Reliance on topographical coding alone may overestimate survival in pancreatic ductal adenocarcinoma

In epidemiological and outcomes research, "pancreatic cancer" is frequently used interchangeably with pancreatic ductal adenocarcinoma (PDAC), the most common and most lethal pancreatic cancer subtype. However, relying solely on an anatomic site code may capture a mix of pancreatic tumor types with better prognoses than PDAC, leading to inflated survival estimates.

We used the Surveillance, Epidemiology, and End Results (SEER) registry to compare data derived from site-only case ascertainment (C25.x, excluding C25.4) versus histologically confirmed PDAC using ICD-O-3 morphology codes. Only microscopically confirmed cases were included. We quantified differences in patient demographics, tumor characteristics, and overall survival between the two case-selection strategies. Survival estimates were generated using Kaplan-Meier methods and multivariable Cox regression models.

A total of 283,465 patients with C25 tumors (excluding C25.4) were identified. Based on morphology coding, 233,421 (82.3%) tumors were classified as PDAC and 50,044 (17.7%) as non-PDAC histology. The 5-year survival rate in the C25 cohort (excluding C25.4) was 12.1%, whereas survival among PDAC patients alone was 6.0%. For surgically resected patients, the 5-year survival rate in the C25 cohort (excluding C25.4) was 35.3%, compared to 21.0% for those with PDAC histology. In multivariable Cox regression analysis, PDAC histology was significantly associated with reduced overall survival (HR 2.57, 95% CI 2.54-2-61, p < 0.001).

Case definitions based on the C25 site code alone lead to overestimation of true PDAC survival. Histology-based selection is necessary for accurate research on PDAC, underscoring the importance of precise tumor classification in registry-based research.