- Email: [email protected]
Hepatocellular Carcinoma in Cirrhotic Liver Disease
Hepatocellular Carcinoma in Cirrhotic Liver Disease: Functional Computed Tomography With Perfusion Imaging in the Assessment of Tumor Vascularization1 Davide Ippolito, MD, Sandro Sironi, MD, Massimo Pozzi, MD, Laura Antolini, MD, Laura Ratti, MD, Chiara Alberzoni, MD Eugenio Biagio Leone, MD, Franca Meloni, MD, Maria Grazia Valsecchi, MD, Ferruccio Fazio, MD
Rationale and Objectives. Our goal was to prospectively determine the value of perfusion computed tomography (CT) in the quantitative assessment of tumor-related angiogenesis in cirrhotic patients with hepatocellular carcinoma (HCC). Materials and Methods. Forty-seven patients met all the following inclusion criteria: 1) Child-Pugh class A or B liver cirrhosis; 2) presence of a single lesion suspected as HCC at screening ultrasound examination; and 3) lesion diameter between 1 and 3 cm. All patients underwent contrast-enhanced ultrasound, pre- and post-contrast triple-phase CT, and perfusion computed tomographic study using multidetector 16-slice CT. Six parameters related to the blood microcirculation and tissue perfusion were measured for the focal liver lesion and cirrhotic parenchyma: perfusion (P), tissue blood volume (BV), hepatic perfusion index (HPI), arterial perfusion (AP), portal perfusion (PP), and time to peak (TTP). Perfusion parameters were described with quartile values of their distribution; univariate paired and unpaired Wilcoxon signed rank tests were used for statistical analysis. Results. HCC was diagnosed in 21 of the 47 patients; in the remaining 26, HCC was not found at contrast-enhanced ultrasound and multidetector 16-slice computed tomographic study. The values of perfusion parameters measured within tumor tissue were: P (ml/s/ 100 g): median ⫽ 47.0 (first quartile ⫽ 36.0, third quartile ⫽ 61.4); BV (ml/100 mg): median ⫽ 24.0 (first quartile ⫽ 18.7, third quartile ⫽ 29.3); HPI (%): median ⫽ 78.4 (first quartile ⫽ 62.9, third quartile ⫽ 100); AP (ml/min): median ⫽ 45.9 (first quartile ⫽ 39.0, third quartile ⫽ 60.1); PP (ml/min): median ⫽ 9.0 (first quartile ⫽ 0.0, third quartile ⫽ 24.5); and TTP (seconds): median ⫽ 18.7 (first quartile ⫽ 16.3, third quartile ⫽ 26.5). The corresponding values calculated in cirrhotic surrounding parenchyma were P (ml/s/100 g): median ⫽ 11.5 (first quartile ⫽ 9.4, third quartile ⫽ 13.9); BV (ml/100 mg): median ⫽ 10.7 (first quartile ⫽ 7.1, third quartile ⫽ 14.2); HPI (%): median ⫽ 10.6 (first quartile ⫽ 8.7, third quartile ⫽ 11.9); AP (ml/min): median ⫽ 13.2 (first quartile ⫽ 10.1, third quartile ⫽ 15.5); PP (ml/min) median ⫽ 55.2 (first quartile ⫽ 40.1, third quartile ⫽ 79.5); and TTP (seconds): median ⫽ 41.7 (first quartile ⫽ 38.9, third quartile ⫽ 44.6). P, BV, HPI, and AP values were higher (P ⬍ .001), whereas PP and TTP were lower (P ⬍ .001) in HCC relative to the surrounding liver. Values of perfusion parameters in the cirrhotic liver of patients with and without HCC were not significantly different (P ⬎ .001). Conclusion. In cirrhotic patients with HCC, perfusion computed tomographic technique can provide quantitative information about tumor-related angiogenesis. Key Words. Computed tomography; perfusion computed tomography; functional computed tomography; liver tumor; blood flow; contrast-enhanced ultrasound. ©
AUR, 2008
Acad Radiol 2008; 15:919 –927 1
From the School of Medicine, University of Milano-Bicocca, Milan, Italy (D.I., S.S., L.A., C.A., E.B.L., M.G.V., F.F.); Departments of Diagnostic Radiology (D.I., S.S., C.A.), Clinical Medicine (M.P., L.R.), Clinical Medicine Prevention and Biotechnology, Section of Medical Statistics (L.A., M.G.V.), and Pathology (E.B.L.), University of Milano-Bicocca, Via Pergolesi 11, H. S. Gerardo Monza 20052, Milan, Italy; Department of Diagnostic Radiology, H Vimercate, Milan, Italy (F.M.); CNR, Institute of Molecular Bioimaging and Physiology, Milan, Italy (F.F.); and Department of Nuclear Medicine, Institute H. S. Raffaele, Milan, Italy (F.F.). Received September 23, 2007; accepted February 9, 2008. Address correspondence to: D.I. e-mail: [email protected]
© AUR, 2008 doi:10.1016/j.acra.2008.02.005
919
IPPOLITO ET AL
As many diseases induce early changes in tissue hemodynamic status, quantitative tissue perfusion imaging could have the ability to characterize pathologic states, establish a diagnosis, and map the response to treatment (1). In patients with liver cirrhosis, a spectrum of focal lesions, including benign regenerative nodules, dysplastic nodules, and hepatocellular carcinoma (HCC) lesions, develop; differences in their respective blood supplies can assist in their detection and characterization (2). Although regenerative nodules receive the majority of blood supply from the portal vein, the evolution from a low-grade dysplastic nodule to frank HCC is associated with a progression toward increasing arterial blood supply, mainly due to tumor-related arterial neovascularization (angiogenesis) (3,4). This is a complex process involving endothelial cell proliferation, capillary formation, coordinated remodeling of extracellular tumor stroma, and anastomosis with the pre-existing host vasculature (5). Tumor angiogenesis may have important implications in the diagnosis and treatment of liver tumors; therefore, the development of clinically applicable techniques enabling its characterization and quantification would be important in the management of these neoplasms (6). Functional computed tomography (CT) with perfusion imaging is a new application in which a quantitative map of tissue perfusion is created from dynamic computed tomographic data and displayed using a color scale, allowing quantification of perfusion in absolute units at high spatial resolution (7–10). Several earlier studies reporting a correlation between contrast enhancement parameters and histologic measurements of angiogenesis have suggested the possible use of perfusion computed tomographic technique as a marker of tumor angiogenesis (11,12). The purpose of this study was to determine the value of functional CT with perfusion imaging in the quantitative assessment of tumor-related arterial angiogenesis in patients with cirrhotic liver disease and HCC, as determined by the European Association for the Study of the Liver (EASL) criteria.
Academic Radiology, Vol 15, No 7, July 2008
sound examination, and (3) maximum lesion diameter between 1 and 3 cm. Excluded from the study were patients with (1) Child-Pugh class C cirrhosis, (2) two or more focal liver lesions at surveillance ultrasound examination, and (3) a single nodule exceeding 3 cm in diameter. Further exclusion criteria were (4) hepatic metastases, (5) thrombosis of the main portal vein branches, and (6) serum creatinine value above 1.5 mg/dl. Eighty-two of the 129 patients initially considered had to be excluded because of one or more of the previously mentioned exclusion criteria. Thus, the final study group consisted of the remaining 47 patients (31 men, age range, 35⫺79 years, mean ⫾ SD, 50 ⫾ 7.3; 16 women, age range, 39⫺67 years, mean ⫾ SD, 47 ⫾ 4.9). Cirrhosis was alcohol-related in six cases, hepatitis C virus (HCV) RNA positive in 33 cases, hepatitis B virus (HBV) DNA positive in 7 cases, and cryptogenic in the remaining case. In agreement with the Child-Pugh classification, 30 patients were classified as Child A, and 17 as Child B. All 47 patients underwent multidetector computed tomographic (MDCT) study, which was performed according to the standard protocol further described in this section. Before being enrolled, all subjects gave their informed consent after the nature of the procedure had been fully explained, in accordance with the regulations of the institutional review board that approved our study. Contrast-Enhanced Ultrasound Contrast-enhanced ultrasound (CEUS) examinations were performed using a Sequoia 512 6.0 unit (Acuson, Mountain View, California), equipped with a convex 3- to 5-MHz transducer and agent detection imaging software. After baseline evaluation, the contrast study was performed using a second-generation contrast agent: 2.4 ml of sulfur hexafluoride microbubbles (SonoVue; Bracco, Milan, Italy) was administered, followed by rapid bolus injection of 5 ml of saline solution. All the images were obtained maintaining the position of the probe on the region of interest and part of the surrounding liver parenchyma.
MATERIALS AND METHODS Patients Between June 2005 and November 2006, a total of 129 consecutive patients with cirrhotic liver disease and high risk for HCC were considered for inclusion in this prospective study. The inclusion criteria of the study were (1) Child-Pugh class A or B liver cirrhosis, (2) presence of a single lesion suspected as HCC at surveillance ultra-
920
MDCT Imaging and Perfusion Computed Tomographic Protocol In all of the 47 patients included in the study, CT was performed using MDCT (Brilliance; Philips Medical Systems, Eindhoven, The Netherlands) with preand post-contrast triple-phase (arterial, portal venous, and equilibrium phases) scans, after intravenous bolus injection of 120 ml of nonionic iodinated contrast ma-
Academic Radiology, Vol 15, No 7, July 2008
PERFUSION CT IN TUMOR VASCULARIZATION ASSESSMENT
terial (Xenetix 350; Guerbet, Aulnay, France) at a rate of 4.5 ml/s, using a 18-gauge catheter positioned into the antecubital vein. The scans were obtained through the liver in a craniocaudal direction with a 2-mm collimation in all phases (pitch of 0.75). Bolus tracking software was used to set individual acquisition times for the arterial, portal, and equilibrium phases. Singlelevel monitoring low-dose scanning (120 kVp, 30 mA) was initiated with contrast material injection. Contrast material enhancement was automatically calculated by placing the region of interest cursor over the abdominal aorta, and the level of trigger threshold was set at an increase of 120 Hounsfield units (HU). Eleven seconds after the trigger threshold had been reached, the arterialphase scanning began automatically. The dynamic imaging consisted of three phases (arterial, portal venous, and equilibrium phases). The mean scanning time delay of arterial phases was 31 seconds. The portal venous and equilibrium phases were acquired after 55 and 120 seconds, respectively, after the trigger threshold had been reached. Perfusion computed tomographic scanning began after selection of the appropriate transverse level; single-location (2.4-cm width) cine computed tomographic scanning (40 scans; eight slices/scan) was performed with the following parameters: 80 kV, 120 mA, 512 ⫻ 512 matrix, 3-mm slice thickness, 1-second scan time, after intravenous bolus injection of 50 ml of nonionic iodinated contrast agent (Xenetix 350; Guerbet) at a rate of 6 ml/s, in addition to the previous 120 ml. To avoid motion artifacts, patients were clearly informed of a possible flushing sensation commonly associated with a rapid bolus of iodinated contrast agent. Moreover, as image quality could be degraded in patients deeply breathing during acquisition, resulting in inaccurate perfusion parameters values, a band compressing the abdomen that limited breath-related liver excursions was used; oxygen inhalation (4 ml/min) was administered to help the patients hold their breath during the examination. Image Analysis and Quantification of Perfusion Parameters The functional computed tomographic images were then transferred to an image workstation, and the data were analyzed using dedicated perfusion software (Philips Brilliance Workspace 2.0; Philips Medical Systems), which generated a quantitative map of the liver perfusion displayed on the monitor using a color scale. The para-
metric map images were created using the highest spatial resolution pixel-by-pixel calculation technique. Perfusion was assessed by the dedicated computed tomographic software based on the maximum slope model: perfusion was therefore calculated as the average slope of the tissue enhancement divided by the peak enhancement in the aorta, as described initially by Miles and Griffiths (8,9). The dedicated perfusion software permitted obtaining six quantitative perfusion parameters using two different perfusion presets: (1) general perfusion and (2) liver perfusion. ●
●
The general perfusion preset was used to obtain quantitative perfusion parameters related to the arterial vascularization. A region of interest was placed on the aorta, as a surrogate for hepatic arterial inflow to determine the enhancement value of arterial input. After positioning of region of interest in the aorta, the following general perfusion parameters were obtained: perfusion (P, ml/s/100 g), which represents the blood flow per unit volume/mass of tissue per second; tissue blood volume (BV, ml/100 g), which is the blood volume contained in 100 g of tissue; and time to peak (TTP, seconds) which is the time to reach the maximum value in contrast material concentration. Using the dedicated liver perfusion preset, we placed a region of interest on the aorta, on the portal vein (or main portal branches), and on the spleen parenchyma to determine the value of portal inflow. Thus, the following perfusion parameters were obtained: arterial perfusion (AP, ml/min), which is the arterial fractional blood flow; portal perfusion (PP, ml/min), which is the portal venous fractional blood flow; and hepatic perfusion index (HPI, %), which represents, of total blood liver flow (arterial perfusion ⫹ portal perfusion), the percentage of arterial origin [AP/AP ⫹ PP].
The regions of interest were placed on each location where an HCC lesion had been detected on the color map (Figs. 1 and 2) and positioned within the tumor. Regions of interests were also placed on the cirrhotic liver parenchyma surrounding the HCC (Figs 1 and 2). In those patients in whom the presence of a focal lesion suspected as HCC had not been confirmed at the MDCT study, regions of interest were positioned in definite areas of the cirrhotic liver parenchyma, and perfusion parameters were calculated. The resulting temporal changes in contrast enhancement were then analyzed to quantify a range of
921
IPPOLITO ET AL
Academic Radiology, Vol 15, No 7, July 2008
Figure 1. Perfusion computed tomographic study performed in a 57-year-old woman with Child B liver cirrhosis and 14-mm focal lesion of hepatocellular carcinoma in the left hepatic lobe. (a) Raw data from multidetector 16-slice computed tomographic scan image shows regions of interest (ROIs) placed on the tumor (T1) and on the surrounding liver parenchyma (T2). (b) Arterial perfusion color map; ROIs are also shown at the same position. (c) Enhancement curves derived from analyses of ROIs. Pixel values in Hounsfield units; A1 ⫽ aorta; V1 ⫽ portal vein; S1 ⫽ spleen; T1 ⫽ tumor; T2 ⫽ surrounding liver parenchyma. Use of dual-input single-compartment modeling techniques for derivation of perfusion parameters required enhancement data from both inputs, aorta and portal vein, as well as the spleen.
parameters that reflects the functional status of tissue perfusion. Standard of Reference According to EASL criteria (13,14), for lesions larger than 2 cm, HCC diagnosis was made on the basis of coincidental findings (arterial hypervascularity) at two different imaging modalities (CEUS and MDCT), with or without increased ␣-fetoprotein (AFP) serum levels. In all the patients where the focal liver lesion detected at CEUS and MDCT showed equivocal imaging features (ie, not coincidental for HCC), particularly in patients with lesions between 1 and 2 cm, fine-needle aspiration biopsy (FNAB) was performed for lesion confirmation. Those patients, in whom the focal liver lesion detected at screening ultrasound was not found at CEUS and MDCT, were followed up with the standard surveillance program
922
at 6-month intervals, consisting of further ultrasound examinations, AFP measurements, and, in selected cases, FNAB. Statistical Analysis The perfusion parameters P, BV, HPI, AP, PP, and TTP were described with the quartiles values of their distribution in the cirrhotic liver parenchyma and within the HCC tissue. The distributions of the individual differences of the parameters between the HCC tissue and the liver parenchyma were graphically represented by boxplot representations. The univariate paired and unpaired Wilcoxon signed rank tests were used to assess whether there were a differential expression of the perfusion parameters between the cirrhotic liver parenchyma and HCC and between the cirrhotic liver in patients with and without HCC. Bonferroni correction was used to take multiple
Academic Radiology, Vol 15, No 7, July 2008
PERFUSION CT IN TUMOR VASCULARIZATION ASSESSMENT
Figure 2. Perfusion computed tomographic (CT) study performed in a 74-year-old man with Child B liver cirrhosis and 11-mm focal lesion of hepatocellular carcinoma in the eighth segment of the liver. (a) Raw data from multidetector 16-slice CT scan image shows regions of interest (ROIs) positioned on the tumor (T1), and on the surrounding liver parenchyma (T2). (b) Arterial perfusion color map; ROIs are also shown at the same position.
comparisons into account, considering P values ⱕ.05/5 indicative of a statistically significant difference (15). In the comparison between the cirrhotic liver and HCC, for P, BV, HPI, and AP, the alternative hypothesis postulated a greater value for the HCC than cirrhotic liver parenchyma, due to a possible increase of the arterial microvessel density within the tumor. For PP and TTP, the alternative hypothesis postulated a lower value for the HCC than cirrhotic liver parenchyma, due to a possible increase of the arterial inflow within the tumor. Statistical analysis was carried out with software SAS (version 8.02, SAS Institute Inc., Cary, North Carolina).
RESULTS Patients with HCC HCC diagnosis was made in 21 of the 47 patients included in the study. In 15 of 21 cases, the diagnosis of HCC was based on imaging features (arterial vascularization) that these nodules showed on both CEUS and MDCT study. These 15 nodules had a maximum diameter between 2 and 3 cm (mean, 2.4). The remaining 6 of 21 patients had a liver nodule between 1 and 2 cm in diameter (mean, 1.3) that was identified at both CEUS and MDCT. The imaging features of these nodules, however, were considered typical of HCC in only two of the six cases. Regardless of imaging findings, a FNAB was performed in all these six cases, which confirmed the presence of HCC lesions in
all of them. Overall, in the 21 patients with HCC, the median AFP serum level was 12.0 UI/ml. Perfusion Parameters in Patients with HCC In the 21 patients with HCC, the descriptive analysis of perfusion parameters measured within the tumor tissue showed the following results: P (ml/s/100 g): median ⫽ 47.0 (first quartile ⫽ 36.0, third quartile ⫽ 61.4); BV (ml/100 mg): median ⫽ 24.0 (first quartile ⫽ 18.7, third quartile ⫽ 29.3); HPI (%): median ⫽ 78.4 (first quartile ⫽ 62.9, third quartile ⫽ 100); AP (ml/min): median ⫽ 45.9 (first quartile ⫽ 39.0, third quartile ⫽ 60.1); PP (ml/min): median ⫽ 9.0 (first quartile ⫽ 0.0, third quartile ⫽ 24.5); and TTP (seconds): median ⫽ 18.7 (first quartile ⫽ 16. 3, third quartile ⫽ 26.5) (Table 1, Fig 3).The corresponding perfusion values calculated in the cirrhotic liver parenchyma surrounding the HCC lesion were P (ml/sec/100 g): median ⫽ 10.4 (first quartile ⫽ 9.3, third quartile ⫽ 13.2); BV (ml/s/100 g): median ⫽ 11.7 (first quartile ⫽ 9.5, third quartile ⫽ 11.9); HPI (%): median ⫽ 16.4 (first quartile ⫽ 13.8, third quartile ⫽ 18.3); AP (ml/min): median ⫽ 10.4 (first quartile ⫽ 9.5, third quartile ⫽ 11.9); PP (ml/min): median ⫽ 55.2 (first quartile ⫽ 40.1, third quartile ⫽ 79.5); and TTP (seconds): median ⫽ 44.6, (first quartile ⫽ 40.3, third quartile ⫽ 51.8) (Table 1, Fig. 3). Parameters showed modest correlations except for P and TTP in cirrotic liver parenchyma and P and AP in focal lesions. Using univariate paired Wilcoxon signed rank test, P, BV, HPI, and AP values were shown to be significantly higher (P ⬍ .001) in the
923
IPPOLITO ET AL
Academic Radiology, Vol 15, No 7, July 2008
Table 1 Descriptive Characteristics of Analyzed Perfusion Parameters
Hepatocellular Carcinoma
Background Liver in Patients with Hepatocellular Carcinoma
Background Liver in Patients without Hepatocellular Carcinoma
Parameter
Median
First Quartile
Third Quartile
Median
First Quartile
Third Quartile
Median
First Quartile
Third Quartile
XP BV HPI AP PP TTP
47.0 24.0 78.4 45.9 9.0 18.7
36.0 18.7 62.9 39.0 0 16.3
61.4 29.3 100 60.1 24.5 26.5
10.4 11.7 16.4 10.4 55.2 44.6
9.3 9.5 13.8 9.5 40.1 40.3
13.2 11.9 18.3 11.9 79.5 51.8
11.5 10.7 10.6 13.2 72.5 41.7
9.4 7.1 8.7 10.1 62.8 38.9
13.9 14.2 11.9 15.5 92.7 44.6
AP, arterial perfusion (ml/min); BV, tissue blood volume (ml/100 g); HPI, hepatic perfusion index (%); PP, portal perfusion (ml/min); TTP, time to peak (TTP, seconds); XP, hepatic perfusion (mL/s/100 g).
Figure 3. Box plots of group perfusion parameters in which boundary of boxes closest to zero indicates 25th percentile, line within boxes marks median, and boundary of boxes farthest from zero indicates 75th percentile. Errors bars below and above boxes indicate minimum and maximum values, respectively. Perfusion (P). Tissue blood volume (BV). Hepatic perfusion index (HPI). Arterial perfusion (AP). Time to peak (TTP). Portal perfusion (PP). P, BV, HPI, and AP values were higher, whereas PP and TTP values were lower, in hepatocellular carcinoma (HCC) relative to the surrounding liver. The values of perfusion parameters as measured in the cirrhotic liver parenchyma of patients with and without hepatocellular carcinoma were not significantly different.
HCC lesion than in the surrounding cirrhotic liver parenchyma. Conversely, PP and TTP values were found to be significantly lower (P ⬍ .001) in HCC relative to adjacent liver parenchyma (Fig 4). Patients without HCC In 26 of the 47 patients included in the study, the focal liver lesion suspected as HCC at screening ultrasound was
924
not found at subsequent CEUS and MDCT study, and the median AFP serum level was 9.3 UI/ml. All of these 26 patients were forwarded, undergoing the standard surveillance program at 6-month intervals, consisting of further ultrasound examinations and AFP measurements. At follow-up ultrasound studies, in all of these 26 patients, the focal liver lesions had the same features shown on the baseline ultrasound. Further, in 14 of these patients,
Academic Radiology, Vol 15, No 7, July 2008
PERFUSION CT IN TUMOR VASCULARIZATION ASSESSMENT
DISCUSSION
Figure 4. Box plots of difference of perfusion parameters (perfusion [P], tissue blood volume [BV], hepatic perfusion index [HPI], arterial perfusion [AP], portal perfusion [PP], and time to peak [TTP]) between hepatocellular carcinoma and surrounding liver (21 patients). The lower boundary of boxes indicates 25th percentile, line within boxes marks median, and the higher boundary of boxes indicates 75th percentile. Errors bars below and above boxes indicate minimum and maximum values, respectively.
FNAB under ultrasound guidance was performed: the results of histopathologic analysis indicated the presence of dysplastic/regenerative nodules, without evidence of tumor cells. Perfusion Parameters in Patients without HCC In the 26 patients without HCC, the descriptive analysis of the perfusion parameters measured in definite regions of cirrhotic liver parenchyma showed the following results: P (ml/s/100 g): median ⫽ 11.5 (first quartile ⫽ 9.4, third quartile ⫽ 13.9); BV (ml/100 mg): median ⫽ 10.7 (first quartile ⫽ 7.1, third quartile ⫽ 14.2); HPI (%): median ⫽ 10.6 (first quartile ⫽ 8.7, third quartile ⫽ 11.9); AP (ml/min): median ⫽ 13.2 (first quartile ⫽ 10.1, third quartile ⫽ 15.5); PP (ml/min): median ⫽ 72.5 (first quartile ⫽ 62.8, third quartile ⫽ 92.7); and TTP (seconds): median ⫽ 41.7 (first quartile ⫽ 38.9, third quartile ⫽ 44.6) (Table 1). The univariate unpaired Wilcoxon signed rank test showed that P, BV, HPI, AP, and TTP values measured in the cirrhotic liver parenchyma of patients without HCC were not significantly different (P ⬎ .001) from those of cirrhotic liver surrounding tumor tissue in patients with HCC (Fig. 3).
Perfusion imaging provides the ability to detect regional and global alterations in organ blood flow (6 –10). Functional CT is a relatively recently developed imaging technology that enables quantification of tissue hemodynamics. By acquiring a continual cine scan at a fixed tissue location after an injection of contrast medium, a timeattenuation curve is generated. On the basis of the linear relationship between the CT-measured enhancement of the contrast medium and the concentration of the contrast medium in the blood, analysis of the time-attenuation curve with appropriate kinetic modeling enables quantification of various aspects of tissue perfusion. In addition, automated image analysis on a pixel-by-pixel basis generates parametric images (functional maps) in which quantitative data for tumor tissue perfusion are displayed. This imaging technique has been used for quantification of several perfusion parameters in normal and pathologic tissues (16). It has also been hypothesized that perfusion parameters as measured with functional CT may constitute physiologic markers related to tumor angiogenesis (17,18). In hepatic nodules, this phenomenon refers to the formation of new arterial vessels, the unpaired arteries, that is, arteries not associated with portal vein branches (6,19). These are crucial to tumor growth, as their number increases from cirrhotic nodules to low- and high-grade dysplastic nodules with the greatest numbers in HCCs (19,20). In the present study, we investigated the value of several tissue perfusion parameters obtained with functional CT in the quantitative assessment of HCC-related angiogenesis. Our results show that the values of the perfusion parameters calculated were significantly different in HCC tissue compared to surrounding cirrhotic liver. In particular, P, BV, HPI, and AP values were found to be higher in HCC than in the surrounding cirrhotic liver tissue, whereas PP and TTP values were lower in HCC than in adjacent liver. All of the six quantitative perfusion parameters evaluated in our study provided stable values. In the assessment of angiogenesis induced by tumor, however, we believe that HPI and AP may likely be considered the most relevant parameters, as they specifically reflect the growth of new arterial blood vessels in the liver nodule. Our findings are in line with those in earlier experimental and clinical studies (21–23). Fournier et al. (21), in a rat model, showed that HCC presented specific changes in perfusion parameters compared to normal liver as mea-
925
IPPOLITO ET AL
sured by CT. Findings in the present work also are consistent with those recently reported in experimental studies where perfusion CT was used for assessing liver tumor response to treatment by evaluating perfusion changes (22,23). Those studies validated the idea that functional CT can provide quantification of residual viable tumor perfusion. In an animal model, Kan et al. (22) found that functional CT enabled accurate quantification of changes in liver tumor perfusion during and after an embolization procedure, thus helping optimize the therapeutic outcomes. In another work (23), the same researchers reported the ability of functional CT to assess changes in liver tumor perfusion in response to antiangiogenic treatment. Results in our study also show that there were no significant differences between perfusion measurements obtained within the cirrhotic liver parenchyma adjacent to an HCC and those of cirrhotic liver tissue in patients without HCC. This indicates that perfusion changes due to HCC growth do not involve the surrounding liver. Interestingly, such finding differs from that of secondary liver malignancy, as it is well known from previous studies with perfusion CT (6,11) that liver metastases are associated with abnormally high arterial perfusion in the surrounding liver. These observations suggest the concept that the use of computed tomographic measurements of perfusion in the liver adjacent to a focal lesion may help in determining its nature. Swensen et al. (24) reported that in malignant lung nodules, a linear correlation could be found between the degree of enhancement at CT and the intensity of angiogenesis, as determined with histopathologic assessment of microvessel density. Similar results were obtained in a later study by Zhang and Kono (25), who compared perfusion measurements with dynamic CT between benign and malignant pulmonary nodules. Dugdale et al. (26) used CT to measure perfusion within lymphoma masses; they also found a positive correlation between perfusion values and intermediate- or high-grade tumors. In comparing results in previous series with ours, however, it should be underlined that while in those studies one dynamic slice per scan was used, in our study eight dynamic slices per scan were used. This gave the chance to assess more lesions or lesions smaller than those previously evaluated with a perfusion computed tomographic technique. There were several limitations in the present study. A certain disadvantage was due to the fact that a liver tissue section of only 2.4 cm in thickness could be examined in each patient. Such limitation, however, should be over-
926
Academic Radiology, Vol 15, No 7, July 2008
come by the advent of MDCT scanners enabling the acquisition of 64 or more sections, which can produce isotropic spatial resolution, while simultaneously delivering exceptional temporal resolution with excellent z-axis coverage (4⫺8 cm) (27). Patient radiation dose, which will increase with the use of such new computed tomographic scanners, is an important issue that may represent a limitation to the use of perfusion computed tomographic technique (27). In the context of oncology, however, the radiation exposure associated with perfusion CT is small compared to the radiotherapy dose that many patients will receive (radiation dose delivered to each patient with our computed tomographic perfusion technique was 10 mSv). Nevertheless, there remains a need to limit the radiation burden associated with perfusion computed tomographic studies. Another limitation in our study was that computed tomographic perfusion parameters were not compared with established markers of tumor vascularity, such as histologic microvessel density counts. Currently, microvascular density count is considered the standard for quantification of angiogenesis in histologic studies. The technique, however, is invasive, which makes it impractical for monitoring treatment.
CONCLUSION Our preliminary study has shown that perfusion CT can provide quantitative information about arterial perfusion of HCC, thus allowing evaluation of tumor-related angiogenesis. It could therefore be used as a complementary diagnostic tool to improve characterization of HCC lesions in cirrhotic patients, beyond that already enabled by conventional imaging techniques. As it is a relatively simple imaging technique, perfusion CT could be integrated into the current computed tomographic protocols, providing an in vivo marker of tumor-related angiogenesis. REFERENCES 1. Miles KA, Hayball M, Dixon AK. Colour perfusion imaging: A new application of computed tomography. Lancet 1991; 337:643– 645. 2. Itai Y, Matsui O. Blood flow and liver imaging. Radiology 1997; 202: 306 –314. 3. Matsui O, Kadoya M, Kameyama T, et al. Benign and malignant nodules in cirrhotic livers: Distinction based on blood supply. Radiology 1991; 178:493– 497. 4. Park YN, Yang CP, Fernandez GJ, Cubukcu O, Thung SN, Theise ND. Neoangio-genesis and sinusoidal capillarization in dysplastic nodules of the liver. Am J Surg Pathol 1998; 22:656 – 662.
Academic Radiology, Vol 15, No 7, July 2008
PERFUSION CT IN TUMOR VASCULARIZATION ASSESSMENT
5. Pluda JM. Tumor-associated angiogenesis: Mechanisms, clinical applications and therapeutic strategies. Semin Oncol 1997; 24:203–218. 6. Pandharipande PV, Krinsky GA, Rusinek H, Lee VS. Perfusion imaging of the liver: Current challenges and future goals. Radiology 2005; 234: 661– 673. 7. Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion obtained with dynamic CT. Radiology 1993; 188:405– 411. 8. Miles KA, Griffiths MR. Perfusion CT: A worthwhile enhancement? Br J Radiol 2003; 76:220 –231. 9. Miles KA. Perfusion CT for the assessment of tumour vascularity: Which protocol? Br J Radiol 2003; 76:36 – 42. 10. Miles KA, Charnsangavej C, Lee FT, Fishman EK, Horton K, Lee TY. Application of CT in the investigation of angiogenesis in oncology. Acad Radiol 2000; 7:840 – 850. 11. Cuenod CA, Leconte I, Siauve N, et al. Deconvolution technique for measuring tissue perfusion by dynamic CT: Application to normal and metastatic liver. Acad Radiol 2002; 9:205–211. 12. Van Beers BE, Leconte I, Materne R, Smith AM, Jamart J, Horsmans Y. Hepatic perfusion parameters in chronic liver disease: Dynamic CT measurements correlated with disease severity. AJR Am J Roentgenol 2001; 176:667– 673. 13. Bruix J, Sherman M, Llovet JM, et al. Clinical management of hepatocellular carcinoma: Conclusions of the Barcelona–2000 EASL Conference. J Hepatol 2001; 35:421– 430. 14. Bruix J, Sherman M. Management of hepatocellular carcinoma. Hepatology 2005; 42:337–356. 15. Hollander M, Wolfe DA. Nonparametric Statistical Methods, 2nd ed. New York: John Wiley and Sons, 1999. 16. Tsushima Y, Funabasama S, Aoki J. Quantitative perfusion map of malignant liver tumors created from dynamic computed tomography data. Acad Radiol 2004; 11:215–223. 17. Passe TJ, Bluemke DA, Siegelman SS. Tumor angiogenesis: Tutorial on implications for imaing. Radiology 1997; 203:593– 600.
18. Li WW. Tumor angiogenesis: Molecular pathology, therapeutic targeting, and imaging. Acad Radiol 2000; 7:800 – 811. 19. 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 intraarterial injection of contrast medium. AJR Am J Roentgenol 1999; 172:969 –976. 20. Kim CK, Lim JH, Park CK, et al. Neoangiogenesis and sinusoidal capillarization in hepatocellular carcinoma: Correlation between dynamic CT and density of tumor microvessels. Radiology 2005; 237:529 –534. 21. 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. 22. Kan Z, Kobayashi S, Phongkitkarun S, Charnsangavej C. Functional CT quantification of tumor perfusion after transhepatic arterial embolization in a rat model. Radiology 2005; 1:144 –150. 23. Kan Z, Phongkitkarun S, Kobayashi S, et al. Functional CT for quantifying tumor perfusion in antiangiogenic therapy in a rat model. Radiology 2005; 237:151–158. 24. Swensen SJ, Brown LR, Colby TV, Weaver AL, Midthun DE. Lung nodule enhancement at CT: Prospective findings. Radiology 1996; 201: 447– 455. 25. Zhang M, Kono M. Solitary pulmonary nodules: Evaluation of blood flow patterns with dynamic CT. Radiology 1997; 205:471– 478. 26. Dugdale PE, Miles KA, Kelly BB, Bunce IH, Legget DAC. CT measurements of perfusion and permeability within lymphoma masses: Relationship to grade, activity and chemotherapeutic response. J Comput Assist Tomogr 1999; 23:540 –547. 27. Boone JM. Multidetector CT: Opportunities, challenges, and concerns associated with scanners with 64 or more detector rows. Radiology 2006; 241:334 –337.
927
Rationale and Objectives. Our goal was to prospectively determine the value of perfusion computed tomography (CT) in the quantitative assessment of tumor-related angiogenesis in cirrhotic patients with hepatocellular carcinoma (HCC). Materials and Methods. Forty-seven patients met all the following inclusion criteria: 1) Child-Pugh class A or B liver cirrhosis; 2) presence of a single lesion suspected as HCC at screening ultrasound examination; and 3) lesion diameter between 1 and 3 cm. All patients underwent contrast-enhanced ultrasound, pre- and post-contrast triple-phase CT, and perfusion computed tomographic study using multidetector 16-slice CT. Six parameters related to the blood microcirculation and tissue perfusion were measured for the focal liver lesion and cirrhotic parenchyma: perfusion (P), tissue blood volume (BV), hepatic perfusion index (HPI), arterial perfusion (AP), portal perfusion (PP), and time to peak (TTP). Perfusion parameters were described with quartile values of their distribution; univariate paired and unpaired Wilcoxon signed rank tests were used for statistical analysis. Results. HCC was diagnosed in 21 of the 47 patients; in the remaining 26, HCC was not found at contrast-enhanced ultrasound and multidetector 16-slice computed tomographic study. The values of perfusion parameters measured within tumor tissue were: P (ml/s/ 100 g): median ⫽ 47.0 (first quartile ⫽ 36.0, third quartile ⫽ 61.4); BV (ml/100 mg): median ⫽ 24.0 (first quartile ⫽ 18.7, third quartile ⫽ 29.3); HPI (%): median ⫽ 78.4 (first quartile ⫽ 62.9, third quartile ⫽ 100); AP (ml/min): median ⫽ 45.9 (first quartile ⫽ 39.0, third quartile ⫽ 60.1); PP (ml/min): median ⫽ 9.0 (first quartile ⫽ 0.0, third quartile ⫽ 24.5); and TTP (seconds): median ⫽ 18.7 (first quartile ⫽ 16.3, third quartile ⫽ 26.5). The corresponding values calculated in cirrhotic surrounding parenchyma were P (ml/s/100 g): median ⫽ 11.5 (first quartile ⫽ 9.4, third quartile ⫽ 13.9); BV (ml/100 mg): median ⫽ 10.7 (first quartile ⫽ 7.1, third quartile ⫽ 14.2); HPI (%): median ⫽ 10.6 (first quartile ⫽ 8.7, third quartile ⫽ 11.9); AP (ml/min): median ⫽ 13.2 (first quartile ⫽ 10.1, third quartile ⫽ 15.5); PP (ml/min) median ⫽ 55.2 (first quartile ⫽ 40.1, third quartile ⫽ 79.5); and TTP (seconds): median ⫽ 41.7 (first quartile ⫽ 38.9, third quartile ⫽ 44.6). P, BV, HPI, and AP values were higher (P ⬍ .001), whereas PP and TTP were lower (P ⬍ .001) in HCC relative to the surrounding liver. Values of perfusion parameters in the cirrhotic liver of patients with and without HCC were not significantly different (P ⬎ .001). Conclusion. In cirrhotic patients with HCC, perfusion computed tomographic technique can provide quantitative information about tumor-related angiogenesis. Key Words. Computed tomography; perfusion computed tomography; functional computed tomography; liver tumor; blood flow; contrast-enhanced ultrasound. ©
AUR, 2008
Acad Radiol 2008; 15:919 –927 1
From the School of Medicine, University of Milano-Bicocca, Milan, Italy (D.I., S.S., L.A., C.A., E.B.L., M.G.V., F.F.); Departments of Diagnostic Radiology (D.I., S.S., C.A.), Clinical Medicine (M.P., L.R.), Clinical Medicine Prevention and Biotechnology, Section of Medical Statistics (L.A., M.G.V.), and Pathology (E.B.L.), University of Milano-Bicocca, Via Pergolesi 11, H. S. Gerardo Monza 20052, Milan, Italy; Department of Diagnostic Radiology, H Vimercate, Milan, Italy (F.M.); CNR, Institute of Molecular Bioimaging and Physiology, Milan, Italy (F.F.); and Department of Nuclear Medicine, Institute H. S. Raffaele, Milan, Italy (F.F.). Received September 23, 2007; accepted February 9, 2008. Address correspondence to: D.I. e-mail: [email protected]
© AUR, 2008 doi:10.1016/j.acra.2008.02.005
919
IPPOLITO ET AL
As many diseases induce early changes in tissue hemodynamic status, quantitative tissue perfusion imaging could have the ability to characterize pathologic states, establish a diagnosis, and map the response to treatment (1). In patients with liver cirrhosis, a spectrum of focal lesions, including benign regenerative nodules, dysplastic nodules, and hepatocellular carcinoma (HCC) lesions, develop; differences in their respective blood supplies can assist in their detection and characterization (2). Although regenerative nodules receive the majority of blood supply from the portal vein, the evolution from a low-grade dysplastic nodule to frank HCC is associated with a progression toward increasing arterial blood supply, mainly due to tumor-related arterial neovascularization (angiogenesis) (3,4). This is a complex process involving endothelial cell proliferation, capillary formation, coordinated remodeling of extracellular tumor stroma, and anastomosis with the pre-existing host vasculature (5). Tumor angiogenesis may have important implications in the diagnosis and treatment of liver tumors; therefore, the development of clinically applicable techniques enabling its characterization and quantification would be important in the management of these neoplasms (6). Functional computed tomography (CT) with perfusion imaging is a new application in which a quantitative map of tissue perfusion is created from dynamic computed tomographic data and displayed using a color scale, allowing quantification of perfusion in absolute units at high spatial resolution (7–10). Several earlier studies reporting a correlation between contrast enhancement parameters and histologic measurements of angiogenesis have suggested the possible use of perfusion computed tomographic technique as a marker of tumor angiogenesis (11,12). The purpose of this study was to determine the value of functional CT with perfusion imaging in the quantitative assessment of tumor-related arterial angiogenesis in patients with cirrhotic liver disease and HCC, as determined by the European Association for the Study of the Liver (EASL) criteria.
Academic Radiology, Vol 15, No 7, July 2008
sound examination, and (3) maximum lesion diameter between 1 and 3 cm. Excluded from the study were patients with (1) Child-Pugh class C cirrhosis, (2) two or more focal liver lesions at surveillance ultrasound examination, and (3) a single nodule exceeding 3 cm in diameter. Further exclusion criteria were (4) hepatic metastases, (5) thrombosis of the main portal vein branches, and (6) serum creatinine value above 1.5 mg/dl. Eighty-two of the 129 patients initially considered had to be excluded because of one or more of the previously mentioned exclusion criteria. Thus, the final study group consisted of the remaining 47 patients (31 men, age range, 35⫺79 years, mean ⫾ SD, 50 ⫾ 7.3; 16 women, age range, 39⫺67 years, mean ⫾ SD, 47 ⫾ 4.9). Cirrhosis was alcohol-related in six cases, hepatitis C virus (HCV) RNA positive in 33 cases, hepatitis B virus (HBV) DNA positive in 7 cases, and cryptogenic in the remaining case. In agreement with the Child-Pugh classification, 30 patients were classified as Child A, and 17 as Child B. All 47 patients underwent multidetector computed tomographic (MDCT) study, which was performed according to the standard protocol further described in this section. Before being enrolled, all subjects gave their informed consent after the nature of the procedure had been fully explained, in accordance with the regulations of the institutional review board that approved our study. Contrast-Enhanced Ultrasound Contrast-enhanced ultrasound (CEUS) examinations were performed using a Sequoia 512 6.0 unit (Acuson, Mountain View, California), equipped with a convex 3- to 5-MHz transducer and agent detection imaging software. After baseline evaluation, the contrast study was performed using a second-generation contrast agent: 2.4 ml of sulfur hexafluoride microbubbles (SonoVue; Bracco, Milan, Italy) was administered, followed by rapid bolus injection of 5 ml of saline solution. All the images were obtained maintaining the position of the probe on the region of interest and part of the surrounding liver parenchyma.
MATERIALS AND METHODS Patients Between June 2005 and November 2006, a total of 129 consecutive patients with cirrhotic liver disease and high risk for HCC were considered for inclusion in this prospective study. The inclusion criteria of the study were (1) Child-Pugh class A or B liver cirrhosis, (2) presence of a single lesion suspected as HCC at surveillance ultra-
920
MDCT Imaging and Perfusion Computed Tomographic Protocol In all of the 47 patients included in the study, CT was performed using MDCT (Brilliance; Philips Medical Systems, Eindhoven, The Netherlands) with preand post-contrast triple-phase (arterial, portal venous, and equilibrium phases) scans, after intravenous bolus injection of 120 ml of nonionic iodinated contrast ma-
Academic Radiology, Vol 15, No 7, July 2008
PERFUSION CT IN TUMOR VASCULARIZATION ASSESSMENT
terial (Xenetix 350; Guerbet, Aulnay, France) at a rate of 4.5 ml/s, using a 18-gauge catheter positioned into the antecubital vein. The scans were obtained through the liver in a craniocaudal direction with a 2-mm collimation in all phases (pitch of 0.75). Bolus tracking software was used to set individual acquisition times for the arterial, portal, and equilibrium phases. Singlelevel monitoring low-dose scanning (120 kVp, 30 mA) was initiated with contrast material injection. Contrast material enhancement was automatically calculated by placing the region of interest cursor over the abdominal aorta, and the level of trigger threshold was set at an increase of 120 Hounsfield units (HU). Eleven seconds after the trigger threshold had been reached, the arterialphase scanning began automatically. The dynamic imaging consisted of three phases (arterial, portal venous, and equilibrium phases). The mean scanning time delay of arterial phases was 31 seconds. The portal venous and equilibrium phases were acquired after 55 and 120 seconds, respectively, after the trigger threshold had been reached. Perfusion computed tomographic scanning began after selection of the appropriate transverse level; single-location (2.4-cm width) cine computed tomographic scanning (40 scans; eight slices/scan) was performed with the following parameters: 80 kV, 120 mA, 512 ⫻ 512 matrix, 3-mm slice thickness, 1-second scan time, after intravenous bolus injection of 50 ml of nonionic iodinated contrast agent (Xenetix 350; Guerbet) at a rate of 6 ml/s, in addition to the previous 120 ml. To avoid motion artifacts, patients were clearly informed of a possible flushing sensation commonly associated with a rapid bolus of iodinated contrast agent. Moreover, as image quality could be degraded in patients deeply breathing during acquisition, resulting in inaccurate perfusion parameters values, a band compressing the abdomen that limited breath-related liver excursions was used; oxygen inhalation (4 ml/min) was administered to help the patients hold their breath during the examination. Image Analysis and Quantification of Perfusion Parameters The functional computed tomographic images were then transferred to an image workstation, and the data were analyzed using dedicated perfusion software (Philips Brilliance Workspace 2.0; Philips Medical Systems), which generated a quantitative map of the liver perfusion displayed on the monitor using a color scale. The para-
metric map images were created using the highest spatial resolution pixel-by-pixel calculation technique. Perfusion was assessed by the dedicated computed tomographic software based on the maximum slope model: perfusion was therefore calculated as the average slope of the tissue enhancement divided by the peak enhancement in the aorta, as described initially by Miles and Griffiths (8,9). The dedicated perfusion software permitted obtaining six quantitative perfusion parameters using two different perfusion presets: (1) general perfusion and (2) liver perfusion. ●
●
The general perfusion preset was used to obtain quantitative perfusion parameters related to the arterial vascularization. A region of interest was placed on the aorta, as a surrogate for hepatic arterial inflow to determine the enhancement value of arterial input. After positioning of region of interest in the aorta, the following general perfusion parameters were obtained: perfusion (P, ml/s/100 g), which represents the blood flow per unit volume/mass of tissue per second; tissue blood volume (BV, ml/100 g), which is the blood volume contained in 100 g of tissue; and time to peak (TTP, seconds) which is the time to reach the maximum value in contrast material concentration. Using the dedicated liver perfusion preset, we placed a region of interest on the aorta, on the portal vein (or main portal branches), and on the spleen parenchyma to determine the value of portal inflow. Thus, the following perfusion parameters were obtained: arterial perfusion (AP, ml/min), which is the arterial fractional blood flow; portal perfusion (PP, ml/min), which is the portal venous fractional blood flow; and hepatic perfusion index (HPI, %), which represents, of total blood liver flow (arterial perfusion ⫹ portal perfusion), the percentage of arterial origin [AP/AP ⫹ PP].
The regions of interest were placed on each location where an HCC lesion had been detected on the color map (Figs. 1 and 2) and positioned within the tumor. Regions of interests were also placed on the cirrhotic liver parenchyma surrounding the HCC (Figs 1 and 2). In those patients in whom the presence of a focal lesion suspected as HCC had not been confirmed at the MDCT study, regions of interest were positioned in definite areas of the cirrhotic liver parenchyma, and perfusion parameters were calculated. The resulting temporal changes in contrast enhancement were then analyzed to quantify a range of
921
IPPOLITO ET AL
Academic Radiology, Vol 15, No 7, July 2008
Figure 1. Perfusion computed tomographic study performed in a 57-year-old woman with Child B liver cirrhosis and 14-mm focal lesion of hepatocellular carcinoma in the left hepatic lobe. (a) Raw data from multidetector 16-slice computed tomographic scan image shows regions of interest (ROIs) placed on the tumor (T1) and on the surrounding liver parenchyma (T2). (b) Arterial perfusion color map; ROIs are also shown at the same position. (c) Enhancement curves derived from analyses of ROIs. Pixel values in Hounsfield units; A1 ⫽ aorta; V1 ⫽ portal vein; S1 ⫽ spleen; T1 ⫽ tumor; T2 ⫽ surrounding liver parenchyma. Use of dual-input single-compartment modeling techniques for derivation of perfusion parameters required enhancement data from both inputs, aorta and portal vein, as well as the spleen.
parameters that reflects the functional status of tissue perfusion. Standard of Reference According to EASL criteria (13,14), for lesions larger than 2 cm, HCC diagnosis was made on the basis of coincidental findings (arterial hypervascularity) at two different imaging modalities (CEUS and MDCT), with or without increased ␣-fetoprotein (AFP) serum levels. In all the patients where the focal liver lesion detected at CEUS and MDCT showed equivocal imaging features (ie, not coincidental for HCC), particularly in patients with lesions between 1 and 2 cm, fine-needle aspiration biopsy (FNAB) was performed for lesion confirmation. Those patients, in whom the focal liver lesion detected at screening ultrasound was not found at CEUS and MDCT, were followed up with the standard surveillance program
922
at 6-month intervals, consisting of further ultrasound examinations, AFP measurements, and, in selected cases, FNAB. Statistical Analysis The perfusion parameters P, BV, HPI, AP, PP, and TTP were described with the quartiles values of their distribution in the cirrhotic liver parenchyma and within the HCC tissue. The distributions of the individual differences of the parameters between the HCC tissue and the liver parenchyma were graphically represented by boxplot representations. The univariate paired and unpaired Wilcoxon signed rank tests were used to assess whether there were a differential expression of the perfusion parameters between the cirrhotic liver parenchyma and HCC and between the cirrhotic liver in patients with and without HCC. Bonferroni correction was used to take multiple
Academic Radiology, Vol 15, No 7, July 2008
PERFUSION CT IN TUMOR VASCULARIZATION ASSESSMENT
Figure 2. Perfusion computed tomographic (CT) study performed in a 74-year-old man with Child B liver cirrhosis and 11-mm focal lesion of hepatocellular carcinoma in the eighth segment of the liver. (a) Raw data from multidetector 16-slice CT scan image shows regions of interest (ROIs) positioned on the tumor (T1), and on the surrounding liver parenchyma (T2). (b) Arterial perfusion color map; ROIs are also shown at the same position.
comparisons into account, considering P values ⱕ.05/5 indicative of a statistically significant difference (15). In the comparison between the cirrhotic liver and HCC, for P, BV, HPI, and AP, the alternative hypothesis postulated a greater value for the HCC than cirrhotic liver parenchyma, due to a possible increase of the arterial microvessel density within the tumor. For PP and TTP, the alternative hypothesis postulated a lower value for the HCC than cirrhotic liver parenchyma, due to a possible increase of the arterial inflow within the tumor. Statistical analysis was carried out with software SAS (version 8.02, SAS Institute Inc., Cary, North Carolina).
RESULTS Patients with HCC HCC diagnosis was made in 21 of the 47 patients included in the study. In 15 of 21 cases, the diagnosis of HCC was based on imaging features (arterial vascularization) that these nodules showed on both CEUS and MDCT study. These 15 nodules had a maximum diameter between 2 and 3 cm (mean, 2.4). The remaining 6 of 21 patients had a liver nodule between 1 and 2 cm in diameter (mean, 1.3) that was identified at both CEUS and MDCT. The imaging features of these nodules, however, were considered typical of HCC in only two of the six cases. Regardless of imaging findings, a FNAB was performed in all these six cases, which confirmed the presence of HCC lesions in
all of them. Overall, in the 21 patients with HCC, the median AFP serum level was 12.0 UI/ml. Perfusion Parameters in Patients with HCC In the 21 patients with HCC, the descriptive analysis of perfusion parameters measured within the tumor tissue showed the following results: P (ml/s/100 g): median ⫽ 47.0 (first quartile ⫽ 36.0, third quartile ⫽ 61.4); BV (ml/100 mg): median ⫽ 24.0 (first quartile ⫽ 18.7, third quartile ⫽ 29.3); HPI (%): median ⫽ 78.4 (first quartile ⫽ 62.9, third quartile ⫽ 100); AP (ml/min): median ⫽ 45.9 (first quartile ⫽ 39.0, third quartile ⫽ 60.1); PP (ml/min): median ⫽ 9.0 (first quartile ⫽ 0.0, third quartile ⫽ 24.5); and TTP (seconds): median ⫽ 18.7 (first quartile ⫽ 16. 3, third quartile ⫽ 26.5) (Table 1, Fig 3).The corresponding perfusion values calculated in the cirrhotic liver parenchyma surrounding the HCC lesion were P (ml/sec/100 g): median ⫽ 10.4 (first quartile ⫽ 9.3, third quartile ⫽ 13.2); BV (ml/s/100 g): median ⫽ 11.7 (first quartile ⫽ 9.5, third quartile ⫽ 11.9); HPI (%): median ⫽ 16.4 (first quartile ⫽ 13.8, third quartile ⫽ 18.3); AP (ml/min): median ⫽ 10.4 (first quartile ⫽ 9.5, third quartile ⫽ 11.9); PP (ml/min): median ⫽ 55.2 (first quartile ⫽ 40.1, third quartile ⫽ 79.5); and TTP (seconds): median ⫽ 44.6, (first quartile ⫽ 40.3, third quartile ⫽ 51.8) (Table 1, Fig. 3). Parameters showed modest correlations except for P and TTP in cirrotic liver parenchyma and P and AP in focal lesions. Using univariate paired Wilcoxon signed rank test, P, BV, HPI, and AP values were shown to be significantly higher (P ⬍ .001) in the
923
IPPOLITO ET AL
Academic Radiology, Vol 15, No 7, July 2008
Table 1 Descriptive Characteristics of Analyzed Perfusion Parameters
Hepatocellular Carcinoma
Background Liver in Patients with Hepatocellular Carcinoma
Background Liver in Patients without Hepatocellular Carcinoma
Parameter
Median
First Quartile
Third Quartile
Median
First Quartile
Third Quartile
Median
First Quartile
Third Quartile
XP BV HPI AP PP TTP
47.0 24.0 78.4 45.9 9.0 18.7
36.0 18.7 62.9 39.0 0 16.3
61.4 29.3 100 60.1 24.5 26.5
10.4 11.7 16.4 10.4 55.2 44.6
9.3 9.5 13.8 9.5 40.1 40.3
13.2 11.9 18.3 11.9 79.5 51.8
11.5 10.7 10.6 13.2 72.5 41.7
9.4 7.1 8.7 10.1 62.8 38.9
13.9 14.2 11.9 15.5 92.7 44.6
AP, arterial perfusion (ml/min); BV, tissue blood volume (ml/100 g); HPI, hepatic perfusion index (%); PP, portal perfusion (ml/min); TTP, time to peak (TTP, seconds); XP, hepatic perfusion (mL/s/100 g).
Figure 3. Box plots of group perfusion parameters in which boundary of boxes closest to zero indicates 25th percentile, line within boxes marks median, and boundary of boxes farthest from zero indicates 75th percentile. Errors bars below and above boxes indicate minimum and maximum values, respectively. Perfusion (P). Tissue blood volume (BV). Hepatic perfusion index (HPI). Arterial perfusion (AP). Time to peak (TTP). Portal perfusion (PP). P, BV, HPI, and AP values were higher, whereas PP and TTP values were lower, in hepatocellular carcinoma (HCC) relative to the surrounding liver. The values of perfusion parameters as measured in the cirrhotic liver parenchyma of patients with and without hepatocellular carcinoma were not significantly different.
HCC lesion than in the surrounding cirrhotic liver parenchyma. Conversely, PP and TTP values were found to be significantly lower (P ⬍ .001) in HCC relative to adjacent liver parenchyma (Fig 4). Patients without HCC In 26 of the 47 patients included in the study, the focal liver lesion suspected as HCC at screening ultrasound was
924
not found at subsequent CEUS and MDCT study, and the median AFP serum level was 9.3 UI/ml. All of these 26 patients were forwarded, undergoing the standard surveillance program at 6-month intervals, consisting of further ultrasound examinations and AFP measurements. At follow-up ultrasound studies, in all of these 26 patients, the focal liver lesions had the same features shown on the baseline ultrasound. Further, in 14 of these patients,
Academic Radiology, Vol 15, No 7, July 2008
PERFUSION CT IN TUMOR VASCULARIZATION ASSESSMENT
DISCUSSION
Figure 4. Box plots of difference of perfusion parameters (perfusion [P], tissue blood volume [BV], hepatic perfusion index [HPI], arterial perfusion [AP], portal perfusion [PP], and time to peak [TTP]) between hepatocellular carcinoma and surrounding liver (21 patients). The lower boundary of boxes indicates 25th percentile, line within boxes marks median, and the higher boundary of boxes indicates 75th percentile. Errors bars below and above boxes indicate minimum and maximum values, respectively.
FNAB under ultrasound guidance was performed: the results of histopathologic analysis indicated the presence of dysplastic/regenerative nodules, without evidence of tumor cells. Perfusion Parameters in Patients without HCC In the 26 patients without HCC, the descriptive analysis of the perfusion parameters measured in definite regions of cirrhotic liver parenchyma showed the following results: P (ml/s/100 g): median ⫽ 11.5 (first quartile ⫽ 9.4, third quartile ⫽ 13.9); BV (ml/100 mg): median ⫽ 10.7 (first quartile ⫽ 7.1, third quartile ⫽ 14.2); HPI (%): median ⫽ 10.6 (first quartile ⫽ 8.7, third quartile ⫽ 11.9); AP (ml/min): median ⫽ 13.2 (first quartile ⫽ 10.1, third quartile ⫽ 15.5); PP (ml/min): median ⫽ 72.5 (first quartile ⫽ 62.8, third quartile ⫽ 92.7); and TTP (seconds): median ⫽ 41.7 (first quartile ⫽ 38.9, third quartile ⫽ 44.6) (Table 1). The univariate unpaired Wilcoxon signed rank test showed that P, BV, HPI, AP, and TTP values measured in the cirrhotic liver parenchyma of patients without HCC were not significantly different (P ⬎ .001) from those of cirrhotic liver surrounding tumor tissue in patients with HCC (Fig. 3).
Perfusion imaging provides the ability to detect regional and global alterations in organ blood flow (6 –10). Functional CT is a relatively recently developed imaging technology that enables quantification of tissue hemodynamics. By acquiring a continual cine scan at a fixed tissue location after an injection of contrast medium, a timeattenuation curve is generated. On the basis of the linear relationship between the CT-measured enhancement of the contrast medium and the concentration of the contrast medium in the blood, analysis of the time-attenuation curve with appropriate kinetic modeling enables quantification of various aspects of tissue perfusion. In addition, automated image analysis on a pixel-by-pixel basis generates parametric images (functional maps) in which quantitative data for tumor tissue perfusion are displayed. This imaging technique has been used for quantification of several perfusion parameters in normal and pathologic tissues (16). It has also been hypothesized that perfusion parameters as measured with functional CT may constitute physiologic markers related to tumor angiogenesis (17,18). In hepatic nodules, this phenomenon refers to the formation of new arterial vessels, the unpaired arteries, that is, arteries not associated with portal vein branches (6,19). These are crucial to tumor growth, as their number increases from cirrhotic nodules to low- and high-grade dysplastic nodules with the greatest numbers in HCCs (19,20). In the present study, we investigated the value of several tissue perfusion parameters obtained with functional CT in the quantitative assessment of HCC-related angiogenesis. Our results show that the values of the perfusion parameters calculated were significantly different in HCC tissue compared to surrounding cirrhotic liver. In particular, P, BV, HPI, and AP values were found to be higher in HCC than in the surrounding cirrhotic liver tissue, whereas PP and TTP values were lower in HCC than in adjacent liver. All of the six quantitative perfusion parameters evaluated in our study provided stable values. In the assessment of angiogenesis induced by tumor, however, we believe that HPI and AP may likely be considered the most relevant parameters, as they specifically reflect the growth of new arterial blood vessels in the liver nodule. Our findings are in line with those in earlier experimental and clinical studies (21–23). Fournier et al. (21), in a rat model, showed that HCC presented specific changes in perfusion parameters compared to normal liver as mea-
925
IPPOLITO ET AL
sured by CT. Findings in the present work also are consistent with those recently reported in experimental studies where perfusion CT was used for assessing liver tumor response to treatment by evaluating perfusion changes (22,23). Those studies validated the idea that functional CT can provide quantification of residual viable tumor perfusion. In an animal model, Kan et al. (22) found that functional CT enabled accurate quantification of changes in liver tumor perfusion during and after an embolization procedure, thus helping optimize the therapeutic outcomes. In another work (23), the same researchers reported the ability of functional CT to assess changes in liver tumor perfusion in response to antiangiogenic treatment. Results in our study also show that there were no significant differences between perfusion measurements obtained within the cirrhotic liver parenchyma adjacent to an HCC and those of cirrhotic liver tissue in patients without HCC. This indicates that perfusion changes due to HCC growth do not involve the surrounding liver. Interestingly, such finding differs from that of secondary liver malignancy, as it is well known from previous studies with perfusion CT (6,11) that liver metastases are associated with abnormally high arterial perfusion in the surrounding liver. These observations suggest the concept that the use of computed tomographic measurements of perfusion in the liver adjacent to a focal lesion may help in determining its nature. Swensen et al. (24) reported that in malignant lung nodules, a linear correlation could be found between the degree of enhancement at CT and the intensity of angiogenesis, as determined with histopathologic assessment of microvessel density. Similar results were obtained in a later study by Zhang and Kono (25), who compared perfusion measurements with dynamic CT between benign and malignant pulmonary nodules. Dugdale et al. (26) used CT to measure perfusion within lymphoma masses; they also found a positive correlation between perfusion values and intermediate- or high-grade tumors. In comparing results in previous series with ours, however, it should be underlined that while in those studies one dynamic slice per scan was used, in our study eight dynamic slices per scan were used. This gave the chance to assess more lesions or lesions smaller than those previously evaluated with a perfusion computed tomographic technique. There were several limitations in the present study. A certain disadvantage was due to the fact that a liver tissue section of only 2.4 cm in thickness could be examined in each patient. Such limitation, however, should be over-
926
Academic Radiology, Vol 15, No 7, July 2008
come by the advent of MDCT scanners enabling the acquisition of 64 or more sections, which can produce isotropic spatial resolution, while simultaneously delivering exceptional temporal resolution with excellent z-axis coverage (4⫺8 cm) (27). Patient radiation dose, which will increase with the use of such new computed tomographic scanners, is an important issue that may represent a limitation to the use of perfusion computed tomographic technique (27). In the context of oncology, however, the radiation exposure associated with perfusion CT is small compared to the radiotherapy dose that many patients will receive (radiation dose delivered to each patient with our computed tomographic perfusion technique was 10 mSv). Nevertheless, there remains a need to limit the radiation burden associated with perfusion computed tomographic studies. Another limitation in our study was that computed tomographic perfusion parameters were not compared with established markers of tumor vascularity, such as histologic microvessel density counts. Currently, microvascular density count is considered the standard for quantification of angiogenesis in histologic studies. The technique, however, is invasive, which makes it impractical for monitoring treatment.
CONCLUSION Our preliminary study has shown that perfusion CT can provide quantitative information about arterial perfusion of HCC, thus allowing evaluation of tumor-related angiogenesis. It could therefore be used as a complementary diagnostic tool to improve characterization of HCC lesions in cirrhotic patients, beyond that already enabled by conventional imaging techniques. As it is a relatively simple imaging technique, perfusion CT could be integrated into the current computed tomographic protocols, providing an in vivo marker of tumor-related angiogenesis. REFERENCES 1. Miles KA, Hayball M, Dixon AK. Colour perfusion imaging: A new application of computed tomography. Lancet 1991; 337:643– 645. 2. Itai Y, Matsui O. Blood flow and liver imaging. Radiology 1997; 202: 306 –314. 3. Matsui O, Kadoya M, Kameyama T, et al. Benign and malignant nodules in cirrhotic livers: Distinction based on blood supply. Radiology 1991; 178:493– 497. 4. Park YN, Yang CP, Fernandez GJ, Cubukcu O, Thung SN, Theise ND. Neoangio-genesis and sinusoidal capillarization in dysplastic nodules of the liver. Am J Surg Pathol 1998; 22:656 – 662.
Academic Radiology, Vol 15, No 7, July 2008
PERFUSION CT IN TUMOR VASCULARIZATION ASSESSMENT
5. Pluda JM. Tumor-associated angiogenesis: Mechanisms, clinical applications and therapeutic strategies. Semin Oncol 1997; 24:203–218. 6. Pandharipande PV, Krinsky GA, Rusinek H, Lee VS. Perfusion imaging of the liver: Current challenges and future goals. Radiology 2005; 234: 661– 673. 7. Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion obtained with dynamic CT. Radiology 1993; 188:405– 411. 8. Miles KA, Griffiths MR. Perfusion CT: A worthwhile enhancement? Br J Radiol 2003; 76:220 –231. 9. Miles KA. Perfusion CT for the assessment of tumour vascularity: Which protocol? Br J Radiol 2003; 76:36 – 42. 10. Miles KA, Charnsangavej C, Lee FT, Fishman EK, Horton K, Lee TY. Application of CT in the investigation of angiogenesis in oncology. Acad Radiol 2000; 7:840 – 850. 11. Cuenod CA, Leconte I, Siauve N, et al. Deconvolution technique for measuring tissue perfusion by dynamic CT: Application to normal and metastatic liver. Acad Radiol 2002; 9:205–211. 12. Van Beers BE, Leconte I, Materne R, Smith AM, Jamart J, Horsmans Y. Hepatic perfusion parameters in chronic liver disease: Dynamic CT measurements correlated with disease severity. AJR Am J Roentgenol 2001; 176:667– 673. 13. Bruix J, Sherman M, Llovet JM, et al. Clinical management of hepatocellular carcinoma: Conclusions of the Barcelona–2000 EASL Conference. J Hepatol 2001; 35:421– 430. 14. Bruix J, Sherman M. Management of hepatocellular carcinoma. Hepatology 2005; 42:337–356. 15. Hollander M, Wolfe DA. Nonparametric Statistical Methods, 2nd ed. New York: John Wiley and Sons, 1999. 16. Tsushima Y, Funabasama S, Aoki J. Quantitative perfusion map of malignant liver tumors created from dynamic computed tomography data. Acad Radiol 2004; 11:215–223. 17. Passe TJ, Bluemke DA, Siegelman SS. Tumor angiogenesis: Tutorial on implications for imaing. Radiology 1997; 203:593– 600.
18. Li WW. Tumor angiogenesis: Molecular pathology, therapeutic targeting, and imaging. Acad Radiol 2000; 7:800 – 811. 19. 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 intraarterial injection of contrast medium. AJR Am J Roentgenol 1999; 172:969 –976. 20. Kim CK, Lim JH, Park CK, et al. Neoangiogenesis and sinusoidal capillarization in hepatocellular carcinoma: Correlation between dynamic CT and density of tumor microvessels. Radiology 2005; 237:529 –534. 21. 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. 22. Kan Z, Kobayashi S, Phongkitkarun S, Charnsangavej C. Functional CT quantification of tumor perfusion after transhepatic arterial embolization in a rat model. Radiology 2005; 1:144 –150. 23. Kan Z, Phongkitkarun S, Kobayashi S, et al. Functional CT for quantifying tumor perfusion in antiangiogenic therapy in a rat model. Radiology 2005; 237:151–158. 24. Swensen SJ, Brown LR, Colby TV, Weaver AL, Midthun DE. Lung nodule enhancement at CT: Prospective findings. Radiology 1996; 201: 447– 455. 25. Zhang M, Kono M. Solitary pulmonary nodules: Evaluation of blood flow patterns with dynamic CT. Radiology 1997; 205:471– 478. 26. Dugdale PE, Miles KA, Kelly BB, Bunce IH, Legget DAC. CT measurements of perfusion and permeability within lymphoma masses: Relationship to grade, activity and chemotherapeutic response. J Comput Assist Tomogr 1999; 23:540 –547. 27. Boone JM. Multidetector CT: Opportunities, challenges, and concerns associated with scanners with 64 or more detector rows. Radiology 2006; 241:334 –337.
927