Back to Annual Meeting Program

Background: Venous resection may be required to achieve complete resection of pancreatic cancers (PC). We have previously shown that histologic invasion of the superior mesenteric vein-portal vein (SMV-PV) is associated with poor prognosis following resection. Using high-definition multidetector computed tomography (CT), we sought to evaluate the ability of two commonly-used sets of radiographic criteria to predict the need for SMV-PV resection at pancreatectomy and the histologic presence of SMV-PV invasion.
Methods: All patients who underwent pancreaticoduodenectomy for PC between 2004 and 2011 at the authors' institution were identified. Preoperative pancreatic protocol CT images were re-reviewed to characterize the interface between the tumor and SMV-PV (no interface, abutment [≤180 degrees], encasement [>180 degrees], occlusion) and the appearance of the SMV-PV using Ishikawa criteria (normal, smooth shift, unilateral narrowing, bilateral narrowing without collaterals, bilateral narrowing with collaterals). Findings were correlated to the need for venous resection at pancreatectomy and to the presence of histologic venous invasion.
Results: 266 patients underwent pancreaticoduodenectomy and met inclusion criteria, of whom 99 required concomitant resection of the SMV-PV. Greatest sensitivity for predicting SMV-PV resection was achieved by an interface threshold of abutment (sensitivity 91.9%, negative predictive value 87.9%), whereas excellent specificity was reached with a threshold of encasement (97.6%, positive predictive value 89.7%). Among patients who underwent venous resection, vessel encasement was associated with a 78.3% rate of histologic SMV-PV invasion; this rate increased to 90% when the vein was occluded. The Ishikawa system, while more detailed, offered no advantage in predicting the need for SMV-PV resection and was less accurate in predicting histologic venous invasion. Subset analyses performed for patients who received neoadjuvant chemoradiation and for those who did not yielded similar findings.
Conclusions: A simple radiographic classification system that categorizes the extent of the tumor-SMV-PV interface accurately predicts the need for SMV-PV resection at pancreatectomy, and correlates with the pathologic involvement of the resected vein. To assist in treatment planning, a standardized description of this anatomic relationship should be routinely performed for patients with borderline resectable tumors.
Correlation of Radiographic Assessment of Tumor-Vessel Relationship to Surgical and Pathologic Outcomes

	Tumor-Vessel Interface*				Vessel Appearance**
Radiographic-Surgical Correlation
CT cut-off	≥ 0	≥ 1	≥ 2	3	≥ 1	≥ 2	≥ 3	≥ 4	5
Number of patients	266	200	39	11	266	166	96	30	29
Sensitivity	N/A	91.9	35.4	10.1	N/ A	85.9	66.7	27.3	27.3
Specificity	N/A	34.7	97.6	99.4	N/A	51.5	82.0	98.2	98.8
PPV	N/A	45.5	89.7	90.9	N/A	51.2	68.8	90.0	93.1
NPV	N/A	87.9	71.8	65.1	N/A	86.0	80.6	69.5	69.6
Accuracy	N/A	56.0	74.4	66.2	N/A	71.8	76.3	71.8	72.2
Radiographic-Pathologic Correlation
CT value	0	1	2	3	1	2	3	4	5
Number of venous resection	8	56	25	10	14	19	39	0	27
Histologic venous invasion (%)	33.3	64.6	78.3	90.0	54.5	52.9	69.7	--	84.6

CT - computed tomography; PPV - positive predictive value; NPV - negative predictive value. *Tumor-SMV-PV interface scale - 0: no interface, 1: abutment (≤ 180°), 2: encasement (> 180°), 3: occlusion. **SMV-PV appearance based on the Ishikawa system - 1: normal, 2: smooth shift, 3: unilateral narrowing, 4: bilateral narrowing without collaterals, 5: bilateral narrowing or occlusion with collaterals.