Patients
Our institutional review board approved this retrospective study and waived the requirement for written informed consent.
This study included consecutive patients who underwent hepatectomy for HCC after MR imaging from April 2012 through December 2013. The flowchart of the patient selection process is shown in Figure 1. Choi et al. reported that HCCs smaller than 1.5 cm in size less frequently showed typical features on MR imaging than larger HCCs, and T2 hyperintensity, in particular, was not observed in small HCCs [12]. Therefore, we excluded HCCs < 1.5 cm. Large HCCs (> 10 cm) were excluded because the margin of these tumors often abutted the subcapsular region, making it difficult to analyze the peritumoral liver parenchyma.
During the study period, we evaluated 61 consecutive patients. Ultimately, 39 HCCs in 37 patients (25 males, 12 females; mean age, 69.3 ± 6.64 years) were selected. Tumor size was 1.5–4.8 cm (mean HCC size and standard deviation was 24.8 ± 8.40 mm). Twenty patients were infected with hepatitis C virus (HCV) (5 patients with hepatitis B virus (HBV)), and 12 patients did not have hepatitis B or C virus (alcoholic liver disease, eight patients; nonalcoholic steatohepatitis, three patients; unknown etiology, one patient). The grading of liver dysfunction was evaluated using the Child–Pugh classification; all patients had class A liver function. The interval between the date of MR imaging and hepatectomy ranged from 3 d to 58 d (median, 23 d).
MR imaging techniques
All MR images were obtained with a Magnetom Trio 3T MR system (Siemens Healthcare, Erlangen, Germany) using a six-channel body array coil and a six-channel spine matrix coil.
T2*WIs were obtained as part of the routine liver protocol in our institution. The sequence and scan parameters of the T2*WIs are shown in Table 1. We also obtained T1W1s (in-phase and opposed-phase), fat-suppressed T2WIs, diffusion-weighted images, and DCE-MR images routinely, and diagnosed the target nodule as HCC.
Image analysis
We defined a PTHR as follows: low intense rim or band encompassing more than one-third of the tumor circumference on T2*WIs (Fig. 2). The purpose of this study was to compare T2*WIs and histopathologic features, but we reviewed T2*WIs compared with fat-suppressed T2-WIs because target nodules were sometimes difficult to detect (Fig. 3) and to distinguish the nodules from other lesions such as cysts on T2*WI alone [11]. Two readers (Y.T. with 11 years and Y.O. with 6 years of experience in abdominal imaging) confirmed the target nodule on T2WIs and then assessed the presence or absence of a PTHR on T2*WIs. Both readers were blinded to patients’ clinical information and histopathologic features, except for the diagnosis of HCC. Disagreement between the two readers was resolved by discussion and reaching consensus. All images were reviewed with EV Insite software (PSP Corp., Tokyo, Japan). The frequency of a PTHR was calculated after the assessment.
Histopathologic analysis
Slides of formalin-fixed, paraffin-embedded sections from resected liver specimens were stained with Prussian blue stain. Semiquantitative analysis was performed by one pathologist (T.U. with 20 years of experience in liver pathology), who was blinded to patients’ clinical information and imaging results, except for the diagnosis of HCC. One representative specimen per nodule containing the boundary between the tumor and the background liver parenchyma was analyzed while considering the MR findings because it was difficult to analyze the entire tumor circumference in large tumors. Iron deposition in the peritumoral liver parenchyma or background liver parenchyma (not adjacent to the tumor) was graded by the pathologist. The grade of iron deposition was as follows: grade 0, iron granules were absent or barely discernible in a high-power field (×400 magnification); grade 1, granules were easily confirmed at ×400 magnification or barely discernible at ×250 magnification; grade 2, granules were resolved at ×100 magnification; grade 3, granules were resolved at ×25 magnification; grade 4, masses were visible at low power (×10 magnification) or with the naked eye [13, 14].
Multispectral imaging can produce more accurate images than red, green, and blue (RGB) imaging because the wavelength range including visible light can be divided into a large number of channels. Therefore, we measured optical density using the Vectra 3 multispectral imaging system (PerkinElmer, Inc., Waltham, MA, USA). This quantitative analysis of iron deposition was performed on the same slide used in the semiquantitative analysis. Scan images of the whole specimen with Prussian blue staining were digitalized by the multispectral imaging system according to a previously published protocol [15]. Each of five individual fields (669 × 500 μm each) of peritumoral liver parenchyma or outer liver parenchyma for multispectral acquisition were selected randomly using Phenochart image viewer software, version. 1.0.4 (PerkinElmer, Inc.) to scan at high-power resolution (×200 magnification) (Fig. 4). High-power multispectral acquisitions were analyzed using inForm image analysis software, version. 2.1 (PerkinElmer, Inc.) to extract the optical density of the Prussian blue staining. The mean optical density of five regions of peritumoral liver parenchyma or outer liver parenchyma was used for analysis.
Subsequently, we reviewed the pathology reports and investigated tumor grade, presence of a fibrous capsule, and degree of liver fibrosis. The degree of liver fibrosis in the background liver was graded according to the new Inuyama criteria as follows: no fibrosis (F0), portal fibrosis widening (F1), portal fibrosis widening with bridging fibrosis (F2), bridging fibrosis and lobular distortion (F3), or cirrhosis (F4) [16].
Statistical analysis
All statistical analyses were performed using IBM SPSS Statistics for Windows, version 25.0 (IBM Corp., Armonk, NY, USA) and Bell Curve for Excel (Social Survey Research Information Co. Ltd., Tokyo, Japan). Inter-reader agreement for the presence or absence of a PTHR on T2*WIs was assessed by calculating the kappa statistic [17]. A kappa value of ≤ 0.20 indicated poor agreement, 0.21–0.40 indicated fair agreement, 0.41–0.60 indicated moderate agreement, 0.61–0.80 indicated good agreement, and 0.81–1.00 indicated excellent agreement. The correlation between the semiquantitative and quantitative methods (peritumoral iron deposition and iron deposition in the background liver parenchyma, respectively) were analyzed using Spearman’s rank correlation coefficient. The association between PTHR and continuous variables in the potential contributing factors, such as peritumoral iron deposition (quantitative analysis), age, iron deposition in the background liver parenchyma (quantitative analysis), size, etiology, degree of fibrosis, and tumor grade, was analyzed using the Mann–Whitney U-test. The association between the PTHR and binary variables in the potential contributing factors, such as sex and a histopathologic fibrous capsule, was analyzed using the χ2 test. P < 0.05 was considered to indicate a statistically significant difference.