| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Memorial Sloan-Kettering Cancer Center, 1275 York
Ave., New York, NY 10021.
2 Present address: Department of Radiology, Mount Sinai Hospital of Queens,
Queens, NY.
3 Present address: Department of Diagnostic Imaging, Brown University and Rhode
Island Hospital, Providence, RI.
Received June 29, 2004;
accepted after revision May 29, 2005.
Presented at the 2005 annual meeting of the American Roentgen Ray Society,
New Orleans, LA.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. A list of all women with breast cancer diagnosed between January 2000 and June 2003 was cross-referenced with radiologic and medical records to identify patients who had undergone MRI of the liver and who had had a hepatic lesion too small to characterize reported on previous CT performed after the diagnosis of breast cancer. The cause reported at MRI for each hepatic lesion too small to characterize seen on CT was recorded as benign, malignant, indeterminate, or no lesion seen on MRI. Subsequent relevant imaging examinations and medical records were reviewed.
RESULTS. Seventy-six (1.4%) of 5,440 women underwent MRI that included the liver. In 38 (50%) of the women a hepatic lesion too small to characterize was found on CT performed before MRI. The reported MRI diagnoses of hepatic lesions too small to characterize on CT included benign lesions (n = 22 women), lesions that remained indeterminate (n = 11 women), no lesion seen on MRI corresponding to the site of a lesion too small to characterize on CT (n = 8 women), and metastatic lesions (n = 2 women). Four women had more than one reported type of hepatic lesion. Subsequent findings of imaging, biopsy, or both performed on eight of 11 women with indeterminate lesions supported benign diagnoses in all eight women.
CONCLUSION. In women with newly diagnosed breast cancer and no definite liver metastasis on initial CT, immediate further evaluation of hepatic lesions too small to characterize with MRI offers only marginal benefit. In only approximately 5% of such women will lesions too small to characterize be shown on MRI to represent metastasis.
Keywords: breast cancer • CT • liver • MRI
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MRI has been shown to effectively allow radiologists to differentiate with moderate to good interobserver agreement benign and malignant hepatic lesions 2 cm or smaller in patients with indeterminate findings on CT [5]. MRI is often ordered by clinicians for further characterization of hepatic lesions considered too small to characterize by our radiologists, but the utility of the information obtained from evaluation of such small lesions in patients with known breast cancer is unclear. We undertook this study to assess the value of hepatic MRI for differentiating benign and malignant causes of hepatic lesions too small to characterize detected on CT in patients with breast cancer.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Patients
The tumor registry at our institution, a dedicated cancer center, was
searched for all women with the diagnosis of breast cancer made between
January 2000 and June 2003. The resulting list was cross-referenced with our
radiology information system to identify all women who had undergone abdominal
MRI evaluation after the initial diagnosis of breast cancer. The radiologic
and medical records for this subset of women were examined to identify women
who had had hepatic lesions too small to characterize detected on CT performed
before MRI and after the diagnosis of breast cancer.
Data Collection
The time interval between initial diagnosis of breast cancer and CT
examination and the interval between CT and subsequent MRI examination were
recorded. The CT scans and subsequent MR images and the associated official
radiology reports were reviewed to classify the reported MRI results for each
hepatic lesion too small to characterize as benign (e.g., cyst, hemangioma,
focal nodular hyperplasia, focal fat), malignant, or of indeterminate cause.
For each hepatic lesion too small to characterize deemed to be of
indeterminate cause on the first MRI evaluation, the results of subsequent
imaging evaluations and any biopsy of the lesion were recorded. The data were
analyzed on a per-patient basis and on the basis of the number of lesion
diagnoses reported on MRI for each patient. For example, a patient with four
cysts was recorded as having one lesion diagnosis (i.e., cyst), and a patient
with two hemangiomas and one cyst was recorded as having two hepatic lesion
diagnoses (hemangioma and cyst). The official radiology reports and clinical
records were reviewed for assessment of the presence of extrahepatic
metastasis.
Imaging Technique
All MRI examinations were performed at our institution on 1.5-T scanners
according to various protocols that included at least axial in-phase and
opposed-phase T1-weighted gradient-echo images; axial fat-saturated fast
spin-echo T2-weighted images; coronal single-shot fast spin-echo T2-weighted
images; and dynamic multiphase gadolinium-enhanced axial T1-weighted
gradient-echo images. Axial images were acquired at 7-mm slice thickness and
coronal images at 8-mm thickness. Most of the CT examinations were performed
at our institution on MDCT scanners at 7.0- or 7.5-mm collimation. Scans were
obtained during the portal venous phase after power injection of 150 mL of IV
contrast material.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Thirty-eight (50%) of the 76 women had at least one hepatic lesion reported as too small to characterize on previous CT. The records of women with hepatic lesions not deemed too small to characterize on CT and women with hepatic lesions detected only on previous sonographic examinations were not analyzed further in this study. The mean time interval between initial diagnosis of breast cancer and CT for the 38 women was 28.8 weeks (range, 0-149 weeks). The mean time interval between CT and subsequent MRI was 4.9 weeks (range, 0-30 weeks). The mean age at diagnosis of breast cancer for these 38 women was 51.0 years (range, 29.9-73.6 years). At the time of diagnosis, breast cancer was pathologic stage I in 12 (32%) of the women, stage IIA in 10 (26%), stage IIB in 10 (26%), stage IIIA in 2 (5%), stage IIIB in 2 (5%), stage IIIC in 1 (3%), and stage IV in 1 (3%). Histologic types of breast cancer included infiltrating ductal carcinoma (n = 27 [71%]), lobular carcinoma with infiltrating ductal carcinoma or ductal carcinoma in situ (n = 7 [18%]), lobular carcinoma not otherwise specified (n = 2 [5%]), mucinous adenocarcinoma (n = 1 [3%]), and carcinoma not otherwise specified (n = 1 [3%]).
Hepatic Lesions Too Small to Characterize
One hepatic lesion too small to characterize was reported on CT in 15 (39%)
of the women and more than one in 23 (61%). The specific diagnoses of hepatic
lesions too small to characterize reported on MRI included cyst (n =
19 women) (Figs. 1A,
1B,
2A, and
2B), hemangioma (n = 5
women) (Figs. 2A and
2B), and metastasis (n
= 2 women) (Figs. 3A and
3B). In 11 (30%) of the women,
the hepatic lesions too small to characterize on CT remained of indeterminate
cause on MRI. In eight (21%) of the women, no lesion was found on MRI at the
site corresponding to at least one of the hepatic lesions too small to
characterize on CT. Of these eight women, three also had a cyst, hemangioma,
or indeterminate lesion on MRI corresponding to a hepatic lesion too small to
characterize on CT, and five had no hepatic lesion on MRI corresponding to any
of the hepatic lesions found on CT. Four patients had more than one type of
hepatic lesion on MRI.
|
|
|
|
|
|
Extrahepatic Metastasis
Extrahepatic metastatic lesions were present in 19 (50%) of the women.
Eighteen of these women had axillary lymph node metastasis, and one had chest
wall metastasis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In addition to metastasis, diagnostic considerations for hepatic lesions too small to characterize on CT include cyst, hemangioma, and biliary hamartoma. Biliary hamartoma is a focal, disorderly collection of bile ducts. It is less frequently considered a diagnostic possibility on CT than are cyst and hemangioma but was reported in 2.8% of 707 consecutive autopsies [11]. Biliary hamartoma typically manifests as lesions smaller than 1 cm on CT that simulate cysts or metastatic lesions. The lesions are commonly multiple but can be solitary [12]. The appearance of biliary hamartomas on MRI has been described as hypointense on T1-weighted images and hyperintense on T2-weighted images with signal intensity similar to that of splenic tissue [13].
Hepatic lesions too small to characterize on CT may be further evaluated with sonography, dedicated CT, or MRI. Eberhardt et al. [14] reported that 66% of hepatic lesions were identified on focused sonograms performed with reference to a specific lesion on a preceding CT examination, whereas only 32% of hepatic lesions were detected without review of previous CT scans. Eberhardt et al. also found that on focused hepatic sonography, small hepatic lesions in 93% of women can be further characterized as cysts, solid or metastatic lesions, or hemangiomas.
Hollett et al. [15] reported improved detection of malignant hepatic neoplasms measuring 1.5 cm or less in 37% of patients on dual phase (arterial and portal venous phase) helical CT of the liver. Images acquired during the arterial phase of enhancement were more useful in evaluation of patients with hypervascular primary neoplasms. In accordance with the CT protocol for assessment for hepatic metastasis in breast cancer patients used at our institution during this study, one standard phase (i.e., portal venous phase) series was obtained on initial CT. It is possible that delayed imaging on initial CT may have helped to identify some hemangiomas. However, this retrospective study was not designed to determine the accuracy of CT in the characterization of small hepatic lesions.
In a study in which multiple logistic regression analysis was used to assess the importance of CT features such as lesion shape, margin sharpness, homogeneity, attenuation, and size among benign and malignant small hepatic lesions, the investigators [3] found that well-circumscribed lesions 5 mm or less in diameter were benign in 96% of cases. At our institution at the time of this study, the standard section thickness used for viewing images on a PACS workstation in a routine protocol to evaluate the extent of disease was 7.0 or 7.5 mm. Given that lesions measuring less than one half of the diameter of the section thickness are affected by partial volume averaging, lesions up to 1.5 cm in diameter may be difficult or impossible to characterize at 7.5-mm section thickness. In our study, thinner sections may have been useful on initial CT for evaluating some of the larger hepatic lesions too small to characterize, but because of less partial volume averaging with surrounding liver, thinner sections also probably contain even more hepatic lesions too small to characterize than do thicker sections.
Heavily T2-weighted fast spin-echo images and gadolinium contrast-enhanced MR images often allow characterization of even tiny lesions (e.g., less than 5 mm) as cysts, as has been shown with renal cysts [16]. In one study [5], MRI was shown to be an effective means of characterizing hepatic lesions less than 2 cm deemed indeterminate on CT as benign or malignant, both when interpreted in a routine clinical setting (area under the receiver operating characteristic curve [Az] = 0.94) and when interpreted by two expert reviewers (Az = 0.84 and 0.88). In contrast, our data suggest that MRI is of limited utility for evaluation of hepatic lesions deemed too small to characterize on CT of patients with known breast cancer. Explanations for the different conclusion of our study may be that all our patients had known cancer and that many of the hepatic lesions in our study were tiny, whereas only 27 of 48 patients in the other study had a history of cancer, and only 14 had lesions smaller than 1 cm.
In our study, hepatic lesions too small to characterize on MRI represented benign lesions alone or lesions that remained too small to characterize or of indeterminate cause in the large majority of the women. In only 5% of women with hepatic lesions too small to characterize were the lesions shown to be malignant on MRI. Thus in 95% of the women who underwent MRI evaluation of hepatic lesions too small to characterize on CT, the results of MRI did not alter medical management because therapy is not changed for benign or indeterminate liver lesions. Advances in CT and MRI technology allow use of thinner sections and thereby allow detection of ever smaller hepatic lesions. However, substantial limitations in characterization of such small hepatic lesions persist. The high prevalence of benign lesions in our study is supported by the findings of Khalil et al. [6], who reported that hepatic lesions too small to characterize on initial CT after diagnosis of breast cancer represent benign entities in 92.7-96.9% of women without other definite metastatic lesions of the liver. Similarly, hepatic lesions measuring 1 cm or less on CT were benign in 82% of 2,978 patients with cancer in a study by Schwartz et al. [2]. Jones et al. [1] reported that among 51% of 209 patients with cancer, hepatic lesions measuring less than 1.5 cm on CT were probably benign. The higher prevalence of benign lesions in our study may be related to our use of more advanced CT scanners with resultant differences in contrast timing, motion, and lesion conspicuity.
Noone et al. [17] found that 11 (32%) of 34 patients with newly diagnosed breast cancer who underwent MRI for suspected breast cancer metastasis had only benign liver lesions. Twenty-one other patients had malignant liver lesions: 19 had metastasis from breast cancer, one had metastasis from carcinoid tumor, and one had hepatocellular carcinoma. Those authors concluded that MRI is useful in the evaluation of breast cancer patients with clinically suspected liver metastasis at initial presentation. The much larger percentage of benign lesions in our study is at least partly explained by our inclusion of only hepatic lesions deemed too small to characterize on previous MRI. In the study by Noone et al., metastatic lesions measured up to 13 cm (mean, 1.5 cm). In addition, it is likely that different referring physicians have different thresholds for ordering abdominal MRI in this clinical context.
Hepatic lesions found on imaging to have decreased in size or disappeared may represent treated metastatic rather than benign lesions [3]. In our study, neither of the two women with hepatic lesions too small to characterize not visible on follow-up imaging was undergoing chemotherapy. Treated hepatic metastatic lesions of breast cancer have been reported in some cases to mimic hemangiomas on MRI [18]. One of five women with hemangiomas in our study underwent chemotherapy until 10 weeks before MRI. The hepatic lesion too small to characterize reported as hemangioma on MRI in this patient conceivably could have represented a treated metastatic lesion, further decreasing the utility of MRI in the evaluation of hepatic lesions too small to characterize in this patient population. The other four women with hemangiomas diagnosed on MRI in our study were not undergoing chemotherapy.
This retrospective study had several limitations. Only a small fraction of women with newly diagnosed breast cancer underwent MRI. The result was marked selection bias. However, MRI is typically requested by clinicians for only women believed on clinical grounds to be at moderate or high risk of hepatic metastasis. The small fraction of women who underwent MRI merely reflects our actual clinical practice. The study was not designed to assess the accuracy of CT or MRI in the detection or characterization of hepatic lesions too small to characterize but rather to assess our actual clinical practice at a dedicated oncology center. The sizes of the hepatic lesions too small to characterize were not mentioned in most cases in the official reports. As with all imaging findings, interobserver and intraobserver variability exist for reporting hepatic lesions too small to characterize on CT. This study was not designed to assess that variability. At our institution, we do not routinely monitor each CT scan, nor do we immediately repeat scans of patients with hepatic lesions too small to characterize at imaging. Such maneuvers might have decreased the number of hepatic lesions deemed too small to characterize. No tissue diagnosis was available for the majority of the small hepatic lesions in this study, as this represented actual clinical practice.
Radiologists ideally would be able to characterize every hepatic lesion too small to characterize. Unfortunately, many of these hepatic lesions remain uncharacterized, and the clinician must make decisions on the basis of available information. We undertook this study and another study of 1,012 women with breast cancer who underwent contrast-enhanced CT including the liver [6] to provide clinicians with an informed estimate of the likelihood of malignancy of hepatic lesions too small to characterize on CT and with help in evaluating these lesions. Hepatic lesions too small to characterize often are found on CT in women with newly diagnosed breast cancer. Given that the overwhelming majority of such lesions are benign in women without obvious liver metastasis, immediate further evaluation with MRI offers marginal benefit beyond peace of mind in cases in which a benign diagnosis can be made. Our findings, in conjunction with the high cost and limited availability of MRI and in accordance with the clinical practice guidelines of the National Comprehensive Cancer Network, led us to conclude that MRI is not warranted in the evaluation of hepatic lesions too small to characterize in every woman with newly diagnosed breast cancer.
Acknowledgments
We thank Chinyere Onyebuchi for assistance in data handling and
analysis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
