statistics were generated and both univariate and multivariate analyses (Cox regression) for median overall survival (OS) were performed. Utilizing the classification and regression tree (CART) algorithm, we developed a decision tree model based on the multivariate analysis results to examine pathways to survival based on HPV status, with a 70-30 training-test split. For post-model development, we calculated the median survival time at parent nodes within the decision tree. Results: After excluding cases with missing data, 1564 remained, with 1349 being male (86.3%) and 1306 being white (83.5%). 1176 (75.2%) had a CCI of 0 and 897 (57.4%) were HPV-positive (HPV-P). Lung metastases were seen in 761 (48.7%) followed by involvement of distant lymph nodes (468, 29.9%), and bone (455, 29.1%). 1068 (68.3%) had a single site of metastasis. 191 (12.2%) received no therapy, 316 (20.2%) received systemic treatment alone, 155 (9.9%) received local treatment alone, and 827 (52.9%) received combined systemic and local treatments (Combo). HPV-P patients who received Combo had an OS of 27.7 months. In the multivariate models, HPV status, CCI, tumor size, time to initiation of treatment, number of metastatic sites, and treatment type were significant predictors of survival. The decision tree nodes with the greatest impact on survival were HPV status, treatment type, and number of metastatic sites. Patients who were HPV-P, received Combo, had one metastatic site of disease, and had involvement of distant lymph nodes had a median OS of 53.7 months. HPV-P patients with non-distant lymph node single site disease who received Combo had a median OS of 28.0 months. The remainder of patient nodes had median OS under 14 months. Conclusion: Significant survival differences are observed among metastatic oropharynx cancer patients, with HPV status, metastatic burden, and treatment type playing crucial roles. Aggressive combined modality treatment could greatly benefit certain HPV-positive patient subgroups. Utilizing decision tree methodology facilitated the creation of an interpretable model for outcome prediction, enabling personalized clinical decision-making for our study population.
