| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiological Sciences (R.I., A.L., C.C., P.R., F.M., F.P., R.P.) and Experimental Medicine and Pathology–Medical Biostatistics (I.N.), University of Rome La Sapienza, Policlinico Umberto I, Rome, Italy; and Department of Radiology, Osaka University Graduate School of Medicine, Osaka, Japan (T.M., M.H.). Received July 29, 2003; revision requested October 10; final revision received April 21, 2004; accepted May 26. Address correspondence to R.I., Via Arturo Graf 40, 00137 Rome, Italy (e-mail: r_iannaccone@yahoo.it).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Local ethical committee approval and patient consent were obtained. One hundred ninety-five patients (129 men, 66 women; mean age, 61 years; age range, 39–78 years) with 250 HCCs underwent multi–detector row helical CT of the liver. A quadruple-phase protocol that included unenhanced, hepatic arterial, portal venous, and delayed phases was performed. Analysis of images from hepatic arterial and portal venous phases combined, hepatic arterial and portal venous phases with the unenhanced phase, hepatic arterial and portal venous phases with the delayed phase, and all phases combined was performed separately by three independent radiologists. Relative sensitivity, positive predictive value, and area under the receiver operating characteristic curve (Az) were calculated for each reading session.
RESULTS: Mean sensitivity and positive predictive values, respectively, for HCC detection were 88.8% (666 of 750 readings) and 97.8% (666 of 681 readings) for the combined hepatic arterial and portal venous phases, 89.2% (669 of 750 readings) and 97.8% (669 of 684 readings) for hepatic arterial and portal venous phases with the unenhanced phase, 92.8% (696 of 750 readings) and 97.3% (696 of 715 readings) for hepatic arterial and portal venous phases with the delayed phase, and 92.8% (696 of 750 readings) and 97.3% (696 of 715 readings) for all four phases combined. The reading sessions in which delayed phase images were available for interpretation showed significantly (P < .05) superior sensitivity and Az values.
CONCLUSION: Unenhanced phase images are not effective for HCC detection. Because of the significant increase in HCC detection, a delayed phase can be a useful adjunct to biphasic CT in patients at risk for developing HCC.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Several authors have investigated the role of unenhanced and delayed phase CT images in the detection of HCC. However, the role of such images remains controversial. Oliver etal (4) reported that unenhanced phase CT images can depict additional HCC nodules that are not identified on hepatic arterial and portal venous phase images, whereas other researchers have demonstrated that unenhanced phase images are not effective for HCC detection (6–8). With regard to delayed phase CT, several authors have emphasized the advantages related to the use of such images in the detection of additional tumors, as well as in the characterization of controversial nodules (9–12). However, Choi et al (13) demonstrated that the sensitivity of CT in HCC detection is not improved by adding a delayed phase to biphasic scanning. At present, delayed phase CT is not routinely used in Western countries when performing imaging in patients at risk for developing HCC (3–6).
Because the acquisition of images at an additional phase leads to increased patient exposure to ionizing radiation and an increase in the time and cost of CT examination and image interpretation, it is important to clarify the advantages of unenhanced and/or delayed phase CT imaging to warrant their use. The development of multi–detector row helical CT since the publication of many of these reports has resulted in additional technologic improvements that affect the capabilities of contrast material–enhanced CT.
We therefore undertook this study to determine, by using multi–detector row helical CT, the added value of obtaining unenhanced and delayed phase scans in addition to biphasic (hepatic arterial phase and portal venous phase) scans in the detection of HCC in patients with cirrhosis.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Of the 195 patients, 81 were found to have HCC, with a total of 250 foci seen at imaging. To document proof of HCC, we reviewed medical and surgical records for all patients for reports of subsequent pathologic examination and surgery. Specifically, proof of focal lesions was obtained at partial surgical resection of 87 nodules in 40 patients and at biopsy of 41 nodules in 41 patients. Therefore, all patients had at least one pathologically confirmed lesion. Typically, in those patients with biopsy-proved HCC, biopsy of only one nodule was performed. Patients with multiple nodules were considered to have multifocal HCC when the other lesions had the same imaging appearance as the biopsy-proved HCC nodule.
The remaining tumors (122 nodules in 41 patients), in which biopsy was not performed, were confirmed at follow-up with a combination of CT and magnetic resonance (MR) imaging for a minimum of 12 months (mean follow-up time, 14 months; range, 12–18 months). In 14 of these patients, 52 tumors were also confirmed to be HCC by means of a response (ie, substantial reduction in size of the lesion) to transcatheter arterial chemoembolization. Proof of tumor burden is summarized in Figure 1.
|
-fetoprotein serum levels less than 20 µg/L.
There were HCC lesions in the resected specimens that were not identified at any imaging phase. Because proof of lesions in 41 of 81 patients found to have HCC was not obtained by using surgical confirmation, the sensitivity for lesion detection obtained in this study refers to a relative sensitivity. Indeed, none of our patients underwent transplantation, which would be the only pathologic proof used to obtain true sensitivity. Follow-up CT was performed with a scanning technique identical to that used in the first CT examination. MR imaging was performed with a 1.5-T unit (Magnetom Vision Plus; Siemens, Erlangen, Germany), and T2-weighted (repetition time msec/echo time msec, 
/90) half-Fourier single-shot turbo spin-echo and T1-weighted (160/2.3) fast low-angle shot sequences were performed before and after intravenous injection of 0.1 mmol of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany) per kilogram of body weight. Unenhanced, hepatic arterial, portal venous, and delayed phase MR images were obtained in all patients.
CT Technique
Quadruple-phase CT (ie, unenhanced, hepatic arterial, portal venous, and delayed phases) was performed by using a multi–detector row helical scanner (Somatom Plus 4 Volume Zoom; Siemens, Forcheim, Germany) equipped with flying spot technology, adaptive array matrix, and a gantry rotation time of 0.5 second.
All patients were asked to drink approximately 600 mL of tap water as an oral contrast agent before CT scanning. Each patient received intravenous nonionic contrast material (iohexol, Omnipaque 300; Amersham Health, Oslo, Norway) by means of a power injector (Envision CT; Medrad, Indianola, Pa) at a rate of 5 mL/sec. To minimize variability in enhancement of the liver related to differences in patient size, the volume of contrast material was calculated according to the body weight of the patient (2 mL of contrast material per kilogram of body weight). The ensuing average volume of contrast material was equal to 134 mL (range, 110–182 mL).
Four complete acquisitions of the entire liver were obtained in a craniocaudal direction with the following parameters: section collimation, 2.5 mm; effective section thickness, 3.0 mm; reconstruction interval, 3.0 mm; table feed, 12.5 mm/sec; 165 effective mAs; and 120 kVp. Each acquisition required a scanning time of 10 seconds and was obtained during one breath hold.
In all patients, unenhanced scanning (ie, the first pass) was performed. To determine the time of peak aortic enhancement, a bolus injection of 20 mL of contrast material was administered, and sequential dynamic sections were acquired every 2 seconds, starting from the hepatic hilum. On the basis of findings of a previous study on multi–detector row helical CT (15), in which the test bolus technique and the same imaging parameters as in our study were used, we calculated the start time for the hepatic arterial phase (ie, the second pass) by adding 14 seconds to the time of peak aortic enhancement calculated at the hepatic hilum. The ensuing average start time for the hepatic arterial phase was 34 seconds (range, 30–38 seconds). The portal venous phase and the delayed phase (ie, the third and fourth passes, respectively) scans were acquired at 60 and 180 seconds, respectively, after the start of contrast material injection.
Image Analysis
Image analysis was performed separately and independently by three gastrointestinal radiologists (P.R., R.I., and F.M., with 21, 5, and 4 years of experience, respectively). Four different readings were performed for each patient in the following order: hepatic arterial and portal venous phase images combined; hepatic arterial and portal venous phase images in conjunction with unenhanced phase images; hepatic arterial and portal venous phase images in conjunction with delayed phase images; and images from all four phases combined. To minimize memory bias, the images were presented in random order to each of the readers at each session. The interval between the reading sessions was 4 weeks. The three radiologists knew that all patients had cirrhosis, but they were unaware of correlative imaging findings, tumor burden, or other clinical information (eg, -fetoprotein levels).
Commercially available hardware (Kayak PC workstation; Hewlett Packard, Palo Alto, Calif) and software (Vitrea 2; Vital Images, Minneapolis, Minn) were used for image analysis, which consisted of the evaluation of magnified transverse CT images directly on a high-resolution monitor. All images were initially reviewed by using liver window settings (window level, 50–100 HU; window width, 170 HU), and then the window setting was adjusted as needed in each case.
For objectivity and reproducibility of the image analysis performed in this study, the criteria for HCC were provided. The radiologists defined a lesion as HCC if a nodular focus of homogeneous or heterogeneous enhancement was detected during the hepatic arterial phase, with lower attenuation during the portal venous and delayed phases (2). In addition, the following lesions were regarded as HCCs: a hypovascular and hypoattenuating nodule that did not meet the criteria for a cyst (ie, a sharply delineated round or oval lesion with attenuation near that of water and no contrast enhancement of the wall or contents) or the criteria for focal confluent fibrosis (ie, a focal hypoattenuating wedge-shaped lesion radiating from the portal fissure and associated with parenchymal atrophy, with overlying capsular retraction and lack of displacement of vessels) during the unenhanced phase (16,17); or a nodule with discrete capsular enhancement during the delayed phase (11).
During each reading session, the investigators separately and independently recorded the number and sizes of focal lesions and assigned a confidence level for the diagnosis of HCC. Lesion size was estimated by measuring the maximum diameter on transverse CT images with an electronic ruler. Diagnostic confidence for each lesion was subjectively scored on a five-point scale (score of 0, no HCC; 1, HCC probably absent; 2, HCC possibly present; 3, HCC probably present; 4, HCC definitely present). Before interpreting the images, the three investigators were informed that the categorization of confidence levels of 2 or higher represented a positive diagnosis of HCC. All lesions assigned a confidence level of 2 or higher and confirmed to be HCC were considered true-positive diagnoses. All lesions assigned a confidence level of 0 or 1 when a lesion was actually proved to be HCC were considered false-negative diagnoses. In the few patients (n = 7) with more than eight lesions, analysis of the eight most representative lesions was performed to prevent the inclusion of these patients from biasing the statistical results.
As a subjective analysis, the three readers were asked to document all HCCs in which a capsule (ie, partial or complete hyperattenuating rim around a nodule) could be depicted on delayed phase images only.
Statistical Analysis
Interobserver variability was evaluated by calculating the 
statistic for multiple readers with the nonweighted binary statistic. A value of 0.01–0.20 was judged as minor agreement; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, high; and 0.81–1.00, excellent. Sensitivity and positive predictive values for all four reading sessions (ie, hepatic arterial and portal venous phases combined; hepatic arterial and portal venous phases in conjunction with the unenhanced phase; hepatic arterial and portal venous phases in conjunction with the delayed phase; and all four phases combined) were also calculated. The McNemar test was also used to compare the sensitivity for each reading session.
For imaging for each phase, alternative free-response receiver operating characteristic (ROC) curve analysis was performed. Although the conventional ROC method allows only one response per image, the alternative free-response ROC method allows an observer response for all of the lesions present, and we analyzed all 250 lesions in this study (18). An alternative free-response ROC curve was fitted to each reader’s five-point scale confidence rating by using a maximum-likelihood estimation (ROCKIT 0.9B; C. E. Metz, University of Chicago, Ill, 1998). The diagnostic accuracy of imaging for each reading session for each reader and their composite data were estimated by calculating the area under the ROC curve (Az). Differences between the imaging techniques in terms of the mean Az values were analyzed statistically by using (a) two-factor analysis of variance with only one observation in each cell and (b) multiple comparison (Dunnett pairwise multiple comparisons t test). In this latter test, the comparison is not performed for all possible pairs; rather, one group is chosen as a control group and the differences are tested between the control group and the other groups. Because the purpose of our study was to evaluate the added value of obtaining unenhanced and/or delayed phase CT scans in addition to the biphasic CT scans, the first reading session (hepatic arterial and portal venous phases combined) was chosen as the control group against which the other three reading sessions were compared.
For all statistical analyses, a two-tailed P value of less than .05 was considered to indicate a statistically significant difference. All statistical analyses were performed by using commercially available software (SPSS 11.0.0 for Windows; SPSS, Chicago, Ill).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
values among the three observers showed excellent agreement at all four reading sessions (Table 1). Two-factor analysis of variance results showed a statistically significant difference in Az values among the four reading sessions (P = .02) (Table 2). The reading sessions in which delayed phase CT images were available for interpretation (ie, session with hepatic arterial, portal venous, and delayed phases and session with all four phases combined) showed statistically significantly greater Az values than did the reading session in which hepatic arterial phase and portal venous phase images combined were analyzed (P = .03) (Table 3).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
A further important consideration regarding the role of delayed phase CT is related to its ability to depict HCC capsule. In our series, we found that 25 (10%) of 250 HCCs had a capsule, which could be detected only on delayed phase CT images. This finding is important for patient care because the presence of a capsule has recently been demonstrated to be a favorable prognostic factor for having massive tumor necrosis after transcatheter arterial chemoembolization (20).
Consequently, because of the substantial improvement in the rate of detection of HCC, as well as the ability to depict the presence of tumor capsule, we believe that a delayed phase CT scan should always be acquired in patients at risk for developing HCC.
Because of the identification of only one additional tumor that could not be seen on hepatic arterial phase and portal venous phase images (but clearly detected on the delayed phase images), our results indicate that the addition of unenhanced CT is not an effective adjunct for HCC detection. This is especially true for lesions smaller than 20 mm in diameter, which are extremely difficult to detect on unenhanced images because of the small difference in tumor and liver attenuation. Although a direct comparison between our research results and previously published data is difficult because of important differences in CT scanners, imaging parameters, delay times, and rate of administration of contrast material, it is likely that advances in CT technology (with the acquisition of thinner sections and improved spatial resolution at multi–detector row helical CT) and optimization of contrast material administration (with the routine use of the test bolus technique and fast contrast material injection rates) have improved the capabilities of hepatic arterial phase and portal venous phase imaging for HCC detection, thus making the identification of additional nodules on unenhanced CT images less frequent. In fact, it is possible that HCC nodules might not have be seen on hepatic arterial phase images in previous studies with single–detector row helical CT and fixed delay times because of suboptimal timing of the hepatic arterial phase, if scanning was performed too early (with inadequate enhancement of tumors) (4).
Although our results indicate that unenhanced images do not add any significant advantage in terms of HCC detection, one should also consider that, in everyday clinical practice, such images can play an important role in the detection and differentiation of uncertain lesions (such as cysts, regenerative nodules, focal confluent fibrosis, and focal sparing of fatty infiltration) from HCC nodules. Further studies are needed to determine the potential role of unenhanced images in the characterization of focal lesions in patients with cirrhosis. Moreover, results of one study have demonstrated that the addition of unenhanced imaging offers increased ability to detect HCC, as well as increased diagnostic confidence in the assessment of viable tumors, compared with use of biphasic helical CT alone, in patients who have been treated with transcatheter arterial chemoembolization (19).
Two inherent limitations may explain the high sensitivity rate in our series. First, our patient population was not a screening population. Rather, all patients included in our study had at least one HCC nodule identified at a previous US examination. In our hospital, patients with cirrhosis are screened with US and are referred for CT examination when there is suspicion of HCC at clinical and/or US examination. Therefore, although the readers were unaware of correlative imaging findings, HCC burden, and other clinical information (eg, -fetoprotein levels), the possibility of a bias of the readers to overcall HCCs cannot be ruled out. Second, we might have missed a substantial number of lesions because we did not have pathologic correlation for all individual focal lesions that we thought were HCC. However, all our patients had at least one pathologically confirmed lesion (either at partial surgical resection or biopsy), each additional nodule seen at unenhanced and/or delayed phase CT had pathologic confirmation, and all unproved lesions had correlative imaging (one MR imaging examination within 1 month of the first CT examination and one follow-up CT examination more than 12 months after the first CT). Therefore, because of these inherent limitations in our methods, the detection rates reported in this study represent relative and not absolute sensitivities and cannot be compared with results of prior studies addressing sensitivity (21–23). However, the purpose of our study was not to test the sensitivity of CT as a screening tool in patients with cirrhosis without clinical suspicion of HCC (as it has already been investigated by other authors [21–23]) but to evaluate the added value of unenhanced and/or delayed phase imaging to hepatic arterial and portal venous phase CT imaging.
Another potential criticism of our study is related to the lack of explant correlation for all individual focal lesions that we believed were representative of HCC. We cannot rule out that some of the hyperattenuating lesions that we interpreted as HCCs were actually false-positive cases, the rate of which has been recently reported to be 3% in a large population of patients undergoing transplantation who were screened at CT for hepatocellular carcinoma (24).
In conclusion, results of our study demonstrate that the use of unenhanced phase imaging is not an effective adjunct for HCC detection. In contrast, because of the significant increase in rate of detection of HCC (especially for lesions 20 mm in diameter or smaller), delayed phase CT can be a useful adjunct to biphasic CT in patients at risk for developing HCC.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, R.P.; study concepts and design, R.I.; literature research, R.I., F.M., F.P.; clinical studies, R.I., F.M., F.P.; data acquisition, R.I., F.M., F.P.; data analysis/interpretation, P.R., R.I., F.M.; statistical analysis, M.H., I.N.; manuscript definition of intellectual content, editing, and revision/review, all authors; manuscript preparation and final version approval, R.I.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J.-J. Chung, J. S. Yu, J. H. Kim, M.-J. Kim, and K. W. Kim Nonhypervascular Hypoattenuating Nodules Depicted on Either Portal or Equilibrium Phase Multiphasic CT Images in the Cirrhotic Liver Am. J. Roentgenol., July 1, 2008; 191(1): 207 - 214. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Willatt, H. K. Hussain, S. Adusumilli, and J. A. Marrero MR Imaging of Hepatocellular Carcinoma in the Cirrhotic Liver: Challenges and Controversies Radiology, May 1, 2008; 247(2): 311 - 330. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H.-J. Jang, T. K. Kim, P. N. Burns, and S. R. Wilson Enhancement Patterns of Hepatocellular Carcinoma at Contrast-enhanced US: Comparison with Histologic Differentiation Radiology, September 1, 2007; 244(3): 898 - 906. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C-S Li, R-C Chen, H-Y Tu, L-S Shih, T-A Zhang, J-M Lii, W-T Chen, S-J Duh, and L-C Chiang Imaging well-differentiated hepatocellular carcinoma with dynamic triple-phase helical computed tomography Br. J. Radiol., August 1, 2006; 79(944): 659 - 665. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. Schima, R. Hammerstingl, C. Catalano, L. Marti-Bonmati, E. J. Rummeny, F. T. Montero, A. Dirisamer, B. Westermayer, M. Bellomi, D. Brisbois, et al. Quadruple-phase MDCT of the liver in patients with suspected hepatocellular carcinoma: effect of contrast material flow rate. Am. J. Roentgenol., June 1, 2006; 186(6): 1571 - 1579. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. R. Wilson and P. N. Burns An Algorithm for the Diagnosis of Focal Liver Masses Using Microbubble Contrast-Enhanced Pulse-Inversion Sonography. Am. J. Roentgenol., May 1, 2006; 186(5): 1401 - 1412. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
