ORIGINAL ARTICLE

Liver Computed Tomographic Perfusion in the Assessment of Microvascular Invasion in Patients With Small Hepatocellular Carcinoma Dong Wu, MD, PhD,*†‡ Ming Tan, MD,*†‡ Meiling Zhou, MD,*†‡ Huichuan Sun, MD, PhD,§∥ Yuan Ji, MD, PhD,¶ Lingli Chen, MD,¶ Gang Chen, MD,*†‡ and Mengsu Zeng, MD, PhD*†‡ Objectives: Detecting microvascular invasion (mVI) in patients with hepatocellular carcinoma is a diagnostic challenge. The present study aimed to acquire a series of quantitative perfusion parameters from liver computed tomography (CT) with a 320-slice scanner in patients with small hepatocellular carcinoma (sHCC) and study its efficacy in identifying mVI. Materials and Methods: Fifty-six patients who underwent hepatic resection for sHCC (≤3 cm) were preoperatively examined with a 320-detector row CT scanner. Histopathological analyses of liver biopsies confirmed that 18 patients had sHCC with mVI and that 38 patients had sHCC without mVI. Hepatic artery flow, portal vein flow (PVF), and perfusion index were measured in both tumor and normal liver tissues. Nonparametric receiver operating characteristic curve analysis was performed to quantify the accuracy of the perfusion CT parameters. Results: The tumor PVF (PVFtumor), difference in PVF between tumor and liver tissue (ΔPVF), and the ΔPVF/liver PVF ratio (rPVF) were significantly higher in sHCC with mVI than in sHCC without mVI (P = 0.0094, P = 0.0018, and P = 0.0007, respectively; Wilcoxon signed rank test). The PVFtumor, ΔPVF, and rPVF correctly predicted mVI in 73.2% (sensitivity, 66.7%; specificity, 76.3%; cutoff, 103.8 mL per 100 mL/min), 76.8% (sensitivity, 66.7%; specificity, 81.6%; cutoff, −53.65 mL per 100 mL/min), and 83.9% (sensitivity, 77.8%; specificity, 86.8%; cutoff, −0.38) of a total of 56 patients with sHCC, respectively. Other parameters were not significantly different between the groups. Conclusions: Liver CT perfusion provides a noninvasive, quantitative method that can predict mVI in patients with sHCC through measurement of 3 perfusion parameters: PVFtumor, ΔPVF, and rPVF. Key Words: hepatocellular carcinoma, perfusion, x-ray computed tomography, microvascular invasion (Invest Radiol 2015;50: 188–194)

P

rimary liver cancer is a significant cause of cancer-related deaths, and hepatocellular carcinoma (HCC) accounts for 70% to 85% of total liver cancers worldwide.1 Screening of HCC is effectively performed by ultrasonography, magnetic resonance imaging (MRI), and computed tomography (CT).2 Selected patients with HCC are candidates for therapy, such as hepatic resection, radiofrequency ablation, transcatheter arterial chemoembolization, and transplantation. Nevertheless, tumor recurrence is 70% at 5 years after resection and 15% to 30% after liver transplantation, leading to tumor-related death.3 During the past decade, many studies have determined that portal vein invasion,

Received for publication April 2, 2014; and accepted for publication, after revision, August 13, 2014. From the *Department of Radiology, Zhongshan Hospital of Fudan University; †Department of Medical Imaging, Shanghai Medical College, Fudan University; ‡Shanghai Institute of Medical Imaging; §Department of Liver Surgery, Zhongshan Hospital of Fudan University; kShanghai Institute of Liver Cancer; and ¶Department of Pathology, Zhongshan Hospital of Fudan University, Shanghai, China. Supported by Shanghai Institute of Medical Imaging (SHIMI-GI-2013-008). Conflicts of interest and sources of funding: none declared. Reprints: Mengsu Zeng, MD, PhD, 180 Fenglin Rd, Xuhui District, Shanghai, China 200032. E-mail: [email protected]. Copyright © 2014 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0020-9996/15/5004–0188

188

www.investigativeradiology.com

whether macrovascular or microvascular invasion (mVI), was the strongest variable related to tumor recurrence and survival.4–6 Some authors have proposed that patients with mVI in the liver specimen after resection should be considered for a liver transplant.7,8 As such, assessment of mVI in patients with HCC is pivotal in the disease management algorithm. Hepatocellular carcinoma has a complex blood supply, with both the hepatic artery and portal vein supplying blood to the liver. Between hepatic regenerative nodules, dysplasia nodules, as well as the early development of HCC and small HCC (sHCC), hepatic artery flow (HAF) and portal vein flow (PVF) undergo a series of complex changes.9 Portal vein invasion of the major venous branches is usually detected by ultrasonography, MRI, and CT. However, the detection and assessment of mVI before resection are challenging, and a conventional dynamic contrast CT or MRI may not be sufficient for precise detection. Hence, novel radiologic techniques are needed to improve the diagnosis of mVI in patients with HCC, but they have rarely been examined to date.10 Perfusion CT imaging of the liver is a relatively new technique that captures serial images after a contrast bolus injection and enables analysis of temporal changes of tumor hemodynamics and vessel attenuation.11 Thus, it allows for quantitative measurements in patients with HCC.12–15 A newer perfusion CT system with 320 detector rows is now clinically available. This system is capable of performing isotropic volumetric imaging in a single rotational acquisition, covering 16 cm in the z-axis direction. It has recently been used to successfully obtain images of most or all of the liver.14 Functional 320-slice liver perfusion CT imaging is able to provide accurate blood flow values of normal liver tissue and HCC as well as to quantitatively evaluate HAF and PVF changes associated with the disease. The present study hypothesized that the probability of detecting mVI in patients with advanced HCC was high.16 To test this hypothesis, we focused on sHCC to improve the accuracy of the perfusion parameter measurements: these tumors are relatively uniform, allowing straightforward selection of a region of interest (ROI) that can cover the whole tumor. The present study aimed to acquire a series of quantitative perfusion parameters from 320-slice liver CT in patients with sHCC and study its efficacy in identifying mVI. In addition, a receiver operating curve (ROC) analysis was performed to determine the accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), false-positive rate (FPR), and false-negative rate (FNR).

MATERIALS AND METHODS Patients This was a retrospective study of a total of 562 patients with suspected liver tumors who underwent CT perfusion imaging between June 2011 and May 2013; of these, 213 patients underwent hepatic resection for primary HCC. Only 56 patients were enrolled per the following inclusion criteria (Fig. 1): (1) age of 18 years or older, (2) histologically confirmed primary HCC, and (3) tumor size of 3 cm or less (the leading medical, oncological, and hepatological associations in China regard tumors of ≤ 3 cm in size as sHCC17). The exclusion Investigative Radiology • Volume 50, Number 4, April 2015

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Investigative Radiology • Volume 50, Number 4, April 2015

320-Slice Liver CT Perfusion in Patients With sHCC

respiration. The scanning protocol included three phases: the first 11 volumes were acquired every 2 seconds (to identify the hepatic artery peak time), followed by 7 volumes during every 3 seconds (to identify the portal vein peak time), and 5 volumes during every 5 seconds (for morphological scanning and to ensure that peak enhancement of the hepatic tissue was imaged in all patients). This resulted in a total examination time of 66 seconds from the time the x-ray was turned on; the total x-ray exposure time was 11.5 seconds. In our patients, the mean (SD) peak times for the hepatic artery and portal vein were 18.67 (3.01) seconds and 37.89 (6.38) seconds, respectively. All volumes were acquired with the following parameters: 100 kV(p), 100 mA, and 0.5-second rotation speed. Each volume was reconstructed at a 0.5-mm thickness with 0.5-mm spacing, providing a total of 7360 (23 volumes  320 images) images.16 The dose-length product for the 320-slice liver CT perfusion was recorded from the CT scanner readout. The effective dose was calculated using a k factor of 0.015.18

Perfusion CT Analysis

FIGURE 1. Enrollment of the study patients.

criteria were as follows: (1) patients with more than 1 tumor; (2) portal vein tumor thrombosis detected by any routine imaging modality, including ultrasonography, dynamic contrast CT, and MRI; (3) any evidence of tumor metastasis; (4) any prior recurrence of HCC; and (5) history of prior therapy, including chemotherapy, radiotherapy, molecular targeting agents, and liver transplantation or hepatectomy. The patients’ demographic and clinical characteristics were collected, including age, sex, etiology of liver disease, history of hepatitis, hepatitis B virus markers, liver function, α-fetoprotein, and intraoperative findings. The study was approved by the hospital’s institutional ethics committee.

Perfusion CT Scanning Perfusion CT was performed on an average of 3.4 days (range, 1–11 days) before hepatic resection using a 320-slice multidetector CT scanner (Aquilion ONE; Toshiba Medical Systems Corporation, Otawara, Japan). A noncontrast helical CT of the liver was initially performed as a baseline study. After this, a 320-slice liver perfusion CT acquisition was performed using the dynamic volume scan mode with 16-cm z-axis coverage and no table movement. Approximately 40 mL of nonionic iodinated contrast medium was injected intravenously (Iopamiron 370; Bayer Health Care, Guangdong, China), followed by 30 mL of saline solution using a power injector at a rate of 8 mL/s through a 20-gauge cannula placed in an antecubital vein. The CT perfusion acquisition protocol was initiated simultaneously with the start of the contrast injection. The first volume acquisition took place 8 seconds after contrast administration. All patients were asked to breathe gently during the entire acquisition, and restraining bands were placed around the abdomen to limit movement of the abdomen during © 2014 Wolters Kluwer Health, Inc. All rights reserved.

All CT perfusion images were transferred to an image processing workstation (Vitrea fX, v6.0; Vital Images, Minnetonka, MN). Before perfusion analysis, nonrigid (deformable) registration was performed on the workstation to account for breathing mismatch between volumes. A default reference template was selected by the software package, although visual inspection was also used to ensure that a reference template with minimal motion artifact was chosen. The time necessary for the entire registration process of a single examination was approximately 15 minutes. After the registration, the perfusion data image sets were reconstructed with a slice thickness of 5 mm and a perfusion map resolution of 3 mm. Body Perfusion software V4.74 (Toshiba Medical Systems) was used for CT perfusion analysis, which applied the dual-input maximum slope analysis method.19–22 Regions of interest were placed on the abdominal aorta at the level of celiac axis, main portal vein, liver, and spleen to generate time-density curves. Subsequently, functional maps were generated automatically, with each pixel value for HAF (milliliter per 100 mL/min), PVF (milliliter per 100 mL/min), and perfusion index (PI; as a fraction) represented using a color scale. The HAF, PVF, and PI were measured in the tumor and in an area with normal liver tissue (no tumor area) in each patient by 2 radiologists who had more than 15 years of experience in abdominal imaging (W.D. and T.M.). Tumor ROIs were manually drawn to include the whole tumor, in a single image plane in which the tumor diameter and contrast enhancement were maximal. The vast majority of tumors appeared circular in the image plane in which their diameter was maximal; hence, a circular ROI was used. For those tumors with an irregular appearance, the border of the tumor was traced manually. Normal liver ROIs were drawn in the same lobe of the liver as the tumor and were of the same size as the tumor ROI. The HAF, PVF, and PI of the tumor and normal liver tissue were calculated automatically by the software. The total liver flow (HFliver = HAFliver + PVFliver), total tumor flow (HFtumor = HAFtumor + PVFtumor), difference in flow between tumor and liver (ΔHF = HFtumor − HFliver), relative flow (rHF = ΔHF/HFliver), difference in HAF (ΔHAF = HAFtumor − HAFliver), relative HAF (rHAF = ΔHAF/HAFliver), difference in PVF (ΔPVF = PVFtumor − PVFliver), relative PVF (rPVF = ΔPVF/PVFliver), difference in PI (ΔPI = PItumor − PIliver), and relative PI (rPI = ΔPI/PIliver) were calculated.

Histopathology Histopathological analysis was carried out on the entire tumor and on the surrounding liver tissue (within 10 mm of the tumor capsule). After surgery, each tumor specimen was sectioned into small pieces (10 mm  5 mm  3 mm) and fixed in 10% neutralized formalin. During processing, the resected liver specimens were embedded in www.investigativeradiology.com

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

189

Investigative Radiology • Volume 50, Number 4, April 2015

Wu et al

and then used for observing mVI. Microvascular invasion was defined as the presence of tumor emboli in a portal vein radicle, large capsule vessel, or vascular space lined by endothelial cells.23 All pathological diagnoses were performed by 2 pathologists (J.Y. and C.L.), who were blinded to the radiologic findings.

TABLE 1. Baseline Patient Characteristics Clinical Features

sHCC Without sHCC With mVI (n = 38) mVI (n = 18)

Age, y Mean (SD) Sex (%) Male Female Origin of liver disease (%) Hepatitis B Other Symptoms (%) Periodic physical examination Right upper quadrant abdominal pain Other History of hepatitis B (%) ≤3 y 10–30 y >30 y Hepatitis B virus markers (%) HBsAg (+) Anti-HBs (+) HBeAg (+) Anti-HBe (+) Anti-HBc (+) Child-Pugh classification* (%) A B Preoperation log10 AFP,† mean (SD), u/mL Postoperation log10 AFP,† mean (SD), u/mL

t/x2

P

50.5 (13.4)

49.0 (12.8)

0.490 0.626

37 (97.4) 1 (2.6)

16 (88.9) 2 (11.1)

1.732 0.188

37 (97.4) 1 (2.6)

18 (100) 0 (0.0)

1.000

23 (60.5)

13 (72.2)

8 (21.1)

2 (11.1)

7 (18.4)

3 (16.7)

11 (29) 16 (42.15) 7 (18.4)

10 (55.6) 7 (33.4) 2 (11.1)

23 (60.5) 0 (0.0) 6 (15.8) 18 (47.4) 22 (57.9)

16 (88.9) 1 (5.6) 3 (16.7) 14 (77.8) 17 (94.4)

Statistical Analyses The patients were divided into 2 groups on the basis of whether there was pathological evidence of mVI. All statistical analyses were performed using Statistical Package for the Social Sciences 13.0 software (SPSS Inc, Chicago, IL). Quantitative data are expressed as mean (SD) if normally distributed or as median and interquartile range if the distribution was skewed (determined using a test of normality). Categorical data are described as absolute frequencies. The Fisher exact test or Pearson χ2 test was used for categorical data, and the independent samples t test was used for continuous data, as appropriate. For comparisons of the CT perfusion parameters between the groups, the independent samples t test was used for normally distributed data (which are presented as means [SDs]), and the univariate paired Wilcoxon signed rank test was used for data with a skewed distribution (presented as medians and interquartile ranges). Receiver operating characteristic curve analysis was performed to quantify the accuracy of the CT perfusion parameters. Threshold, accuracy, sensitivity, specificity, PPV, NPV, FPR, and FNR were calculated. For all statistical determinations, a value of P < 0.05 (double sided) was considered statistically significant.

0.702

0.384

6.167 8.440 7.333 7.000 7.735

0.046 0.026 0.030 0.021 0.078 0.239

37 (97.4) 1 (2.6) 1.47 (0.31)

16 (88.9) 2 (11.1) 2.12 (0.39) −1.798 0.078

1.27 (0.28)

1.84 (0.27) −1.578 0.123

RESULTS Patient Characteristics The mean (SD) age of the 56 patients (53 males and 3 females) enrolled was 50 (10.7) years. The disposition of the patient population is shown in Figure 1. All patients tolerated the CT perfusion examination with no adverse effects. Eighteen patients (32.1%) had mVI, and the remaining 38 patients (67.9%) showed no histopathological evidence of mVI. The mean (SD) diameter of the tumor was 22 (13 mm) (range, 5–30 mm). Twenty-six lesions occurred in the left lobe of the liver, and 30 lesions occurred in the right lobe. There were no significant differences in the following general and clinical characteristics between the 2 groups: age, etiology of the liver disease, history of hepatitis, liver function, sex, and α-fetoprotein (P > 0.05). However, hepatitis B virus markers such as HBsAg, anti-HBs, and anti-HBe were higher in the mVI group than in the patients without mVI (P < 0.05). The baseline characteristics for both groups are shown in Table 1.

*There was no Child-Pugh classification C in any of the patients’ medical records. †The value of AFP shows good normal distribution characteristics after taking the logarithm. AFP indicates α-fetoprotein; anti-HBs, antibody to hepatitis B surface antigen; anti-HBc, antibody to hepatitis B core antigen; anti-HBe, antibody to hepatitis B e antigen; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; mVI, microvascular invasion; sHCC, small hepatocellular carcinoma.

Perfusion Parameters paraffin with the greatest dimension. The formalin-fixed, paraffinembedded tissue blocks were cut into 4-μm–thick sections, deparaffinized, and rehydrated. Each section was stained with hematoxylin and eosin

In both groups, the HAF and PI were significantly higher in sHCC than in normal liver parenchyma and PVF was significantly lower in sHCC than in normal liver parenchyma (Table 2; P < 0.05).

TABLE 2. Differences in HAF, PVF, and PI Between sHCC and Normal Liver Parenchyma sHCC With mVI (n = 18) HAF PVF PI

sHCC Without mVI (n = 38)

sHCC

Liver

P

sHCC

Liver

P

61.4 (37.0, 86.8) 108.1 (78.5, 122.9) 44.1 (31.1, 59.1)

28.9 (21.3, 33.0) 147.4 (112.9, 171.2) 14.8 (10.0, 21.4)

0.0001 0.0075 0.001

72.0 (58.3, 123.7) 69.2 (50.0, 88.7) 54.0 (47.2, 69.6)

30.3 (23.7, 37.8) 143.7 (117.5, 171.4) 16.2 (13.8, 20.2)

0.001 0.001 0.001

Data are presented as median values (25th and 75th percentiles). HAF indicates hepatic artery flow (milliliter per 100 mL/min); Liver, normal liver parenchyma; mVI, microvascular invasion; PI, perfusion index (%); PVF, portal vein flow (milliliter per 100 mL/min); sHCC: small hepatocellular carcinoma.

190

www.investigativeradiology.com

© 2014 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Investigative Radiology • Volume 50, Number 4, April 2015

320-Slice Liver CT Perfusion in Patients With sHCC

TABLE 3. Perfusion Parameters in sHCC Patients With and Without mVI

Group HAFtumor PVFtumor PItumor HFtumor HAFliver PVFliver PIliver HFliver ΔHAF ΔPVF ΔPI rHAF rPVF rPI ΔHF rHF

sHCC With mVI (n = 18)

sHCC Without mVI (n = 38)

P

61.4 (37.0, 86.8) 108.1 (78.5, 122.9) 44.1 (31.1, 59.1) 183.6 (126.2, 254) 28.9 (21.3, 33) 147.4 (112.9, 171.2) 14.8 (10.0, 21.4) 170.4 (147.3, 196.5) 41.9 (18.5, 59.4) −32.0 (−64.0, −8.6) 22.8 (17.4, 42.3) 1.57 (0.70, 2.95) −0.25 (−0.37, −0.08) 1.59 (0.98, 2.60) 5.2 (−28.1, 46.0) 0.04 (−0.17, 0.28)

72.0 (58.3, 123.7) 69.2 (50.0, 88.7) 54.0 (47.2, 69.6) 135.9 (113.7, 200.7) 30.3 (23.7, 37.8) 143.7 (117.5, 171.4) 16.2 (13.8, 20.2) 173.4 (143.4, 206.0) 46.1 (35.2, 75.9) −72.8 (−106.3, −59.4) 37.3 (26.1, 50.1) 1.64 (1.28, 3.00) −0.54 (−0.67, −0.43) 2.35 (1.60, 3.25) −31.9 (−62.7, 6.8) −0.21 (−0.37, 0.04)

0.1908 0.0094 0.0576 0.2414 0.3984 0.9513 0.5245 0.7600 0.2414 0.0018 0.0896 0.4332 0.0007 0.1635 0.0646 0.0849

Data are presented as median values (25th and 75th percentiles). HF, ΔHF, HAF, ΔHAF, PVF, and ΔPVF values are expressed as milliliter per100 mL/min. PI, rHAF, rPVF, rPI, and rHF values are expressed as fractions (no units). HAFliver indicates hepatic artery flow of liver; HAFtumor, hepatic artery flow of tumor; HFliver, total liver flow; HFtumor, total tumor flow; mVI, microvascular invasion; PIliver, perfusion index of liver; PItumor, perfusion index of tumor; PVFliver, portal vein flow of liver; PVFtumor, portal vein flow of tumor; rHAF, relative hepatic artery flow (ΔHAF/HAFliver); rHF, relative flow (ΔHF/HFliver); rPI, relative perfusion index (ΔPI/PIliver); rHF, relative portal vein flow (rPVF = ΔPVF/PVFliver); sHCC: small hepatocellular carcinoma; ΔHAF, difference of hepatic artery flow (HAFtumor − HAFliver); ΔHF, difference in flow between tumor and liver (HFtumor − HFliver); ΔPI, difference of perfusion index (PItumor − PIliver); ΔPVF, difference of portal vein flow (PVFtumor − PVFliver).

The perfusion parameters for each group are presented in Table 3. The PVFtumor, ΔPVF, and rPVF were significantly higher (Ρ = 0.0094, Ρ = 0.0018, and Ρ = 0.0007, respectively) in the mVI group than in the group that did not have mVI. There were no other statistically significant differences in the other perfusion parameters tested between the 2 groups (Figs. 2 and 3). For the ROC analysis, a cutoff value of 103.8 mL per 100 mL/min for PVFtumor correctly predicted mVI in 73.2% of sHCC with the following parameters: sensitivity, 66.7%; specificity, 76.3%; PPV, 57.1%; NPV, 82.9%; FPR, 42.9%; and FNR, 17.1%. In addition, a cutoff value of −53.65 mL per 100 mL/min for ΔPVF correctly predicted mVI in 76.8% of sHCC with the following parameters: sensitivity, 66.7%; specificity, 81.6%; PPV, 63.2%; NPV, 83.8%; FPR, 36.8%; and FNR, 16.2%. Lastly, a cutoff value of −0.38 for rPVF correctly predicted mVI in 83.9% of sHCC with the following parameters: sensitivity, 77.8%; specificity, 86.8%; PPV, 73.7%; NPV, 89.2%; FPR, 26.3%; and FNR, 10.8% (Fig. 4; Table 4). There were no significant differences in the 3 parameters per the ROC area under the curve analysis (95% confidence interval), but the sensitivity and specificity were higher for rPVF than for the other 2 parameters. The dose-length product for the perfusion CT was 922.9 mGy·cm. Per the calculation method recommended by the International Commission on Radiological Protection,18 the dose was 13.8 mSv.

DISCUSSION Vascular invasion is an advanced phase in tumor progression. In HCC, the presence of mVI is a marker of aggressive biological tumor behavior that dramatically changes the disease prognosis, particularly after potential curative therapy.14 The risk for recurrence and death is almost doubled in HCC patients with mVI, and mVI is an independent risk factor for both events.24,25 Portal vein invasion of the major venous branches is usually detected by ultrasonography, MRI, and CT. However, the detection and assessment of mVI before resection are challenging. In most cases, information about mVI comes from resection or liver specimens after transplant. A prediction of mVI before therapy without the need for

FIGURE 2. A 56-year-old man was diagnosed with HCC in the right lobe by ultrasonography during a routine physical examination. A, Late arterial phase CT; an sHCC was seen in the right frontal lobe (segment V). B to D, Liver perfusion CT maps of HAF, PVF, and PI, respectively. B, The calibration bar maximum value (obscured) is 200. Normal liver parenchyma PVF was 131.7 mL per 100 mL/min, HCC PVF was 60.4 mL per 100 mL/min, ΔPVF was −71.3 mL per 100 mL/min, and rPVF was −0.54. The histopathological specimen showed no tumor thrombus in the peritumoral portal vein branch.

© 2014 Wolters Kluwer Health, Inc. All rights reserved.

www.investigativeradiology.com

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

191

Investigative Radiology • Volume 50, Number 4, April 2015

Wu et al

FIGURE 3. A 50-year-old man was diagnosed with HCC in the right lobe by ultrasonography during a routine physical examination. A, Time-activity curves determined from ROIs placed on the abdominal aorta at the level of celiac axis, the main portal vein, the liver, and the spleen. The peak value of the artery was 762 Hounsfield units (HU), and the time to peak was 20 seconds. The peak value of the portal vein was 198 HU, and the time to peak was 39 seconds. The peak value of the spleen was 172 HU, and the time to peak was 28 seconds. B, Late arterial phase CT; an sHCC was seen in the right lobe (segment VIII). C to E, Perfusion maps of HAF, PVF, and PI, respectively. C, The calibration bar maximum value (obscured) is 200. Normal liver parenchyma PVF was 141.5 mL per 100 mL/min, HCC PVF was 95.7 mL per 100 mL/min, ΔPVF was −45.8 mL per 100 mL/min, and rPVF was −0.32. F, Histologic specimen demonstrating tumor thrombus in the peritumoral portal vein branch (hematoxylin and eosin staining, magnification 5).

biopsy would greatly improve the management of patients with cirrhosis and HCC because this could guide treatment and predict prognosis.2,5,6 Many studies have tried to identify clinical, imaging, or serum parameters that are able to predict mVI. Regarding imaging technologies, most studies have focused on the morphological changes and enhanced features of HCC.26–28 Some serum markers such as α-fetoprotein and protein induced by vitamin K absence or antagonist-II are overexpressed in patients with HCC and mVI.29,30 However, these parameters can be abnormally high with advanced fibrosis without HCC, so they are not sufficiently specific.31 Perfusion CT has been used widely in the clinical examination of brain diseases.32 Recent technological advances in CT hardware and software have allowed for its application in other body regions, including the liver.13–16,33,34 Whereas conventional CT focuses on morphology, perfusion CT scanning performs quantitative assessment of the changes in the blood flow of the microenvironment. Although some previous

studies have carried out perfusion CT of the liver using 64-slice scanners,35 these instruments allow a z-axis coverage of only 4 cm; this limits the region of the liver in which the tumor can be located for simultaneous imaging of the tumor and portal vein, and thus accurate perfusion calculations.12 The present study used a 320-slice scanner; with this instrument, a single rotation can cover 16 cm in the z axis, enabling dynamic imaging of a much larger portion of the liver. This scanner allows images of both the tumor and portal vein to be captured simultaneously regardless of the tumor location. A recent study was able to obtain images of the whole upper abdomen in patients with and without liver diseases,16 whereas other researchers have used this scanner for perfusion imaging of the whole brain36 and renal tumors.37 Nonetheless, 320-slice CT is not the only approach to extending the z-axis scan range. Periodic spiral-mode CT represents an alternative technique that can extend the z-axis coverage and overcome this limitation of conventional 64-slice CT scanning. For example, Goetti et al38,39 have successfully carried out perfusion analysis using a spiral-mode

FIGURE 4. The ROC analysis of PVF, ΔPVF, and rPVF for preoperative prediction of mVI. A, The ROC for PVF. A critical PVF value of 103.8 mL per 100 mL/min was identified (n = 56; area under the ROC curve, 0.73; 95% confidence interval, 0.56-0.89; P = 0.0094); sensitivity of 0.667 and specificity of 0.763. B, The ROC for ΔPVF. A critical ΔPVF value of −53.65 mL per 100 mL/min was identified (n = 56; area under the ROC curve, 0.77; 95% confidence interval, 0.64–0.91; P = 0.0018); sensitivity of 0.667 and specificity of 0.816. C, The ROC for r PVF. A critical rPVF value of −0.38 was identified (n = 56; area under the ROC curve, 0.80; 95% confidence interval, 0.65-0.94; P = 0.0007); sensitivity of 0.778 and specificity of 0.868.

192

www.investigativeradiology.com

© 2014 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Investigative Radiology • Volume 50, Number 4, April 2015

TABLE 4. ROC Analysis of PVFtumor, ΔPVF, and r PVF in the Assessment of mVI in Patients With sHCC

Threshold Accuracy, % Sensitivity, % Specificity, % Positive predictive value, % Negative predictive value, % False-positive rate, % False-negative rate, %

PVFtumor

ΔPVF

rPVF

103.8

−53.65

−0.38

Milliliter Milliliter per 100 mL/min per 100 mL/min 73.2 66.7 76.3 57.1 82.9 42.9 17.1

76.8 66.7 81.6 63.2 83.8 36.8 16.2

83.9 77.8 86.8 73.7 89.2 26.3 10.8

mVI indicates microvascular invasion; PVFtumor, portal vein flow of tumor; ROC, receiver operating curve; rPVF, relative portal vein flow (rPVF = ΔPVF/ PVFliver); sHCC, small hepatocellular carcinoma; ΔPVF, difference of portal vein flow (PVFtumor − PVFliver).

single-source 64-slice CT with a scan range of 6.7 cm as well as a dualsource 128-slice CT with a scan range of 14.8 cm; using the latter technique, 98% of all hepatic metastases were successfully imaged in their patients.38 Others have also demonstrated the utility of periodic spiral CT40,41 and have shown that a periodic spiral scan mode is equivalent to a standard dynamic mode for quantitative evaluation of tissue flow.42 Thus, our findings regarding the utility of perfusion parameters such as rPVF for the preoperative diagnosis of mVI in sHCC may be applicable to other methodologies that extend the zaxis coverage to permit scanning of the majority of the liver. The present study involved the retrospective analysis of the efficacy of 320-slice perfusion CT in detecting mVI in patients with sHCC. This is the first study to report the accuracy of using perfusion parameters obtained using a 320-slice CT scanner for determining mVI. The results indicate that PVFtumor, ΔPVF, and rPVF could predict mVI in the majority of patients with sHCC. These findings may provide a basis for more accurate clinical diagnosis of mVI, improving clinical decision making in patients with sHCC. A particular advantage of our approach is that rPVF would be expected to be relatively stable with regard to the use of different scanning equipment or alterations in the injection protocol. Therefore, our approach would likely be widely transferable to departments using different scanning equipment and injection protocols. The study results also show that HCC with or without mVI had a higher HAF and a lower PVF than normal liver tissue did. The PI was also higher in HCC than in normal tissue, and these data are similar to previously reported results.43,44 However, previous studies of blood supply in HCC were subject to qualitative analysis. For instance, Hayashi et al9 observed the arterial supply of HCC with hepatic arteriography CT and hepatic portovenography CT through hepatic artery catheterization. It has been demonstrated previously that different degrees of cirrhosis can affect the perfusion CT parameters of normal liver parenchyma.45 For the same scanner and protocol, PVF would directly depend on the peak enhancement of the portal vein, which would be sensitive to user variation and partial volume effects. In addition, PVF would be influenced by the injection protocol used. The present study injected 40 mL of contrast medium at a rate of 8 mL/s; this is because the maximum slope model used for our analysis favors shorter injection times, with flow underestimated for longer injection times. This is evident from comparison of the normal liver PVF value of approximately145 mL per 100 mL/min (at an injection rate of 8 mL/s) measured © 2014 Wolters Kluwer Health, Inc. All rights reserved.

320-Slice Liver CT Perfusion in Patients With sHCC

in our study with values of approximately 115 mL per 100 mL/min at 6 mL/s16 and approximately 95 mL per 100 mL/min at 5 mL/s41 obtained in previous studies. This would limit the utility of PVF as a parameter for assessing the presence or absence of mVI. However, a major advantage of rPVF is that it is a self-normalizing quantity that has the potential to overcome all of these effects. Therefore, the standardization of parameters such as ΔPVF and rPVF in future studies would be expected to reduce the differences observed between individuals caused by different equipment or perfusion CT protocol and thereby enhance their clinical utility in the preoperative detection of mVI. In a previous study, approximately 10.7% of sHCC (≤2 cm) showed mVI.46 In the present study, 32.1% of the patients with sHCC (≤3 cm) had mVI. Therefore, this technique is clinically significant because it can be used to decide therapeutic methods, monitor treatment efficacy, and establish prognosis by accurately assessing mVI before any surgical procedure. In this study, PVF was significantly lower in HCC tissue than in normal liver parenchyma. However, in the patients with HCC and mVI, PVF in the tumor was significantly higher when compared with the patients with HCC but no mVI. There was no significant difference in HAF and PI between the groups. This suggests that PVF is increased when there is mVI in the HCC. This may be caused by the reason that, when the portal vein is invaded, hepatic artery-portal vein fistula formation occurs and the portal vein is arterialized, resulting in increased mVI. Another possibility is that if portal vein-hepatic venous fistula formation occurs, PVF might directly enter into the hepatic vein without passing through the capillary network, resulting in increased PVF. This might allow for microemboli to flow into the portal vein and for distant metastasis to occur. Another possible mechanism is that, when small branch mVI occurs, adhesion of vascular endothelial cells gets reduced. This could increase vascular permeability and reduce portal vein resistance. However, all of these possible mechanisms require further investigation. To reduce the radiation dose administered to the patient, 320-slice hepatic perfusion CT uses the maximum slope analysis method to shorten the exposure time. Thus, only blood flow perfusion parameters (HAF and PVF) were calculated. Other perfusion parameters, such as blood volume and mean transit time, could not be calculated. It is unknown whether these parameters may also be helpful in the assessment of mVI in sHCC. Our study is not without its limitations. First, the sample size was small. Second, our software only permitted the use of ROIs drawn in a single image plane (in which the tumor diameter was maximal); it was not possible to use volumes of interest for the analysis of perfusion parameters,47 which would have represented a more robust approach. As a result, if mVI was not present in the image plane evaluated but was present in other image planes (in which tumor diameter was not maximal), the region with mVI would not have been evaluated during the perfusion analysis, potentially leading to false-negative results (because mVI would have been identified by histopathology, which assessed the entire tumor). Third, variations in the degree of the disease condition between the study participants may also have had some effect on the results. Hence, further research with larger sample sizes are required to validate the study results. In conclusion, 320-slice liver CT perfusion provides a noninvasive, quantitative method that can predict mVI in patients with sHCC through measurement of 3 perfusion parameters: PVFtumor, ΔPVF, and rPVF. This technique could facilitate the determination of the appropriate therapeutic methods as well as the extent of the resection procedure and predict prognosis. REFERENCES 1. Caldwell S, Park SH. The epidemiology of hepatocellular cancer: from the perspectives of public health problem to tumor biology. J Gastroenterol. 2009;44 (suppl 19):96–101.

www.investigativeradiology.com

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

193

Investigative Radiology • Volume 50, Number 4, April 2015

Wu et al

2. Wong R, Frenette C. Updates in the management of hepatocellular carcinoma. Gastroenterol Hepatol (N Y). 2011;7:16–24. 3. Bruix J, Sherman M. Management of hepatocellular carcinoma. Hepatology. 2005;42:1208–1236. 4. Iwatsuki S, Dvorchik I, Marsh JW, et al. Liver transplantation for hepatocellular carcinoma: a proposal of a prognostic scoring system. J Am Coll Surg. 2000; 191:389–394. 5. Ikai I, Arii S, Kijiro M, et al. Reevaluation of prognostic factors for survival after liver resection in patients with hepatocellular carcinoma in a Japanese nationwide survey. Cancer. 2004;101:796–802. 6. Shirabe K, Kajiyama K, Harimoto N, et al. Prognosis of hepatocellular carcinoma accompanied by microscopic portal vein invasion. World J Gastroenterol. 2009; 15:2632–2637. 7. Sala M, Llovet JM, Vilana R, et al. Initial response to percutaneous ablation predicts survival in patients with hepatocellular carainoma. Hepatology. 2004;40: 1352–1360. 8. Fuks D, Dokmak S, Paradis V, et al. Benefit of initial resection of HCC followed by transplantation in case of recurrence: an intention-to-treat analysis. Hepatology. 2012;55:132–140. 9. Hayashi M, Matsui O, Ueda K, et al. Correlation between the blood supply and grade of malignancy of hepatocellular nodules associated with liver cirrhosis: evaluation by CT during intra-arterial injection of contrast medium. Am J Roentgenol. 1999;172:969–976. 10. Akihiro N, Kengo Y, Yoshiki A, et al. Radiologic delectability of minute portal venous invasion in hepatocellular carcinoma. Am J Roentgenol. 2008;190:81–87. 11. Pandharipande PV, Krinsky GA, Rusinek H, et al. Perfusion imaging of the liver: current challenges and future goals. Radiology. 2005;234:661–673. 12. Miles KA, Hayball M, Dixon AK. Colour perfusion imaging: a new application of computed tomography. Lancet. 1991;337:643–645. 13. Miles KA, Charnsangavej C, Lee FT, et al. Application of CT in the investigation of angiogenesis in oncology. Acad Radiol. 2000;7:840–850. 14. Ippolito D, Sironi S, Pozzi M, et al. Perfusion CT in cirrhotic patients with early stage hepatocellular carcinoma: assessment of tumor-related vascularization. Eur J Radiol. 2010;73:148–152. 15. Petralia G, Fazio N, Bonello L, et al. Perfusion computed tomography in patients with hepatocellular carcinoma treated with thalidomide: initial experience. J Comput Assist Tomogr. 2011;35:195–201. 16. Kanda T, Yoshikawa T, Ohno Y, et al. Perfusion measurement of the whole upper abdomen of patients with and without liver diseases: initial experience with 320detector row CT. Eur J Radiol. 2012;81:2470–2475. 17. Chinese Society of Liver Cancer, Chinese Anti-Cancer Association Chinese Society of Clinical Oncology, Chinese Anti-Cancer AssociationLiver Cancer Study Group, Chinese Society of Hepatology, Chinese Medical Association. Expert consensus on the norms of local ablation therapy for hepatocellular carcinoma [in Chinese]. Zhonghua Gan Zang Bing Za Zhi. 2011;19:257–259. 18. Stamm G, Nagel HD. CT-Expo—a novel program for dose evaluation in CT [in German]. Rofo. 2002;174:1570–1576. 19. Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion obtained with dynamic CT. Radiology. 1993;188:405–411. 20. Dugdale PE, Miles KA. Hepatic metastases: the value of quantitative assessment of contrast enhancement on computed tomography. Eur J Radiol. 1999; 30:206–213. 21. Blomley M, Coulden R, Dawson P. Liver perfusion studied with ultrafast CT. J Comput Assist Tomogr. 1995;19:424–433. 22. Tsushima Y, Blomley JK, Kusano S, et al. The portal component of hepatic perfusion measured by dynamic CT: an indicator of hepatic parenchymal damage. Dig Dis Sci 1999;44:1632–1638. 23. Rodriguez-Peralvarez M, Luong TV, Andreana L, et al. A systematic review of microvascular invasion in hepatocellular carcinoma: diagnostic and prognostic variability. Ann Surg Oncol. 2013;20:325–339. 24. Huang ZY, Liang BY, Xiong M, et al. Long-term outcomes of repeat hepatic resection in patients with recurrent hepatocellular carcinoma and analysis of recurrent types and their prognosis: a single-center experience in China. Ann Surg Oncol. 2012;19:2515–2525.

194

www.investigativeradiology.com

25. Chan SC, Fan ST, Chok KS, et al. Survival advantage of primary liver transplantation for hepatocellular carcinoma within the up-to-7 criteria with microvascular invasion. Hepatol Int. 2011;6:646–656. 26. Kim BK, Han KH, Park YN, et al. Prediction of microvascular invasion before curative resection of hepatocellular carcinoma. J Surg Oncol. 2008;97:246–252. 27. Kaibori M, Ishizaki M, Matsui K, et al. Predictors of microvascular invasion before hepatectomy for hepatocellular carcinoma. J Surg Oncol. 2010;102:462–468. 28. Witjes CD, Willemssen FE, Verheij J, et al. Histological differentiation grade and microvascular invasion of hepatocellular carcinoma predicted by dynamic contrast-enhanced MRI. J Magn Reson Imaging. 2012;36:641–647. 29. Eguchi S, Takatsuki M, Hidaka M, et al. Predictor for histological microvascular invasion of hepatocellular carcinoma: a lesson from 229 consecutive cases of curative liver resection. World J Surg. 2010;34:1034–1038. 30. Pawlik TM, Delman KA, Vauthey JN, et al. Tumor size predicts vascular invasion and histologic grade: implications for selection of surgical treatment for hepatocellular carcinoma. Liver Transpl. 2005;11:1086–1092. 31. Sterling R, Wright E, Morgan T, et al. Frequency of elevated hepatocellular carcinoma (HCC) biomarkers in patients with chronic hepatitis C and advanced fibrosis with and without HCC. Hepatology. 2010;52(suppl 1):328A. 32. Faggioni L, Neri E, Bartolozzo C. CT perfusion of head and neck tumors: how we do it. Am J Roentgenol. 2010;194:62–69. 33. Kambadakone AR, Sahani DV. Body perfusion CT: technique, clinical applications, and advances. Radiol Clin North Am. 2009;47:161–178. 34. Spira D, Schulze M, Saulter A, et al. Volume perfusion-CT of the liver: insights and applications. Eur J Radiol. 2012;81:1471–1478. 35. Meijerink MR, Waesberghe JHTM, Weide L, et al. Total-liver-volume perfusion CT using 3-D image fusion to improve detection and characterization of liver metastases. Eur Radiol. 2008;18:2345–2354. 36. Shankar JJ, Lum C. Whole brain CT perfusion on a 320-slice CT scanner. Indian J Radiol Imaging. 2011;21:209–214. 37. Chen C, Liu Q, Hao Q, et al. Study of 320-slice dynamic volume CT perfusion in different pathologic types of kidney tumor: preliminary results. PLoS One. 2014; 9:e85522. 38. Goetti R, Leschka S, Desbiolles L, et al. Quantitative computed tomography liver perfusion imaging using dynamic spiral scanning with variable pitch: feasibility and initial results in patients with cancer metastases. Invest Radiol. 2010;45: 419–426. 39. Goetti R, Reiner CS, Knuth A, et al. Quantitative perfusion analysis of malignant liver tumors: dynamic computed tomography and contrast-enhanced ultrasound. Invest Radiol. 2012;47:18–24. 40. Morsbach F, Pfammatter T, Reiner CS, et al. Computed tomographic perfusion imaging for the prediction of response and survival to transarterial radioembolization of liver metastases. Invest Radiol. 2013;48:787–794. 41. Wang X, Xue H-D, Jin Z-Y, et al. Quantitative hepatic CT perfusion measurement: comparison of Couinaud’s hepatic segments with dual-source 128-slice CT. Eur J Radiol. 2013;82:220–226. 42. Haberland U, Klotz E, Abolmaali N. Performance assessment of dynamic spiral scan modes with variable pitch for quantitative perfusion computed tomography. Invest Radiol. 2010;45:378–386. 43. Tsushima Y, Fubasama S, Aoki J, et al. Quantitative perfusion map of malignant liver tumors, created from dynamic computed tomography data. Acad Radiol. 2004;11:215–223. 44. Fournier LS, Cuenod CA, de Bazelaire C, et al. Early modifications of hepatic perfusion measured by functional CT in a rat model of hepatocellular carcinoma using a blood pool contrast agent. Eur Radiol. 2004;14:2125–2133. 45. Van Beers BE, Leconte I, Materne R, et al. Hepatic perfusion parameters in chronic liver disease: dynamic CT measurements correlated with disease severity. Am J Roentgenol. 2001;176:667–673. 46. Lu XY, Xi T, Lau WY, et al. Pathobiological features of small hepatocellular carcinoma: correlation between tumor size and biological behavior. J Cancer Res Clin Oncol. 2011;137:567–575. 47. Reiner CS, Goetti R, Eberli D, et al. CT perfusion of renal cell carcinoma: impact of volume coverage on quantitative analysis. Invest Radiol. 2012;47:33–40.

© 2014 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Liver computed tomographic perfusion in the assessment of microvascular invasion in patients with small hepatocellular carcinoma.

Detecting microvascular invasion (mVI) in patients with hepatocellular carcinoma is a diagnostic challenge. The present study aimed to acquire a serie...
4MB Sizes 0 Downloads 4 Views