Yin-Chen Tsai1,2, Jou-Ho Shih3, Hsuen-En Hwang1,2, Nai-Chi Chiu1,2, Rheun-Chuan Lee1,2, Hsiou-Shan Tseng2,4, Chien-An Liu5,6. 1. Department of Radiology, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, Taipei, Taiwan. 2. School of Medicine, National Yang Ming Chiao Tung University, No. 155, Sec. 2, Linong Street, Taipei, 112, Taiwan. 3. Department of Genetics, Stanford University School of Medicine, Stanford University, Stanford, CA, 94305, USA. 4. Department of Medical Imaging, Cheng Hsin General Hospital, No. 45, Cheng Hsin St., Beitou District, Taipei, Taiwan. 5. Department of Radiology, Taipei Veterans General Hospital, No. 201, Sec. 2, Shipai Rd., Beitou District, Taipei, Taiwan. caliu@vghtpe.gov.tw. 6. School of Medicine, National Yang Ming Chiao Tung University, No. 155, Sec. 2, Linong Street, Taipei, 112, Taiwan. caliu@vghtpe.gov.tw.
Abstract
OBJECTIVE: To evaluate whether parenchyma-to-lipiodol ratio (PLR) and lesion-to-lipiodol ratio (LLR) on C-arm cone-beam computed tomography (CBCT) can predict 1-year tumor response in patients with hepatocellular carcinoma (HCC) treated with conventional transcatheter arterial chemoembolization (cTACE). METHODS: This retrospective analysis included 221 HCC target lesions within up-to-seven criteria in 80 patients who underwent cTACE with arterial-phase CBCT and unenhanced CBCT after cTACE from 2015 to 2018. PLR and LLR of every tumor slice were obtained through mean density division of liver parenchyma and tumor enhancement with intratumoral lipiodol deposition. The cutoff values (COVs) of maximal PLR and LLR of every tumor were analyzed using Youden's index. The reliability of COV, correlations between the related parameters, and 1-year progression were assessed through interobserver agreement and multivariate analysis. COV validity was verified using the chi-square test and Cramer's V coefficient (V) in the validation cohort. RESULTS: Standard COVs of PLR and LLR were 0.149 and 1.4872, respectively. Interobserver agreement of COV for PLR and LLR was near perfect (kappa > 0.9). Multivariate analysis suggested that COV of PLR is an independent predictor (odds ratio = 1.23532×1014, p = 4.37×10-7). COV of PLR showed strong consistency, correlation with 1-year progression in prediction model (V = 0.829-0.776; p < 0.0001), and presented as an effective predictor in the validation cohort (V = 0.766; p < 0.0001). CONCLUSION: The COV of PLR (0.149) assessed through immediate post-embolization CBCT is an objective, effective, and approachable predictor of 1-year HCC progression after cTACE. KEY POINTS: • The maximal PLR value indicates the least lipiodol-distributed region in an HCC tumor. The maximal LLR value indicates the least lipiodol-deposited region in the tumor due to incomplete lipiodol delivery. PLR and LLR are concepts like signal-to-noise ratio to characterize the lipiodol retention pattern objectively to predict 1-year tumor progression immediately without any quantification software for 3D image analysis immediately after cTACE treatment. • COV of PLR can facilitate the early prediction of tumor progression/recurrence and indicate the section of embolized HCC, providing the operator's good targets for sequential cTACE or combined ablation. • The validation cohort in our study verified standard COVs of PLR and LLR. The validation process was more convincing and delicate than that of previous retrospective studies.
OBJECTIVE: To evaluate whether parenchyma-to-lipiodol ratio (PLR) and lesion-to-lipiodol ratio (LLR) on C-arm cone-beam computed tomography (CBCT) can predict 1-year tumor response in patients with hepatocellular carcinoma (HCC) treated with conventional transcatheter arterial chemoembolization (cTACE). METHODS: This retrospective analysis included 221 HCC target lesions within up-to-seven criteria in 80 patients who underwent cTACE with arterial-phase CBCT and unenhanced CBCT after cTACE from 2015 to 2018. PLR and LLR of every tumor slice were obtained through mean density division of liver parenchyma and tumor enhancement with intratumoral lipiodol deposition. The cutoff values (COVs) of maximal PLR and LLR of every tumor were analyzed using Youden's index. The reliability of COV, correlations between the related parameters, and 1-year progression were assessed through interobserver agreement and multivariate analysis. COV validity was verified using the chi-square test and Cramer's V coefficient (V) in the validation cohort. RESULTS: Standard COVs of PLR and LLR were 0.149 and 1.4872, respectively. Interobserver agreement of COV for PLR and LLR was near perfect (kappa > 0.9). Multivariate analysis suggested that COV of PLR is an independent predictor (odds ratio = 1.23532×1014, p = 4.37×10-7). COV of PLR showed strong consistency, correlation with 1-year progression in prediction model (V = 0.829-0.776; p < 0.0001), and presented as an effective predictor in the validation cohort (V = 0.766; p < 0.0001). CONCLUSION: The COV of PLR (0.149) assessed through immediate post-embolization CBCT is an objective, effective, and approachable predictor of 1-year HCC progression after cTACE. KEY POINTS: • The maximal PLR value indicates the least lipiodol-distributed region in an HCC tumor. The maximal LLR value indicates the least lipiodol-deposited region in the tumor due to incomplete lipiodol delivery. PLR and LLR are concepts like signal-to-noise ratio to characterize the lipiodol retention pattern objectively to predict 1-year tumor progression immediately without any quantification software for 3D image analysis immediately after cTACE treatment. • COV of PLR can facilitate the early prediction of tumor progression/recurrence and indicate the section of embolized HCC, providing the operator's good targets for sequential cTACE or combined ablation. • The validation cohort in our study verified standard COVs of PLR and LLR. The validation process was more convincing and delicate than that of previous retrospective studies.