| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Seoul National University Bundang Hospital, Seoul
National University College of Medicine, Institute of Radiation Medicine,
Seoul National University Medical Research Center, 300 Gumi-dong, Bundang-gu,
Seongnam-si, Gyeonggi-do 463-707, Korea.
2 Department of Radiology and the Institute of Radiation Medicine, Seoul
National University College of Medicine, Clinical Research Institute, 28
Yongon-dong, Chongno-gu, Seoul, 110-744, Korea.
3 Department of Radiology, University of Ulsan College of Medicine 388-1,
Poongnap-dong, Songpa-ku, Seoul 138-736, Korea.
Received November 22, 2002;
accepted after revision November 10, 2004.
Address correspondence to J. K. Han
(hanjk{at}radcom.snu.ac.kr).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Helical CT scans of 60 patients with hepatic abscess were analyzed for the presence of complete or partial obstruction of the portal or hepatic veins and for attenuation differences in the surrounding parenchyma. Clinical (age, sex, underlying disease, and microorganism) and CT (obstruction of the portal or hepatic vein and number, location, and size of abscesses) findings were analyzed statistically for possible associations with each of regional parenchymal hyper- and hypoattenuation by using the chi-square test and multivariate logistic regression analysis.
RESULTS. Regional parenchymal hyperattenuation was identified in 40 patients (67%). More patients with portal vein obstruction showed regional parenchymal hyperattenuation than patients without portal vein obstruction (22/27 patients vs 18/33, p = 0.028), and more patients with hepatic vein obstruction showed regional parenchymal hypoattenuation than those without hepatic vein obstruction (11/21 vs 3/39, p = 0.0003). Multivariate logistic regression analysis showed that portal venous obstruction was the only statistically significant predictor of regional parenchymal hyperattenuation (p = 0.032; odds ratio, 3.7) and that parenchymal hypoattenuation was associated with hepatic venous obstruction (p = 0.001; odds ratio, 44.9).
CONCLUSION. Parenchymal hypo- and hyperattenuation are frequently observed in the hepatic region surrounding an abscess on dynamic CT. Moreover, these parenchymal attenuation differences are associated with regional portal or hepatic vein obstruction.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Helical CT is currently used as one of the primary diagnostic tests in patients with a suspected hepatic abscess. However, most CT studies of hepatic abscesses were published before the introduction of helical CT. Therefore, we believed that an investigation of the hemodynamic alterations associated with hepatic abscesses and of their presentations on dynamic CT was warranted. This study was undertaken to describe regional attenuation differences of the hepatic parenchyma surrounding hepatic abscesses and to determine their possible associations with obstruction of the regional portal or hepatic veins.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Pyogenic abscess was confirmed by positive cultures from percutaneous aspirates in 51 of the 60 patients (Table 1). Forty-four of these 51 patients had monomicrobial abscesses and seven had polymicrobial abscesses. In three patients, amebic abscess was diagnosed by positive serologic testing; these patients responded to percutaneous drainage and amebicidal therapy. In the remaining six patients, whose cultures were negative for microbial growth, the diagnosis of hepatic abscess was based on clinical and CT findings. All six of these patients responded to percutaneous drainage and antimicrobial therapy.
Of the 60 patients analyzed, 12 were documented to have diabetes mellitus. Other factors predisposing to hepatic abscess were documented in 28 patients, which included four of the 12 diabetes mellitus patients; these included a history of bilioenteric anastomosis for biliary or pancreatic malignancy (n = 12), hepatobiliary lithiasis (n = 8), a history of other malignant tumor (n = 6) including gastrointestinal tract cancer (n = 4), and abdominal infectious disease (n = 2, acute cholecystitis and colonic diverticulitis in each patient). Of the remaining 32 patients, eight had diabetes mellitus and 24 patients had no documented predisposing factor.
CT Acquisition
The CT examinations were performed using different scanners (Somatom Plus S
and Somatom Plus 4, Siemens Medical Solutions; HiSpeed Advantage, GE
Healthcare; MX 8000, Marconi Medical Systems), and CT techniques also varied
because of the retrospective nature of the study. In general, examinations
were performed using a spiral technique with 5- to 10-mm collimation and 5- to
10-mm reconstruction intervals. The X-ray tube voltage was 120-140 kV, and the
current varied between 150 and 220 mA. During the study period, our standard
protocol for dynamic CT consisted of a total volume of 100-150 mL of nonionic
IV contrast material (300-370 mg I/mL) administered by power injection at a
rate of 3 mL/sec, with a scanning delay of 30-40 sec for the hepatic artery
phase (HAP) and 60-80 sec for the portal venous phase (PVP). Of the 60
patients analyzed, 52 underwent dual-phase (HAP and PVP) CT, and the remaining
eight underwent single-phase (PVP) CT.
CT Analysis
Four abdominal radiologists participated in the analysis. Two reviewers
analyzed the presence of portal or hepatic vein obstruction, and the other two
analyzed the presence of regional parenchymal hyper- or hypoattenuation. All
four were blinded to the study concept when they analyzed CT images, and the
two pairs of radiologists were blinded to each other's results. All CT images
were displayed using a PACS (Marotech). Window and level setting manipulation
was permitted during the analysis.
CT images were analyzed independently by two radiologists who determined the number of abscesses and whether the regional portal or hepatic vein was obstructed around the abscess. Clustered lesions were counted as a solitary lesion. The regional portal vein or the hepatic vein was considered to be obstructed in the presence of at least one of the following CT findings: (a) thrombosis was identified as an intravascular tubular low-attenuation lesion during the PVP, (b) the portal vein or the hepatic vein was replaced or displaced by the abscess and not identified at all while vessels in the other hepatic regions were depicted by contrast-enhancement, or (c) the portal vein or hepatic vein in contact with the abscess was stenotic as compared with vessels in other hepatic regions. Because two or more of findings (a), (b), and (c) could be observed at different segments of the same vessel, these findings were analyzed separately. Findings (a) or (b) were considered to indicate a complete obstruction; and finding (c), without findings (a) or (b), was considered to indicate a partial obstruction. A consensus evaluation by the two reviewers was obtained for images that caused reviewer disagreements. One of the two reviewers determined abscess size by measuring the largest transverse diameter on CT and classified the location of the abscess as left liver, right anterior section, or right posterior section. A lesion in the caudate lobe was classified as one in the right posterior section.
In addition to and independently from the analyses just described, all CT images were also analyzed for the presence of each of parenchymal hyper- and hypoattenuation area in the hepatic region surrounding the abscess, as compared with the attenuation of other hepatic regions. Because results were analyzed statistically on a per-patient basis, for patients with multiple abscesses, regional parenchymal hyper- and hypoattenuation were determined to be present if observed in a hepatic region adjacent to the single largest abscess. If both regional parenchymal hyper- and hypoattenuation were present in the same patient, this patient was allocated to both the hyper- and hypoattenuation categories. This analysis was performed independently by the two other radiologists. If the two reviewers disagreed, they together performed region-of-interest (circular, 1 cm in diameter) measurements on the mean attenuation number (in Hounsfield units) in hepatic regions devoid of large vessels. The threshold attenuation difference was 10 H [8, 9] in these patients. One of the two reviewers recorded the temporal phases (HAP, PVP, or both phases) of the CT scan during which regional parenchymal hyper- or hypoattenuation appeared.
Statistical Analysis
Clinical and CT findings (the 10 variables listed in
Table 1) were statistically
analyzed to identify possible associations with each type of regional
parenchymal attenuation difference (hyper- or hypoattenuation) observed on CT.
In each patient with multiple abscesses, only the largest abscess was included
in the statistical analysis because it was impossible to determine which of
the multiple abscesses in any given hepatic region was responsible for a
regional parenchymal attenuation difference. In terms of portal or hepatic
vein obstruction, reviewers' responses were collapsed into a binary answer
(i.e., obstructed or not obstructed) regardless of whether the obstruction was
complete or partial. A regional parenchymal attenuation difference was
determined to be present if this finding was observed during at least one of
the two temporal phases (HAP and PVP) of the CT scanning. Bivariate
associations between variables were analyzed using the chi-square test. To
identify independent variables associated with regional parenchymal
hyperattenuation (or hypoattenuation) after controlling for all other factors,
the 10 listed variables were subjected to stepwise multivariate logistic
regression analysis. The presence of regional parenchymal hyperattenuation (or
hypoattenuation) was treated as the outcome variable. Relationship strengths
were ranked using odds ratios.
Kappa statistics were used to evaluate interobserver agreement with respect
to binary responses concerning the presence of portal or hepatic vein
obstruction and of regional parenchymal hyper- or hypoattenuation. Strengths
of agreements were interpreted according to the guidelines suggested by Landis
and Koch [10] as follows:
almost perfect, 
= 0.81-1.00; substantial, = 0.61-0.80;
moderate, = 0.41-0.60; fair, = 0.21-0.40; and slight,
= < 0.20.
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The regional portal vein was considered to be completely (n = 20) or partially (n = 7) obstructed in 27 patients (45%). In these 27 patients, portal venous thrombosis was observed in 17, the regional portal vein was not identified in five, and stenosis of the portal vein was observed in 18. Stenotic portal veins were surrounded by a thin low-attenuation cuff, although this finding was indistinguishable from dilated bile ducts in six patients (Figs. 1A, 1B, 2A, 2B, and 2C). The regional hepatic vein was considered to be completely (n = 17) or partially (n = 4) obstructed in 21 patients (35%). In these patients, hepatic venous thrombosis was observed in 12 patients, and the regional hepatic vein was not identified in eight. Stenosis of the hepatic vein was observed in five patients; however, the perivascular low-attenuation cuff was not observed around the stenotic hepatic vein (Figs. 3A, 3B, and 3C). In 11 patients, both portal venous and hepatic venous obstructions were observed.
|
|
|
|
|
Regional parenchymal hyperattenuation was identified in 40 (67%) of the 60 patients analyzed. In 52 patients who underwent dual-phase CT, this finding was observed in 36 (69%) patients during HAP (n = 14), PVP (n = 5), or both phases (n = 17). The observed regional parenchymal hyperattenuation always involved the subcapsular region of the liver and usually appeared as a wedge shape with straight margins (Figs. 1A, 1B, 2A, 2B, and 2C). Regional parenchymal hypoattenuation was identified in 14 (23%) of the 60 patients analyzed during HAP (n = 2), PVP (n = 7), or both phases (n = 5). Regional parenchymal hypoattenuation areas were all located peripheral to the abscess and had ill-defined margins (Figs. 3A, 3B, and 3C). In 10 patients, both hyperattenuation and hypoattenuation of the regional parenchyma were observed in different areas adjacent to the abscess.
Associations Between Regional Parenchymal Attenuation Differences and Selected Variables
Table 1 presents the
percentages of patients who showed regional hyper- or hypoattenuation in the
surrounding parenchyma by categories of various factors, with the results of
chi-square testing. Bivariate analysis using the chi-square test revealed that
more patients with portal vein obstruction had regional parenchymal
hyperattenuation than those without portal vein obstruction (22/27 patients vs
18/33, p = 0.028), and that more patients with hepatic vein
obstruction had regional parenchymal hypoattenuation than those without
hepatic vein obstruction (11/21 vs 3/39, p = 0.0003). However, no
other significant association was found between any combination of the 12
variables, including regional parenchymal hyper- or hypoattenuation, listed in
Table 1.
Stepwise multivariate logistic regression analysis revealed that of the 10 variables examined, the presence of portal venous obstruction was the only significant predictor of regional parenchymal hyperattenuation (p = 0.032; odds ratio, 3.7; 95% confidence interval [CI], 1.1, 12.0). The presence of hepatic venous obstruction was the most significant predictor of regional parenchymal hypoattenuation (p = 0.001; odds ratio, 44.9; 95% CI, 4.6, 435.6). The likelihood of regional parenchymal hypoattenuation also increased when the abscess was located in the right anterior section rather than in the left liver (p = 0.030; odds ratio, 14.5; 95% CI, 1.3, 162.9). None of the other eight variables was found to be predictive of regional parenchymal hypoattenuation.
Reviewer Disagreements
Reviewer disagreement concerning the number of abscesses (two adjacent
lesions vs a clustered solitary lesion) occurred in three instances. After a
consensus evaluation, these lesions were considered to be single abscesses in
each patient. Table 2 summarizes the kappa statistics for interobserver agreement with respect to
binary responses for the presence of obstruction of the portal or hepatic vein
and of regional parenchymal differences, and the results of consensus
evaluation. Interobserver agreement was almost perfect or substantial for all
CT findings analyzed except for two variables (nonidentification of the portal
vein and stenosis of the hepatic vein), for which interobserver agreement was
moderate.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
One clinical implication of our observations is that the CT findings of intrahepatic vascular obstruction and associated hemodynamic alteration around a hepatic focal lesion do not necessarily indicate malignancy. Because such findings typically have been described to be helpful in the diagnosis of a malignant tumor, especially hepatocellular carcinoma [5, 11], they may lead to a misdiagnosis [12] if other CT findings of a hepatic abscess are atypical.
Our results concerning the association between portal venous obstruction and regional parenchymal hyperattenuation validate the hypothesis proposed by several researchers [1-4] who attributed parenchymal hyperenhancement around a hepatic abscess to an arterioportal shunt resulting from portal venous compromise.
In cases of diminished hepatic venous flow, the hemodynamics is complex and can differ according to the site and the chronicity of the occlusion [5]. When the hepatic vein is acutely obstructed, which is likely in the case of a hepatic abscess, it is generally believed that the portal vein becomes a draining vein, thus resulting in a functional arterioportal shunt, which appears as parenchymal hyperattenuation in the congested hepatic region during the HAP [13, 14]. Although statistically not significant, regional parenchymal hyperattenuation was observed in 16 of 21 patients with hepatic venous obstruction. Three of these patients had CT findings that were relevant to this hypothesis. The findings of one of these patients are illustrated in Figures 4A, 4B, and 4C. In the other two patients, the vertex of the wedge-shaped hyperattenuating area pointed to the thrombosed hepatic vein [5, 15]. However, in the remaining 13 patients, we could not recognize any distinguishing feature of regional parenchymal hyperattenuation that might be specific to this patient subgroup.
|
|
|
Our statistical analysis showed that hepatic venous obstruction is associated with regional parenchymal hypoattenuation, but not with hyperattenuation. This is not a surprising result when one considers the complexities of hemodynamics in cases of hepatic venous compromise. An experimental study [16] recently revealed that acute occlusion of the hepatic vein can induce hypoattenuation in the affected parenchyma, probably because an arterioportal shunt occurs at the presinusoidal level (through the transplexal route), and the reversed opacified portal flow escapes from (or does not enter) the congested segment and is then dispersed into adjacent segments. The parenchymal hyperenhancement that results from an arterioportal shunt, which is an idea derived mainly from studies [13, 14] using angiography or CT during arteriography, might not be observed in dynamic CT after IV contrast enhancement because reduced portal venous enhancement outweighs increased arterial enhancement in a congested hepatic region. Although this complex facet of hemodynamics that results in regional parenchymal hypoattenuation has not been described in human patients with hepatic venous obstruction, regional parenchymal hypoattenuation has been recently described in acutely contested hepatic segments after living donor liver transplantation [17]. This may be observed only when the hepatic vein is acutely obstructed but collateral drainage has not had time to develop, which is likely if the obstruction is caused by acute inflammation such as is caused by a hepatic abscess.
Our multivariate statistical analysis showed an unexpected association between abscess location and regional parenchymal hypoattenuation; however, the implications of this association are not clear.
In our patients, it was virtually impossible to determine whether a given hepatic abscess was the result of pylephlebitis or vice versa. However, we believe that portal or hepatic vein obstruction occurred as a result of hepatic abscess development in most patients because none of our patients with these venous obstructions had a documented abdominal infectious disease likely to have caused pylephlebitis [18]. Regional venous thrombosis is a well-known complication of various local infectious diseases throughout the body [19-22]. Furthermore, it has been reported that thrombosis of the portal [4, 23, 24] and hepatic [23, 25] veins might be the result of an intrahepatic inflammatory lesion. Hanazaki et al. [24] proposed that a hepatic abscess may lead to infectious damage of the portal vein, which might result in thrombosis. This hypothesis was reinforced by Gabata et al. [3], who reported marked periportal inflammation and stenosis of the portal venules surrounding hepatic abscesses. In our study, diffuse stenosis of the regional portal vein, with a periportal low-attenuation cuff, was frequently (11/18 patients) found to be accompanied by thrombosis at distal branches. We believe that these CT findings may represent different stages or degrees of infectious phlebitis.
In our study, obstruction of the portal or hepatic veins did not always induce a parenchymal attenuation difference, and a parenchymal attenuation difference was not always accompanied by portal or hepatic vein obstruction. This can be explained by several factors, as follows: First, hemodynamic alterations in a hepatic region can certainly occur without a large vascular obstruction identifiable on CT. Smaller peripheral vessels might be obstructed that are not visualized on CT. Second, in addition to vascular obstruction, the compressive effect of an abscess on the regional parenchyma might have a role in the determination of parenchymal attenuation differences [26]. The degree of this compressive effect may be correlated with abscess size; however, no significant association was observed in our study between abscess size and parenchymal attenuation differences. Finally, the parenchymal attenuation differences observed in our study may not necessarily indicate hemodynamic alterations in the normal hepatic parenchyma. Without histopathologic data, it is unclear whether parenchymal attenuation differences reflect histopathologic changes, such as edema or inflammation, in some patients.
The major limitation of our study is that the variables analyzed did not include clinical and laboratory findings describing abscess inflammatory activity, and therefore our statistical analysis might represent a compilation of snapshots of abscesses at different stages of evolution and resolution. Regional parenchymal hyperattenuation has been reported to disappear after antibiotic therapy for hepatic abscess [3]. It is unclear whether the disappearance of a parenchymal attenuation difference is accompanied by a reinstatement of blood flow in an obstructed vessel. We did not stratify inflammatory activity because all of our patients required a percutaneous drainage procedure, which entails high levels of inflammatory activity in most patients. The second limitation of our study concerns the variability of the CT protocols and iodine concentrations in contrast material. Because many patients were scanned before the introduction of the bolus tracking technique, the temporal phase of CT scans varied. These factors could not be controlled for because of the retrospective nature of our study. Moreover, this limitation precluded a consistent quantitative analysis of regional attenuation values. Third, we pooled the results of parenchymal attenuation differences between the HAP and the PVP for the statistical analysis for simplification reasons and to compensate for inconsistencies in the temporal phases of CT scanning. This limitation precluded a more systematic analysis by substratifying data according to the temporal phases of the CT scanning. For example, there may have been more uncounted regional parenchymal hyperattenuation in the eight patients who underwent single-phase CT. Therefore, a more elaborate study, such as a prospective study using a consistent CT protocol, is needed to confirm our results. The fourth limitation is that we did not consider other factors affecting hepatic attenuation on dynamic CT. These factors include the presence of liver cirrhosis and the patient's hemodynamic status, the latter of which is likely to be unstable in such acutely ill patients.
In conclusion, obstruction of the portal or hepatic vein and parenchymal hypo- and hyperattenuation are frequently observed in the hepatic region surrounding an abscess on dynamic CT. Parenchymal hyper- or hypoattenuation of the hepatic region surrounding an abscess was found to be associated with obstruction of the regional portal vein or the hepatic vein, respectively, although parenchymal hyperattenuation can be observed without visible portal venous thrombosis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  H. J. Lee, K. W. Kim, H. S. Mun, J. H. Kim, G. W. Song, S. Hwang, and S. G. Lee Uncommon Causes of Hepatic Congestion in Patients After Living Donor Liver Transplantation Am. J. Roentgenol., September 1, 2009; 193(3): 772 - 780. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
