Study design and patient population
This monocentric observational retrospective study was approved by a local ethics and institutional review board committee (CERIM- 2308 − 375). According to the retrospective design and the noninterventional nature of the study, the requirement for written informed consent was waived.
From January 2013 to December 2018, all consecutive liver-transplanted patients with HCC for whom IAT (TACE or TARE) was performed as a bridging therapy (BT) or downstaging therapy (DT) under dual-phase open trajectory-CBCT guidance were included. Criteria for exclusion were 1) patients without PII or liver explant anatomopathological analysis results available in their medical records, 2) patients for whom IAT was performed without per-interventional DP-CBCT imaging or with images hampered by respiratory or metallic artifacts, and 3) patients lost to follow-up. Patients were included in the LT list according to the combination of Milan criteria and alpha-fetoprotein (AFP) score (5, 24, 33). IAT was indicated as BT for patients within the Milan Criteria when the estimated waiting time was ≥ 6 months, while DT was aimed to reduce the tumor load and bring patients outside the Milan Criteria into the LT eligibility criteria (9, 24). All medical decisions were discussed in a multidisciplinary liver tumor board meeting including at least a hepatologist, a hepatobiliary surgeon, a pathologist, a liver-subspecialized radiologist, and an interventional radiologist.
Preinterventional imaging
All patients underwent preinterventional dynamic contrast-enhanced imaging: contrast-enhanced CT and/or multiphase MRI with a preintervention delay of less than two months. MRI exams were performed using a 1.5-T system (Avanto, Siemens Healthcare, Erlangen, Germany) or 3-T system (Skyra; Siemens Healthcare, Erlangen, Germany) equipped with a phased-array torso coil with an 18-channel system. All MR exams followed a standardized imaging protocol detailed in Table 1. The dynamic contrast-enhanced 3D VIBE T1-weighted sequence included four repeated series (bolus-triggered arterial, portal, venous and late phases) after injection of 0.2 mL/kg gadoterate meglumine (Dotarem; Guerbet, Aulnay-sous-Bois, France) or 0.1 mL/kg gadobenate dimeglumine (Multihance; Bracco Imaging, Milan, Italy) at a rate of 2 mL/sec. If necessary, delayed hepatobiliary phase acquisitions were acquired after a mean delay of 90 min following gadobenate dimeglumine injection T1-weighted sequences (34–36).
Table 1
Liver MRI examination protocol
Sequence | ET (ms) | RT (ms) | Flip angle (°) | Slice thickness (mm) | Echo length train |
Breath-hold fat-suppressed TSE T2-weighted sequence | 2600 | 83 | 129 | 5 | 23 |
Breath-hold HASTE T2- weighted sequence | 1000 | 118 | 120 | 3.5 | 156 |
breath-hold in-phase and out-of-phase T1-weighted sequences | 131 | 2.43–3.69 | 70 | 5 | NA |
Fat-suppressed T1-WI sequences | 117 | 2.78 | 70 | 5 | NA |
Diffusion weighted imaging single-shot, spin-echo-planar imaging (3 b-value 50, 400 and 800 sec/mm2) | 2500 | 138 | 150 | 5 | 23 |
Dynamic breath-hold 3D VIBE T1-weighted sequences | 3.31 | 1.25 | 15 | 3 | NA |
Abbreviations: TSE: turbo spin-echo; HASTE: half-Fourier acquisition single-shot; IP: in-phase; OP: out-of-phase; ET: echo time; RT: repetition time |
CT exams were acquired on a 64-detector row CT scanner (Discovery CT750 HD system, General Electric Healthcare, Milwaukee, WI). Patients were examined by the following protocol: unenhanced, late-arterial and portal phase acquisitions after intravenous injection of 1.5 ml/kg of nonionic contrast agent (Iomeron, 350-Iomeprol; Bracco Imaging, Milan, Italy) at an injection rate of 3–4 ml/s. A bolus-tracking technique was performed with automated scan triggering (SmartPrep; General Electric Healthcare, Milwaukee, WI). An elliptic region of interest (ROI) was positioned in the descending thoracic aorta at the diaphragm level, and the threshold enhancement value was set at 100 HU. The late arterial phase was acquired with a delay of 20 seconds after aortic enhancement threshold timing. CT acquisition and reconstruction parameters included tube current range (mA): 150–650 (mean: 455); tube voltage (kVp): 120; rotation time: 0.7 s; pitch: 1.375; automatic exposure control: Auto mA-Smart mA; noise index: 25; field of view: large body; reconstruction kernel: standard; section thickness: 0.625 mm. Native raw data of the acquired images were reconstructed using model-based iterative reconstruction, and then the reconstructed images were transferred and archived in the institutional picture archiving and communication system.
Per-interventional imaging
All IAT procedures were performed in an angiographic suite equipped with a flat panel detector C-arm angiographic system (Allura Xper FD20 and Allura Clarity; Philips Healthcare, Best, The Netherlands) under CBCT guidance. Patients were placed in a supine position on the angiography table. The celiac trunk and then the hepatic artery were catheterized by a 5 French Cobra catheter, and nonselective hepatic digital subtracted angiography was performed initially to display the global arterial anatomy of the liver. The dual-phase open trajectory protocol was applied for all CBCT scans, including two consecutive 5-second C-arm rotations after a single intra-arterial contrast medium injection (25, 37, 38). The early scan (five seconds after contrast injection) served to identify tumor-feeding arteries, whereas the delayed scan (17 seconds after contrast injection) displayed parenchymal and tumoral enhancement (25, 37, 38). Open trajectory acquisition was used to ensure the maximum coverage of the liver parenchyma. Each rotation acquires 312 frames (60 frames/s) covering a 240° clockwise arc. The flat panel detector displayed an FOV of 250 x 250 x 193 mm with a matrix size of 384 x 384 x 297 pixels. The acquired 3D volumetric CBCT images had an isotropic resolution of 0.65 mm. Intra-arterial contrast medium (iodixanol 320 mg iodine/ml, Visipaque; GE Healthcare AS, Oslo, Norway) was injected following this protocol: (a) for nonselective injection, 20 ml of contrast medium was injected at a rate of 2 ml/s; (b) for selective (hemi-hepatic) injection, 10 ml of contrast medium was injected at a rate of 1 ml/s. Patients were instructed to hold their breath during acquisition with free breathing between the 2 phases to avoid motion artifacts (37, 38).
CBCT scan projections were automatically transferred to a dedicated 3D workstation (Xtravision, Philips Medical Systems, Best, the Netherlands) for analysis of multiplanar and 3D reconstruction of acquired data.
Imaging analysis
PII data, referred to as the “gold standard”, were reviewed in consensus by two abdominal interventional radiologists with 8 years of experience (HD and AG). The two readers were blinded to the clinical data, per-interventional imaging data and explant anatomopathological analysis. HCC tumors were diagnosed on PII data according to the Liver Imaging Reporting and Data System (LI-RADS) criteria (39). We only considered LI-RADS 5 tumors planned for IAT according to multidisciplinary board meeting decisions.
Second, per-interventional CBCT images were reviewed for each patient in consensus by both readers to compare the detectability of all \(\ge 1 \text{c}\text{m}\)-hypervascular nodules included in the field of view on either arterial or late CBCT phases by reference to the PII. For TARE group patients, we evaluated CBCT scans acquired during work-up procedures.
Following this evaluation, patients were categorized into two groups: 1) incidentaloma +: group in whom at least one HVI was detected on per-interventional CBCT, and 2) incidentaloma -: group with no HVI detected on per-interventional CBCT compared to the PII.
Anatomopathological analysis
Following each LT, liver explant was analyzed by a 15-year-experenced liver-subspecialized pathologist (JC). Explants were macroscopically examined and then systematically dissected into 10 mm sections after formalin fixation. The number, size and location of all visible nodules were recorded; then, these nodules were sampled and fixed in paraffin for histological investigation. Active tumoral lesions were identified as well as focal benign nodules. As part of the study, anatomopathological analysis was performed retrospectively based on detailed pathological reports of each liver explant. Three histopathological criteria associated with poor prognosis (HPP) were recorded for each explant: capsular effraction, macroscopic and/or microscopic vascular invasion and the presence of liver metastasis (5, 7, 14).
Patient follow-up and prognosis evaluation
Our institutional post-LT follow-up protocol included immediate postoperative monitoring until patient recovery and then regular surveillance every 6 months. Biological monitoring included liver function tests and serum alpha-fetoprotein (AFP) levels (5, 40). Radiological monitoring consisted of a multiphasic contrast-enhanced CT scan performed every 6 months during the first two years and then annually thereafter. TR was tracked by serum AFP levels and chest and abdominal CT scans. When TR was suspected, liver MRI and 18FDG PET-CT were performed. TR was confirmed on the combination of imaging criteria or on pathologic data after percutaneous biopsy.
Statistical analysis
Data collection and statistical analysis were performed using Microsoft Excel 365 (Microsoft Corporation, Redmond, WA, USA) and SPSS Version 25.0 (SPSS Inc., Chicago, Illinois, USA). Categoric variables are expressed herein as frequencies. The normal distribution of continuous variables was verified using the Shapiro‒Wilk test. Normally distributed continuous variables are expressed as the means ± SDs and ranges. Other continuous variables are expressed as medians, Q1 (25%) and Q3 (75%).
The number of detected nodules on per-interventional CBCT and PII was compared using the nonparametric Wilcoxon test. Similarly, the number of detected nodules on per-interventional CBCT, of all nodules and of active-HCC nodules on liver explants were compared using the Wilcoxon test. Bivariate analysis using the Spearman test was performed to assess the correlation between the number of detected nodules on per-interventional imaging and the number of active-HCC nodules.
The correlation between the presence of HVI on per-interventional CBCT and HPP was investigated using Fisher’s exact test.
Finally, recurrence-free survival (RFS), overall survival (OS) and tumor recurrence-related survival (TRRS) were evaluated using Kaplan‒Meier analysis and compared using the log-rank test.
For all statistical tests, a two-sided p value of 0.05 was considered statistically significant.