| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, MRI Section, University of Michigan, 1500 E. Medical
Center Dr., UH B2B311, Ann Arbor, MI 48109-0030.
2 Center for Statistical Consulting and Research, University of Michigan, Ann
Arbor, MI 48109-0030.
3 Department of Medicine, University of Michigan, Ann Arbor, MI
48109-0030.
Received July 24, 2002;
accepted after revision September 11, 2002.
Address correspondence to R. C. Carlos.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Sixty examinations of 58 patients with biopsy
proof of lesions suggestive of hepatocellular carcinoma on MR imaging were
retrospectively reviewed. The signal intensity of the lesion on T2-weighted
imaging and dynamic gadoliniumenhanced imaging, the size of the lesion, and
the number of suspicious lesions were recorded; in addition, patient age and
sex, 
-fetoprotein level, and hepatitis C viral genotype were noted. The
association between malignancy and each predictor variable was evaluated using
the chi-square test or the two-group t test. The final logistic
regression model included the variables that were shown to have a significant
association with malignancy and the clinically relevant predictors. We used
the adjusted odds ratios to measure the strength of each association. The
discriminant ability of the model for detecting hepatic malignancy was
assessed using receiver operating characteristic curve analysis.
RESULTS. The prevalence of hepatic malignancy in our study
population was 64%. The area under the receiver operating characteristic curve
for the logistic regression model was 0.82. Venous washout (odds ratio = 9.2),
-fetoprotein level (odds ratio = 3.2), and number of lesions (odds
ratio = 1.5) were significant predictors for malignancy (p <
0.05). When arterial enhancement and venous washout were either both present
or both absent, -fetoprotein level contributed little to the prediction
of malignancy.
CONCLUSION. The MR characteristics of hepatic lesions during the
dynamic venous phase in conjunction with the serum -fetoprotein level
and number of lesions are predictors of hepatic malignancy. The use of these
predictors can facilitate explicit estimation of malignancy in individuals
with underlying cirrhosis, potentially improving clinical decision-making.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In patients with cirrhosis, one of the major roles of imaging is to reveal hepatocellular carcinoma and reliably enable the radiologist to differentiate malignant nodules from other masses. A standard MR examination, consisting of T1- and T2-weighted sequences and a dynamic gadolinium-enhanced sequence, is designed to detect and characterize hepatic lesions. MR criteria for lesion characterization include arterial enhancement and washout patterns [3, 4] and, to a lesser extent, signal characteristics on T2-weighted imaging [5, 6, 7, 8, 9]. Although the classic MR imaging features of hepatocellular carcinoma have been described, the imaging characteristics of benign tumors overlap with those of malignant tumors [3, 10, 11, 12, 13].
A prior study evaluating the ability of MR imaging to characterize hepatic lesions included cysts and hemangiomas [14]; however, the area of difficulty in MR imaging interpretation is differentiating between solid benign and solid malignant tumors. The value of obtaining a clinical history and of performing MR imaging has also been reported [15, 16], particularly for evaluating incidental liver lesions. The authors of these reports found that in the absence of a clinical history of cancer, none of the detected hepatic lesions were malignant. In both studies, whether the patients had a history of underlying liver disease was not recorded. For their study, Tello et al. [16] developed a prediction rule for assessing liver masses. These researchers used T2 relaxivity combined with clinical and demographic characteristics to detect malignancy in patients with or without a history of malignancy, but these researchers did not record whether patients had a history of liver disease. However, measuring T2 relaxivity is not part of MR imaging in routine clinical practice. To our knowledge, no previous study has quantitatively evaluated the contribution of MR imaging characteristics from a routine examination as predictors of hepatocellular carcinoma in patients with cirrhosis.
Unlike the objectives of previous studies, the purposes of this study were to identify predictors of hepatic malignancy and develop a predictive model for diagnosing hepatic malignancy in patients with known cirrhosis. We therefore included in this study only patients with known cirrhosis and results from a prior biopsy of suspicious hepatic nodules; consequently, the underlying prevalence of malignancy for our study population increased.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Fifty-eight patients (14 women and 44 men; mean age, 55.4 years; age range, 42–84 years) with known cirrhosis underwent 60 MR examinations. We included a total of 61 index lesions with pathologic diagnosis of a core needle biopsy sample. One patient had two index lesions. Two other patients had the same lesion biopsied twice at 6-month intervals, and these lesions were treated as independent lesions.
Biopsy Technique
The biopsies were performed under sonographic or CT guidance. Samples were
obtained using a semiautomated 18- or 20-gauge needle. Specimens were fixed in
formalin and sent to the pathology department for histologic analysis. The
results of the original report were reviewed for diagnosis. All specimens were
considered diagnostic.
MR Imaging Techniques
All MR studies were performed on a 1.5-T Signa scanner (General Electric
Medical Systems, Milwaukee, WI) before biopsy. The following sequences were
performed in all patients: an axial T1-weighted sequence using either a
spin-echo technique (TR/TE, <500/12) or spoiled gradientrecalled echo in-
and out-of-phase techniques (TR/TE range: out-of-phase, <180/1.8–2.4;
in-phase, <180/4.2–4.8); an axial T2-weighted fast spin-echo sequence
with respiratory triggering (TR/TE, <5000/90); and a dynamic
contrast-enhanced two-dimensional or three-dimensional T1-weighted spoiled
gradient-echo sequence. The parameters for the two-dimensional spoiled
gradient-echo sequence with fat suppression were a TR of less than 200 msec, a
TE of 1.2 msec, and a flip angle of 70°. The parameters for the
three-dimensional spoiled gradient-echo sequence with spectral fat suppression
were a TR of less than 6 msec, a TE of 1.2 msec, and a flip angle of
12°.
Dynamic MR imaging was performed before IV injection of 20 mL of gadolinium (Magnevist [gadopentetate dimeglumine], Berlex Laboratories, Wayne, NJ; or Omniscan [gadodiamide], Nycomed Amersham, Oslo, Norway) and subsequently during the arterial dominant, portal venous, and delayed phases. For the early studies, dynamic imaging was performed using a two-dimensional T1-weighted spoiled gradient-echo sequence, and the arterial phase images were acquired after a fixed delay of 15 sec from the start of contrast injection. The later dynamic studies were performed using a three-dimensional T1-weighted spoiled gradient-echo sequence, and the arterial phase images were timed using automated contrast-bolus detection software (SmartPrep; General Electric Medical Systems).
MR Image Analysis
Index lesions were identified by correlating the findings from CT or
sonography used for biopsy guidance with those from MR imaging. Two
experienced radiologists who were unaware of the patient's history, the
clinical MR interpretation, and the biopsy results independently determined
the signal intensity of all biopsy-proven index lesions on T2-weighted
contrast-enhanced arterial, venous, and delayed phase images. The signal
intensity was graded using a scale composed of categories; the index lesion
was graded as hypointense, isointense, or hyperintense to the liver on each of
the imaging sequences performed. Disagreement between observers (1.8%) was
resolved by consensus review. In addition, the size of the index lesion and
the total number of suspicious lesions visualized were recorded. Lesions
varied in size from 0.5 to 16.7 cm (mean, 4.4 cm). The mean number of lesions
was 2.6. All images were reviewed on a commercially available workstation
(Advantage Windows Workstation; General Electric Medical Systems).
Statistical Analysis
Potential predictors of hepatocellular carcinoma in the setting of
cirrhosis that we considered included variables identified by other
researchers [17,
18,
19,
20] and other variables
thought to be clinically relevant. The variables studied for association with
hepatic malignancy in the setting of cirrhosis included the following: patient
age at the time of the MR imaging examination, patient sex,
-fetoprotein level before the MR imaging examination, presence or
absence of hepatitis C viral genotype, size of the index lesion, number of
suspicious lesions, and signal intensity of the index lesion on each of the
imaging sequences described earlier.
Because of high skewedness, the data for -fetoprotein level were
log-transformed (base 10) for our analysis. The number of lesions was also
skewed to the right because a few patients had a large number of lesions.
Thus, to minimize the effect of the data from a few patients with a large
number of lesions unduly influencing our prediction model, we considered each
patient who had more than five liver lesions (n = 12 patients) to
have six lesions. The average -fetoprotein level before the MR imaging
examination was 379 ng/mL (range, 1.9–13,746 ng/mL). Hepatitis C viral
genotype was detected in 45 (78%) of 58 patients.
Each variable that was a potential predictor for hepatic malignancy was screened first for its relationship to hepatic malignancy status. The relationships between malignancy status and each predictor variable were evaluated. The chi-square test was used when the predictors were categoric variables, such as lesion imaging characteristics, and the two-group t test was used when the predictors were continuous variables, such as age. A logistic regression model was constructed to predict a malignant lesion in which the potential predictors included the value of the sequence-specific MR imaging characteristics and the value of each of the clinical and demographic factors that had been found to be significantly associated with malignancy status from the initial screening of the potential predictors. The final logistic regression model included variables found to be significantly associated with malignant lesions as well as variables that were believed to be clinically relevant predictors. The adjusted odds ratio and its confidence interval were obtained from the final model as a measure of the association between the predictor and malignancy status. The discriminatory performance of the model was assessed by the receiver operating characteristic (ROC) curve analysis. The area under the ROC curve was calculated for the final model. Subsequently, the probability of malignancy was estimated using the final logistic regression model on the basis of combinations of imaging, clinical, and demographic data [21].
We did not validate the final model using a method such as the split sample method, for which the data set would be divided into a model development data set and a validation data set, because our sample was not large enough. Instead, we used the bootstrap method to obtain more reliable estimates of the standard errors of our estimated odds ratios and their 95% confidence intervals [22, 23]; for this test, we drew 500 bootstrap samples of 50 lesions each with replacement. Bootstrapping is the method by which we draw a repeated number of random samples with replacement from our observed sample to obtain a more reliable estimate of the standard error of the estimated odds ratio of the predictor variable. The precision of an estimated odds ratio is dependent on the sampling distribution of the odds ratio obtained from the sample drawn from the underlying population (in our analysis, suspicious liver masses in patients with cirrhosis). Multiple random samples of the underlying population yield a more precise estimate of the distribution of the odds ratio than a single sample. Because we had only a single sample of 61 liver lesions, we could simulate the creation of multiple random samples using our single sample as a proxy for the population. Multiple samples of 50 lesions each were randomly drawn with replacement from our observed sample of 61 liver lesions, and the odds ratio was calculated and tallied for each bootstrap sample, which in turn provided an estimate of the standard errors of the estimated odds ratio more reliable than would be provided if the bootstrap method had not been applied to the data set.
All statistical analyses were performed using software (Stata version 6.0; Stata, College Park, TX), and statistical significance was set at a p value of 0.05.
Classification of Malignancy
At our institution, high-grade dysplastic nodules and frank hepatocellular
carcinomas are similarly aggressively managed. Therefore, we considered
dysplastic nodules to be malignant lesions, and these nodules were grouped
with hepatocellular carcinomas and metastases a priori. Of the 61 lesions, 39
(64%) were malignant (37 hepatocellular carcinomas and two metastatic
adenocarcinomas), three were high-grade dysplastic nodules (small cell
dysplasia), and 19 represented nonmalignant lesions (regenerative
nodules).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-fetoprotein level, presence or absence of the hepatitis C viral
genotype, the size of the lesion, the number of lesions, patient age and sex,
and sequence-specific imaging characteristics were considered as potential
predictors of hepatic malignancy. Initially, the sequence-specific imaging
characteristics were considered categoric variables; for each sequence, two
dichotomous variables were created for lesion hyperintensity and for
hypointensity relative to the liver. Thus, the reference group for each
sequence was the isointense (to liver) category. After an exploratory analysis
of the bivariate relationship between malignancy status and each of the
sequence-specific imaging characteristic variables, the imaging
characteristics were regrouped as follows. We collapsed hypointensity and
isointensity during the arterial phase into a single category representing no
arterial enhancement; hyperintensity represented arterial enhancement. We
similarly collapsed hyperintensity and isointensity on venous phase images
into a single category, representing no venous washout; hypointensity
indicated venous washout. For T2-weighted imaging, hypointensity and
isointensity were also collapsed into a single category.
The variable screening analysis revealed only venous washout to be
statistically significant (p = 0.01). The number of lesions was
marginally significant (p = 0.10). Arterial enhancement (p =
0.13), T2 hyperintensity (p = 0.65), -fetoprotein level
(p = 0.19), hepatitis C viral genotype (p = 0.61), patient
age (p = 0.21), patient sex (p = 0.27), and lesion size
(p = 0.95) were not significant.
Final Predictive Model
To preserve parsimony of the regression model, we included only the
variables that were initially significant (p < 0.05) or marginally
significant (p < 0.10) and specific imaging characteristics of the
index lesion that were used for clinical decision-making, such as
-fetoprotein level, arterial enhancement, and lesion size
[17]. The final logistic
regression model for malignant lesions in the setting of cirrhosis included
dummy variables for arterial enhancement, venous washout, -fetoprotein
level, the size of the index lesion, and the number of lesions.
The final model had an area under the ROC curve of 0.82
(Fig. 1), indicating very good
discrimination. The model attained a sensitivity of 88% and a specificity of
53%. In the final model, venous washout, the number of lesions, and the
-fetoprotein level were significant predictors (p < 0.04).
The adjusted odds ratio for -fetoprotein level (in log base 10) was
3.24 (Table 1)—that is,
for every 10-fold increase in -fetoprotein level, the odds of
malignancy more than triples. The odds ratio for venous washout was 9.24,
which indicates that the presence of venous washout increases the risk of
malignancy almost 10-fold. The odds ratio for the number of lesions was
1.54—that is, each additional lesion increases the risk of malignancy
1.54 times.
|
|
The model predicts that for patients with an index lesion that is 4.4 cm
and three lesions (average number of lesions per patient in this sample)
having both arterial enhancement and venous or delayed phase washout (Fig.
2A,
2B) increases the probability
of malignancy to at least 65% regardless of the -fetoprotein level
(Fig. 3). Furthermore, an
-fetoprotein value of more than 20 ng/mL increases the probability of
malignancy to more than 85%. For a patient with the same characteristics but
with imaging findings that do not show arterial enhancement and venous
washout, the probability of malignancy decreases to less than 35% as long as
the -fetoprotein value remains less than 160 ng/mL. Furthermore, if the
-fetoprotein level is within the normal range (<20 ng/mL), the
probability of malignancy is less than 20%.
|
|
|
The two enhancement patterns described represent the extreme cases, for
which the contribution of -fetoprotein level to the estimation of
malignancy is small. In the average patient with imaging findings showing
either arterial enhancement (Fig.
4A,
4B) or venous washout—but
not both—the contribution of the -fetoprotein level to the
estimation of malignancy is substantial
(Fig. 3).
|
|
The effect of lesion multiplicity with both arterial enhancement and venous
washout shows that as the number of liver lesions increases, the probability
of malignancy increases as well (Fig.
5). The effect of the number of lesions is particularly pronounced
at normal -fetoprotein levels (defined as < 20 ng/mL). A solitary
lesion with arterial enhancement and venous washout has a probability of
malignancy of between 35% and 98%, depending on the -fetoprotein level.
If five or more lesions are present, the probability of malignancy rises to a
minimum of approximately 80% regardless of the -fetoprotein level.
|
The latter clinical scenario of a large number of lesions in the presence
of a normal -fetoprotein value was reflected in our clinical
population. Of the 12 patients with five or more lesions, eight had a normal
-fetoprotein level (<20 ng/mL). Of these eight individuals, six
(75%) had malignant lesions. Of the remaining four patients without a normal
-fetoprotein level, all four (100%) had malignant lesions.
Of the possible predictors of malignancy studied, only venous washout remains significant at the 95% confidence level—that is, the 95% confidence intervals around the estimates of the odds ratios did not include 1 after 500 bootstrap samples were drawn.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Underlying hepatic nodules in the cirrhotic liver and the extensive overlap
of the imaging characteristics for benign and malignant nodules limit the
sensitivity and specificity of many imaging modalities
[3,
4,
10,
11,
12]. A retrospective review of
269 patients with cirrhosis who underwent liver transplantation was undertaken
to determine the best screening algorithm for detecting occult hepatocellular
carcinoma in the cirrhotic liver
[28]. In this population, the
evaluation before transplantation consisted of focused hepatic sonography,
gadolinium-enhanced MR imaging, and serum -fetoprotein testing. Results
indicated that MR imaging was more sensitive than sonography (68% vs 43%,
respectively) for detecting hepatocellular carcinoma, although both were
extremely specific (99%). A greater percentage of patients with malignant
lesions had an elevated -fetoprotein level (
20 ng/mL) compared with
those without a malignant lesion. Although Gogel et al.
[28] concluded from these data
that screening in patients with cirrhosis should include serum
-fetoprotein monitoring and gadolinium-enhanced MR imaging at regular
intervals, these researchers found that neither test proved sufficiently
sensitive independently.
Another group of researchers verified the limited sensitivity and
specificity of MR imaging in detecting small hepatocellular carcinomas and
small dysplastic nodules before transplantation
[29]. This study and another
[28] independently considered
imaging characteristics as predictors for malignancy without incorporating the
clinical and demographic characteristics of the patients. Our analysis reveals
that controlling for the presence of cirrhosis, patient age and sex, and
hepatitis C viral genotype does not add significant predictive power for
determining lesion malignancy. However, the final predictive model using
-fetoprotein level testing and a variety of imaging characteristics of
the index lesion have a greater sensitivity (88%) than was previously reported
for MR imaging, but at lesser specificity (53%).
Ecologic studies examining the likelihood of malignancy in hepatic masses detected on MR imaging revealed that if a patient does not have a history of malignancy or underlying hepatic disease, the identified hepatic mass is unlikely to be malignant [15, 16]. Tello et al. [16] combined patient history and demographic information with T2 relaxivity to develop a prediction rule for characterizing hepatic lesions detected on MR imaging. In this population of patients with and without a history of underlying nonhepatic malignancy, a prediction rule using T2 relaxivity, history of malignancy, and gadolinium enhancement pattern (hemangiomalike or not) attained an area under the ROC curve of 0.95. However, Tello et al. incorporated only T2 relaxivity, an uncommon measurement of T2 signal, and a limited manifestation of enhancement patterns in their regression model. Moreover, the range of lesions included a large number of hemangiomas and cysts that, in general, are less challenging to diagnose. Other groups of researchers reported that the specificity of MR imaging for differentiating cysts and hemangiomas from hepatocellular carcinomas and metastases is 100% [9, 30].
Individuals with cirrhosis are inherently different from the populations in the studies we have discussed. Selecting patients with cirrhosis increased the likelihood of hepatocellular carcinoma in our study population and also limited the prognostic contribution of clinical history to lesion characterization. The use of MR imaging characteristics that can be obtained from a nonstandard examination limits the generalizability of the previously developed prediction model across different clinical practices. Therefore, we sought to develop a prediction model based on a routine MR imaging examination of the liver combined with available clinical and demographic data. We excluded hepatic cysts and cavernous hemangiomas because these lesions are usually detected and characterized with sufficient confidence to avoid needle biopsy.
Our analysis focused on differentiating benign solid masses from malignant
solid masses in the cirrhotic liver. The incorporation of imaging
characteristics, such as arterial enhancement and venous washout, and
-fetoprotein level, patient age, lesion size, and number of lesions
yielded a prediction model with an area under the ROC curve of 0.82,
indicating very good discriminatory ability. In addition, interesting trends
emerged from the data that are not captured by sensitivity and specificity
alone and that may increase the usefulness of our analysis.
The combination of arterial phase enhancement and venous phase washout was
extremely predictive of malignancy, even at normal serum -fetoprotein
levels. The combination of a lack of arterial phase enhancement and a lack of
venous phase washout markedly reduced the probability of malignancy regardless
of the serum -fetoprotein level. On the other hand, we found that in
the presence of either arterial phase enhancement or venous phase
washout—but not both—the -fetoprotein level greatly affects
the probability of malignancy. In addition, our results indicate that
increasing the number of lesions increases the probability of malignancy
independent of enhancement characteristics and can double the probability of
malignancy at low to normal levels of -fetoprotein. We found that
lesion size was not predictive of malignancy, after controlling for dynamic
enhancement characteristics and -fetoprotein level.
Current screening recommendations for individuals with cirrhosis include
serial sonography and -fetoprotein level testing every 6 months, with
additional cross-sectional imaging on CT or MR imaging when a suspicious
nodule is detected or the -fetoprotein level increases
[17]. Bruix et al.
[17] have suggested the
following imaging diagnostic criteria for hepatocellular carcinoma, without
requiring additional biopsy: either images obtained from two coincident
techniques showing a focal lesion larger than 2 cm with arterial
hypervascularization or images obtained from one technique showing a focal
lesion larger than 2 cm with arterial hypervascularization and an
-fetoprotein level of greater than 400 ng/mL. However, our study
suggests that the most robust imaging characteristic that suggests
hepatocellular carcinoma is present in patients with cirrhosis is venous
washout, although arterial enhancement was a borderline significant predictor
of malignancy during the model development. Additionally, with the appropriate
pattern of enhancement, a single imaging test appears to enable one to predict
whether a malignancy is present, even in patients with a normal or only mildly
elevated -fetoprotein value.
We chose to focus on the imaging sequences that comprise a typical liver MR
imaging protocol performed routinely in general clinical practice to increase
the usefulness of the prediction model developed. Although the prediction
model in its present iteration may not be sufficiently sensitive to detect all
malignant lesions, this model may aid in estimating the probability of
malignancy and may provide guidance in clinical decision-making. In
particular, the model may suggest which lesions should be aggressively pursued
or expectantly managed, particularly if the visualized nodule is not easily
accessible for percutaneous biopsy. In addition, because MR imaging in
patients with cirrhosis often reveals multiple lesions, the model can
potentially facilitate the explicit assignment of probability estimates of
malignancy in MR imaging reports and could potentially be used to direct
subsequent biopsy toward the lesion with findings most suspicious for
malignancy. The potential predictive capability of the model may be useful,
particularly in individuals with chronic viral hepatitis who may present with
a transient elevation of the -fetoprotein value during inflammatory
disease flares and also with persistent elevations without hepatocellular
carcinoma [17]. In these
patients, the potential specificity of venous washout for malignancy may
prevent unnecessary biopsies from being performed.
An acknowledged limitation of our study rests on the method used for lesion verification. Although the diagnoses of all lesions considered were proven by needle biopsy findings, considering dysplastic nodules as benign or malignant is problematic. Frank hepatocellular carcinoma often coexists within predominantly dysplastic nodules; however, these carcinomatous foci may be missed as a result of sampling error. Nonetheless, at our institution, the preferred clinical approach to high-grade dysplastic nodules is aggressive management, often radiofrequency ablation. Therefore, we considered dysplastic nodules as malignant lesions.
The prevalence of disease in our retrospective study exceeded 58%, whereas
prospective studies with explant correlation had a disease prevalence of
8–15% [28,
29,
31]. Clearly, the elevated
prevalence of malignancy reflects a clinical management bias in selecting
patients for percutaneous biopsy. Patients with lesions that had equivocal
findings may have been clinically followed up with serum -fetoprotein
testing and repeated imaging rather than immediate biopsy. However, the
inclusion of lesions with verification on follow-up imaging would have
increased the number of benign solid masses, potentially limiting the
assignment of the correct pathologic diagnosis, particularly given the
spectrum of malignant potential in these masses. The rate of malignant
transformation can be rapid
[27] and independent of
concomitant alterations in imaging characteristics. Although MR imaging does
not detect most hepatocellular carcinomas in a cirrhotic liver before
transplantation [28,
29], whether this diminished
detection leads to worse outcomes is uncertain because these tumors are
usually small. Variability in contrast administration may also have limited
the detection of hepatocellular carcinoma during dynamic contrast imaging.
The number of patients available for inclusion in the generation of the final predictive model precluded direct validation of the model. However, we have simulated validation by applying bootstrap methodology. Using this method, we found that venous phase washout is the most robust predictor of hepatic malignancy. The ultimate predictive ability of the model remains to be independently validated.
Acknowledgments
In conclusion, the dynamic enhancement characteristics of hepatic lesions
in conjunction with serum -fetoprotein level and lesion multiplicity
can facilitate explicit estimation of the probability of malignancy in
individuals with underlying cirrhosis. If arterial enhancement and venous
washout are either both present or both absent, prediction of malignancy can
rely predominantly on the enhancement characteristics. If only one of these
characteristics is present, the -fetoprotein level has a greater
contribution toward estimating the probability of hepatic malignancy. Such
estimation can potentially aid clinical management.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  B. B. Frericks, C. Loddenkemper, A. Huppertz, S. Valdeig, A. Stroux, M. Seja, K.-J. Wolf, and T. Albrecht Qualitative and Quantitative Evaluation of Hepatocellular Carcinoma and Cirrhotic Liver Enhancement Using Gd-EOB-DTPA Am. J. Roentgenol., October 1, 2009; 193(4): 1053 - 1060. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Willatt, H. K. Hussain, S. Adusumilli, and J. A. Marrero MR Imaging of Hepatocellular Carcinoma in the Cirrhotic Liver: Challenges and Controversies Radiology, May 1, 2008; 247(2): 311 - 330. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Bixby, R. Tello, and E. Kuligowska Presence of a Yolk Sac on Transvaginal Sonography Is the Most Reliable Predictor of Single-Dose Methotrexate Treatment Failure in Ectopic Pregnancy J. Ultrasound Med., May 1, 2005; 24(5): 591 - 598. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. K. Hussain, I. Syed, H. V. Nghiem, T. D. Johnson, R. C. Carlos, W. J. Weadock, and I. R. Francis T2-weighted MR Imaging in the Assessment of Cirrhotic Liver Radiology, March 1, 2004; 230(3): 637 - 644. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
); it
rises to more than 80% if 
). In average patient with either
arterial enhancement and no venous washout (
) or no arterial
enhancement and venous washout (
), but not both, contribution of
; five lesions, 
) on probability
of hepatic malignancy. In presence of arterial phase enhancement and venous
phase washout, number of lesions present has marked effect on probability of
malignancy; however, as