Radiology estimates of viable tumor percentage in hepatocellular carcinoma ablation cavities correlate poorly with pathology assessment.

Context: No study has evaluated radiology/pathology correlation of percentage viable tumor (PVT) estimates in ablated hepatocellular carcinoma (HCC) to examine the reliability of radiologic estimates.

Objective: To determine how well interdisciplinary PVT estimates correlate and identify pathologic factors that influence this correlation.

Design: Pathologists and radiologists established blinded PVT estimates in 22 HCC ablation cavities. Paired sample t tests examined the differences between the interdisciplinary estimates.

Results: Fifteen cavities had pathologic viable tumor (VT) (68%) and 6 had radiographic VT (22%). Radiology's sensitivity for detecting VT was 40% and the specificity was 100%. Pathology detected significantly more VT than radiology (pathology mean = 22.3% versus radiology mean = 2.6%; P = .005). Five cavities had tumor growth in a discontinuous rim pattern, 7 in a nodular pattern, and 3 in a solid pattern. Radiology did not detect VT in cavities with a discontinuous rim pattern (sensitivity = 0%) but did detect VT in 3 cavities with a nodular pattern (sensitivity = 43%), and in all cavities with a solid pattern (sensitivity = 100%). There was no significant difference in PVT estimates in cavities 3.5 cm or larger (P = .07), but there was a significant difference in cavities smaller than 3.5 cm (P = .01).

Conclusion: This study clarifies that the risk of underestimation by imaging is greatest in small lesions (<3.5 cm), though the sensitivity of detection depends primarily on the tumor growth pattern within the cavity. This underestimation raises the question of whether basing treatment decisions on a radiologic impression of complete ablation is valid.