Table 4.

Sensitivity and specificity for predicting vascular invasion by tumor using 50% tumor contiguity with the circumference of a vessel as threshold

Authors	Sensitivity (%)	Specificity (%)	Comment/Proposed modification(s)
Lu et al. [17]	84	98	In this study, vessel constriction was considered a sign of vascular invasion, regardless of the degree of contact between vessel and tumor. Tumor contiguity with 50% of the circumference of a vessel was used as a criterion only in cases without vessel constriction.
O’Malley et al. [19]	46	99	In this study, the grading system proposed by Lu et al. was modified to exclude vessel constriction as a sign of vascular invasion. This may account for the lower sensitivity reported. Raising the threshold to 75% decreased sensitivity to 38%, but increased specificity to 100%.
Nakayama et al. [20]—veins only	71	86	The criteria used by Lu et al. were found to be optimal.
Nakayama et al. [20]—arteries only	78	79	Raising the threshold to 75% decreased sensitivity to 61%, but increased specificity to 92%.
Li et al. [21]—veins only	49	100	Including venous obliteration, venous stenosis, or teardrop SMV as signs of venous invasion increased sensitivity to 92% while specificity remained 100%. Note that these criteria are identical to those proposed by Lu et al.
Li et al. [21]—arteries only	97	91	Using a combination of tumor involvement of >50% of the circumference of an artery and irregularity of the arterial wall or stenosis of the artery reduced sensitivity to 79%, but increased specificity to 99%.