Preoperative evaluation of patients awaiting liver transplantation: Comparison of multiphasic contrast-enhanced 3D magnetic resonance to helical computed tomography examinations

Authors

  • William B. Eubank MD, MPH,

    Corresponding author
    1. Department of Radiology, University of Washington School of Medicine, Seattle, Washington
    2. Department of Radiology (S-113-RAD), Puget Sound VA Health Care System, Seattle, Washington
    • Department of Radiology (S-113-RAD), Puget Sound VA Health Care System, 1660 South Columbian Way, Seattle, WA 98108-1597
    Search for more papers by this author
  • Kimberly L. Wherry MD,

    1. Department of Radiology, University of Washington School of Medicine, Seattle, Washington
    Search for more papers by this author
  • Jeff H. Maki MD, PhD,

    1. Department of Radiology, University of Washington School of Medicine, Seattle, Washington
    2. Department of Radiology (S-113-RAD), Puget Sound VA Health Care System, Seattle, Washington
    Search for more papers by this author
  • Hakan Sahin MD,

    1. Department of Radiology, University of Washington School of Medicine, Seattle, Washington
    Search for more papers by this author
  • Charles P. Funkhouser PhD,

    1. Department of Mathematics, California State University at Fullerton, Fullteron, California
    Search for more papers by this author
  • Udo P. Schmiedl MD, PhD

    1. Department of Radiology, University of Washington School of Medicine, Seattle, Washington
    Search for more papers by this author

  • This article is a US Government work and, as such, is in the public domain in the United States of America.

Abstract

Purpose

To determine the feasibility of using a multiphasic magnetic resonance (MR) examination to evaluate the hepatic arterial anatomy and parenchyma in patients awaiting orthotopic liver transplantation (OLT).

Materials and Methods

Twenty consecutive patients awaiting OLT underwent multiphasic MR (using a T1-weighted 3D gadolinium-enhanced gradient-echo (GRE) sequence and two separate injections of contrast material) and computed tomography (CT) imaging; both imaging studies were performed within a 1-week period for each patient. Quantitative and qualitative assessment of the hepatic arterial system on MR data was performed. Two independent observers classified the hepatic arterial anatomy and evaluated the hepatic parenchyma from the MR data. The prospective CT interpretation was used as the gold standard.

Results

Overall qualitative rating of hepatic arterial system-to-background contrast on MR data was good to excellent (average pooled score of 2.00 ± 0.27), with no significant difference between the two observers after the first or second injections of contrast material. Classification of hepatic arterial anatomy by MR angiography (MRA) and CT angiography (CTA) was concordant in 85% (17/20) of patients and discordant in 15% (3/20) of patients. Focal parenchymal lesions were detected in 25% (5/20) of patients by MR and CT; however, two lesions in one patient with multiple lesions were detected only with MR.

Conclusion

Multiphasic T1-weighted 3D gadolinium-enhanced MR examination can provide comprehensive evaluation of the hepatic arterial anatomy and parenchyma in patients awaiting OLT. MR may offer an advantage over CT in the detection of focal parenchymal lesions. J. Magn. Reson. Imaging 2002;16:565–575. Published by Wiley-Liss, Inc.

PREOPERATIVE IMAGING GOALS in the evaluation of candidates for orthotopic liver transplantation (OLT) include the accurate depiction of hepatic arterial anatomy and the detection and characterization of parenchymal lesions (1). Preoperative imaging plays a key role in evaluating factors that contraindicate or potentially expedite surgical transplantation (2–4). Multiphasic helical CT has proven useful in achieving these comprehensive goals (4–10). Computed tomography angiography (CTA) for the evaluation of hepatic arterial anatomy has been validated against conventional angiography and has replaced conventional angiography at many institutions (6). Drawbacks of CT include the risks associated with iodinated contrast material and exposure to ionizing radiation. Because renal dysfunction often complicates the course of liver transplantation candidates (11, 12), iodinated contrast-associated nephrotoxicity is an important consideration in these patients.

Dynamic magnetic resonance imaging (MRI) of the liver after IV injection of gadolinium chelates has been proven a useful alternative to CT for the evaluation of hepatic parenchyma and vascular anatomy in patients with end-stage liver disease (13–23). Gadolinium chelates can be safely used in patients with impaired renal function (24, 25). Most investigators using dynamic MRI of the liver have focused on either the evaluation of the hepatic parenchyma, primarily with a 2D gradient-echo (GRE) sequence (13–18), or the hepatic vasculature using a breath-held contrast-enhanced 3D GRE sequence (19–23).

We hypothesized that a multiphasic MR examination could replace multiphasic CT, the conventional imaging modality used in the preoperative evaluation of candidates for OLT. To our knowledge, a prospective comparison between MR and CT in the preoperative evaluation of patients awaiting OLT has not been performed. The purpose of this study was to determine the feasibility of using a multiphasic MR examination to primarily evaluate the hepatic arterial anatomy and parenchyma in patients awaiting OLT and compare these findings to the conventional method of pretransplantation evaluation, namely, multiphasic CT examination.

MATERIALS AND METHODS

Patients

Patients awaiting liver transplantation were recruited for this prospective study. Inclusion criteria were 1) compatibility for MR and contrast-enhanced CT imaging (creatinine ≤ 1.5), and 2) the logistical ability to schedule CT and MR within a 2-week interval. Twenty consecutive patients (11 women and 9 men; average age, 51.2 years; age range, 41–72) meeting these criteria were enrolled in the study from February 1999 through March 2000. All patients signed a written consent form approved by the investigational review board of our institution. The underlying hepatic disease in these patients was hepatitis C (N = 15), cryptogenic cirrhosis (N = 3), alcoholic cirrhosis (N = 1), and progressive sclerosing cholangitis (N = 1). The patients had a moderate degree of liver compromise with an average Child-Turcotte-Pugh score of 7.8 (range, 6–13) (26). The evaluation for potential liver transplant candidacy begins early on in the patient's disease process at our institution, in most cases well before the patient is put on the transplantation list. Patients are followed with imaging at 6-month intervals until transplantation occurs or they are removed from the list. Thirteen patients had CT and MR performed on the same day, three had MR performed before CT (mean time interval ± SD, 1.3 days ± 0.6), and four had CT performed before MR (mean time interval ± SD, 5.5 days ± 1.7).

CT Imaging

All examinations were performed using a HiSpeed Advantage Scanner (General Electric Medical Systems, Milwaukee, WI). Unenhanced imaging of the liver was first performed using a 7-mm collimation at a pitch of 1:1 within a single breath hold (maximum, 30–35 seconds). We used the technique for hepatic CTA and imaging during the portal venous-dominant phase that has been previously described (6). Briefly, a timing bolus examination (20 mL of contrast material) was performed to determine the optimal timing for arterial-dominant phase imaging. Enhanced imaging of the liver was performed after the IV injection of 180 mL of contrast material at 5 mL/second using a 3-mm collimation and pitch between 1:1 and 1:7 for arterial-dominant phase imaging and a 7-mm collimation and pitch of 1:1 for portal venous-dominant phase imaging (60 seconds after the start of contrast injection). Postprocessing of the CTA data was performed using the maximum intensity projection (MIP) and shaded surface display (SSD) reconstruction algorithms on a GE Advantage workstation (General Electric Medical Systems, Milwaukee, WI). The source images (3-mm collimated axial slices reconstructed at 1-mm intervals) and the 3D reconstructions from the CTA data were used primarily to classify the hepatic arterial anatomy and evaluate the hepatic parenchyma for hypervascular lesions. The data acquired in the portal venous-dominant phase was used to evaluate hepatic morphology and parenchyma for focal lesions, estimate liver volume and degree of cirrhosis, and assess the sequellae of portal venous hypertension (asites, portosystemic venous collaterals, venous thrombosis, and splenomegaly). Two attending radiologists with experience in abdominal imaging, who were on clinical service on the day the CT examinations were performed, interpreted all data prospectively.

MRI

All MRI was performed with a 1.5-T MR system (Signa Horizon Echospeed, version 6.6; General Electric Medical Systems, Milwaukee, WI) with a peak gradient strength of 23 mT/m and a maximal slew rate of 20T/m/second. Imaging was performed with a torso phased-array receive coil.

Unenhanced imaging in the axial plane consisted of the following: 1) a respiratory-triggered, fat-saturated T2-weighted fast spin-echo (FSE) sequence (TR/effective TE, 5217–9230/105; echo train length, 8; matrix, 256 × 192, slice thickness, 6–7 mm; gap, 3 mm; averages (NEX), 2) and 2) a T1-weighted 3D breath-hold GRE sequence (TR/TE, 7.7/3, with minimum full echo; flip angle, 20°; rectangular field of view (FOV), 340–480 × 200–380 mm; in-plane matrix, 256 × 192 (interpolated by zero filling); true slice thickness in the z direction, 7–8 mm (interpolated to 3.5–4 mm); bandwidth, 31.2 kHz; NEX, 1). In a single breath-hold duration (25–31 seconds), 20–36 partitions in the z direction with a true slice thickness of 7–8 mm (interpolated to 3.5- to 4.0-mm-thick overlapping slices) were acquired to ensure complete coverage of the liver (slab thickness of 140–288 mm (mean, 180 mm)). Interpolated voxel size was under 1.9 × 2.0 × 4.0 mm.

For enhanced imaging, two multiphasic T1-weighted 3D breath-hold GRE examinations using separate injections of gadolinium chelate were performed on each patient. The first was acquired in the axial plane using the same parameters as the unenhanced 3D GRE sequence, and the second (beginning 7–8 minutes after the first injection of gadolinium chelate) was acquired in the coronal plane to cover the hepatic vascular anatomy. Parameters for the coronal 3D GRE examination were selected to optimize depiction of arterial anatomy (TR/fractional TE, 6.1/1.3; flip angle, 20°; rectangular FOV, 440–480 × 260–400 mm; in-plane matrix, 256 × 192; bandwidth, 31.3 kHz; NEX, 1). In a single breath-hold duration (24–31 seconds), a 60- to 96-mm slab was divided into 20–24 partitions in the z direction with a true thickness of 3–4 mm (interpolated to 1.5–2 mm) to cover the hepatic arterial anatomy. Prior to the contrast-enhanced examination, a timing examination, using a test bolus of 1 mL of gadopentetate dimeglumine (Magnevist; Berlex Laboratories, Wayne, NJ) and 10 cc of saline flush at 2 cc/second through an 18-gauge venous cannula, was performed to determine the transit time and optimal delay between start of injection and image acquisition (27). This delay time was used to acquire the first postcontrast 3D data set after both injections of contrast material. The mean transit time for arrival of test bolus to aorta was 25.3 ± 4.5 seconds, with transit times ranging from 15–32 seconds. The axial 3D GRE examination consisted of three 3D volume data sets acquired during the arterial-dominant (first postcontrast 3D data set), portal venous-dominant (10 seconds after end of acquisition of arterial-dominant phase), and equilibrium (5 minutes postinjection) phases of contrast redistribution following the hand injection of 19 mL of gadopentetate dimeglumine and 20 mL of saline flush at 2 cc/second. The coronal 3D GRE examination consisted of only arterial-dominant and portal venous-dominant phase imaging after a second hand injection of 20 mL gadolinium chelate and saline flush at 2 cc/second. These patients received a total mean dose of 0.24 mmol/kg (range, 0.16–0.48 mol/kg).

The source images from the arterial phase of the second multiphasic (coronal) 3D breath-hold GRE examination were processed into 3D reconstructions at an independent work station using MIP, SSD, and volume rendering (VR) algorithms. Multiple projections from each 3D reconstruction algorithm were printed separately on film.

Image Interpretation: MRI

MR data were evaluated both quantitatively and qualitatively to assess their feasibility in the depiction of hepatic arterial anatomy and parenchymal lesions using this dual-injection multiphasic approach.

Quantitative Assessment

Region-of-interest (ROI) analysis was performed by one observer (W.B.E) at the independent workstation (General Electric Medical Systems, Milwaukee, WI). Signal intensity (SI) measurements of the following vessels and hepatic parenchyma were obtained from the source images of the 3D volume acquisitions at baseline (precontrast), arterial-dominant, and portal venous-dominant phases after the first injection, and at the arterial-dominant phase after the second injection of contrast material: aorta, common hepatic artery (CHA), main portal vein (MPV), and suprarenal inferior vena cava (IVC). Vessels were identified on the arterial-dominant phase images after the first contrast injection, and identical ROIs were placed on the corresponding images from the precontrast and portal venous-dominant phases. Care was taken to place ROIs within the same anatomic segment of these vessels on the coronally acquired images. To assess the relative enhancement of the portal vein to the hepatic artery for the arterial-dominant phase data set after each injection, a venous-to-arterial enhancement ratio (22) was calculated by using the following formula: (SIpostV − SIpreV)/(SIpostA − SIpreA), where V is MPV and A is CHA. A ratio of 0 represents no portal venous enhancement relative to hepatic arterial enhancement, and a ratio of 1 represents equal venous and arterial enhancement.

Qualitative Assessment

Two independent observers (U.P.S and K.L.W) with experience in body MR and cross-sectional imaging and blinded to the CT data evaluated the MR data on the workstation. The observers evaluated the arterial-dominant phase data sets for artifacts (patient motion, respiration, pulsation, aliasing, or susceptibility) and the contrast of the arterial anatomy to the background soft tissue. Artifacts were scored on a four-point scale (0 = absent, 1 = mild, 2 = moderate, 3 = severe). Contrast of the arterial anatomy to the background was scored on a four-point scale (0 = poor, 1 = moderate, 2 = good, 3 = excellent) for the following segments: celiac trunk, superior mesenteric artery (SMA), CHA, left gastric artery (LGA), right hepatic artery (RHA), and left hepatic artery (LHA). A multiphasic examination was considered diagnostic when the observers gave a qualitative score of 2 or higher (good or excellent) for the arterial contrast of the celiac trunk and CHA, and the SI of the aorta and CHA was greater on the arterial-dominant than on the portal venous-dominant phase images.

After reviewing the source images from the arterial-dominant phase acquisitions and the 3D reconstructions, the observers defined the arterial anatomic type according to Michels' classification (28), summarized as follows:

  • Type I: conventional anatomy (proper hepatic artery divides into RHA and LHA

  • Type II: LHA replaced to LGA

  • Type III: replaced RHA to SMA

  • Type IV: replaced RHA and LHA

  • Type V: accessory LHA replaced to LGA

  • Type VI: accessory RHA replaced to SMA

  • Type VII: accessory RHA and LHA

  • Type VIII: replaced RHA and accessory LHA, or replaced LHA and accessory RHA

  • Type IX: proper hepatic artery arises from SMA

  • Type X: proper hepatic artery arises from LGA

Any discrepancy in arterial classification between the two observers was decided by a consensus reading. The observers reviewed the source images from the portal venous-dominant phase for assessment of patency of the portal venous system.

Both unenhanced and enhanced MR data were evaluated for focal hepatic parenchymal lesions. The observers indicated the number and location of detected lesions and the SI characteristics and enhancement pattern (homogenous, peripheral, heterogenous) on unenhanced and enhanced data, respectively. The observers also indicated the most likely diagnosis for each detected lesion and the likelihood that the lesion was malignant (low, moderate, or high).

Interpretation of the MR data with respect to arterial classification, portal vein patency, and focal hepatic parenchymal lesions was compared with the prospective interpretation of the CT data. For the cases in which CTA and MRA were discordant in the depiction of the arterial anatomy, an experienced observer in cross-sectional imaging (W.B.E.) retrospectively reviewed the CT and MR data. At the time this article was written, confirmation of imaging findings was available in 10 patients: two underwent orthotopic liver transplantation with histopathohologic examination of the explanted liver (interval between initial imaging and transplantation, 7 and 9 months), seven had follow-up multiphasic helical CT examinations (average follow-up interval, 14.4 months; range, 10–22 months), and one underwent CT-guided biopsy of a focal hepatic parenchymal lesion (performed at 4 months; further follow-up CT at 17 months after initial imaging).

Statistical Analysis

A two-tailed Student's t-test was used to compare the venous-to-arterial enhancement ratios and the intraobserver qualitative assessment of arterial contrast-to-background on data acquired during the arterial-dominant phase after the first and second injections of contrast material. Weighted kappa values (κ) were calculated to assess interobserver agreement on the qualitative assessment of arterial contrast-to-background. The level of interobserver agreement was based on the following scale for kappa values: poor, <.4; good, ≥.4 and <.75; excellent, ≥.75.

RESULTS

A diagnostic multiphasic 3D GRE examination was obtained in 90% (18/20) of the patients. In two patients, the arterial phase data set was acquired too early (SI ratio of aorta-to-portal vein was greater on the portal venous-dominant phase than on the arterial-dominant phase) after both injections of contrast material. The observers also rated the contrast of the celiac axis and the CHA to background poor to moderate (score of 1 or less) in both of these cases.

Quantitative Assessment

The mean SI values of liver and major vessels acquired at the baseline, arterial-dominant, and portal venous-dominant phases after the first injection, and at the arterial-dominant phase after the second injection are plotted in Figure 1. The mean SI ratio of the aorta to portal vein was over twofold higher on the arterial-dominant phase after the first (ratio of 2.4) and second (ratio of 2.1) injections of contrast. There was no significant difference (P =.70) between the venous-to-arterial enhancement ratios on arterial-dominant phase acquisition after the first contrast injection, 0.42 ± 0.33 (mean ± SD), and the second contrast injection, 0.42 ± 0.17 (mean ± SD).

Figure 1.

Mean SI measurements on MR for ▪, aorta; Δ, main portal vein; •, suprahepatic IVC; and ♦, liver. The mean SI ratio of the aorta-to-portal vein was greater than 2 for the arterial-dominant phases after both injections of gadolinium chelate.

Qualitative Assessment

Image artifacts on enhanced MR data were rated absent to mild by the observers; the mean scores ± SD for observers one and two were 0.70 ± 0.80 and 0.61 ± 0.70, respectively. The subjective values for contrast between the arterial system and background soft tissue on the arterial-dominant phase after both injections of contrast material are shown in Table 1. The overall contrast rating was good to excellent (average pooled score of 2.00 ± 0.27 after the first injection and 2.06 ± 0.31 after the second injection), and there was no significant difference for any of the arterial segments between the first and second multiphasic examinations. There was a trend toward poorer contrast-to-background values as vessel caliber decreased.

Table 1. Qualitative Assessment of Arterial-to-Background Contrast on the Arterial-Dominant Phase of MR Examinations
Vessel1st injection (axial)κ2nd injection (coronal)κP
  1. Data are the mean ± SD. Mean score based on a four-point scale (0, poor; 1, moderate; 2, good; and 3, excellent).

  2. SMA, superior mesenteric artery; CHA, common hepatic artery; LGA, left gastric artery; RHA, right hepatic artery; LHA, left hepatic artery.

Celiac trunk     
 Observer 12.30 ± 1.08.332.40 ± 0.94.29.63
 Observer 22.20 ± 1.06 2.20 ± 0.89 1.0
SMA     
 Observer 12.30 ± 1.08.332.45 ± 0.94.23.45
 Observer 22.17 ± 1.04 2.22 ± 0.94 .80
CHA     
 Observer 12.25 ± 1.12.532.30 ± 0.98.47.80
 Observer 21.95 ± 1.00 2.05 ± 1.00 .63
LGA     
 Observer 11.95 ± 1.28.352.00 ± 1.17.31.77
 Observer 21.40 ± 1.05 1.30 ± 0.86 .68
RHA     
 Observer 12.00 ± 1.21.672.05 ± 1.05.53.75
 Observer 21.80 ± 1.01 1.90 ± 1.07 .58
LHA     
 Observer 12.00 ± 1.21.642.10 ± 1.07.49.54
 Observer 21.75 ± 1.07 1.80 ± 1.05 .82

Evaluation of Hepatic Arterial Anatomy and Portal Venous Patency

Hepatic arterial classifications based on MR and CT data are presented in Table 2. Overall, MR angiography (MRA) and CTA were concordant in 17/20 (85%) and discordant in 3/20 (15%) patients. The arterial anatomy was classified as conventional (Michels' type I) by CTA in 12/20 (60%) and nonconventional in 8/20 (40%) patients. Among patients classified as having conventional hepatic arterial anatomy, MRA and CTA were concordant in all 12 (100%) patients. Conventional arterial anatomy was confirmed in one patient at the time of transplantation. Among patients classified as having nonconventional hepatic arterial anatomy, MRA and CTA were concordant in 5/8 (63%) (Fig. 2) and discordant in 3/8 (37%) (Fig. 3) patients.

Table 2. Categorization of Hepatic Arterial Anatomy and Agreement Between CTA and MRA
Michels' classification by CTANo. of patientsConcordant (CTA vs MRA)Discordant (CTA vs MRA)
  • Number in parentheses is percent.

  • a

    MRA interpreted as type V.

  • b

    MRA interpreted as type III.

  • c

    MRA interpreted as type VI.

I12 (60)120
II2 (10)11a
III1 (5)10
IV1 (5)01b
V1 (5)10
VI1 (5)10
VII1 (5)01c
IX1 (5)10
Total2017 (85)3 (15)
Figure 2.

A 70-year-old woman with cirrhosis secondary to hepatitis C and an accessory RHA arising from the SMA (Michels' type VI anatomy). a: Volume-rendered 3D reconstruction from the arterial-dominant phase of 3D GRE MR examination (TR/fractional TE, 6.1/1.3; flip angle, 20°) shows accessory RHA (solid arrow) arising from the SMA. Note conventional bifurcation of the proper hepatic artery (open arrow) that arises from the celiac axis. b: Volume-rendered 3D reconstruction from the arterial-dominant phase of contrast-enhanced CT shows comparable arterial anatomy as in a.

Figure 3.

A 54-year-old man with cryptogenic cirrhosis and accessory RHA and LHA (Michels' type VII anatomy). a: Axial source image from arterial-dominant phase of 3D GRE MR examination (TR/TE, 7.7/3; flip angle, 20°) shows poorly opacified small-caliber accessory LHA within the ligamentum venosum (arrow). b: Coronal source image from arterial-dominant phase of 3D GRE MR examination (TR/TE, 6.1/1.3; flip angle, 20°) shows better opacification of the accessory LHA (arrow) than in a. Both observers of MR data missed this accessory artery. c: Axial image from arterial-dominant phase of CT shows well opacified accessory LHA within the ligamentum venosum (arrow), arising from the LGA. This accessory artery was observed at prospective CT interpretation.

Discordance between prospective CT and MR interpretation in the classification of hepatic arterial anatomy (N = 3) involved a small-caliber (<3 mm) aberrant LHA in all cases. One patient had a replaced RHA arising from the SMA (Michels' type III); a replaced LHA was diagnosed prospectively by CT but was not confirmed either retrospectively by CT or MR. A second patient had accessory RHA and LHA (Michels' type VII) by prospective and retrospective interpretation of CT. However, the small-caliber accessory LHA was missed by MR (Fig. 3). Both RHAs and LHAs were confirmed at transplantation surgery in this patient. In the last case, there was discordance in the interpretation of Michels' classification, specifically the failure to recognize the middle hepatic artery. This patient had a dominant LHA arising from the LGA and a small-caliber artery supplying segment four of the liver (middle hepatic artery) arising from the celiac trunk that were seen prospectively by CT and MR. The prospective CT interpretation correctly classified the anatomy as a replaced LHA arising from the LGA with a middle hepatic artery arising from the celiac trunk (Michels' type II), and the MR interpretation incorrectly classified the anatomy as an accessory LHA (Michels' type V). Therefore, of the three patients with discordant findings between CTA and MRA, two were due to perceptive errors (one each for prospective CT and MR interpretation) and the last to a discrepancy in the interpretation of the classification scheme.

One patient had a thrombus partially filling the portal venous confluence that was diagnosed by both MR observers and at prospective CT interpretation. Presence of the thrombus was confirmed by interval enlargement of the clot on 7-month follow-up CT.

Evaluation of Hepatic Parenchyma

In the majority of patients (15/20 (75%)), no focal lesions were detected by MR or CT. Absence of focal lesions was confirmed in six patients at transplantation (N = 2) or follow-up CT (N = 4). In the remaining five patients (25%), seven focal lesions were detected by CT and nine by MR (Table 3). In one patient with multiple lesions, two of the lesions were detected only by MR (Fig. 4). Two patients had hypervascular lesions on the arterial-dominant phase of both MR and CT examinations. One of these patients underwent CT-guided percutaneous biopsy of a single lesion due to a high suspicion of hepatocellular carcinoma (HCC) (Fig. 5). The other patient had several small (<1 cm), peripheral, homogeneously enhancing nodules that became less conspicuous on follow-up CT. These lesions were considered to most likely represent small arterioportal shunts or foci of inflammation.

Table 3. Imaging Characteristics of Focal Hepatic Parenchymal Lesions Detected by CT or MR
Pt no./lesion no./size (cm)Appearance on unenhanced examEnhancement of lesionLikely Dx/suspicion for HCCa
CTMR (T1/T2)CTMR
  • a

    Likely diagnosis and likelihood of lesion representing an HCC was identical for both CT and MR when lesion was detected by both modalities.

  • b

    No change in lesion at 12-month follow-up CT.

  • c

    Both lesions were associated with capsular retraction and progressive enhancement on delayed images.

  • d

    Patient had 7-month follow-up CT that showed decrease in conspicuity of subcentimeter hypervascular nodular lesions consistent with perfusion abnormalities.

  • e

    Patient with primary sclerosing cholangitis underwent a CT-guided biopsy that revealed no malignancy and a 17-month follow-up CT that showed no progression of this lesion.

  • DN, dysplastic nodule; RN, regenerating nodule; HCC, hepatocellular carcinoma.

1/1/1HypoHypo/hyperCyst/lowb
2/1/1IsoHyper/isoRN or DN/low
2/2/1IsoHyper/isoRN or DN/low
2/3/1.5HyperHyper/isoRN/low
2/4/1.5HyperHyper/isoRN/low
2/5/2.5HypoIso/iso++Confluent fibrosis/lowc
3/1/3.0HypoIso/iso++Confluent fibrosis/lowc
4/1/<1 cm, severalIsoIso/iso+, early+, earlySmall HCC vs. Arterioportal shunt/moderated
5/1/2.5IsoIso/hyper+, hetero+, heteroHCC/highe
Figure 4.

A 42-year-old man with cirrhosis secondary to alcohol and hepatitis C. a: Axial source image from unenhanced 3D GRE MR examination (TR/TE, 7.7/3; flip angle, 20°) shows a 1-cm hyperintense lesion in segment six (arrow). b: SI measurements from the axial source image from arterial-dominant phase of 3D GRE MR examination (TR/TE, 7.7/3; flip angle, 20°) did not vary more than 10% compared with the precontrast images, indicating no enhancement of the lesion in a. This lesion and another one with similar MRI characteristics were not detected at prospective interpretation of multiphasic helical CT examination (not shown). These lesions were considered to have low malignant potential by MR observers; however, no confirmation was obtained.

Figure 5.

A 57-year-old man with primary sclerosing cholangitis. a: Axial source image from arterial-dominant phase of 3D GRE MR examination (TR/TE, 7.7/3; flip angle, 20°) shows a focal hyperenhancing lesion (arrow) in segment eight. b: Axial image from arterial-dominant phase of CT shows lesion (arrow) with similar enhancement characteristics. The patient underwent a CT-guided percutaneous biopsy that revealed no malignancy and had a 17-month follow-up CT that showed no progression of this lesion.

DISCUSSION

In our study, MRA and CTA were concordant for depiction of arterial anatomy in 17/20 (85%) patients, and of the nine hepatic parenchymal lesions detected, two were detected only by MR. Successful acquisition during the arterial-dominant phase of the 3D gadolinium-enhanced fast GRE sequence is necessary for the accurate evaluation of arterial anatomy and the detection of HCC (13–15, 27, 29). Since arterial phase imaging depends on individual variation in time of bolus arrival to the targeted arterial system, we and other investigators (22, 23, 27, 30) have used a timing bolus examination to tailor the delay-to-acquisition time for each multiphasic examination. The acquisiton of the arterial-dominant phase was successful, by both qualitative and quantitative criteria, in 90% of the multiphasic MR examinations performed in this study. Poor timing of acquisition was the reason for nondiagnostic examinations in two patients. One potential limitation of our technique is the manual injection of contrast injection and saline flush. The use of a power injector would likely give more reproducible results; however, this equipment was not available during the time period our patients were imaged.

The types of hepatic arterial supply demonstrated in our study closely parallel Michels' study (28) of meticulous cadaveric dissections: 60% and 40% of our patients were classified as having conventional and nonconventional arterial anatomy, respectively, vs. 55% and 45%, respectively, in Michels' study. In order to investigate the source of error in the three cases for which CTA and MRA were discordant for depiction of the arterial anatomy, the CT and MR data in these cases were retrospectively reviewed by an experienced observer in cross-sectional imaging (W.B.E.). Our results are also similar to those of Kopka et al (19), who correlated MRA with digital subtraction angiography in 60 patients and found that small accessory arteries were missed at MRA in 3/60 patients. Small-caliber aberrant hepatic arteries are generally ligated at transplantation surgery, and their presence usually does not impact the preoperative planning for arterial anastomosis performed at transplant surgery (4).

Focal parenchymal lesions were detected in 25% (5/20) of patients in our study. This detection rate is over double that found in the study by Smith et al (5), who detected focal lesions in 10% (5/50) of patients who underwent preoperative liver transplant evaluation with dual-phase CT angiography. However, all of the lesions mentioned in that study had at least some hypervascular component on the arterial-dominant phase, and it is not clear whether only hypervascular lesions were considered. In our study, 2/20 (10%) patients had hypervascular lesions suspicious for HCC seen on the arterial-dominant phase of both CT and MR examinations, which is more in line with the lesion detection rate by Smith et al (5). Two lesions in our study were detected only by MR; both were consistent with regenerating or dysplastic nodules based on imaging findings. Several investigations (9, 10) have shown that multiphasic helical CT is relatively insensitive (59%–71% sensitivity) for the detection of small HCCs and even less sensitive (39%) for dysplastic nodules when compared to the pathologic findings of the explanted liver. Dynamic gadolinium-enhanced MR examination of the liver, using a T1-weighted GRE multislice 2D technique, has been shown to have higher sensitivity for detecting small (<2 cm) HCCs in patients with cirrhosis than multiphasic helical CT (13, 14). However, MRI has been shown to have limited ability to detect small (<2 cm) HCCs (sensitivity, 47%) and dysplastic nodules (sensitivity, 15%) in patients without known HCC when compared to explanted liver pathologic results (31).

Each patient in our study underwent two separate multiphasic MR examinations (contrast boluses separated by 7–8 minutes). This imaging strategy was implemented to optimize evaluation of both the hepatic parenchyma and arterial anatomy. The parameters of the first multiphasic sequence, acquired in the axial plane, were selected to optimize the signal-to-noise ratio (SNR) and the contrast-to-noise ratio (CNR) of the hepatic parenchyma. The second multiphasic examination, acquired in the coronal plane, was designed to optimize the depiction of small-caliber vessels. Furthermore, coronal acquistion is the most efficient way to cover the relevant anatomy and provide 3D reconstructions for hepatic MRA (32). Contrast of the hepatic arterial system after the first and second gadolinium injections was not significantly different by qualitative or quantitative assessment. This suggests that there is no compromise of arterial depiction after the second injection, as the gadolinium chelate from the first injection rapidly redistributes from the intravascular to interstitial space. However, we do feel it is important to perform the parenchymal evaluation first, since it is not clear whether focal lesions can be accurately evaluated after a second bolus of contrast material.

There are several limitations of this study. We used CTA as the gold standard for the evaluation of hepatic arterial anatomy since it has replaced conventional angiography in the preoperative evaluation of OLT candidates at our institution. We recognize this is an imperfect gold standard. Both CTA and MRA are limited by spatial resolution compared with conventional angiography. This should be considered a preliminary study, since surgical or histological confirmation of hepatic arterial anatomy and parenchymal lesions was scant. At the time this article was written, only two patients had undergone transplantation and one patient had a percutaneous biopsy of a suspicious parenchymal lesion. Finally, the number of patients in this study is small, limiting the number of patients with focal parenchymal lesions and nonconventional hepatic arterial anatomy.

Technological developments in both CT and MR will likely lead to even greater improvement in detection of lesions in patients with chronic liver disease. The use of multidetector helical CT may improve detection of hypervascular lesions, due to its ability to provide even greater temporal resolution of image acquisition (33, 34). Improvement in gradient strength and parallel processing techniques (29, 35) will further speed MR image acquisition as well, decreasing artifacts due to motion. Improvement in lesion detection may impact patients awaiting liver transplantation by placing those with lesions suspicious for HCC higher on the waiting list or prompting closer follow-up surveillance.

CONCLUSIONS

A multiphasic MR examination using a T1-weighted 3D gadolinium-enhanced fast GRE sequence can provide comprehensive evaluation of the hepatic arterial anatomy and parenchyma in liver transplantation candidates and may offer an advantage over CT in the detection of parenchymal lesions. For patients at risk for allergic reaction or nephrotoxicity to iodinated contrast agents, a multiphasic MR examination can replace CT for presurgical evaluation.

Ancillary