| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (G.A.K., V.S.L., M.T.N., N.M.R., J.C.W.), Pathology (N.D.T.), and Transplantation (G.R.M., L.W.T.) and the Kaplan Comprehensive Cancer Center (G.A.K., N.D.T.), New York University Medical Center, 530 First Ave, New York, NY 10016. Received March 7, 2000; revision requested April 25; revision received May 25; accepted June 5. G.A.K. supported by an RSNA Seed Grant. Address correspondence to G.A.K. (e-mail: glenn.krinsky@med.nyu.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: MR imaging (1.5 T) was performed within 0–117 days (mean, 30 days) before liver transplantation in 77 patients. Two readers retrospectively evaluated gradient-echo (GRE) (echo time [TE], 
9 and 4–5 msec) and turbo short inversion time inversion-recovery or T2-weighted images for low-signal-intensity nodules. Whole-explant pathologic correlation was available in every case.
RESULTS: At explantation, 28 (36%) of 77 patients had HCC, 25 (32%) had dysplastic nodules, and nine (12%) had both; 35 (45%) patients had siderotic nodules. The sensitivity of GRE imaging with 9-msec or longer TE for the detection of siderotic nodules was 80% (28 of 35) but decreased to 31% (11 of 35) with 4–5-msec TE. Frequency of HCC was not significantly higher (P = .27) in patients with (43% [15 of 35]) than in patients without (31% [13 of 42]) siderotic nodules. Frequency of dysplastic nodules also was not significantly higher (P = .42) in patients with (37% [13 of 35]) than in patients without (29% [12 of 42]) siderotic nodules.
CONCLUSION: Sensitivity of MR imaging for the detection of siderotic nodules was improved with use of GRE pulse sequences with longer TEs of 9 msec or greater (80%) versus 4–5 msec (31%); however, there was no significant increased frequency of HCC or dysplastic nodules in patients with pathologically proved siderotic nodules.
Index terms: Liver, cirrhosis, 761.794 • Liver, MR, 761.121412, 761.121413 • Liver, nodules • Liver, transplantation, 761.45 • Liver neoplasms, 761.32
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In a recent article, Ito et al (12) found that the frequency of HCC in patients with iron deposition in regenerative nodules was significantly higher than that in patients without iron in regenerative nodules. On the basis of their findings, they hypothesized that "the occurrence of hepatocellular carcinoma (HCC) may be associated causally with iron deposition in large regenerative nodules in patients with cirrhosis" (12). However, there were several potential sources of errors in that study, including a variety of MR imaging units, coils (phased array or body), pulse sequences, and TEs and limited explant correlation (27 of 196 cases).
The purpose of this retrospective study was to determine the sensitivity and specificity of GRE (TE, 4–5 and 9 msec) and turbo short inversion time inversion-recovery (STIR) or turbo spin-echo T2-weighted MR imaging for the diagnosis of siderotic nodules and to determine by means of whole-explant pathologic correlation whether patients with these nodules have a higher frequency of dysplastic nodules and/or HCC.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
-1-antitrypsin deficiency (n = 1), eosinophilic hepatitis (n = 1), primary biliary cirrhosis (n = 1), and hemochromatosis (n = 1). The patient with hemochromatosis was excluded from the study. Seven other patients also were excluded because an iron-sensitive (susceptibility) pulse sequence (TE, 9 msec; flip angle, 
45°) was not performed in these subjects. Therefore, the study population consisted of 77 patients. All studies were clinically indicated, and our standardized routine hepatic imaging protocol was performed. Informed consent was obtained prior to the gadolinium-enhanced study. Our study was approved by our institutional review board.
MR Imaging Technique
MR imaging was performed with a 1.5-T system (Magnetom Vision; Siemens, Erlangen, Germany), with a quadrature phased-array multicoil. In 73 of 77 patients, breath-hold T1-weighted spoiled GRE images (fast low-angle shot [FLASH], 160– 220/4–5 [repetition time msec/TE msec], 70°–90° flip angle) were acquired in the transverse plane with a section thickness of 5–8 mm, no intersection gap, a 128–192 x 256 matrix, and a rectangular field of view with the largest dimension of 30–40 cm. This resulted in a 15–24-second breath hold. In four patients, who were unable to suspend respiration, a T1-weighted magnetization-prepared GRE pulse sequence (turbo FLASH, 11/4.2/300 [repetition time msec/TE msec/inversion time msec], 15° flip angle) was performed with a similar voxel size. A multiphase, dynamic gadolinium-enhanced examination was performed as well, but these images were not evaluated as part of this study.
In 73 patients, a breath-hold turbo STIR sequence (4,000–5,000/58 or 76/165 [repetition time msec/effective TE msec/inversion time msec], flip angle of 150°–180°, and echo train length of 33) was performed with an 8-mm section thickness and a 2-mm intersection gap. This required two noninterleaved breath-hold acquisitions for sufficient anatomic coverage. The matrix and field of view were similar to those of the previously mentioned T1-weighted images. In the four patients who were unable to suspend respiration, a non–breath-hold turbo spin-echo T2-weighted pulse sequence (5,000/99 [repetition time msec/effective TE msec], echo train length of 11) was performed in the transverse plane with an 8-mm section thickness, a 2-mm intersection gap, three signals acquired, and a similar matrix and field of view.
Finally, in all 77 patients, a breath-hold flow-sensitive GRE sequence with a relatively long TE (9 msec [n = 68], 10 msec [n = 5], or 20 msec [n = 4]) was performed to provide a susceptibility sensitive imaging strategy. This sequence was performed first with selective presaturation above and again with presaturation below the imaging sections to assess patency and direction of portal venous flow. The parameters consisted of 20–200/9–20 (repetition time msec/effective TE msec) and a flip angle of 20°–45° with a first order gradient moment refocusing pulse. A total of at least four 10-mm sections were acquired through the liver with a matrix of 128 x 256, rectangular field of view with the largest dimension of 30–40 cm, and intersection gap of 20 mm. The breath hold ranged from 16 to 22 seconds, depending on the use of the rectangular-field-of-view option. The sequence was performed without breath holding in four patients.
Image Interpretation
MR studies were retrospectively reviewed in consensus by two radiologists (G.A.K., V.S.L.) by using hard-copy images of all the nonenhanced pulse sequences described above. Each pulse sequence was evaluated independently per patient (not in a randomized fashion). Cirrhotic nodules that were hypointense to the background hepatic parenchyma were considered siderotic and were graded as present or absent but not individually counted. Because of the large number and small size of these nodules, it was not possible to evaluate each pulse sequence against the explanted liver on a nodule per nodule basis. Therefore, the nodules were evaluated on a per-patient basis with the explanted liver. The readers were blinded to the pathologic results at the time of pulse sequence review.
Pathologic Analysis
Explanted livers were sectioned sequentially at 5–8-mm intervals by one of two hepatopathologists (N.D.T.) so that they corresponded as closely as possible to the MR imaging planes. Dysplastic nodules and HCC nodules were identified grossly as those that were distinct from surrounding regenerative nodules in terms of size, texture, color, or degree of bulging beyond the cut surface of the liver (13). All livers were photographed, and all distinctive nodules were sampled; this included staining with hematoxylin-eosin as well as Prussian blue for the identification of iron.
Low-grade dysplastic nodules were defined as nodules showing normal architecture and cytology or diffuse large cell change (13). High-grade dysplastic nodules were defined on the basis of the presence of one of the following: diffuse small cell change, pseudogland formation, nodule-in-nodule lesions with small cell dysplasia, iron resistance in siderotic nodules, fatty change, clear cell change, or Mallory body clustering (13).
Hepatocyte iron within nodules was scored by the hepatopathologist on a 0–4 scale as follows: 0 indicated none; 1, minimal or scant; 2, mild; 3, moderate; and 4, severe.
Statistical Analysis
Sensitivity and specificity of MR imaging for the detection of siderotic nodules were computed on the basis of findings at pathologic examination. The 
2 test was used to compare the frequency of HCC and dysplastic nodules between the patients with and those without siderotic nodules at explantation. A P value of less than or equal to .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Thirty-five (45%) of the 77 patients had pathologically proved siderotic nodules. Of these 35 patients, 34 had either hepatitis C, hepatitis B, alcohol abuse, or a combination of these causes. The other patient had cryptogenic cirrhosis.
GRE MR imaging with a TE of 9 msec or longer enabled detection of siderotic nodules in 28 of 35 patients (sensitivity, 80%) (Fig 1b), whereas imaging with a shorter TE of 4–5 msec (Fig 2a) enabled detection of siderotic nodules in only 11 of 35 patients (sensitivity, 31%) (Table). For the seven patients with siderotic nodules not detected with MR imaging with a TE of 9 msec or longer, the iron scores were minimal (score 1) (n = 6) or mild (score 2) (n = 1). When the six patients with minimal amounts of iron (score 1) were excluded, the sensitivity of GRE MR imaging with a TE of 9 msec or longer for the detection of siderotic nodules increased to 95% (21 of 22 patients).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The frequency of HCC in patients with pathologically proved siderotic nodules was 43% (15 of 35 patients). For patients without siderotic nodules, the frequency of HCC was 31% (13 of 42 patients). Although the frequency of HCC was slightly higher in patients with siderotic nodules, this difference did not achieve statistical significance (P = .27). No siderotic nodules contained foci of HCC at pathologic examination.
Dysplastic nodules were present in 25 (32%) of 77 explanted livers. The frequency of dysplastic nodules in patients with pathologically proved siderotic nodules (37% [13 of 35 patients]) was not significantly greater (P = .42) than that in patients without siderotic nodules (29% [12 of 42 patients]).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
9 msec) and turbo STIR or turbo T2-weighted spin-echo MR imaging for the diagnosis of siderotic nodules. Our findings demonstrated that breath-hold GRE MR imaging with a TE of 9 msec or longer was both sensitive (80%) and specific (95%) for the diagnosis of iron-containing siderotic nodules. These study results confirm the findings of previous studies (3–7) but with a much larger patient population, with explant correlation in every case, and the use of a phased-array coil. The poor sensitivity (31%) of breath-hold GRE MR imaging with a TE of 4–5 msec for the detection of siderotic nodules limits its use as a susceptibility sensitive sequence. This is important because these pulse sequences are widely used in conjunction with dynamic gadolinium-enhanced examinations for lesion detection and characterization. Therefore, for the detection of siderotic nodules, a pulse sequence with a longer TE is clearly needed. The optimal pulse sequence for the detection of these nodules and hemochromatosis is not known.
Low-signal-intensity nodules were seen in 26 (74%) of 35 patients when using predominantly the breath-hold turbo STIR pulse sequence. However, these nodules were also present in 15 (36%) of the 42 patients without pathologically proved siderotic nodules. Therefore, the presence of low-signal-intensity nodules on images obtained with this pulse sequence was not specific for siderotic nodules and was seen with ordinary regenerative nodules. Although the detection of magnetic susceptibility effects with turbo spin-echo T2-weighted pulse sequences is minimized by the use of multiple 180° refocusing pulses, the sensitivity of the pulse sequences we used for the detection of siderotic nodules might have improved with decreased echo train lengths and longer effective TEs.
Ito et al (12) found a significantly increased frequency of HCC in patients with siderotic nodules and hypothesized that there may exist a causal relationship between siderotic regenerative nodules and HCC. Our study could not replicate these findings. Whereas the prevalence of siderotic nodules detected at MR imaging in patients with cirrhosis was the same (36% in both studies), we did not find a significantly increased frequency of HCC or dysplastic nodules. However, in their study different MR machines, coils (phased array or body), and pulse sequences were used at different institutions, and there was no histologic proof of siderotic nodules. Because explant correlation was not available for many of the cases, the control group of patients without HCC may have had small undetected HCC, and this could have changed the results of their study.
Although the presence of iron is associated with an increased incidence of hepatocellular carcinoma in patients with hemochromatosis (1,2), to our knowledge there has been no pathologic explant study in which the results demonstrated an increased frequency of HCC in patients with siderotic regenerative nodules.
Patients with dysplastic nodules have a higher frequency of HCC (14), but the frequency of HCC in patients with siderotic dysplastic nodules is not known. Although Terada and Nakanuma (15) observed a higher frequency of hyperplastic hepatocellular foci (now called high-grade dysplastic nodules) within siderotic dysplastic nodules in an autopsy series of 34 patients (an additional 10 patients had surgical resection), no mention of an association with HCC was made. Although HCC has been shown to occur within dysplastic nodules at both MR imaging (16,17) and pathologic examination (18–21), iron has not been proved to be the cause of the malignant transformation.
Results of previous studies have shown an association between hepatic iron excess at pathologic examination and cirrhosis (22) and HCC (23–25) in patients without genetic hemochromatosis. Although these authors state that excess iron is a risk factor for developing HCC, they do not specifically discuss focally increased iron in cirrhotic nodules or prove a causal relationship between iron excess and HCC.
It is unclear why iron accumulates within regenerative or dysplastic nodules in the absence of genetic hemochromatosis or hemosiderosis. One theory is that transferrin receptor proteins may be more active in these lesions (26). Because these proteins probably mediate cellular iron uptake, greater activity may result in selective iron accumulation (26).
This study had recognized limitations, including its retrospective design. The sensitivity of MR imaging for the diagnosis of siderotic nodules may have increased if all patients had been examined with a GRE pulse sequence with a longer TE (ie, 15 msec) (11). Our use of a limited number of sections (four or more sections; TE, 9 msec) for determining the presence of siderotic nodules also might be viewed as a shortcoming, because segmental iron deposition has been reported at MR imaging (27). However, long TEs place constraints on the number of sections available for a given repetition time. With our MR system, this would require three to four breath holds to evaluate the entire liver. This may not be practical in a cohort of patients with cirrhosis, some of whom have pulmonary compromise from ascites or the hepatopulmonary syndrome. Because we performed consensus review of pulse sequences, the interobserver variability for the detection of siderotic nodules is unknown. Finally, our series with explant correlation had a high prevalence of HCC (36%), and the results may not be extrapolated to a screening population with a lower prevalence of disease.
In conclusion, the sensitivity of MR imaging for the detection of siderotic nodules was improved with the use of GRE pulse sequences with longer TEs (9 msec [80%] vs 4–5 msec [31%]); however, we found no significant increase in the frequency of HCC or dysplastic nodules in patients with pathologically proved siderotic nodules.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, G.A.K.; study concepts, all authors; study design, G.A.K., V.S.L.; definition of intellectual content, all authors; literature research, M.T.N.; clinical studies, G.A.K., N.M.R., V.S.L.; data acquisition, G.A.K., N.M.R., V.S.L.; data analysis, G.A.K., V.S.L.; statistical analysis, V.S.L.; manuscript preparation and editing, all authors; manuscript review, G.A.K., V.S.L.; manuscript final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  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]
|
|
|
|
|
|
T. Parikh, S. J. Drew, V. S. Lee, S. Wong, E. M. Hecht, J. S. Babb, and B. Taouli Focal Liver Lesion Detection and Characterization with Diffusion-weighted MR Imaging: Comparison with Standard Breath-hold T2-weighted Imaging Radiology, March 1, 2008; 246(3): 812 - 822. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. A. Krinsky and G. Israel Nondysplastic Nodules That Are Hyperintense on T1-Weighted Gradient-Echo MR Imaging: Frequency in Cirrhotic Patients Undergoing Transplantation Am. J. Roentgenol., April 1, 2003; 180(4): 1023 - 1027. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. L. Baron and M. S. Peterson From the RSNA Refresher Courses: Screening the Cirrhotic Liver for Hepatocellular Carcinoma with CT and MR Imaging: Opportunities and Pitfalls RadioGraphics, October 1, 2001; 21(90001): S117 - 132. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
