Understanding Breakthrough Cancer Pain Through Bayesian Network Analysis: Interactions Between Opioid Therapy, Cancer Severity, and Pain Phenotypes

Beyond its immediate physical burden, inadequately controlled cancer pain can substantially impair functional status, psychological well-being, treatment adherence, and health-related quality of life. Breakthrough cancer pain (BTcP) represents a particularly challenging manifestation of cancer-related pain. Its clinical presentation is heterogeneous, with considerable variability in episode frequency, intensity, duration, and time to peak severity. Disease burden, metastatic involvement, ongoing anticancer therapies, baseline pain characteristics, and individual clinical conditions may contribute to substantial variability in pain presentation and therapeutic requirements. Bayesian networks provide an alternative framework by combining a directed acyclic graph, which represents the dependency structure among variables, with conditional probability distributions that quantify their relationships. Using this approach, we examined how these relationships varied according to BTcP status at baseline, anticancer treatment, metastatic disease severity, and chronic pain phenotype. Rather than estimating causal treatment effects, the analysis was designed to characterize conditional dependencies and explore clinically relevant patterns that may contribute to a more individualized understanding of BTcP.