| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gastrointestinal Imaging |
1 From the Department of Radiology (S.H.P., P.N.K., K.W.K., S.W.L., S.E.Y., H.K.H., M.G.L.), Division of Hepatobiliary Surgery and Liver Transplantation, Department of Surgery (S.H., S.G.L.), and Department of Diagnostic Pathology (E.S.Y.), University of Ulsan College of Medicine, Asan Medical Center, 388-1 Poongnap-Dong, Songpa-Gu, 138-040 Seoul, Korea; Department of Radiology, Eulji University School of Medicine, Eulji Hospital, Seoul, Korea (S.W.P.); and Department of Pathology, Sungkyunkwan University School of Medicine, Samsung Medical Center, Seoul, Korea (E.Y.C.). Received March 2, 2005; revision requested April 27; revision received May 16; accepted June 13; final version accepted July 5. Address correspondence to P.N.K. (e-mail: pnkim{at}amc.seoul.kr).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: Institutional review board approval and informed consent were obtained. A total of 154 candidates, including 104 men (mean age, 30.2 years ± 10.3 [standard deviation]) and 50 women (mean age, 31.8 years ± 11.2), underwent same-day unenhanced CT and ultrasonography-guided liver biopsy. Histologic degree of macrovesicular steatosis was determined. Three liver attenuation indices were derived: liver-to-spleen attenuation ratio (CTL/S), difference between hepatic and splenic attenuation (CTL–S), and blood-free hepatic parenchymal attenuation (CTLP). Regression equations were used to quantitatively estimate the degree of macrovesicular steatosis. Limits of agreement between estimated macrovesicular steatosis and the reference standard were calculated. Receiver operating characteristic analyses were used to determine the performance of each index for qualitative diagnosis of macrovesicular steatosis of 30% or greater. The cutoff value that provided a balance between sensitivity and specificity and the highest cutoff value that yielded 100% specificity were determined.
Results: Limits of agreement were –14% to 14% for CTL/S and CTL–S and –13% to 13% for CTLP. Performance in diagnosing macrovesicular steatosis of 30% or greater was not significantly different among indices (P > .05). Cutoff values of 0.9, –7, and 58 were determined for CTL/S, CTL–S, and CTLP, respectively, and provided a balance between sensitivity and specificity. Cutoff values of 0.8, –9, and 42 were determined for CTL/S, CTL–S, and CTLP, respectively, and yielded 100% specificity for all indices, with corresponding sensitivities of 82%, 82%, and 73% for CTL/S, CTL–S, and CTLP, respectively.
Conclusion: Diagnostic performance of unenhanced CT for quantitative assessment of macrovesicular steatosis is not clinically acceptable. Unenhanced CT, however, provides high performance in qualitative diagnosis of macrovesicular steatosis of 30% or greater.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Liver biopsy is used as the standard method for quantitative assessment of the degree of hepatic steatosis in donor candidates (3). Although liver biopsy is known to be a safe procedure with a low complication rate in healthy living donors (14,15), it is still an invasive procedure that may cause morbidity (16). It has been suggested that unenhanced computed tomography (CT) might be useful in the noninvasive quantification of the degree of hepatic steatosis in experimental and in vivo human studies (17–19). To our knowledge, only a few reports have been published regarding the role of CT for the evaluation of hepatic steatosis in potential LDLT donors (12,20,21). In one report, researchers suggested that unenhanced CT might preclude liver biopsy in only a small percentage of liver donor candidates who have an unacceptable degree of macrovesicular steatosis (12). This study, however, had a small sample size and a wide range of time (ie, up to 4 weeks) between liver biopsy and CT (12). Therefore, the purpose of our study was to determine prospectively the diagnostic performance of unenhanced CT in the assessment of macrovesicular steatosis in potential donors for LDLT by using same-day biopsy as the reference standard.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Study Population
Between August and December 2004, 158 consecutive potential donors for LDLT were enrolled in our prospective study. Ultrasonography (US)-guided liver biopsy was performed after CT on the same day unless there were any abnormal findings other than fatty liver on CT images that might preclude liver donation. Four donor candidates were excluded from liver biopsy because of diffuse intrahepatic bile duct dilatation (n = 1), numerous tiny cystlike lesions in the liver that suggested biliary hamartoma (n = 1), and uncharacterized focal low attenuation lesions in the liver (n = 2). The remaining 154 donor candidates included 104 men (mean age, 30.2 years ± 10.3 [standard deviation]) and 50 women (mean age, 31.8 years ± 11.2).
CT Scanning
In all donor candidates, CT was performed with a multi–detector row helical scanner (Lightspeed 16; GE Medical Systems, Milwaukee, Wis). In addition to contrast material–enhanced CT of the liver, which was used to obtain hepatic CT angiograms, unenhanced CT was performed to acquire transverse images through the liver during a single breath hold. Scanning and reconstruction parameters for unenhanced CT images were as follows: beam collimation, 8 x 2.5 mm; table speed, 27 mm per rotation; beam pitch, 1.35; gantry rotation time, 0.6 second; 120 kVp; 150 mAs; section thickness, 5 mm; and reconstruction interval, 5 mm. CT images were reviewed on a picture archiving and communication system workstation (PetaVision; Asan Medical Center, Seoul, Korea) by a radiologist (K.W.K., with 8 years of experience) who was blinded to histologic findings.
For each candidate, three different liver attenuation indices were obtained by using the unenhanced CT images. The first attenuation index—the liver-to-spleen attenuation ratio (CTL/S)—was calculated as L/S, where L is the hepatic attenuation and S is the splenic attenuation. The second attenuation index—the difference between the hepatic and splenic attenuation (CTL–S)—was calculated as L – S. The third attenuation index—the blood-free hepatic parenchymal attenuation (CTLP)—was calculated as [L – 0.3(0.75P + 0.25A)]/0.7, where P is the portal venous blood attenuation and A is hepatic arterial blood attenuation.
Calculations for CTLP were derived as follows: Although the regions of interest (ROIs) that were used for measurement of hepatic attenuation were placed with care to exclude macroscopic hepatic vessels, because of the rich sinusoids in the liver, in vivo hepatic attenuation represents a combination of both hepatic parenchyma and blood within the sinusoids. The liver contains about 12% of the total blood volume (22), and about 25%–30% of the hepatic volume in vivo is accounted for by blood (23). We therefore assumed that L = (0.7 · CTLP) + (0.3 · hepatic blood attenuation). Because the portal vein provides approximately 75% of the blood flow to the liver and the hepatic artery provides the other 25% (24), we calculated the hepatic blood attenuation as 0.75P + 0.25A.
To obtain the three indices, the hepatic attenuation was measured by averaging the Hounsfield units of three ROIs. Each ROI formed a 1.5 x 1.5-cm square, and the ROIs were placed at three different sites in the right hepatic lobe approximately between hepatic segments V, VI, VII, and VIII, which were defined according to the Couinaud system (Fig 1a). The locations of the ROIs were chosen to match the biopsy sites as closely as possible. Each ROI was placed with special care so as to exclude macroscopic hepatic vessels, and contrast-enhanced CT images were used to more precisely identify the locations of hepatic vessels.
|
US-guided Needle Biopsy and Histologic Analysis
Five board-certified radiologists, two of whom were authors (S.H.P., S.E.Y.), performed US-guided liver biopsy by using a freehand technique; each of the five radiologists had performed at least 100 US-guided liver biopsies. Biopsy was performed with an 18-gauge needle (Stericut 18G coaxial; TSK Laboratory, Tochigi, Japan). In all subjects, biopsy specimens were obtained twice at two different sites in the right hepatic lobe; these sites were located approximately between hepatic segments V, VI, VII, and VIII (Fig 1b). Biopsy sites were devoid of macroscopic vessels and were approximately 1 cm apart. Each biopsy specimen was approximately 1.5 cm in length. An experienced liver pathologist (E.S.Y., with 20 years of experience) who was blinded to the radiologic findings reviewed the histologic findings. The slides for histologic review were prepared with hematoxylin-eosin staining. The degree of macrovesicular and microvesicular steatosis was quantified by using a scale that was based on percentages (ie, the amount of liver parenchyma that was replaced by macrovesicular and microvesicular steatotic droplets). For our study, the degree of hepatic steatosis at histologic analysis was used as the reference standard.
|
To determine and compare the performance of CTL/S, CTL–S, and CTLP for the qualitative distinction of macrovesicular steatosis of 30% or greater (ie, the unacceptability limit for donation) and macrovesicular steatosis of less than 30%, we performed a receiver operating characteristic (ROC) analysis. The area under the ROC curves and 95% confidence intervals were calculated and compared among the three attenuation indices (26,27). For CTL/S, CTL–S, and CTLP, the cutoff values (ie, values less than or equal to the cutoff values used to indicate positivity for macrovesicular steatosis of 30% or greater) that provided a balance between sensitivity and specificity for the diagnosis of macrovesicular steatosis of 30% or greater (ie, minimal false-negative and false-positive results) were determined by selecting the coordinate that was nearest to the left upper corner (ie, [0, 1]) on each ROC curve. The sensitivities and specificities at these cutoff levels were compared among the three indices by using the McNemar test. The corresponding positive predictive and negative predictive values for each index were also calculated.
For CTL/S, CTL–S, and CTLP, the highest cutoff value that yielded 100% specificity for the diagnosis of macrovesicular steatosis of 30% or greater was determined. Sensitivities, positive predictive values, and negative predictive values at each of these cutoff points were calculated, and the sensitivities were compared among the three indices by using the McNemar test. Because both macrovesicular steatosis and microvesicular steatosis may influence hepatic attenuation at CT, the distribution of the degree of microvesicular steatosis for each degree of macrovesicular steatosis and the linear correlation between the degree of microvesicular steatosis and the degree of macrovesicular steatosis were assessed to help estimate the confounding effect of microvesicular steatosis on hepatic attenuation. The degree of microvesicular steatosis was also compared between subjects with macrovesicular steatosis of 30% or greater and those with macrovesicular steatosis of less than 30% by using the Mann-Whitney U test. A P value of .05 was considered to indicate a statistically significant difference.
For statistical analyses involving three pairwise comparisons, a P value less than .017 for each pairwise comparison was used to indicate a statistically significant difference and to account for an increase in 
error. Statistical analyses were performed by using two commercially available software programs (SPSS 11.5 for Windows, SPSS, Chicago, Ill, and MedCalc 7.4, MedCalc Software, Mariakerke, Belgium).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
In our study population, the number of cases decreased as the degree of macrovesicular steatosis increased; therefore, the majority of negative cases (ie, cases with macrovesicular steatosis of less than 30%) were distributed closer to 0% than to 30%. We believe that this distribution characteristic rendered the distinction between macrovesicular steatosis of 30% or greater and macrovesicular steatosis of less than 30% more clear-cut when compared with negative cases, which were more evenly distributed throughout the 0%–30% range. Therefore, high diagnostic performance, especially specificity, was partly attributed to the skewed distribution of the cases for each degree of macrovesicular steatosis in our study population. Such a distribution characteristic, however, is probably the nature of the LDLT donor candidate population, as is seen in other studies (12,20). Thus, we believe that our results are pertinent to the clinical situation.
Despite the high diagnostic performance of unenhanced CT for the qualitative assessment of macrovesicular steatosis, we do not think that determining the appropriateness of a potential hepatic donor based solely on unenhanced CT findings without liver biopsy is appropriate. As seen in our results and in those of other studies (12), the accuracy of quantitative estimation with CT is limited.
Occasionally, pathologic abnormalities of the hepatic parenchyma that preclude liver donation may go undetected at clinical examination, laboratory testing, and imaging (3,12,15). In a small number of cases, iron overload can increase hepatic attenuation, which may lead to the masking or underestimation of hepatic steatosis at CT (12,28). Our results, however, suggest that unenhanced CT may be useful in selecting LDLT candidates with moderate to severe macrovesicular steatosis who are unacceptable for liver donation regardless of other factors, thereby preventing unnecessary liver biopsy and the possible health risks and costs of biopsy in unacceptable donor candidates.
If we were to use unenhanced CT to screen out donor candidates with macrovesicular steatosis of 30% or greater, putting more emphasis on the shortage of liver donors than on the adverse effects of unnecessary biopsy, adopting strict criteria to minimize false-positive results (and thus not excluding eligible donor candidates) would be reasonable. Also, because of the low prevalence of moderate to severe macrovesicular steatosis in a potential living donor population, as seen in our study population (7%) and in that of others (6.8%) (20), a small decrease in specificity would produce a relatively large number of false-positive results and lead to a relatively large decrease in positive predictive value—that is, a decrease in the positive predictive value of CTL/S from 100% to 67% would be seen when specificity decreases from 100% to 97% and a decrease in the positive predictive value of CTLP from 100% to 61% would be seen when specificity decreases from 100% to 95%. Therefore, using 0.8, –9, and 42 as cutoff values for CTL/S, CTL–S, and CTLP, respectively, in the diagnosis of macrovesicular steatosis of 30% or greater would be more appropriate than adopting cutoff values of 0.9, –7, and 58 for CTL/S, CTL–S, and CTLP, respectively.
In general, our results coincided with those of a few previous studies (12,20). In a previous study (12), by using a CTL–S value of less than –10 to diagnose moderate to severe macrovesicular steatosis, researchers correctly diagnosed four of four donor livers with macrovesicular steatosis of greater than 30% and 38 of 38 donor livers with macrovesicular steatosis of less than 30%. A cutoff value of 1.1 was proposed for CTL/S to diagnose macrovesicular steatosis of 30% or greater, which provided a balance between sensitivity (83% [15 of 18]) and specificity (82% [202 of 248]) (20).
In our study, analysis was focused on the degree of macrovesicular steatosis and hepatic attenuation. Microvesicular steatosis, however, may also cause a decrease in hepatic attenuation at CT. After considering the positive correlation between the degree of macrovesicular steatosis and the degree of microvesicular steatosis in our study population (r = 0.689), it was presumed that the decrease in hepatic attenuation seen in cases with a higher degree of macrovesicular steatosis was partly attributed to an increase in microvesicular steatosis. In fact, the 11 subjects with macrovesicular steatosis of 30% or greater had a significantly higher degree of microvesicular steatosis than the 143 subjects with macrovesicular steatosis of less than 30%. The additional hepatic attenuation difference that was caused by differences in the degree of microvesicular steatosis might also help discriminate between these groups by using unenhanced CT.
Results for the strength of correlation between the degree of macrovesicular steatosis and hepatic attenuation indices, optimal cutoff values for each attenuation index in the diagnosis of macrovesicular steatosis of 30% or greater, and the diagnostic performance of the indices may change a little according to the distribution of microvesicular steatosis for each degree of macrovesicular steatosis. This might explain some of the small differences between our results and those of previous studies, although this is conjecture because the degree of microvesicular steatosis was not addressed in previous studies (12,20).
This study has limitations. First, although we chose macrovesicular steatosis of 30% or greater as the criterion for unacceptable living donors, the importance of hepatic steatosis in LDLT has not yet been clearly determined, and the acceptable ranges may vary according to institutions (7,13,29). Second, compared with the 143 negative cases (ie, cases with macrovesicular steatosis of less than 30%), only 11 positive cases (ie, cases with macrovesicular steatosis of 30% or greater) were noted. Third, at needle biopsy, only a small portion of the liver was obtained; therefore, sampling errors could have occurred because fatty changes are sometimes unevenly distributed in the hepatic parenchyma (30) and because needle biopsy may not be an ideal reference standard. At present, however, needle biopsy appears to be the most reasonable and practical method for objective quantification of hepatic steatosis (9). To improve the comparison between biopsy and CT, we tried to match the biopsy sites and the locations of the ROIs at CT as closely as possible. Fourth, histologic analysis on hemosiderin deposits was not performed.
In conclusion, the accuracy of unenhanced CT in the quantitative assessment of macrovesicular steatosis is not clinically acceptable. Unenhanced CT, however, provides high performance in the qualitative diagnosis of macrovesicular steatosis of 30% or greater and, therefore, will be helpful in avoiding unnecessary liver biopsy in those donor candidates with an unacceptable degree of macrovesicular steatosis.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: LDLT = living donor liver transplantation • ROC = receiver operating characteristic • ROI = region of interest
See also Science to Practice in this issue.
Author contributions: Guarantors of integrity of entire study, S.H.P., P.N.K.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, S.H.P., P.N.K., S.W.P., M.G.L., S.H., S.G.L., E.Y.C.; clinical studies, S.H.P., K.W.K., S.W.L., S.E.Y., S.H., E.S.Y.; statistical analysis, S.H.P.; and manuscript editing, S.H.P., P.N.K., H.K.H., S.G.L.
Authors stated no financial relationship to disclose.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. B. Soder, M. Baldisserotto, and V. D. da Silva Computer-Assisted Ultrasound Analysis of Liver Echogenicity in Obese and Normal-Weight Children Am. J. Roentgenol., May 1, 2009; 192(5): W201 - W205. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Nomura, Y. Ishizaki, K. Suzuki, and S. Kawasaki Development of Hepatic Steatosis After Pancreatoduodenectomy Am. J. Roentgenol., December 1, 2007; 189(6): 1484 - 1488. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Nickkholgh, J. Weitz, J. Encke, P. Sauer, A. Mehrabi, M. W. Buchler, J. Schmidt, and P. Schemmer Utilization of extended donor criteria in liver transplantation: a comprehensive review of the literature Nephrol. Dial. Transplant., September 1, 2007; 22(suppl_8): viii29 - viii36. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. W. Lee, S. H. Park, K. W. Kim, E. K. Choi, Y. M. Shin, P. N. Kim, K. H. Lee, E. S. Yu, S. Hwang, and S.-G. Lee Unenhanced CT for Assessment of Macrovesicular Hepatic Steatosis in Living Liver Donors: Comparison of Visual Grading with Liver Attenuation Index Radiology, August 1, 2007; 244(2): 479 - 485. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. K. Choudhary, L. F. Donnelly, J. M. Racadio, and J. L. Strife Diseases Associated with Childhood Obesity Am. J. Roentgenol., April 1, 2007; 188(4): 1118 - 1130. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Brancatelli Science to Practice: Should Biopsy Be Performed in Potential Liver Donors When Unenhanced CT Shows an Unacceptable Degree of Steatosis for Transplantation? Radiology, April 1, 2006; 239(1): 1 - 2. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
