| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 All authors: Department of Radiology, University of Yamanashi, 1110 Shimokato, Tamaho, Nakakoma, Yamanashi 409-3898, Japan.
Received February 26, 2004;
accepted after revision March 21, 2005.
Address correspondence to T. Ichikawa.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. The injection rates 3 mL/sec and 5 mL/sec were used prospectively in imaging examinations of patients with the suspected diagnosis of hepatocellular carcinoma. Each group consisted of 30 patients by chance. The group that received injections at 3 mL/sec had 35 hepatocellular carcinoma lesions, and the 5 mL/sec group had 41 lesions. In all patients the dose and concentration of contrast material were 100 mL and 350 mg/mL iodine (total dose of iodine, 35 g). In each patient a mini-test-bolus technique was used with an additional 15 mL of contrast material to determine optimal scan delay after administration of contrast material. Receiver operating characteristic analysis was used to assess diagnostic performance with the two injection rates of contrast material.
RESULTS. There were no statistically significant differences between the two groups in regard to area under the curve and sensitivity. These values for the 3 mL/sec group were 0.97 and 28/35 (80%) and for the 5 mL/sec group were 0.96 and 36/41 (88%). However, the specificity and positive predictive values at 3 mL/sec (236/250 [95%] and 28/42 [67%]) were significantly higher than those at 5 mL/sec (227/265 [86%] and 36/73 [49%]) (p < 0.05). These results suggest there were more false-positive findings of contrast-enhanced lesions in cirrhotic livers on hepatic arterial dominant phase images obtained after injection of contrast material at 5 mL/sec than on images obtained after injection at 3 mL/sec.
CONCLUSION. In the detection of small hypervascular hepatocellular carcinoma in cirrhotic liver, the risk of false-positive findings of lesions on hepatic arterial dominant phase images is significantly greater with the higher injection rate (5 mL/sec) than with the medium rate (3 mL/sec).
Keywords: angiography • contrast media • CT • liver disease • MDCT
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Fast MDCT scanners enable faster hepatic CT acquisition and the capability of more imaging passes than can be achieved with the helical CT approach with single-detector CT scanners. Scan speed with MDCT scanners has changed continuously, and the terminology and scan timing of the hepatic arterial dominant phase have changed as well [6-8]. Foley et al. [6] showed that the hepatic arterial dominant phase during biphasic CT with single-detector scanners should be divided into two phases when fast MDCT scanners are used. Those authors termed the first pass the hepatic artery phase and the second pass the late arterial phase or portal venous inflow phase. Although the two arterial phases defined by Foley et al. were included in the acquisition interval considered the hepatic arterial dominant phase during biphasic CT with single-detector scanners, each phase has a different role in hepatic CT. The late arterial phase is essential for detecting hypervascular HCC [6, 8]. Because of rapid changes in hepatic MDCT theory, many factors in the protocol for contrast-enhanced hepatic MDCT, such as volume, concentration, and injection rate of contrast material, should be redetermined more strictly than those in protocols for single-detector CT scanners.
The rate of injection of contrast material is one of the most important technique-related factors in the detection of hypervascular HCC with hepatic arterial dominant phase imaging. It has been reported that enhancement of the aorta, including the hepatic artery, on hepatic arterial dominant phase images increases significantly at higher injection rates of contrast material [9-15]. Because of these results, it has been suggested that in hepatic arterial dominant phase imaging of the liver, higher injection rates of contrast material can improve detection of hypervascular hepatic lesions [5, 16]. However, the best injection rate of contrast material for MDCT continues to be controversial because injection rates have been determined not with MDCT scanners but with single-detector CT scanners.
Another controversy is that the rate of false-positive findings of hypervascular lesions, in which small arterial-portal venous shunts may play a major role, may increase in hepatic arterial dominant phase imaging of cirrhotic livers [17]. An increasing rate of false-positive findings may diminish lesion specificity and thus overall diagnostic performance in the detection hypervascular HCC in cirrhotic liver. Because of the possible increase in false-positive findings of hypervascular lesions and hypervascular HCC on hepatic arterial dominant phase images obtained with a higher rate of injection of contrast material, there has been little agreement about the use of higher injection rates in the evaluation of hypervascular HCC in the presence of liver cirrhosis. The purpose of this study was to evaluate 3 mL/sec and 5 mL/sec injection rates of contrast material and to perform receiver operating characteristic (ROC) analysis to assess diagnostic performance with each injection rate in the detection of hypervascular HCC.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The ages of the 60 patients (43 men and 17 women) enrolled in the study ranged from 37 to 75 years (mean, 65 years), and their body weight ranged from 37 kg to 65 kg (mean, 56 kg). For the 76 small HCC lesions, the diagnosis of HCC was made with confidence on the basis of satisfactory pathologic results in 61 cases (10 at surgery and 51 at needle biopsy). In 15 cases, the diagnosis of HCC was made with confirmatory imaging criteria. The confirmatory imaging criteria were as follows: nodules showing round defect on CT arterial portographic images and round enhancement on CT hepatic arteriographic images, deposition of iodized oil on iodized-oil CT images obtained 4-6 weeks after CT during arterial portography and CT hepatic arteriography, and evidence of growth of the lesion during a follow-up period lasting more than 1 year. Needle biopsy was performed on nine of the 15 lesions that showed regrowth during the follow-up period, and the pathologic diagnosis was HCC. In the multiphasic contrast-enhanced MDCT examinations of 76 lesions in 60 patients, the 5 mL/sec injection rate of contrast material was used in 41 lesions in 30 patients, and the 3 mL/sec rate was used in 35 lesions in 30 patients.
Multivariate analysis of the patient population showed no significant difference in distribution of sex, age, weight, or height in the two groups. Among the 159 of 267 patients who underwent contrast-enhanced MDCT, CT during arterial portography, CT hepatic arteriography, and iodized-oil CT imaging of the liver that showed no evidence of HCC, two groups of 30 patients without HCC underwent imaging, one group at each injection rate. These patients were the control groups for receiver-operating characteristic (ROC) analysis. The 60 patients selected for the control groups also were examined with multiple imaging techniques and measurement of tumor markers for at least 1 year (mean, 17 months) after CT during arterial portography and CT hepatic arteriography and had no evidence of HCC during the follow-up period. When the patients in the control groups were chosen, attention was paid to suitability of images. The backgrounds of the patients were matched to those of the 60 HCC patients in the study with regard to image quality, scan location, degree of contrast enhancement of the liver and intrahepatic vessels, age, and sex.
CT
All multiphasic contrast-enhanced MDCT examinations were performed with a
commercially available MDCT scanner (Aquilion, Toshiba) with 0.5-sec gantry
rotation speed. The detector configuration was 2 x 4 mm, in which four
interspaced helical data sets were obtained from 16 detector rows of 0.5 mm.
All patients underwent unenhanced, hepatic arterial dominant phase, and
delayed phase imaging. Patients received 100 mL of iomeprol IV (Iomeron,
Bracco; 350 mg/mL; total dose of iodine, 35 g) through a power injector
(Nemoto Kyorindo). The contrast material was administered by monophasic
injection technique at a rate of 3 mL/sec or 5 mL/sec to 30 patients with
HCC.
To optimize scan delay after injection of contrast material for hepatic arterial dominant phase scanning with the different injection rates, a preliminary mini-test-bolus technique with generation of a time-attenuation curve was performed with 15 mL of contrast material before the diagnostic hepatic arterial dominant phase imaging of each patient. The 15 mL of contrast material used for the mini-test-bolus technique was not included in the 100 mL of contrast material used for diagnostic MDCT. The acquisition method for the mini-test-bolus technique was conventional, and the injection rate of contrast material was identical to that for the diagnostic hepatic arterial dominant phase imaging of each patient.
Diagnostic hepatic arterial dominant phase imaging was started 15 sec after time to aortic peak observed with the mini-test-bolus technique on the basis of results of start time of the late arterial phase defined in previous use of MDCT scanners [6, 7]. As a result, the scan delays for diagnostic hepatic arterial dominant phase scanning were different for each patient. The multiphasic contrast-enhanced MDCT examination was performed at least 3 min after the end of the mini-test-bolus injection to minimize the effect of the mini-test-bolus injection.
To confirm whether the scan delay determined with the mini-test-bolus technique worked similarly for the scan timing of hepatic arterial dominant phase imaging for each injection rate of contrast material, we performed hepatic arterial dominant phase scanning at each injection rate in 10 patients. These patients were not included in the study analysis. This imaging was performed as a side study with the permission of the institutional review board, and informed consent was obtained before the start of the study protocol.
All scans were acquired in a cephalocaudal direction. The table speed was 11 mm with a pitch of 5.5. All MDCT examinations were performed at 120 kVp and 150 mA. For the reconstructed transverse hepatic arterial dominant phase images, a 5-mm slice thickness and 5-mm interval were used. The acquisition time for the hepatic arterial dominant phase was 8-11 sec. The total length of hepatic arterial dominant phase scanning was 10-17 cm.
Imaging Analysis
To investigate the possibility that different scan delays after injection
of contrast material at different rates affects the contrast of hepatic
arterial dominant phase images and the detectability of hypervascular HCC, two
reviewers worked separately to evaluate all MDCT images for degree of contrast
enhancement of intrahepatic branches of the hepatic artery, portal vein, and
hepatic vein. A 5-point grading system (5 = strong, 4 = good, 3 = moderate, 2
= slight, 1 = none) was used for this purpose. All CT images were subjectively
graded on the basis of strength of contrast enhancement of each kind of
vessel. If there was disagreement between the two reviewers' subjective
grades, a third reviewer was used to achieve majority rule. Hepatic arterial
dominant phase images were assigned grade 1 if no contrast enhancement of the
vessels was seen. If there was strong contrast enhancement of the portal vein
or hepatic vein, which was likely to be seen on portal venous phase CT images
on the basis of our experience, grade 5 was assigned. Degrees of contrast
enhancement between none and strong were subjectively scored 2-4. In cases in
which it was impossible to evaluate the degree of contrast enhancement of any
vessels on an MDCT image, the reviewers were allowed to ask for an MDCT image
obtained at a different scan level.
In daily clinical practice, when we visually decide whether hepatic arterial dominant phase images are acquired at the optimal timing for detection of hypervascular HCC, we confirm that the intrahepatic portal veins are at least moderately enhanced and that no intrahepatic veins are enhanced on the images. Therefore in the present study we defined scan timing as too early if intrahepatic portal veins showed no or slight contrast enhancement on the images (grades 1 and 2 for the portal vein). Scan timing was defined as too late if intrahepatic veins were even slightly enhanced (grades 2-4 for the hepatic vein). On the basis of these definitions of scan timing of hepatic arterial dominant phase images, all hepatic arterial dominant phase images were classified into three groups: too early, optimal, and too late.
For ROC analysis, two radiologists serving as study coordinators initially reviewed all MDCT images with knowledge of clinicopathologic findings. On the basis of clinicopathologic reports, they attempted to determine the number and location of lesions and to correlate the anatomic findings on the images with the pathologically confirmed lesions as accurately as possible to allow detection of false-positive readings. For the blind reading, the study coordinators reprinted the images for all 120 patients (60 with HCC and 60 without HCC) in the study and cut the films as each image was separated. The study coordinators drew lines indicating the hepatic segment numbering system of Couinaud on each interpreted image to prevent mislocation of the lesions by the reviewers. False-positive lesions were any identified by the reviewers that had not been indicated by the study coordinators. The total number of liver segments indicated by the study coordinators on all images was 591.
All MDCT images were independently interpreted in random order by three experienced abdominal radiologists blinded to patient and injection rate of contrast material. The reviewers did not interpret all images obtained in the study but only single images the study coordinators made by cutting the reprinted films for each patient. Image interpretation was conducted on a segment-by-segment basis because the purpose of this study was not to assess lesion sensitivity alone but to investigate the overall diagnostic ability of the reviewers in lesion detection on hepatic arterial dominant phase images obtained at the different rates of injection of contrast material. The reviewers interpreted all 591 liver segments indicated by the study coordinators on the films. The reviewers were aware that all MDCT examinations of each patient were performed for evaluation of possible HCC. However, they were blinded to all other information, such as patient identity, clinical history, results of other imaging examinations, and histopathologic findings. For all MDCT images, each reviewer graded the presence or absence of lesions on a 5-point confidence scale (1 = definitely absent, 2 = probably absent, 3 = equivocal, 4 = probably present, 5 = definitely present). If a lesion was considered present in the liver on any kind of MDCT image, the number and location of tumors were recorded.
Statistical Analysis
To analyze the results of grading of the degree of contrast enhancement for
each type of intrahepatic vessel and the differences in scan timing for
hepatic arterial dominant phase imaging with the different injection rates, we
used the Mann-Whitney U test for all quantitative grading data.
Before ROC analysis, interobserver agreement for image interpretation for each
hepatic arterial dominant phase imaging technique with the different injection
rates was assessed to establish the reliability of the imaging interpretation
in this study. The degree of interobserver agreement in each combination of
two reviewers was calculated with weighted kappa statistics. A kappa value
greater than 0.80 was considered excellent agreement; between 0.61 and 0.80,
good agreement; between 0.41 and 0.60, moderate agreement; between 0.21 and
0.41, fair agreement; and 0.20 or less, poor agreement
[18].
For ROC analysis, composite ROC curves were made for each hepatic arterial dominant phase imaging technique with the different injection rates. Composite ROC curves used to represent the performance of the three reviewers as a group were calculated by averaging the binormal parameter values of the individual curves. The findings were analyzed by maximum likelihood estimation of binormal ROC curve grading data [19]. The diagnostic accuracy of each hepatic arterial dominant phase imaging technique for each reviewer was evaluated by calculation of area under the ROC curve (Az). Az combines true-positive, false-positive, and false-negative ratings and level of confidence and represents a measure of the tradeoff between detecting true-positive versus false-positive lesions (percentage of true-positive findings before the first false-positive finding) over a range of thresholds [20]. Factors with Az values greater than 0.80 were considered to have good diagnostic accuracy on the basis of findings of a previous study in which ROC analysis was used [19]. The Az values of each imaging technique calculated from the ROC curves were analyzed with the jackknife method [21, 22].
Lesions assigned a grade of 4 or 5 (probably or definitely present) were regarded as positive for the presence of HCC. Lesions assigned a grade of 1 or 2 (probably or definitely absent) were regarded as negative. On the basis of these results, the sensitivity, specificity, and positive predictive value for detection of hypervascular HCC with each hepatic arterial dominant phase imaging technique were calculated. The present findings may not confirm absolute sensitivity, specificity, and positive predictive value because the study included only representative lesions of hypervascular HCC and did not include other types of hepatic disease. Therefore the terms "sensitivity," "specificity," and "positive predictive value" used in this study mean relative values. The significance of differences in these relative sensitivity, specificity, and positive predictive values between each hepatic arterial dominant phase imaging technique was estimated with the McNemar test [23].
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
For the scan timing (scan delay) of each hepatic arterial dominant phase imaging technique with the different injection rates (Table 2), scan timing was considered optimal in most of the hepatic arterial dominant phase images at 3 mL/sec (28/30 [93%]) and 5 mL/sec (27/30 [90%]). The scan timing of the hepatic arterial dominant phase images was judged to be too early in two (7%) of 30 patients with the 3 mL/sec injection rate and three (10%) of 30 patients with the 5 mL/sec rate. In no case was scan timing of the hepatic arterial dominant phase images with either injection rate judged to be too late, because no contrast enhancement of hepatic veins was identified on any hepatic arterial dominant phase image. There was no significant difference between the 3 mL/sec and 5 mL/sec rates in regard to scan timing of the hepatic arterial dominant phase images. These results suggest that the individually different scan delays for each injection rate did not affect the contrast of the hepatic arterial dominant phase images, possibly resulting in lesion detectability.
|
Interobserver Agreement for Image Interpretation
The chance-corrected kappa values indicating the confidence levels for
image interpretation between the three reviewers are summarized in
Table 3. The kappa values were
good agreement (
= 0.61-0.80) between any combination of two reviewers
for all hepatic arterial dominant phase images with both injection rates.
|
ROC Analysis
The findings of the composite ROC curves and Az,
sensitivity, specificity, and positive predictive value for each hepatic
arterial dominant phase imaging technique with the different injection rates
are shown in Figure 1 and
Table 4. Both mean
Az values of each hepatic arterial dominant phase imaging
technique were confirmed as good because they were greater than 0.80. There
were no significant differences in Az between the 3 mL/sec
(mean, 0.97 ± 0.01) and 5 mL/sec (mean, 0.96 ± 0.01) injection
rates of contrast material for all reviewers. There were also no significant
differences in the sensitivity values between 3 mL/sec (mean, 28/35 [80%]) and
5 mL/sec (mean, 36/41 [88%]) for all reviewers. Specificity and positive
predictive value, however, were significantly different between the 3 mL/sec
(mean, 236/250 [95%] and 28/42 [67%]) and 5 mL/sec (mean, 227/265 [86%] and
36/73 [49%]) rates. These results suggest that a larger number of
false-positive contrast-enhanced lesions in cirrhotic liver were detected on
hepatic arterial dominant phase images at the 5 mL/sec than at the 3 mL/sec
rate of injection of contrast material.
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Some investigators have suggested the presence of numerous contrast-enhanced false-positive lesions in cirrhotic liver on hepatic arterial dominant phase images. Most of the false-positive lesions may be associated with arterial-portal venous shunts underlying cirrhotic liver and are visualized on hepatic arterial dominant phase images depending on the degree of inflow of arterial blood [17]. In the presence of cirrhosis, hemodynamic changes represented by a dramatic decrease in portal blood flow resulting in an increase in arterial flow occur in various degrees. In the daily practice of hepatic CT interpretation, radiologists frequently encounter such false-positive lesions and are troubled by them on hepatic arterial dominant phase images. Therefore radiologists should thoroughly consider the risk that an increase in contrast-enhanced false-positive lesions on hepatic arterial dominant phase images may diminish overall diagnostic performance at a higher rate of injection of contrast material in imaging of patients with cirrhotic livers.
Unfortunately, previous studies assessing the effects of different injection rates of contrast material in the detection of hypervascular HCC have dealt with only true-positive lesions [5, 16]. To our knowledge, no attempt has been made to assess the presence of contrast-enhanced false-positive lesions represented by small arterial-portal venous shunts. The qualitative results based on results of ROC analysis in the present study showed that the detection rate of contrast-enhanced false-positive lesions on hepatic arterial dominant phase images was significantly higher with an injection rate of 5 mL/sec than with a 3 mL/sec injection rate. Both specificity and positive predictive value were significantly lower at 5 mL/sec than at 3 mL/sec.
We could not determine an optimal injection rate of contrast material for contrast-enhanced MDCT to evaluate hypervascular HCC in patients with cirrhotic livers because our results of ROC analysis showed no significant difference between 5 mL/sec and 3 mL/sec injection rates in regard to either comprehensive diagnostic performance or sensitivity. However, a high rate of injection of contrast material, such as 5 mL/sec, is not always better than a medium injection rate, such as 3 mL/sec, in the evaluation of hypervascular HCC in patients with cirrhotic livers. We recommend use of a medium injection rate for hepatic MDCT imaging in patients with cirrhosis because the specificity and positive predictive value in detection of hypervascular HCC are significantly higher with a medium injection rate than with a high injection rate. Because of the definitely high rate of detection of false-positive lesions with a high injection rate, our recommendation may be especially useful for general institutions employing radiologists who are not experts in interpretation of hepatic CT images.
There were several limitations to the present study. The dose of iodine in the contrast material (35 g/body) seemed to be less than the dose most institutions currently use for hepatic CT, although it may have been acceptable in the present study because of the low body weight of the patients (range, 37-65 kg; mean, 56 kg). The dose of iodine in contrast material is important and should be carefully determined at each institution because iodine dose significantly influences peak enhancement of each organ, the width of the temporal window for each acquisition phase, and the injection rate of contrast material [24, 25]. The study population also seemed too small for acknowledgment of all the results in the present study. In addition, several patients had no pathologic evidence because of the small size of the lesions included in the study, even though follow-up after arterial portographic and hepatic arteriographic CT examinations was conducted with sonograms, contrast-enhanced CT, and iodized-oil CT in all patients.
The selection bias in enrollment of patients in this study was substantial. To some extent the patients were selected and enrolled in the present study at the study coordinators' convenience. However, it may not have been possible to avoid selection bias, because the study coordinators had to pay careful attention to many factors, including suitability of images and the background of the patients, so that all factors were matched between the group with and the group without HCC. Thus selection bias of patients might have been artificially introduced in our results, and therefore the study might not have been randomized. Bias also occurred in the retrospective part of the study because all reviewers knew that each subject might have HCC alone if they had any lesion.
The present findings cannot confirm the absolute sensitivity, specificity, or positive predictive value of hepatic arterial dominant phase imaging in the prospective diagnosis of hypervascular HCC. The study did include patients without HCC but did not include as controls patients with other hypervascular hepatic lesions, such as hemangioma, hepatic adenoma, and focal nodular hyperplasia. Including these conditions in the study would have presented a more real-life situation.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  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]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
