Radiologic partial response poorly predicts pathologic complete response in hepatocellular carcinoma after conversion therapy

Abstract

BACKGROUND

Conversion therapy has increased the proportion of patients with initially unresectable hepatocellular carcinoma (HCC) who become eligible for surgical resection. Radiologic response criteria, including Response Evaluation Criteria in Solid Tumours (RECIST) version 1.1 and modified RECIST (mRECIST), are widely used to evaluate treatment efficacy. However, their ability to predict pathologic complete response (pCR) remains uncertain.

AIM

To evaluate RECIST 1.1 and mRECIST for predicting pathological complete response and radiologic-pathologic discordance after HCC conversion therapy surgery.

METHODS

We retrospectively analysed 66 patients with initially unresectable HCC underwent curative-intent resection after conversion therapy between January 2019 and February 2024. Radiologic response was evaluated using RECIST 1.1 and mRECIST based on the last preoperative contrast-enhanced computed tomography or magnetic resonance imaging scan. pCR was defined as the absence of viable tumour cells in the resected specimens. The predictive performance of radiologic partial response (PR) for pCR was evaluated using sensitivity, specificity, positive predictive value (PPV), negative predictive value, and Cohen’s kappa.

RESULTS

According to mRECIST, 55 patients were classified as having PR and 11 as having stable disease (SD), whereas RECIST 1.1 categorised 37 patients as having PR and 29 as having SD. Overall, 20 patients (30.3%) achieved pCR. Under the mRECIST criteria, all patients with pCR were classified as having PR, yielding a sensitivity of 100% but a low specificity of 23.9% and a PPV of 36.4%. For RECIST 1.1, the sensitivity and specificity were 60.0% and 45.7%, respectively. Agreement between radiologic PR and pCR was poor for both criteria (Cohen’s κ = 0.16 for mRECIST and 0.046 for RECIST 1.1). These findings indicate substantial discordance between radiologic PR and pathological response.

CONCLUSION

Radiologic PR does not reliably predict pCR after HCC conversion therapy. Imaging alone should not guide surgical timing; integrating clinical and biological markers may improve preoperative response assessment.

Keywords: Hepatocellular carcinoma; Conversion therapy; Pathologic complete response; Radiologic-pathologic correlation; Surgical resection

Core Tip: Radiologic response is widely used to assess treatment efficacy in hepatocellular carcinoma after conversion therapy, yet its correlation with pathological response remains uncertain. In this study, radiologic partial response assessed by both Response Evaluation Criteria in Solid Tumours (RECIST) version 1.1 and modified RECIST showed poor concordance with pathologic complete response (pCR). Although all patients achieving pCR were classified as responders by modified RECIST, most radiologic responders still had residual viable tumor. These findings underscore the limitations of imaging-based response assessment and support the integration of additional biomarkers to improve preoperative evaluation and treatment planning.