| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gifu University Hospital Gifu 500-8705, Japan
Dr. Steingruber et al. are to be congratulated for their splendid article [1] on the accuracy of CT during arterial portography and digital subtraction hepatic angiography in the pretransplantation detection of hepatocellular carcinoma in cirrhosis in the July 2003 issue of the AJR [1]. In their study, the accuracy of CT for detecting hepatocellular carcinoma determined by receiver operating characteristic curve analysis was not sufficient compared with that of conventional angiography. I respectfully disagree with this observation. I find three points at issue in their study methods. The first issue involves image quality assurance for CT, the second involves diagnostic criteria on CT, and the third involves histopathologic diagnosis.
First, the CT images in figures 3 and 4 in their article showed heterogeneous insufficient hepatic parenchymal enhancement in spite of intense enhancement in the portal vessels. I assume that the authors might have used contrast material of 300 mg I/mL concentration and may have begun scanning somewhat earlier than at the optimal time. Regarding concentration, although the authors might have wanted to show what iodine concentration was used in their study, most radiologists in Japan use 150 mg I/mL for CT during arterial portography [2, 3].
Regarding the scanning delay, Steingruber et al. [1] used bolus-tracking software and set a region of interest in the portal vein and began scanning when the CT value reached 100–120 H. First, radiologists should remember that peak hepatic parenchymal enhancement is often delayed by cirrhosis with portal hypertension. I have reported [4] that hepatic parenchymal enhancement increased up to 177 H when using contrast material at a concentration of 150–160 mg I/mL; it increased up to 301 H at a concentration of 300–320 mg I/mL. I infer that Steingruber et al. should have set a higher CT value as the threshold or set a region of interest in hepatic parenchyma. I use double-phase CT during arterial portography with MDCT: The first-phase images are obtained especially for recognition of portal venous anatomy and are acquired as soon as the CT attenuation in the portal trunk reaches 100 H. The second-phase images for hepatic parenchymal evaluation are acquired approximately 15 sec after the trigger of bolus-tracking technique. Parenchymal enhancement is homogeneous and intense, and conspicuity of tumors is better with the second-phase images.
Second, Steingruber et al. [1] regarded round perfusion defects smaller than 1 cm as dysplastic nodules on CT, but I have not encountered dysplastic nodules showing perfusion defects in my last 10 years of practice. I regard tiny round perfusion defects as imaging findings of malignant tumors, cysts, hemangiomas, granulomas, peripheral arterioportal shunting, and so on.
Third, Steingruber et al. [1] seemingly were dedicated to achieving rigorous histopathologic evaluation and found 16 well-differentiated tumors (41%) in 39 tumors. This ratio of well-differentiated tumors is some-what greater than I would expect, and I believe their pathologists were prone to characterize borderline lesions as cancerous even when no clear histopathologic and clinical evidence of malignancy was available.
CT during arterial portography is less sensitive to borderline lesions in cirrhotic patients. I have performed CT during arterial portography in patients with cirrhosis, used the histopathologic criteria advocated by the International Working Party [5], and followed up these patients for years, but histopathologically diagnosed well-differentiated tumors were not necessarily invisible on optimized CT during arterial portography.
To conclude, I respectfully suggest that the most important points in discussing the accuracy of CT during arterial portography for detecting hepatocellular nodules in cirrhosis include how optimal the imaging techniques are, how experienced the radiologists are, and how standardized the histopathologic diagnostic criteria are.
References
Innsbruck University Hospital Innsbruck 6020, Austria
We thank Dr. Kanematsu for the insightful comments. The evaluation of the hemodynamics of a hepatocellular carcinoma in the cirrhotic liver is based on both the portal supply as assessed by CT arterioportography and the hepatic arterial supply as assessed by digital subtraction hepatic angiography or CT hepatic arteriography. A combined interpretation using both techniques is clinically more relevant because it enables better sensitivity and specificity in the detection of hepatocellular carcinoma in the cirrhotic liver than either CT arterioportography or digital subtraction hepatic angiography does alone, as we described in our article [1].
Kanematsu raises three issues regarding the accuracy of our CT arterioportography findings. First, we agree that homogeneous enhancement on CT arterioportography is important. However, homogeneous enhancement on CT arterioportography in a cirrhotic liver depends on several factors, including the severity of cirrhosis, the presence of arteriovenous or arterioportal shunts, parenchymal necrosis, fatty infiltration, and parenchymal fibrosis. Both of the patients depicted in our figures 3 and 4 had significant portosystemic shunts diverting a considerable amount of contrast material away from their livers, which resulted in slight heterogeneous enhancement. Furthermore, we used 100–120 H in the portal vein as a threshold for initiating the first phase of CT arterioportography because we know that when using the SmartPrep software (General Electric Medical Systems, Milwaukee, WI) to determine scanning delay, the CT scanner requires at least 5 sec after reaching the required threshold in the portal vein to start scanning. This delay time is referred to as "the obligatory fastest possible time for the transition from SmartPrep to the diagnostic helical scan" [2, 3]. An enhancement of 100–120 H in the portal vein during SmartPrep does not mean that the parenchymal enhancement during CT acquisition taking place at least 5 sec later at the top of the liver and several additional seconds later at the bottom of the liver would be 100–120 H; the enhancement would be considerably more. Thus, our CT arterioportography protocol was not suboptimal.
The contrast medium application protocol presented by Kanematsu using an MDCT scanner is indeed interesting. However, to our knowledge, its accuracy in detecting and—more important—in ruling out hepatocellular carcinoma is still unknown. It would be interesting to learn more about Kanematsu's experience with MDCT arterioportography and how it compares with liver explant histopathologic findings, which are the only reliable gold standard for evaluating specificity—the ability to correctly rule out hepatocellular carcinoma. In this regard, we believe that homogeneous and intense enhancement would not rule out every well-differentiated hepatocellular carcinoma that still has a preserved portal blood supply.
Second, finding a suspicious intrahepatic nodule smaller than 1 cm on CT arterioportography does not necessarily justify the characterization of that nodule as hepatocellular carcinoma. Round perfusion defects smaller than 1 cm were considered to be dysplastic nodules only at the prospective analysis, which was performed according to our clinical practice rules. Our rules require follow-up or biopsy to confirm the diagnosis for a 1-cm intrahepatic lesion. At the retrospective analysis, which was performed to scientifically assess the accuracy of CT arterioportography and digital subtraction hepatic angiography, all hepatocellular carcinomas were included. We believe this method does not compromise the clinical experience of the CT arterioportography reviewers. As for the density of dysplastic nodules, it is well known that they have variable enhancement on CT arterioportography [4, 5].
Third, the standard of reference used in our study was liver explant histopathologic findings. All patients underwent liver transplantation under strict requirements, which included restrictions on the number and size of hepatocellular carcinomas. Thus, only patients with early or limited hepatocellular carcinomas were recruited for the study. A large number of patients were excluded from liver transplantation because of advanced hepatocellular carcinoma. Disqualifying criteria included one hepatocellular carcinoma larger than 5 cm in diameter; two or three hepatocellular carcinomas, one of them exceeding 3 cm; and more than three hepatocellular carcinomas. We believe that this obligatory preoperative patient selection is the source of the relatively high rate of well-differentiated hepatocellular carcinomas in our study. All histopathologic examinations were performed according to standard techniques, and borderline lesions were not considered hepatocellular carcinomas.
In summary, we agree that the evaluation of CT arterioportography for preoperative surveillance of the cirrhotic liver does indeed depend on optimized and validated imaging techniques; on radiologists with profound clinical, technical, and statistical experience; and on comprehensive histopathologic analysis.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
