CT-based vascular invasion in pancreatic ductal adenocarcinoma compared with intraoperative and histological findings

To assess the concordance of CT-based radiological, surgical, and histopathological determination of vascular invasion in pancreatic ductal adenocarcinoma (PDAC).

This retrospective single-center study included 103 treatment-naive PDAC patients (median age 68 years, male 61%) with arterial-/portal venous-contrast enhanced CT undergoing successful primary resection. Three radiologists independently assessed tumor-vessel contact according to NCCN criteria. Radiological vascular invasion was operationally defined as > 180° or contour irregularity versus no invasion defined as no contact or contact ≤ 180° without contour irregularity, to compare to binary intraoperative and histological reference. Intraoperatively, venous and arterial invasion were defined as tumor adherence to or invasion of the vessel wall. Histopathologically, venous invasion (V0/1) was defined according to UICC-TNM. Additionally, as major arteries were not resected, microscopically seen periarterial perineural invasion (Pn_art) served as a surrogate marker for arterial invasion (Pn0/Pn1), intermodal concordance of CT-to-surgery ≤/ > 4 weeks) was compared, Cohen's kappa and McNemar's-tests were used.

Median CT-to-surgery was 17 days (81% within 4 weeks). CT-based interobserver agreement was κ = 0.8 for venous, κ = 0.6 for arterial invasion determination. Intraoperatively, 35% patients showed venous, 5% arterial involvement. Histopathologically, 14% showed venous invasion, 3% Pn_art1-status. CT-surgery venous invasion concordance was 75%, arterial 96%. CT-histopathology venous invasion concordance was 81%, 96% for Pn_art-status. CT-to-surgery interval ≤4 weeks showed higher CT-surgery (77% vs 65%) and CT-histopathology venous invasion concordance (84% vs 60%) compared to > 4 weeks interval.

NCCN-derived, threshold-based radiological venous invasion determination was similar to binary intraoperative and histopathological findings and decreased with longer CT-to-surgery intervals.

NCCN-derived, threshold-based CT venous-invasion determination in PDAC is concordant with binary intraoperative and histopathological invasion, but depends on the selected threshold and decreases with longer CT-to-surgery interval, underlining the clinical need for standardized staging intervals in patients undergoing surgery.

Comparison between CT vascular assessment and surgical/histopathological findings in PDAC remains highly complex, and current guidelines lack CT-to-surgery interval recommendations. Threshold-based CT venous invasion agreed moderately with surgery and strongly with histopathology, while CT-based arterial status correlated strongly with surgery, albeit in markedly low overall incidence. CT-vascular assessment precision decreases with longer time intervals, showing the need for establishing timeframes for preoperative imaging.