| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology II (A.L., R.I., I.C., R.F., F.M., R.P.), Radiology III (P.R.), and Experimental Medicine and Pathology (I.N.), University of Rome-La Sapienza, Policlinico Umberto I, Viale Regina Elena 324, 00161 Rome, Italy. Received December 13, 2001; revision requested February 22, 2002; revision received April 18; accepted June 5. Address correspondence to A.L. (e-mail: andrea.laghi@uniroma1.it).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Multi–detector row CT was performed in 77 patients with 140 foci of hepatocellular carcinoma (HCC). A triple-phase protocol that included an early arterial phase, a late arterial phase, and a portal venous phase was performed. Images were analyzed separately by three radiologists to document the presence and number of HCC nodules. Separate reading sessions were performed for images from the early arterial phase, images from the late arterial phase, images from both arterial phases combined, and images from all three phases. Sensitivity and positive predictive values were calculated for each reading session.
RESULTS: The average sensitivity and positive predictive values, respectively, for the detection of HCC were 48.5% and 96.4% for early arterial phase images, 87.1% and 94.0% for late arterial phase images, 87.1% and 94.0% for images from both arterial phases, and 88.5% and 93.4% for images from all three phases. Analysis of images from both arterial phases together yielded no improvement in either sensitivity or positive predictive value compared with analysis of late arterial phase images alone. Analysis of the combination of late arterial and portal venous phase images resulted in the highest sensitivity value.
CONCLUSION: The acquisition of images during two arterial contrast phases does not provide additional benefit over timed conventional biphasic CT technique.
© RSNA, 2003
Index terms: Computed tomography (CT), multi–detector row, 761.12114 • Liver neoplasms, 761.323 • Liver neoplasms, CT, 761.12114
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Therefore, such lesions have often been difficult to detect with conventional computed tomography (CT) of the liver, in which only portal venous phase imaging was performed because of the long scanning time. During the past decade, the introduction of helical CT technology has made biphasic scanning feasible. In particular, results of several studies have demonstrated that use of a two-phase approach, with acquisition of images in the hepatic arterial dominant phase as well as in the portal venous phase, greatly improves the detection of HCC (1,4,6–9).
At present, with the advent of multi–detector row helical CT scanners, it is possible to achieve acquisition times two to eight times faster than were achievable with previously available equipment (the so-called single–detector row helical CT scanners) (10–13), and, consequently, the entire liver can now be examined in approximately 10 seconds (14). On the basis of this advantage, Foley et al (15) have proposed a new protocol for liver evaluation in which three sets of images are acquired (two sets of images in a single breath hold during the hepatic arterial dominant phase and one set of images during the portal venous phase). The role of "double arterial phase" imaging has also recently been evaluated by Murakami et al (14) in a selected patient population at high risk for developing HCC.
Concern has recently been expressed over the high radiation doses delivered when multi–detector row helical CT is used (16). This concern, combined with the recent trend toward performing multiple acquisitions through the liver during these examinations, warrants continued study to ensure the efficacy of obtaining three series of images during dynamic contrast imaging. One must also consider that, in addition to dynamic contrast material–enhanced images, other images through the liver are often obtained, with a consequent increase in the radiation exposure to the patient. The acquisition of unenhanced liver images has also been advocated as an adjunct to helical biphasic contrast-enhanced CT protocols (1), and delayed equilibrium images are often obtained to clarify findings as well as to enable timing of image acquisition when sensitivity for HCC detection is high (17,18).
We undertook this study to evaluate whether the use of two arterial phase image acquisition series, when combined with portal venous phase imaging at multi–detector row helical CT, would be superior enough to the use of a single arterial phase image acquisition series to warrant the increased radiation dose.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Forty-eight of the 77 patients were found to have HCC, for a total of 140 nodules. To document proof of HCC, we reviewed medical and surgical records of all patients for reports of subsequent pathologic examination and surgery. Specifically, proof of focal lesions was obtained at partial surgical resection of 27 nodules in 13 patients and at biopsy of 35 nodules in 35 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 (78 nodules in 35 patients) were confirmed at follow-up with a combination of CT and magnetic resonance (MR) imaging (range of duration of follow-up, 6–10 months; mean duration of follow-up, 8 months). Twenty-three tumors in 11 patients were additionally confirmed to be HCC by means of a response (ie, substantial reduction in size of the lesion) to transcatheter arterial chemoembolization. For the 29 patients who did not have focal lesions, proof of absence of disease was based on the combination of findings at an MR imaging examination performed within 1 month after CT, findings at a follow-up CT examination performed more than 6 months after the first CT examination (follow-up range, 7–14 months; mean follow-up, 11 months), and negative 
-fetoprotein serum levels.
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), with 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 before and after intravenous injection of 0.1 mmol per kilogram of body weight of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany). Non–contrast-enhanced and arterial and portal venous phase MR images were obtained in all 35 patients in whom proof of disease was established at follow-up CT and MR imaging.
Triple-Phase CT Imaging
Following the technique described by Foley et al (15), triple-phase (ie, early arterial phase, late arterial phase, and portal venous phase) CT imaging was performed.
All examinations were performed with a multi–detector row helical CT scanner (Somatom Plus 4 Volume Zoom; Siemens) equipped with a flying spot, an adaptive array matrix, and a gantry rotation time of 0.5 second. Images were acquired through the liver with the following parameters: section collimation, 2.5 mm; section thickness, 3.0 mm; table feed, 10.8 mm/sec; 165 mAs; 120 kVp; dose exposure, 2.4 mSv (in men) and 3.1 mSv (in women) for each phase. All images were reconstructed at an interval of 3.0 mm. Early arterial phase images were also reconstructed at an interval of 1.0 mm to obtain a high-quality data set for CT arteriographic reconstruction.
To determine the optimal time to initiate imaging after contrast material administration, we injected a test bolus of 20 mL of contrast material at 5 mL/sec and acquired sequential dynamic sections every 2 seconds beginning at the hepatic hilum. The time of peak aortic enhancement was used as the start time for the early arterial phase.
Before the CT examination, patients received approximately 600 mL of tap water as an oral contrast agent. Contrast material (Omnipaque 300; Nycomed, Oslo, Norway) was intravenously administered in all patients by using a power injector (Envision CT; Medrad, Indianola, Pa) at a rate of 5 mL/sec. The appropriate volume of contrast material at 2 mL/kg was calculated according to the body weight of the patients. The average volume of contrast material was equal to 128 mL (range, 120–168 mL).
Three scanning passes were performed through the entire liver. Each scanning pass required an acquisition time of 10 seconds. The first pass (early arterial phase) was performed in a cephalad to caudad direction; then, after a between-phase delay of 4 seconds for table movement, the second pass (late arterial phase) was performed in a caudal to cephalic direction. Image acquisition during the early and late arterial phases was accomplished during a single breath hold of 24 seconds. The average scanning delay was 20 seconds (range, 16–32 seconds) for the early arterial phase and 34 seconds (range, 30–46 seconds) for the late arterial phase. The third scanning pass (portal venous phase) began 60 seconds after the initiation of contrast material injection and required a second breath hold of 10 seconds.
Image Analysis
Analysis of early arterial phase images, late arterial phase images, double arterial phase images (ie, the combination of images from the early and late arterial phases), and triple-phase images (ie, the combination of images from the portal venous phase and the early and late arterial phases) was performed separately and independently by three experienced gastrointestinal radiologists (A.L., P.R., R.P.). Thus, for each patient, four separate readings were made by each radiologist. The early arterial phase images were analyzed in the first reading session, the late arterial phase images were analyzed in the second, the double arterial phase images were analyzed in the third, and the triple-phase images were analyzed in the fourth. Images were randomly presented during each of the four reading sessions to avoid bias from learning effects. The time interval between each reading session was 1 week. The three investigators knew that all patients had chronic hepatitis, but they were unaware of other clinical information (eg, -fetoprotein levels) or correlative US or MR imaging findings.
Each reader was instructed to determine if, in his or her opinion, HCC was present, and if so, how many foci were present. The image readers defined a lesion as HCC if a nodular focus of homogeneous or heterogeneous enhancement that did not meet the criteria for a benign lesion was identified on arterial phase images (20,21). A hypovascular and hypoattenuating lesion 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 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) (22) was also considered to represent HCC.
Image interpretation was performed directly at a dedicated workstation (Kayak PC; Hewlett Packard, Palo Alto, Calif) by using a software package with a volume-rendering algorithm (Vitrea 2.2; Vital Images, Minneapolis, Minn). The size of each lesion was assessed by using an electronic ruler. CT data sets obtained during the early arterial phase were used to generate CT arteriographic images of the hepatic and mesenteric circulation. CT arteriographic images were analyzed during the same reading session as the early arterial phase images by using volume rendering. No hard copy of a CT image was analyzed.
During each reading session, the radiologists recorded the sizes of focal hepatic lesions and assigned a conspicuity level for the diagnosis of HCC. Lesion conspicuity was graded subjectively with a five-point scale: 0, no lesion; 1, lesion probably absent; 2, lesion possibly present; 3, lesion probably present; and 4, lesion definitely present. All lesions assigned a conspicuity level of 2 or higher that were confirmed to be HCC were considered true-positive diagnoses. All areas assigned a conspicuity level of 0 or 1 when a lesion was actually proved were considered false-negative diagnoses. In the few patients (n = 4) with more than eight lesions, analysis of the eight most representative lesions was performed to prevent the inclusion of the data from these patients from significantly biasing the statistical results.
The presence or absence of arterial-portal shunting was evaluated on the early and late arterial phase images. During the arterial phases, under normal circumstances there should be no or minimal enhancement of the portal vein. Diagnosis of arterial-portal shunting was accomplished when an intense accumulation of contrast material was identified within the distal portal vein or portal venous branches that was associated with parenchymal staining with no or minimal contrast enhancement in the proximal portal vein or superior mesenteric and splenic veins (4). Arterial-portal shunts could be differentiated from small HCC nodules by their shape and location (23–25).
Statistical Analysis
Interobserver variability was evaluated by calculating a nonweighted binary 
statistic for multiple readers (26,27). A of 0.01–0.20 was considered to represent minor agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, high agreement; and 0.81–1.00, almost perfect agreement. Sensitivity and positive predictive values for all four reading sessions (ie, early arterial phase images, late arterial phase images, double arterial phase images, and triple-phase images) were also calculated. The McNemar test was used to compare sensitivity values for each reading session. A two-tailed P < .05 was considered to indicate a statistically significant difference. Due to the lack of histologic proof for 78 nodules, specificity and accuracy could not be evaluated. All statistical data were calculated by using commercially available software (SPSS 8.0 for Windows; SPSS, Chicago, Ill).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The values for the three observers showed high to almost perfect agreement for all four reading sessions except the early arterial phase image reading session, in which only moderate agreement ( = 0.59) was observed between reader 2 and reader 3 (Table 1).
|
|
|
|
|
|
|
|
|
|
|
Seven vascular anomalies (right hepatic artery arising from the superior mesenteric artery, n = 4; common hepatic artery arising from the aorta, n = 2; left gastric artery arising from the common hepatic artery, n = 1) were diagnosed with transverse CT images. In all seven cases, volume-rendered three-dimensional images generated from the early arterial phase data set provided panoramic depiction of the vascular anatomy of the hepatic and mesenteric circulation (Fig 5).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
However, there are several variables that may affect the optimal timing of the hepatic arterial dominant phase. In particular, patient size and cardiovascular status may substantially alter the degree of enhancement (14). Therefore, as suggested by Murakami et al (14), we administered a fixed amount of contrast medium per kilogram (2 mL/kg) in each patient to reduce the influence of patient size, and we used the test bolus technique to minimize the effects of different circulation times.
Our study was designed according to a technique first proposed by Foley et al (15), who optimized a triple-pass technique with a multi–detector row CT scanner for the detection of hypervascular neoplasms. The rationale for performing this triple-pass examination is to maximize detection of hypervascular neoplasms. The first pass consists of scanning during the early arterial phase, with either no or minimal admixture of enhanced portal venous blood (15).
As already noted by Foley et al (15), hypervascular tumors can be seen on images obtained during this first pass, but the degree of enhancement is considerably increased on images obtained during the second pass, also called the late arterial phase. When a multi–detector row CT scanner is used, early and late arterial phase images can be acquired during the hepatic arterial dominant phase within a single breath hold. The third pass (acquired 60 seconds after the start of contrast material infusion) corresponds to the portal venous phase of the conventional biphasic protocol used with a single–detector row helical CT scanner. On images obtained during this phase, hepatic veins are enhanced and hypervascular lesions may appear isoattenuating or hypoattenuating.
Foley et al (15) concluded that, although hypervascular neoplasms are best seen during the late arterial phase, the use of an early arterial phase allows for CT arteriography, which can be important for the preoperative detection of vascular anomalies in patients who are candidates for hepatic resection and/or tumor cryoablation or arterial chemoembolization. Although Murakami et al (14) did not take into consideration the value of portal venous phase images in their recent study, they reported a distinct benefit of early arterial phase imaging for tumor detection and reduction of false-positive diagnoses.
Results from our experience with the early arterial phase differ substantially from the results of Murakami et al (14). In particular, in our series, the use of an early arterial phase did not result in any improvement in the detection of HCC nodules. In addition, no reduction of false-positive diagnoses was observed when images from this additional phase were reviewed. One possible explanation for this discrepancy is the different average size of HCC nodules observed in our study. A considerably higher number of small nodules (>20 mm) were observed in our study (107 [76.4%] of 140 nodules) than in the study of Murakami et al (14) (45 [46.8] of 96 nodules).
Rather, in our experience, the highest sensitivity value was obtained when portal venous phase images were available for analysis in the triple-phase image reading session, because four hypovascular HCC nodules (2.8%) were visible only during the portal venous phase. As is well known, the portal venous phase not only has the advantage of depicting hypovascular HCC (1,4) but also is useful in the demonstration of portal venous thrombosis (18,33), the differentiation of neoplasms from vessels, and the identification of varices and shunts (18).
We agree with Foley et al (15) that the use of an early arterial phase plays an important role in the evaluation of vascular structures—not in diagnosis, because much diagnostic information is already present on transverse images, but for the generation of three-dimensional reconstructions that make understanding of vascular anatomy easier for surgeons. Further studies are needed to investigate the potential role of a low-dose early arterial phase that may depict vascular anomalies while entailing a relatively low degree of radiation exposure.
Our results indicate that early arterial phase images are not useful for the detection of HCC. The reason for this may be related to the fact that during the early arterial phase there is insufficient time for a substantial amount of contrast material to reach the tumor. A further explanation is inherent in our CT technique. Scanning the liver in a cephalad to caudad direction during the first pass (early arterial phase) and in a caudal to cephalic direction during the second pass (late arterial phase) probably results in the top of the liver being scanned during both passes when tumors there are least visible (ie, the top of the liver is scanned too early in the first pass and too late in the second).
A potential limitation to our study could be the lack of histologic proof for all nodules believed to be HCC. However, several kinds of confirmatory findings, including MR imaging findings, follow-up CT findings, and response to chemoembolization, were available for all nodules. The presence of other proved nodules, the small size of many nodules, and/or the patient’s clinical history frequently made biopsy of individual lesions impractical or unnecessary. Consequently, the exact number of HCC nodules in each liver could not be determined, and, therefore, specificity and accuracy could not be calculated. In particular, although atypical hemangiomas are uncommon in patients with chronic hepatitis, these lesions could have simulated small HCC foci (22). In addition, absence of disease was proved on the basis of the combination of results from an MR imaging examination, results from a follow-up CT examination, and negative -fetoprotein serum levels; these criteria for proof did not enable us to definitively recognize all possible false-negative diagnoses.
In conclusion, the addition of an early arterial phase to the conventional biphasic CT protocol did not provide any benefit in the detection of HCC in terms of either sensitivity or positive predictive value. However, early arterial phase images were important in the evaluation of the anatomy of the hepatic and mesenteric circulation, although we believe that the increased time, cost, and radiation exposure attendant to the use of this additional phase should be taken into consideration before the early arterial phase is routinely incorporated into the imaging protocol for patients with chronic hepatitis. If one is planning to perform single arterial phase imaging with multi–detector row helical CT, a late arterial phase (average scanning delay in our experience, 34 seconds) should be included.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, R.P.; study concepts and design, A.L., R.I., I.C., R.F., F.M.; literature research, R.I., R.F., F.M.; clinical studies, R.I., I.C., R.F., F.M.; data acquisition, R.I., I.C.; data analysis/interpretation, A.L., P.R., R.P.; statistical analysis, R.I., I.N.; manuscript preparation, A.L., R.I.; manuscript definition of intellectual content, A.L., R.I., P.R., R.P.; manuscript editing, A.L., R.I., R.P.; manuscript revision/review, P.R.; manuscript final version approval, P.R., R.P.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  X. Ma, A. E. Samir, N.-S. Holalkere, and D. V. Sahani Optimal Arterial Phase Imaging for Detection of Hypervascular Hepatocellular Carcinoma Determined by Continuous Image Capture on 16-MDCT Am. J. Roentgenol., September 1, 2008; 191(3): 772 - 777. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Yanaga, K. Awai, T. Nakaura, T. Namimoto, S. Oda, Y. Funama, and Y. Yamashita Optimal Contrast Dose for Depiction of Hypervascular Hepatocellular Carcinoma at Dynamic CT Using 64-MDCT Am. J. Roentgenol., April 1, 2008; 190(4): 1003 - 1009. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Nakaura, K. Awai, Y. Yanaga, Y. Nakayama, S. Oda, Y. Funama, and Y. Yamashita Detection of Early Enhancement of Hypervascular Hepatocellular Carcinoma Using Single Breath-Hold 3D Pixel Shift Dynamic Subtraction MDCT Am. J. Roentgenol., January 1, 2008; 190(1): W13 - W18. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Monzawa, T. Ichikawa, H. Nakajima, Y. Kitanaka, K. Omata, and T. Araki Dynamic CT for Detecting Small Hepatocellular Carcinoma: Usefulness of Delayed Phase Imaging Am. J. Roentgenol., January 1, 2007; 188(1): 147 - 153. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.-J. Kim, J. Y. Choi, J. S. Lim, J. Y. Kim, J. H. Kim, Y. T. Oh, E. H. Yoo, J. J. Chung, and K. W. Kim Optimal scan window for detection of hypervascular hepatocellular carcinomas during MDCT examination. Am. J. Roentgenol., July 1, 2006; 187(1): 198 - 206. [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]
|
|
|
|
|
|
P. T. Johnson and E. K. Fishman IV Contrast Selection for MDCT: Current Thoughts and Practice Am. J. Roentgenol., February 1, 2006; 186(2): 406 - 415. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.-J. Kim, Y. E. Chung, K. W. Kim, J.-J. Chung, J. S. Lim, Y. T. Oh, and J. H. Kim Variation of the Time to Aortic Enhancement of Fixed-Duration Versus Fixed-Rate Injection Protocols Am. J. Roentgenol., January 1, 2006; 186(1): 185 - 192. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. H. Kim, D. Choi, S. H. Kim, J. H. Lim, W. J. Lee, M. J. Kim, H. K. Lim, and S. J. Lee Ferucarbotran-Enhanced MRI Versus Triple-Phase MDCT for the Preoperative Detection of Hepatocellular Carcinoma Am. J. Roentgenol., April 1, 2005; 184(4): 1069 - 1076. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Iannaccone, A. Laghi, C. Catalano, P. Rossi, F. Mangiapane, T. Murakami, M. Hori, F. Piacentini, I. Nofroni, and R. Passariello Hepatocellular Carcinoma: Role of Unenhanced and Delayed Phase Multi-Detector Row Helical CT in Patients with Cirrhosis Radiology, February 1, 2005; 234(2): 460 - 467. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Murakami, T. Kim, M. Hori, and M. P. Federle Double Arterial Phase Multi-Detector Row Helical CT for Detection of Hypervascular Hepatocellular Carcinoma [letter] Radiology, December 1, 2003; 229(3): 931 - 932. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHIC |
