Abstract / Summary
Salivary gland carcinoma (SGC) is a rare and heterogeneous group of head and neck malignancies. Prognosis is influenced by multiple factors, including age, tumor behavior, and comorbidities. Latent variable and clustering methods have increasingly been used to capture disease heterogeneity and improve risk stratification. Herein, we identified SGC patient characteristics and determined treatment patterns and survival outcomes using latent class analysis (LCA). This retrospective cohort study analyzed the medical records of 107 patients diagnosed with SGC at King Faisal Specialist Hospital and Research Center in Riyadh, Saudi Arabia, from 2014 to 2022. Collected data included demographic information, tumor characteristics, treatment modalities (surgery, radiotherapy, chemotherapy), smoking status, and mortality. Categorical variables were reported as counts (%) and continuous variables as mean (SD) or median (IQR), as appropriate. LCA was performed to identify distinct multidimensional clinical phenotypes based on demographic, tumor, comorbidity, and treatment variables. Overall survival was estimated using the Kaplan–Meier method across latent clinical phenotypes. Statistical analyses were conducted in Python, and significance was determined at a 2-sided α level of 0.05. The cohort comprised 107 patients with a mean age of 45.9 ± 20.4 years, and 59.8% were male. The parotid gland was the most common site of tumor origin, and mucoepidermoid carcinoma accounted for 21.5% of cases. Approximately 75% of patients underwent surgery, 45.8% received radiotherapy, and 10.3% received chemotherapy. LCA identified three clinically distinct phenotypes, characterized by different conditional probabilities across demographic, tumor characteristics, comorbidity, and treatment variables. These were the low-risk, surgery-dominant class (Class 1; n = 49), the locally advanced, multimodal class (Class 2; n = 39), and the advanced, high-burden class (Class 3; n = 19). Mortality rates were 34.7% in Class 1, 23.1% in Class 2, and 21.1% in Class 3. This study provides valuable insights into the characteristics of patients with SGC and suggests that LCA may help identify clinically meaningful patient subgroups with distinct survival patterns. However, larger cohorts, more comprehensive clinical data, and longer follow-up are needed to validate the utility of LCA for prognostic stratification in SGC.