Inter-reader agreement of RECIST and mRECIST criteria for assessing response to transarterial chemoembolization in hepatocellular carcinoma.

Abstract

Objectives: To evaluate the reproducibilities of Response Evaluation Criteria in Solid Tumors (RECIST) and modified RECIST (mRECIST) for hepatocellular carcinoma (HCC) patients treated with transarterial chemoembolization (TACE) using contrast-enhanced computed tomography (CT).

Methods: This retrospective study included 105 consecutive patients with confirmed HCC recruited from November 2002 to June 2012. The study protocol has been pre-registered at ( https://osf.io/nxg4q/ ) on the Open Science Framework (OSF) platform. Patients with pre-procedural and follow-up CT scans who had solely received TACE were included. The tumor response evaluation to TACE was conducted using RECIST 1.1 and mRECIST guidelines. Three experienced board-certified abdominal radiologists interpreted CT scans.

Results: For pre-procedure CT, the agreement was more excellent when using RECIST guidelines with a "marginally significant" p-value of 0.056. This trend continued for post-procedural CT scans, with RECIST again showing significantly higher agreement with a p-value of 0.001. When evaluating the four categories of response, Gwet's coefficient was 0.90 (CI = 0.83 to 0.97) for RECIST and 0.80 (CI = 0.63 to 0.90) for mRECIST. Conversely, the Fleiss Kappa analysis demonstrated a higher agreement for the mRECIST guideline. There was an insignificant difference in RECIST and mRECIST guidelines inter-reader agreement when categorizing the tumor response with a p-value of 0.101.

Conclusion: Both guidelines' inter-reader reproducibility in assessing tumor response through CT after the TACE procedure was excellent, with RECIST's reproducibility being very slightly better.