| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ultrasonography |
1 From the Departments of Radiology (D.V.S., S.P.K., S.M.H., M.J.O., E.F.H., S.S., P.R.M.) and Surgery (K.K.T.), Massachusetts General Hospital, White 270, 55 Fruit St, Boston, MA 02114. From the 2001 RSNA scientific assembly. Received June 6, 2003; revision requested July 17; final revision received January 30, 2004; accepted February 17. Address correspondence to D.V.S. (e-mail: dsahani@partners.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: A radiologist (D.V.S.) and a surgeon (K.K.T.) retrospectively identified 79 patients (36 female and 43 male patients; age range, 10–78 years; mean age, 57 years) who had undergone surgical resection for primary liver tumor or metastasis and had also undergone preoperative contrast-enhanced MR imaging within 6 weeks before surgery. MR imaging was performed with a 1.5-T system. Dedicated intraoperative US of the liver was performed or supervised by a gastrointestinal radiologist using a 7.5-MHz linear-array transducer, after adequate hepatic mobilization by the surgeon. Histopathologic evaluation of the 159 resected hepatic lesions served as the reference standard. The lesion distribution included colon cancer metastasis (n = 122), hepatocellular carcinoma (n = 23), cholangiocarcinoma (n = 6), cavernous hemangioma (n = 4), focal nodular hyperplasia (n = 2), hamartoma (n = 1), and metastatic embryonal sarcoma (n = 1).
RESULTS: Of 159 lesions, 138 (86.7%) were identified at both MR imaging and intraoperative US. Twelve additional lesions (7.5%) in 10 patients were detected only at intraoperative US (eight metastases, one hepatocellular carcinoma, one cholangiocarcinoma, one hemangioma, and one biliary hamartoma). Both modalities failed to depict nine lesions (5.6%) (four metastases, four hepatocellular carcinomas, and one cholangiocarcinoma). The sensitivities of MR imaging and intraoperative US for liver lesion depiction were 86.7% and 94.3%, respectively. Surgical management was altered on the basis of the intraoperative US findings in only three of 10 patients (4%).
CONCLUSION: Contrast-enhanced MR imaging is as sensitive as intraoperative US in depicting liver lesions before hepatic resection.
© RSNA, 2004
Index terms: Liver, surgery • Liver neoplasms, 76.314, 76.323, 76.33 • Liver neoplasms, MR, 76.1214, 76.121412, 761.121416, 761.12143 • Liver neoplasms, US, 76.12981 • Magnetic resonance (MR), comparative studies, 76.121411, 76.121412, 76.12143 • Ultrasound (US), comparative studies, 76.12981 • Ultrasound (US), intraoperative, 76.12982
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Intraoperative ultrasonography (US) is a useful adjunct during surgery for identifying liver lesions (2). Several studies have shown that intraoperative US often reveals important information not seen at preoperative imaging and that these additional, unsuspected findings change surgical planning in up to 51% of patients (3–5). As a result, intraoperative US is now used routinely to assist in planning for liver resection, mainly to enable detection of additional tumors and evaluation of the relationship of tumors to major vascular structures.
In the past decade, advances in cross-sectional imaging techniques have greatly improved preoperative patient selection for liver resection. In patients who undergo surgical exploration, the resectability rates for liver metastases from colorectal malignancy have increased from 50% to nearly 80% during this time (5). We hypothesize that the previously reported informational yield of intraoperative US may reflect the less advanced cross-sectional imaging techniques used in the past rather than an inherent benefit of intraoperative US.
Magnetic resonance (MR) imaging with contrast material enhancement is now considered a sensitive preoperative imaging technique for liver lesion depiction. In addition, liver-specific MR imaging contrast agents have shown promising signs of further improving the sensitivity of MR imaging for depicting liver lesions (6,7). We performed a retrospective study to compare intraoperative US and preoperative contrast-enhanced MR imaging for the depiction of liver lesions in patients undergoing hepatic resection.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
From the surgical database, we (D.V.S., K.K.T.) identified 137 patients who had undergone intraoperative US between September 1996 and July 2002. Fifty-eight of 137 patients were excluded for the following reasons: 19 had not undergone MR imaging at our institution, 22 had undergone no preoperative MR imaging, two had technically inadequate MR images, 12 had undergone MR imaging more than 6 weeks before surgery, and three had not undergone liver biopsy or resection. The remaining 79 of 137 patients, who met the study criteria, were included.
The study cohort comprised 36 female patients (age range, 10–75 years; mean age, 55.7 years) and 43 male patients (age range, 31–78 years; mean age, 59.2 years). Overall age range was 10–78 years, and mean age was 57 years. There was no significant difference between the mean ages of male and female patients (t test, P = .73). At pathologic examination, which served as the reference standard for lesion location, number of lesions, and lesion diagnosis, 159 lesions were seen in 79 patients. The lesion distribution included colon cancer metastasis (n = 122), hepatocellular carcinoma (HCC) (n = 23), cholangiocarcinoma (n = 6), cavernous hemangioma (n = 4), focal nodular hyperplasia (n = 2), hamartoma (n = 1), and metastatic embryonal sarcoma (n = 1). Our study was a retrospective review of radiologic and pathologic reports rather than a direct review of images and pathologic specimens. Two authors (D.V.S., S.M.H.) together reviewed all the reports, and the data were compiled by using spreadsheet software (Excel 2000; Microsoft, Redmond, Wash).
MR Imaging
All patients underwent preoperative MR imaging performed with a 1.5-T system (Signa; GE Medical Systems, Milwaukee, Wis) and a phased-array coil. Each study was performed before and after the administration of contrast agents. The following contrast agents were used in our patient population: gadoxetic acid (Eovist; Berlex Laboratories, Montville, NJ) in 21 patients, gadopentetate dimeglumine (Magnevist; Berlex Laboratories) in 18, mangafodipir trisodium (Teslascan; Nycomed, Amersham, NJ) in 35, and superparamagnetic iron oxides (Feridex; Advanced Magnetics, Cambridge, Mass) in five.
Pulse sequences included a nonenhanced breath-hold T1-weighted gradient-echo sequence (150–200/2.1–4.2 [repetition time msec/echo time msec], 80° flip angle, 256 x 160 matrix, 6–9-mm section thickness, and 0–2-mm intersection gap) and a respiratory-triggered fat-saturated T2-weighted fast spin-echo sequence (4,000–6,000/104–108, four signals acquired, 256 x 192 matrix, 8-mm section thickness, and 2-mm intersection gap). Contrast-enhanced T1-weighted breath-hold gradient-echo images were acquired in the transverse plane with and without fat saturation by using the same technical parameters described for the nonenhanced sequence. In patients who received superparamagnetic iron oxides as the contrast agent, contrast-enhanced images were acquired both with the T2-weighted sequence and a T2*-weighted (200/8) sequence. Four radiologists, each with at least 4 years of experience in liver MR imaging, interpreted these studies.
Intraoperative US
At laparotomy, all patients underwent thorough exploration before hepatic resection. The extent of hepatic disease was assessed by means of bimanual palpation, followed by intraoperative US after complete hepatic mobilization. Each intraoperative US examination was performed or supervised by a subspecialty-trained gastrointestinal radiologist (one of four radiologists with at least 5 years of experience in intraoperative US) using one of two 7.5-MHz linear-array transducers (Toshiba Medical Systems, Tochigi-ken, Japan; Logic 700, GE Medical Systems). The liver was evaluated, with knowledge of the MR imaging findings, for the number of lesions, hepatic segmental localization, and the relation of the lesions to the hepatic veins, inferior vena cava (IVC), portal vein branches, and hepatic hilum. The hepatic segments were evaluated sequentially.
Surgery
The surgical procedures were performed by five surgeons, each with a minimum of 5 years of experience in performing liver surgery. The procedures included right hepatectomy, left hepatectomy, segmentectomy, and extended resection. Surgical management was considered modified if the planned resection was changed either to extended or to limited resection or if the surgery was not performed. These modifications were based on the review of records (D.V.S., K.K.T.). If new lesions discovered at intraoperative US were confined to the lobe or hepatic segment designated for resection, the planned surgical procedure was not changed. If new lesions were discovered in a lobe or segment other than that designated for resection, intraoperative biopsy and frozen-section histologic evaluation were performed. Surgical management was modified on the basis of the results of histologic analysis, if necessary.
Statistical Analysis
In the analysis, we used the Couinaud system for segmental division of the liver, which includes eight liver segments. We considered an imaging study or histopathologic analysis to be positive for a given hepatic segment if any lesion was detected within that segment. MR imaging and intraoperative US were considered to have demonstrated a lesion correctly only if its apparent location at imaging correlated precisely with its anatomic location described in the surgical and pathologic reports. On the basis of the available data, we calculated the sensitivity of MR imaging and intraoperative US. (The results of intraoperative US, however, were not directly comparable with those of MR imaging, because intraoperative US was performed with prior knowledge of MR imaging findings.) The 95% confidence interval was calculated for the fraction of patients in whom findings at intraoperative US led to altered management.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
Intraoperative US, however, proved useful to the surgeon in delineating the vascular plane for resection. In one patient with a large mass in the right lobe, the proximity of the mass to the portal vein and the hepatic venous confluence was better appreciated at intraoperative US than at MR imaging, and the clearer US depiction increased the surgeon’s confidence in selecting the proper vascular plane for resection. In three patients with metastases close to the IVC, intraoperative US enabled confident differentiation between extrinsic compression and tumor invasion of the IVC (Fig 2). In another patient with HCC, left portal vein invasion, which was detected at intraoperative US, could not have been predicted with confidence on the basis of preoperative MR images. In eight patients, vascular (venous) involvement was preoperatively identified with MR imaging and further confirmed with intraoperative US, but surgical management was not altered.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In our study, among the 12 additional lesions detected with intraoperative US in 10 patients, 10 lesions were in the left lobe, and seven were smaller than 10 mm. The failure of MR imaging to demonstrate these lesions might be secondary to ghosting artifacts, caused by the heart and aorta, that commonly appear on images of the left liver lobe. This observation suggests a limitation of MR imaging, which should be improved with use of currently available volumetric (three-dimensional) thin-section imaging performed in a single breath hold. In one patient, MR imaging revealed a single 5-cm hepatic mass that was shown at later intraoperative US and histopathologic analysis to represent three adjacent but separate lesions. This apparent error, which lowered the calculated sensitivity of MR imaging, was probably related to the relatively low resolution of the technique employed in our patient series but had no effect on the surgical procedure. This potential limitation may be superseded by further refinement in MR imaging techniques, such as high spatial resolution and thin-section imaging.
In our study, findings at intraoperative US altered surgical management in only three patients, but the identification of hepatic veins was useful for determining the plane of hepatic resection. US also was useful in depicting venous thrombus, showing the relationship between tumor and vessels, and enabling differentiation between extrinsic venous compression and tumor extension into veins. In addition, small simple cysts were confidently diagnosed with intraoperative US. In view of these advantages, we believe that intraoperative US is useful in the surgical planning of hepatic resection, although its use rarely results in the modification of surgical planning because of increased lesion detection. In addition, it is better to screen patients with preoperative MR imaging and exclude those who may not benefit from surgery, before contemplating the use of intraoperative US.
Although intraoperative US after MR imaging depicted additional lesions, it did not reveal nine (5.6%) of 159 lesions. These nine lesions were found at histopathologic analysis in the segments or lobes resected. A few lesions also might have been missed in segments or lobes that were not resected. Such missed lesions are responsible for the lower inherent sensitivity of intraoperative US.
Many studies have shown the superiority of intraoperative US, compared with helical computed tomography (CT) or MR imaging, in the depiction of liver lesions (3,4,9–12). In our study, as well, intraoperative US depicted more lesions than did preoperative MR imaging (94.3% vs 86.7%). Likewise, the results of other investigations have shown that intraoperative US findings modified surgical management in 11%–51% of patients (2–4,9–13). In our study, MR imaging and intraoperative US were discordant for lesion depiction in 10 patients, but surgical management was altered because of additional lesions detected in only three patients (4%). We believe that this disparity between our findings and earlier reports is due to technologic advances in MR imaging and contrast agents. Results similar to ours were reported in the study of Jarnagin et al (14). In their series of 111 patients, hepatic resection either with no change or with slight modification was possible in most of the patients despite new findings at intraoperative US.
The substantial advantages of intraoperative US are identification of hepatic vasculature, segmental localization, and demonstration of the proximity of hepatic lesions to hepatic or portal vessels. Intraoperative US also helps detect occult extension of hepatic tumors into hepatic or portal veins or the IVC (13) and aids in the intraoperative biopsy of liver lesions. In addition, it can help characterize small (subcentimeter) hepatic lesions that could not be characterized at preoperative MR imaging, especially small hepatic cysts (9). The drawbacks of intraoperative US include its lack of specificity, increased surgical procedure time, and cost.
The major reasons for the better sensitivity of MR imaging include the availability of high-field-strength magnets, the use of phased-array coils, inherently high soft-tissue resolution, and multiplanar capability. New techniques have further improved the diagnostic accuracy and sensitivity of MR imaging; these include breath-hold acquisition, fat saturation, three-dimensional imaging, and multiphasic imaging after contrast agent administration.
Liver-specific contrast agents increase the imaging window for evaluation of liver and thus permit thin sections with high spatial resolution to be acquired throughout the liver. Studies have shown that MR imaging with mangafodipir trisodium enhancement was superior to nonenhanced MR imaging and contrast-enhanced CT for the depiction of focal lesions and differentiation of metastases from HCCs (15,16).
Vogl et al (17) showed that contrast-enhanced MR imaging with superparamagnetic iron oxides was equal to CT during arterial portography for the depiction of liver lesions and was more specific but less sensitive than intraoperative US. In a study of MR imaging with gadobenate dimeglumine enhancement, detection of primary hepatic neoplasms improved after the administration of contrast medium (18). A small study in which contrast-enhanced MR imaging with superparamagnetic iron oxides was compared with gadobenate dimeglumine–enhanced MR imaging showed that the former was more sensitive in liver lesion depiction (19). Our aim, however, was not to demonstrate the superiority of any contrast medium or pulse sequence but to compare the overall sensitivity of contrast-enhanced MR imaging with that of intraoperative US.
Our study had several limitations. First, it was a retrospective analysis of the reports of imaging, surgical, and histopathologic findings. Second, intraoperative US was performed with prior knowledge of the MR imaging results, which may have biased the radiologist’s interpretation. In addition, the findings at bimanual examination of the liver also guided the use of intraoperative US for lesion detection. Thus, the true sensitivity of intraoperative US may be slightly lower than reported here. Additionally, intraoperative US was performed by gastrointestinal radiologists with various levels of expertise, which would affect its sensitivity.
MR imaging was performed with three different contrast agents, although the superiority of none of these agents has been clearly established in the literature. Our inclusion of patients who had undergone MR imaging up to 6 weeks before surgery might have influenced the results of intraoperative US, because a few additional lesions may have developed in the interval period. There was also selection bias, in that only patients considered for surgical resection were included in the study, and patients in whom MR findings indicated that surgery was inappropriate were excluded. Therefore, the true sensitivity of MR imaging versus that of intraoperative US cannot be established.
Finally, clustering of lesions and interdependence may have influenced the statistical analysis. The dependence of the findings in the different segments of the same liver, however, is largely a distributional concern affecting P values and confidence intervals. If the distribution of MR imaging findings among liver segments in the same patient differs from that of intraoperative US findings, the sensitivities and specificities that we calculated may be additionally biased. We consider this effect to be minor, however, compared with that of the lack of blinding in the intraoperative US interpretations.
In conclusion, contrast-enhanced MR imaging is as sensitive as intraoperative US in depicting liver lesions before hepatic resection. Intraoperative US continues to be an excellent imaging tool for the planning of liver resection for neoplasms, because it delineates the hepatic vasculature and its relation to the neoplasms. With further advances in preoperative MR imaging techniques, intraoperative US would probably depict additional lesions less often and would rarely change the planned surgical procedure.
|
FOOTNOTES |
|---|
Abbreviations: HCC = hepatocellular carcinoma, IVC = inferior vena cava
Author contributions: Guarantors of integrity of entire study, D.V.S., K.K.T., S.S.; study concepts, D.V.S., M.J.O.; study design, D.V.S., M.J.O., P.R.M.; literature research, D.V.S., S.P.K., S.M.H.; clinical studies, D.V.S., K.K.T., S.M.H., S.P.K.; data acquisition and analysis/interpretation, S.M.H., D.V.S., S.P.K.; statistical analysis, E.F.H.; manuscript preparation and revision/review, D.V.S., S.P.K.; manuscript editing, M.J.O., P.R.M.; manuscript definition of intellectual content and final version approval, D.V.S.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. Torzilli, A. Palmisano, D. Del Fabbro, M. Marconi, M. Donadon, A. Spinelli, P. P. Bianchi, and M. Montorsi Contrast-Enhanced Intraoperative Ultrasonography During Surgery for Hepatocellular Carcinoma in Liver Cirrhosis: Is It Useful or Useless? A Prospective Cohort Study of Our Experience Ann. Surg. Oncol., April 1, 2007; 14(4): 1347 - 1355. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Minagawa, J. Yamamoto, T. Kosuge, Y. Matsuyama, S.-i. Miyagawa, and M. Makuuchi Simplified Staging System for Predicting the Prognosis of Patients With Resectable Liver Metastasis: Development and Validation Arch Surg, March 1, 2007; 142(3): 269 - 276. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Benoist, A. Brouquet, C. Penna, C. Julie, M. El Hajjam, S. Chagnon, E. Mitry, P. Rougier, and B. Nordlinger Complete Response of Colorectal Liver Metastases After Chemotherapy: Does It Mean Cure? J. Clin. Oncol., August 20, 2006; 24(24): 3939 - 3945. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. B. Kruskal and R. A. Kane Intraoperative US of the Liver: Techniques and Clinical Applications. RadioGraphics, July 1, 2006; 26(4): 1067 - 1084. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
