| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, University Clinic Heidelberg, INF 110, Heidelberg
69120, Germany.
2 Department of Surgery, University Clinic Heidelberg, Heidelberg,
Germany.
3 Department of Radiology, German Cancer Research Center (DKFZ), Heidelburg,
Germany.
Received May 24, 2005;
accepted after revision August 22, 2005.
Address correspondence to J. Hansmann
(jochen_hansmann{at}med.uni-heidelberg.de).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Fifteen patients with liver metastases were enrolled before surgery according to the guidelines of our institutional review board and good clinical practice. After mesentericography, unenhanced scans (Volume Zoom) were performed initially. For CTAP, the contrast medium was injected through the superior mesenteric artery. Images were acquired in portal and delayed enhancement. The MR protocol (1.5 T; Magnetom Symphony) started with T1-weighted fast low-angle shot (FLASH) T2-weighted turbo spin echo (TSE). MRAP followed with gadolinium-enhanced dynamic T1-weighted 3D FLASH. Delayed-phase T1-weighted 2D FLASH axial images were performed 2 min after IV injection of the contrast medium. Qualitative and quantitative evaluation of CTAP and MRAP was performed by three blinded radiologists regarding the number of lesions and their size, localization, and differential diagnosis.
RESULTS. The overall sensitivity in detecting liver metastases was 97% with MRAP and 93% with CTAP (p > 0.05, not significant [n.s.]). The specificity was calculated to be 97% for MRAP and 82% for CTAP (p < 0.0001, statistically significant [s.s.]). The differences in sensitivity were more accentuated if only lesions 10 mm or smaller were considered (95% vs 88%, p > 0.05, n.s.), for which the respective specificities were 95% and 80% (p < 0.0014, s.s.). Improvements in sensitivity and specificity were associated with a higher lesion-to-liver contrast-to-noise ratio (59.4 ± 51.0 for MRAP vs 10.4 ± 7.3 for CTAP) and resulted in higher diagnostic confidence in the differential diagnosis of liver lesions (p < 0.001, s.s.) and better interobserver agreement (median kappa value, 0.88 vs 0.63).
CONCLUSION. MRAP proved to be a reliable method in the preoperative detection of small liver metastases in particular, with a higher sensitivity and specificity than CTAP. If organizational difficulties of MRAP can be overcome, MRAP could be considered instead of CTAP in the preoperative invasive evaluation of metastatic liver disease.
Keywords: arterial portography • CT • liver • MRI • oncologic imaging
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The presurgical evaluation of these patients underscores the role of diagnostic imaging because detection of the number, size, and segmental location of lesions directly influences considerations of further therapy. The latter includes resection, transplantation, and regional therapeutic procedures such as transcatheter arterial embolization and radiofrequency ablation.
Intraoperative sonography is the most sensitive technique currently available for evaluating the liver, with sensitivities and specificities greater than 95%. It is therefore accepted as the standard of reference [3, 4].
Because intraoperative sonography requires surgical exposure of the liver, it continues to be important for the small metastases that contraindicate hepatic resection to be detected preoperatively. For preoperative evaluation, CT during arterial portography (CTAP) is currently considered the most sensitive imaging technique for detecting liver metastases, with sensitivities reported to range from 86-97% [5-8] compared with sensitivities between 58% and 85% for contrast-enhanced helical CT or MDCT, respectively [8-10].
In contrast to its high sensitivity in detecting lesions, the specificity of CTAP for characterizing intrahepatic lesions is low. Tumor-mimicking benign perfusion abnormalities and benign lesions have led to a reported incidence of false-positive lesions between 9% and 63% in primary and secondary liver lesions, which decreases the specificity of the examination remarkably. False-positive findings related to benign causes have been even more frequently observed in patients with coexisting hepatic cirrhosis and portal hypertension [5, 11, 12]. Such false-positive information can cause diagnostic uncertainty and substantially misguide the management of malignant liver disease.
Although CTAP has been considered a routine procedure for patients with malignant hepatic disease to assess resectability and determination of a surgical approach in the past [13, 14], most institutions do not perform CTAP as a routine examination before hepatic surgery.
Because of recent advances in MRI techniques, including 3D gradient-echo (GRE) techniques and tissue-specific MR contrast agents—for example, superparamagnetic iron oxide (SPIO), mangafodipir trisodium (MnDPDP), or gadolinium ethoxybenzyl-diethylenetriamine-pentaacetic acid (Gd-EOB-DTPA)—MRI is challenging CTAP as the less invasive, more sensitive, and more specific technique for the preoperative diagnosis of focal liver lesions [15-19]. The use of MRI during arterial portography (MRAP) to detect focal liver lesions is the attempt to combine the advantages of arterioportal contrast medium administration with the high accuracy of MRI. The feasibility of this procedure has been shown previously [20-22].
Because CT and MRI technology have improved dramatically in the last few years, the purpose of our study was to compare CTAP and MRAP in detecting and characterizing malignant liver lesions when a new generation of scanner, optimized sequence technology, and state-of-the-art protocols are used. Sensitivity and specificity were determined on the basis of the results with the gold standard method (intraoperative sonography) in combination with histologic examination.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The patients enrolled in this study had a suspected metastatic liver disease and were all candidates for hepatic resection. The existence of liver metastasis was detected during a previous helical CT, MRI, or sonography examination, predominantly performed in a free-standing radiologic institution. All patients had tolerable renal and liver function, and their cardiovascular status was stable. They denied having had any previous allergic reactions to contrast media and had no general contraindications to MRI. CT and MRI with CTAP and MRAP were performed to determine the extent of metastatic liver disease before surgery and to confirm resectability or alternatively to suggest interventional therapy or chemotherapy.
Between February 2001 and February 2003, 17 consecutive patients with suspected metastatic malignant hepatic tumors referred to our hospital for preoperative workup underwent CT and MRI with CTAP and MRAP, which were performed in the same session. Before enrollment, written informed consent was obtained from each of the patients after the nature of the procedure was fully explained.
Two patients were excluded from the study because of diffuse metastatic liver disease, which was diagnosed during the preoperative workup. A cutoff of more than 20 lesions was selected. The remaining 15 patients (six women and nine men; ages, 41-73 years; mean age, 63 years) had a clinical history of colorectal carcinoma (seven patients); breast cancer (four patients); and malignant melanoma, pancreatic carcinoma, neuroendocrine carcinoma, and cholangiocellular carcinoma (one patient each). Patients with suspected primary liver tumors (e.g., hepatocellular carcinomas or cholangiocellular carcinomas) were not included because of the better tumor differentiation performing CT during hepatic angiography in such patients. Patients with hepatic cirrhosis were not examined either because hepatic resection in such patients is not a proper surgical option.
The following surgical interventions were performed in the study population: hemihepatectomy (five patients), segmentectomy (four patients), explorative laparotomy with intraoperative confirmation of preoperatively suspected peritoneal carcinomatosis (three patients), palliative tumor reduction with the resection of a peritoneal metastasis (one patient), and a large liver metastasis of a neuroendocrine carcinoma before chemotherapy (one patient). Surgery was not performed on one patient initially suspected of having hepatic metastases because MRI revealed three benign lesions with features typical of hemangiomas. This patient was followed for a period of 22 months using sonography and MRI. The lesions showed no change in size or morphology over time and were therefore judged to be hemangiomas without further histopathologic evaluation.
Imaging Technique
CT—For CTAP and MRAP, the femoral superficial artery was
punctured in local anesthesia using the Seldinger technique via an adapted
fluoroscopic system and a 4- or 5-French angiographic Cobra 2 or sidewinder
catheter that was positioned in the proximal superior mesenteric artery.
Angiography was performed to assess the portal supply of the liver, evaluate the patency of the portal vein, and estimate the time delay until the individual's portal liver phase depending on the patient's individual cardiovascular situation. The volume of contrast medium injected during angiography was minimized to prevent reduced contrast between the hepatic parenchyma and lesions on subsequent CT scans and never exceeded 60 mL of iopromide (concentration, 300 mg I/mL).
After placement of the catheter, CT was performed first. The same protocol was used for all patients. For CT, scanning was performed with an MDCT scanner (Somatom Plus 4 Volume Zoom, Siemens Medical Solutions) equipped with an adaptive array matrix and a gantry rotation time of 0.5 sec. Native scans were performed initially, covering the entire liver parenchyma. For CTAP, 70 mL of iopromide (concentration, 370 mg I/mL) was injected through the superior mesenteric artery at a rate of 4 mL/sec with a power injector (Medrad En-vision CT). Acquisition of images was timed to coincide with the portal phase of mesenteric angiography. Delayed contrast-enhanced images were acquired approximately 2 min after CTAP.
The images were obtained in a craniocaudal direction during a single breath-hold helical acquisition of 8-12 sec depending on liver size. Imaging parameters for the CT scan were beam collimation, 4 x 2.5 mm; effective section thickness, 3.0 mm; reconstruction interval, 1.5 mm; table speed, feed/rotation: 15 mm; field of view, to fit; 120 kV and 100 mA. The ensuing beam pitch, calculated by multiplying table speed by gantry rotation time and dividing the product by beam collimation, was 1.5 (30 mm sec x 0.5 sec) / (4 x 2.5 mm). Images were reconstructed with a standard body reconstruction algorithm (kernel B30, medium smooth) available on the CT scanner and routinely used for abdominal CT.
MRI—MRI was performed on a 1.5-T whole-body scanner (Magnetom Symphony, Siemens Medical Solutions) equipped with a high-performance gradient (Quantum) system (maximum gradient strength, 30 mT/m; slew rate, 125 T/msec). A combination of the standard body phased-array coil with spine array coils was used for signal reception.
T1-weighted 2D fast low-angle shot (FLASH) (TR/TE, 128/4.76; flip angle, 70°), T2-weighted turbo spin-echo (TSE) (TR/TE, 3,220/101; TSE factor, 29), heavily T2-weighted TSE (TR/TE, 4,050/265; TSE factor, 33), and T1-weighted 3D FLASH (TR/TE, 3.8/1.3; flip angle, 25°) images were obtained before administration of the contrast medium.
For MRAP, 0.1 mmol/kg of body weight gadopentetate dimeglumine was injected through the superior mesenteric artery at a rate of 2 mL/sec with a power injector (Medrad Spectris MR Injector). Again, the acquisition of MR images during portal angiography was timed to coincide with the portal phase of mesenteric angiography (T1-weighted 3D FLASH with TR/TE, 3.8/1.3; flip angle, 25°). T1-weighted 3D FLASH sequences were repeated 30 and 70 sec later. Delayed-phase axial images were performed 2 min after IV injection (T1-weighted 2D FLASH with TR/TE, 157/6; flip angle, 70°).
All pulse sequences were acquired during breath-holding. The matrix size was 135-210 x 256 and the field of view to fit. A total of 15-20 sections 2.7 mm thick were obtained during T1-weighted 3D FLASH: For all other sequences, the thickness was 6 mm. In each case the intersection gap was 10%. Administration of sedation was not necessary for any of the performed imaging techniques.
Image Analysis
For image analysis, all images of each technique were interpreted
prospectively and evaluated independently by three experienced observers. They
knew the patients were referred for assessment of suspected liver metastases
but were not provided with any other information about the patients. The
images from each technique were interpreted in separate sessions at 4-week
intervals and in a randomized sequence. For characterization of liver lesions,
all images of each CT or MRI examination were reviewed together using all the
sequences available. Examinations performed in free-standing radiologic
institutions were not considered in image evaluation because of heterogeneous
scan protocols and differing image quality. To minimize learning bias, the
patient names were changed. All images were evaluated on a 2K monitor, using a
PACS (GE Healthcare Integrated Imaging Solutions). Because the range of
contrast enhancement during CTAP and MRAP varied markedly, the optimal window
setting in each case was adjusted individually as needed. The image review was
conducted segment by segment with eight autonomous segments according to the
Couinaud classification system
[24].
Each observer recorded the location, table position, and size of the focal lesion (maximum of two perpendicular diameters following the guidelines of the National Cancer Institute) [25]. The observers were asked to specify the character (benign, malignant, or pseudolesion) and diagnostic confidence for each lesion in each patient on a scale of 1 to 5 (1 = very sure, 2 = sure, 3 = intermediate, 4 = uncertain, 5 = very uncertain). A well-delineated rounded or oval space-occupying lesion was characterized as a true lesion. A distinction between benign or malignant lesions was presumed on the basis of commonly accepted characteristics [26, 27]. Because hemangiomas and liver cysts were the only benign liver lesions in our patient population, in brief, a perfusion defect at CTAP was seen as a metastatic liver tumor when signs of a hemangioma (peripheral globular enhancement during delayed enhancement), signs of a liver cyst (hypodensity without contrast enhancement), or typical signs of a pseudolesion were missing. MRI offered more morphologic criteria because of T1- and T2-weighted images. In addition to the perfusion pattern during dynamic MRI, the presence of markedly hyperintense lesions on T2-weighted images enhanced the ability to differentiate hemangiomas (lesions with peripheral enhancement during delayed enhancement) and liver cysts (no enhancement) from malignant tumors (moderate hyperintensity on T2-weighted images with perfusion defect). The criterion for a pseudolesion was either a perfusion defect that was wedged, flat, or irregular in shape or a typical location such as the fossa of the gallbladder, the space anterior to the porta hepatis, or the intersegmental fissure. Liver lesions correctly characterized as metastases with either technique were designated true-positive lesions. We considered liver lesions that were correctly characterized with either imaging technique as benign lesions (e.g., cysts and hemangiomas) or pseudolesions to be true-negative lesions, and malignant lesions misclassified as benign lesions or pseudolesions to be false-negative lesions.
For quantitative image analysis, the lesion-to-liver contrast-to-noise
ratio (CNR) was calculated as described earlier
[28,
29]. In brief, the
lesion-to-liver CNR was determined as follows:
 |
where Sm and Sl are the signal intensities of the lesion and liver parenchyma, respectively. The SD of back-ground noise (SDb) was measured in the phase-encoding direction outside the anterior abdominal wall. To avoid partial volume effects, liver lesions (29/73) with a diameter of 10 mm or larger were studied. For the liver lesions, a circular region of interest (ROI) was drawn to encompass as much of the lesion as possible. ROIs were at least 35 mm2.
In a final step, the reviewers had to evaluate the homogeneity of the liver parenchyma after contrast medium administration on a three-level scale: good = absence of segmental or nonsegmental perfusion defects, moderate = small perfusion inhomogeneities without considerable limitations in the diagnostic value, and poor = distinct perfusion inhomogeneities with considerable limitations in diagnostic accuracy.
The final determination of the number and the diagnosis of each focal hepatic lesion were correlated in a separate consensus review session by two experienced observers who did not participate in the initial interpretation. Findings on CTAP and MRI were correlated with surgical results on a lesion-to-lesion basis, specifically with the findings from intraoperative sonography and histopathologic specimens. Intraoperative sonography (Sonoline Sienna, Siemens Medical Solutions) was routinely performed by either a radiologist or a surgeon in all the patients who underwent surgery using a 7.5-MHz probe without the application of IV contrast medium. The differentiation between a benign lesion and a metastasis was done by morphologic criteria and palpation. In terms of morphology, a hyperechoic lesion led rather to the diagnosis of a hemangioma, an anechoic lesion with dorsal signal enhancement to the diagnosis of a liver cyst, and hypoechoic lesions with a halo were interpreted as metastases. It was not possible for every single lesion to be examined histopathologically because explorative laparotomy or palliative tumor reduction was performed in five patients. Such tumors were considered to have the same disease as those with identical imaging findings that were examined histologically.
In one patient, histopathologic evidence of disease was not obtained because MRI revealed three benign lesions with features typical for hemangiomas. The benign nature of these lesions was confirmed during a 22-month follow-up.
Statistical Analysis
To evaluate the relative preoperative diagnostic values of CTAP and MRAP,
the sensitivity in detecting malignant liver lesions and the specificity in
differentiating between malignant and benign lesions were calculated for each
reviewer and for the total data for both techniques. Differences in the
sensitivity and specificity between both imaging techniques were compared
using the chi-square test for each reviewer and for the composite data. Those
p values smaller than 0.05 were considered statistically
significant.
Weighted kappa statistics for interobserver agreement were used to assess the interobserver variability in terms of the number of malignant lesions detected and the malignant focal hepatic lesions differentiated from benign ones. A kappa value of 0.2 or less indicated positive but poor agreement; a value of 0.21-0.40, fair agreement; a value of 0.41-0.60, moderate agreement; a value of 0.61-0.80, good agreement; and a value of 0.81-1.0, very good agreement.
The diagnostic confidence for the interpretation of each lesion and the homogeneity of liver parenchyma after contrast medium enhancement were analyzed using the Pearson's chi-square test.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Surgically Confirmed Lesions
In total, 73 lesions (23 benign, 50 malignant) were evaluated in 15
patients. Of these, 70 were confirmed by surgery on the basis of
histopathologic examination or intraoperative sonography (or both). Three
lesions in one patient were followed up for a period of 22 months because of a
consensus diagnosis of hemangiomas. The maximum number of lesions in one
patient was 17. Of the 50 malignant lesions, 28 (56%) were 10 mm or smaller,
as were 20 (87%) of the 23 benign lesions.
Sensitivities and Specificities
Table 1 summarizes the
results of the three observers' evaluations of CTAP and MRAP.
Table 2 presents the
sensitivities and specificities for each observer and each technique and the
composite data for the detection of malignant liver lesions. The sensitivity
for detecting malignant liver lesions using MRAP was higher than that with
CTAP. For CTAP, the sensitivity was between 92% and 94% (93.3% for the
composite data); for MRAP, it was between 96% and 98% (97.3% for the composite
data). This difference in sensitivity failed to reach statistical
significance. The specificity for detecting malignant liver lesions with CTAP
was between 81% and 82.5% compared with 96.8% for all the observers with MRAP.
This difference was statistically significant for the composite data and for
each observer.
|
|
When lesions of 10 mm or smaller are considered, the sensitivity with CTAP was between 85.7% and 89.3% (88.1% for the composite data) and with MRAP was between 92.9% and 96.4% (95.2% for the composite data). This difference in sensitivity also failed to reach statistical significance. The specificity for detecting small malignant lesions was between 72.4% and 86.2% (80.4% for the composite data) using CTAP and between 93.1% and 96.6% (95.2% for the composite data) using MRAP. These results were statistically significant for the composite data and for two of the three observers.
The results for the different radiologists with regard to the detection of malignant liver lesions using CTAP were in good agreement. Even the results with MRAP were in very good agreement. The kappa values among the three radiologists were between 0.63 and 0.75 (median, 0.63) for CTAP and between 0.88 and 0.89 (median, 0.88) for MRAP (Table 3).
|
False-Positive and False-Negative Findings
Table 4 lists the
false-positive and false-negative findings of the three reviewers. With CTAP,
19 benign and 15 pseudolesions were misinterpreted as being malignant,
compared with three false-positive results for each lesion type using
MRAP.
|
Quantitative Image Analysis
The quantitative assessments of the mean lesion-to-liver CNRs for 29 liver
lesions (3 10 mm) obtained with CTAP and MRAP were 10.4 ±
7.3 and 59.4 ± 51.0, respectively (Figs.
1A and
1B).
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Cross-sectional imaging techniques play an important role in the selection of potential candidates for hepatic resection among patients with liver metastases, but the question of the most appropriate technique remains controversial [30].
Because of its high sensitivity for detecting hepatic lesions, CTAP is one of the most reliable tools in the preoperative workup for liver metastases. The reported sensitivities for CTAP range from 81% to 97%. In contrast to its high sensitivity for detecting lesions, however, the specificity of CTAP for characterizing lesions is between 37% and 91% [5, 11, 12, 28]. The reason for this lies in the nature of the technique: The rationale behind the use of CTAP is for contrast material to be delivered directly to the liver through the portal vein before it can return to the hepatic artery from the systemic circulation to optimize the detection of the hepatic metastases that do not have portal vein flow and appear as hypodense nodules. The difference in lesion-to-liver attenuation is thus maximized on CTAP compared with CT scans obtained with IV administration of contrast medium. The main drawback of CTAP is that any lesion in the liver without portal flow (such as an adenoma, hemangioma, focal nodular hyperplasia, or small cyst) cannot be differentiated from metastases by CTAP alone [16].
Nontumorous perfusion abnormalities can also cause false-positive findings on CTAP. They appear around the gallbladder, adjacent to the falciform ligament, or in the quadrate lobe of the liver. Nontumorous perfusion abnormalities are the result of aberrant venous drainage or, in the hepatic hilum, to decreased portal inflow caused by nonportal supply through the parabiliary venous system. Tumors located centrally within the liver may compress or invade a portal branch, resulting in an area of hypoperfusion in the parenchyma distal to the mass. The incidence of hemodynamic alterations with a decrease in specificity and accuracy of tumor detection is even higher in patients with cirrhotic liver changes and portal hypertension [31].
The reasons mentioned may account for the low specificity in the characterization of malignant hepatic lesions with CTAP, which was also true for this patient population, for which the overall false-positive findings amounted to 18% because of misinterpretation of 19 benign lesions and 15 pseudolesions by the three radiologists. This corresponds to the 9-63% prevalence of false-positive findings that other authors have reported [5, 11, 12].
MRI is a noninvasive and feasible technique that is available for routine examinations or as an alternative to angiographically assisted CT.
Most studies that have directly compared CTAP and MRI, using either gadolinium or SPIO as the contrast agent, reported no significant differences in sensitivity. They have, however, reported a higher specificity of MRI than of CT in detecting hepatic lesions, especially in terms of the accuracy in discriminating between malignant lesions and pseudotumors such as perfusion defects [7, 11, 16, 32].
Some authors have therefore argued that angiographically assisted helical CT—because of its invasive nature and high false-positive rate—should be replaced by less invasive and more precise contrast-enhanced MRI techniques in the routine preoperative workup for detecting malignant liver lesions.
Tumor detectability can be optimized if contrast material is delivered directly into the liver through the portal vein, bypassing systemic circulation and leading to a high degree of hepatic parenchymal enhancement. This facilitates the detection of hepatic metastases that do not have a portal vein supply. The difference in lesion-to-liver attenuation is thus maximized compared with IV administration of contrast medium.
To combine the advantages of direct contrast medium administration into the portal vein with the high accuracy of MRI, MRI of the liver during portal perfusion (MRAP) can be performed. This technique is conceptually very similar to the CTAP technique. Both rely on a relatively selective enhancement of liver parenchyma, achieved by positioning a catheter in the mesenteric artery to enable intraarterial administration of contrast medium, to increase the contrast between focal lesions and parenchyma. The increased amount of contrast material selectively delivered to the liver parenchyma increases the sensitivities for the detection of hepatic tumors. In contrast, the dose of IV-administered contrast material in conventional CT or MRI is widely distributed throughout the body.
The feasibility of using MRAP to detect focal liver lesions and the improved lesion-to-liver contrast that is achieved were first shown by Pavone et al. [20] in 1991. The few subsequent studies that have directly compared MRAP and CTAP reported that MRAP exhibited advantages in terms of sensitivity and specificity in the evaluation of malignant liver lesions [22, 33, 34].
The use of a state-of-the-art imaging technique is critical for high diagnostic accuracy. Because the MRI and CT techniques used in previous MRAP studies are no longer state of the art and a direct comparison between MRAP and CTAP was often not reported [35, 36], we examined 15 patients with suspected malignant secondary liver lesions preoperatively. We compared the sensitivity and specificity of MRAP and MDCT arterial portography using a preoperative state-of-the-art staging protocol and up-to-date sequence technology. Essential aspects of this study were the use of an MDCT scanner for CTAP and a 3D breath-hold GRE technique with full liver coverage in one pass for MRAP. Both techniques provided optimal spatial resolution in conjunction with image acquisition in the hepatic portal dominant phase of enhancement, creating a maximum lesion-to-liver contrast. Optimal portal phase contrast was achieved by setting the optimal time point for administration of the contrast medium bolus on the basis of the time delay determined on angiography.
|
|
|
|
|
|
|
Despite the use of new imaging technology for CTAP and MRAP, the fact that MRAP provided better results than CTAP in terms of specificity is not astonishing and confirms the results reported in previous studies. The better results are because of the inherent high tissue contrast through T1- and T2-weighted MR images in combination with MRAP and dynamic contrast-enhanced examination because images of all sequences are usually reviewed together. Especially the use of the dynamic FLASH 3D sequence with three acquisitions makes it possible to differentiate among perfusion abnormalities, true benign lesions, and malignant lesions (Figs. 4A, 4B, 4C, 4D, 4E, 4F and 4G). The reason is that the injection of the highly concentrated contrast agent results in fewer perfusion inhomogeneities, thus offering more information than unenhanced and delayed contrast-enhanced CT images after CTAP.
In this study, this resulted in substantially fewer misinterpretations of benign lesions and pseudolesions, causing improved specificity. Furthermore, the characterization of liver lesions using MRAP resulted in a higher diagnostic confidence and a better agreement among the observers, which is reflected in higher kappa values.
The better sensitivity of MRAP compared with CTAP was surprising, however, because the spatial resolution of MRAP based on the reduced matrix is minor compared with MDCT arterial portography (135-210 x 256 vs 512 x 512). The fact that this result was not statistically significant was probably because of the small patient population.
In our opinion, the main reason for the higher sensitivity of MRAP than CTAP in this study is that with MRAP it is possible to inject the full systemic dose of contrast medium in the form of a small highly concentrated bolus of 10-15 mL in a shorter time period than in CTAP. In contrast, the bolus of contrast medium in CTAP is large and the injection time is long. As a result, the arterial phase usually overlaps with the portal venous phase. To prevent this and to reduce the number of artifacts from the highly concentrated contrast medium, the total amount of contrast medium has to be reduced. The consequence is that the concentration of contrast medium is insufficient for evaluation of the liver parenchyma in delayed enhancement (Figs. 4A, 4B, 4C, 4D, 4E, 4F and 4G).
The better use of contrast medium during MRAP had two results: MRAP is better able to obtain a homogeneous parenchymal enhancement of the liver, which is essential for improving lesion detection and avoiding false-positive or false-negative findings; and MRAP provides a better lesion-to-liver CNR than CTAP (59.4 vs 10.4; p < 0.001), which accounts for the higher sensitivity in the detection of liver lesions despite lower spatial resolution (Figs. 1A, 1B).
The higher lesion-to-liver contrast is also the result of the almost total absence of enhancement of liver lesions with contrast medium.
The administration of contrast medium required for catheter positioning in mesenteric angiography during CTAP decreases lesion-to-liver contrast because of the interstitial accumulation of contrast material in the tumor. This effect is still considerable even if the amount of contrast medium injected is minimized as much as possible.
For MRAP, the gadopentetate dimeglumine solution is injected only once, and the liver is imaged during the first portal circulation of the contrast-enhanced blood, during which time the lesion has not been reached by the paramagnetic agent. To our knowledge, iodinated agents show no effect on MR signal intensity. Therefore, a static high degree of lesion-to-liver contrast is achieved with MRAP regardless of the amount of contrast agent injected in angiography.
Our results agree with those of others who found that lesion-to-liver contrast was better and the number of lesions detected higher with MRAP than with conventional MRI [20, 33, 34].
There are some limitations in this study. The first concerns the relatively small number of patients included. Our results thus need to be confirmed by further studies evaluating more patients. Furthermore, pathologic examination of all lesions was not possible. Although, ideally, histologic proof is obtained from all lesions by surgery, pathologic examination of the total liver parenchyma as a standard of reference is only feasible with patients undergoing liver transplantation. For regions of the liver that were not resected, we used a combination of surgical exploration and intraoperative sonography to determine the character of the lesion examined. This combination ensured the highest sensitivity and specificity—more than 95% for each—for detecting hepatic metastases [3, 37, 38].
Further comparative studies with more patients are required before a general proposal for the clinical application of MRAP can be made. The status of this technique in the range of imaging options can be established only with a systemic comparison of MRI with liver-specific contrast media. An appropriate algorithm might begin with the less expensive noninvasive examination of CT or MRI. These methods can detect most intra- and extrahepatic lesions and usually allow the distinction between benign and malignant lesions. Patients referred for hepatic resection could then be studied with MRAP to detect additional small lesions, which might lead to alteration of the planned therapy. Because MRAP is more effective in the detection and evaluation of small liver lesions (i.e., shows both a higher sensitivity and specificity than CTAP), MRAP should be performed instead of CTAP. Limitations and difficulties in performing MRAP are similar to those encountered with CTAP. In addition to the general contraindications to MRI, this technique obviously requires that the MR unit be located near the angiography suite. Despite the complexity of such an invasive procedure compared with a CTAP examination, investments in time and material are moderate and result in an examination that is approximately 20 min longer.
Again, the cost-benefit issue, which was not the subject of our study, is important when comparing a noninvasive with an invasive MRI technique. And because economic pressure favors the use of less expensive strategies, invasive techniques such as MRAP will not be performed routinely but will be reserved for special indications. In this context, a preoperative determination of the number and location of hepatic metastases with a high sensitivity is critical for selecting candidates for hepatic resection and therefore crucial to avoid unnecessary surgical exploration in patients with unresectable tumors. In these cases, MRAP can reduce running costs despite its relative expense.
In conclusion, our results show that MRAP is a safe and reliable imaging technique in clinical practice with a higher sensitivity and specificity than CTAP, especially in the detection and characterization of small liver lesions. It may develop as an alternative to explorative surgery with intraoperative sonography. Future assessment of MRAP requires more comparative studies with MRI with liver-specific contrast media. MRAP could be performed in patients with metastatic disease in which noninvasive routine screening such as CT or MRI has revealed only a small number of metastases that would make surgery a feasible option. Based on the data of this study, if organizational difficulties of MRAP can be overcome, MRAP should replace CTAP in the preoperative invasive evaluation of metastatic liver disease.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
