Intraprocedural 3D Quantification of Lipiodol Deposition on Cone-Beam CT Predicts Tumor Response After Transarterial Chemoembolization in Patients with Hepatocellular Carcinoma

Zhijun Wang, Rongxin Chen, Rafael Duran, Yan Zhao, Gayane Yenokyan, Julius Chapiro, Rüdiger Schernthaner, Alessandro Radaelli, MingDe Lin, Jean-François Geschwind, Zhijun Wang, Rongxin Chen, Rafael Duran, Yan Zhao, Gayane Yenokyan, Julius Chapiro, Rüdiger Schernthaner, Alessandro Radaelli, MingDe Lin, Jean-François Geschwind

Abstract

Purpose: To evaluate whether intraprocedural 3D quantification of Lipiodol deposition on cone-beam computed tomography (CBCT) can predict tumor response on follow-up contrast-enhanced magnetic resonance imaging (CE-MRI) in patients with hepatocellular carcinoma (HCC) treated with conventional transarterial chemoembolization (cTACE).

Materials and methods: This IRB approved, retrospective analysis included 36 patients with 51 HCC target lesions, who underwent cTACE with CBCT. CE-MRI was acquired at baseline and 1 month after cTACE. Overall tumor volumes as well as intratumoral Lipiodol volumes on CBCT were measured and compared with the overall and necrotic (non-enhancing) tumor volumes on CE-MRI using the paired student's t test. Tumor response on CE-MRI was assessed using modified response evaluation criteria in solid tumors (mRECIST). A linear regression model was used to correlate tumor volumes, Lipiodol volumes, and the percentage of Lipiodol deposition on CBCT with the corresponding parameters on CE-MRI. Nonparametric spearman rank-order correlation and trend test were used to correlate the percentage of Lipiodol deposition in the tumor with tumor response.

Result: A strong correlation between overall tumor volumes on CBCT and CE-MRI was observed (R(2) = 0.986). In addition, a strong correlation was obtained between the volume of Lipiodol deposition on CBCT and tumor necrosis (in cm(3)) on CE-MRI (R(2) = 0.960), and between the percentage of Lipiodol deposition and tumor necrosis (R(2) = 0.979). Importantly, the extent of Lipiodol deposition (in percentage of total tumor volume) correlated strongly with tumor response on CE-MRI (Spearman rho = 0.84, p < 0.001).

Conclusions: Intraprocedural 3D quantification of Lipiodol deposition on CBCT can be used to predict tumor response on follow-up CE-MRI.

Keywords: Cone-beam computed tomography; Hepatocellular carcinoma; Quantitative; Three dimensional; Transcatheter arterial chemoembolization; Tumor response.

Figures

Figure 1
Figure 1
3D volumetric semi-automatic evaluation of Lipiodol deposition (Complete Response according to mRECIST criteria) in HCC on a representative case. Segmentation of the tumor (red circle) on CBCT at corresponding slice level as on MR (A, B). 3D segmentation volume rendering on the same slice (C, D). Quantitative color map of Lipiodol deposition on CBCT (E) and tumor viability on follow-up MRI (F). The box represents the location of the background ROI. For the anterior target lesion: the tumor volume on CBCT and on MRI was 4.69cm3 and 4.60 cm3, respectively; the volume of Lipiodol on CBCT and non-enhancement on MRI was 4.66cm3 and 4.60 cm3, respectively. For the posterior target lesion: the tumor volume on CBCT and on MRI was 4.58cm3 and 4.51 cm3, respectively; the volume of Lipiodol on CBCT and non-enhancement on MRI was 4.55cm3 and 4.51 cm3, respectively.
Figure 2
Figure 2
3D volumetric semi-automatic evaluation of Lipiodol deposition (Partial Response according to mRECIST criteria) in HCC on a representative case. Segmentation of the tumor (red circle) on CBCT at corresponding slice level as on MR (A, B). 3D segmentation volume rendering on the same slice (C, D). Quantitative color map of Lipiodol deposition on CBCT (E) and tumor viability on follow-up MRI (E). The box represents the location of the background ROI. The tumor volume on CBCT and on MRI was 6.79cm3 and 6.80 cm3, respectively. The volume of Lipiodol deposition on CBCT and non-enhancement on MRI was 5.83cm3 and 6.12 cm3, respectively.
Figure 3
Figure 3
Correlation between CBCT and MR of tumor volume (A), and % of (B) and in cm3 (C) of Lipiodol deposition on CBCT with non-enhancing regions on MR (B). The linear regression model demonstrated a strong correlation between overall tumor volumes as well as the volume of Lipiodol and tumor necrosis on CBCT and CE-MRI, respectively (Figure 3: A/C). In addition, a strong correlation between the % of Lipiodol deposition on CBCT and % of non-enhancing tumor tissue on CE-MRI was observed (Figure 3: B). Importantly, the extent of Lipiodol deposition (in % of total tumor volume) correlated strongly with tumor response on CE-MRI (D).

Source: PubMed

3
Se inscrever