| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of General Oncologic Surgery (GNM, LLL, LDW) and Division of Radiology (HFM), City of Hope National Medical Center, Duarte, CA 91010.
Correspondence: Address correspondence and reprint requests to: Gary N. Mann, MD, Department of Surgery, Section of Surgical Oncology, University of Washington, 1959 NE Pacific Street, P.O. Box 356410, Seattle, WA 98195; Fax: 206-543-8136; E-mail: gnmann@ u.washington.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: Records and diagnostic images of all patients undergoing enhanced Teslascan MRI (T-MRI) at our institution were reviewed. We assessed the relative sensitivities of contrast-enhanced CT scan (CECT) and T-MRI in detecting lesions, as well as the impact of T-MRI in the decision to operate or not on patients. In those patients taken to surgery, the correlation between T-MRI and intraoperative palpation and intraoperative ultrasound (IOUS) was determined.
Results: Fifty-four patients were noted on CECT to have focal liver lesions and subsequently underwent imaging with T-MRI. The T-MRI correlated with CT findings in 22 patients (41%), upstaged the liver disease in 26, and demonstrated fewer lesions in 6. Only 43 patients were considered operative candidates and T-MRI influenced the operative decision in 32 patients (74%), dissuading operative intervention in 14. In the 25 patients without clear preoperative evidence of unresectability who were taken to the operating room, T-MRI correlated with findings of intraoperative palpation in 19 (76%). In the 20 patients who underwent IOUS, T-MRI correlated with IOUS in 14 patients (70%). IOUS detected an additional nine lesions, all of which were <1 cm. Seventeen patients underwent resection and/or ablation of their liver lesions. Compared with pathology, sensitivities of CECT, T-MRI, and intraoperative evaluation were 61%, 83%, and 93%, respectively. T-MRI failed to predict hepatic-specific unresectability in only one of eight patients, the other seven having extrahepatic disease.
Conclusions: These findings suggest that T-MRI is more sensitive than CECT in the preoperative predicting of the resectability of hepatic lesions. Despite T-MRI accurately correlating with intraoperative surgical findings, IOUS should be performed on all patients prior to a final decision to resect or ablate a focal liver lesion.
Key Words: Magnetic resonance imaging • Mangafodipir trisodium • Liver resection • Hepatic imaging
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Of the estimated 150,000 new colon cancer cases annually, up to one-third will present with or develop liver metastases. Of these, 25%, or approximately 12,000 patients, will be potential candidates for surgical therapy. In those patients treated successfully, 5-year survival rates of 35%–40% have been consistently reported.1–3 The prognosis of primary liver tumors is even worse. In selected series, only 10%–19% of patients evaluated are potential surgical candidates. However, in those patients resected, 5-year survival rates of up to 40% are attainable.4–6
In recent years, there has been an increasing trend in the development and use of noninvasive and minimally invasive methods to identify the patient population best suited for surgical treatment of their liver malignancies.7,8 Currently, contrast-enhanced CT scan (CECT) is the most widely used imaging modality to assess focal liver disease.9 In all but the most experienced hands, the overall reported sensitivity of CECT is 60%–90%, with reduced sensitivity for lesions smaller than 1 cm.10,11 Although CT with arterial portography (CTAP) is more sensitive, it is also more invasive, costly, and less specific, thus potentially dissuading operative treatment of patients who may be resection candidates.12–14 One of the results of this is that, even in the best of hands, only 50%–75% of patients brought to the operating room are ultimately resected,15 subjecting a considerable number of patients to nontherapeutic explorations. More sensitive and specific preoperative, noninvasive, and cost-effective hepatic imaging modalities are necessary.
Magnetic resonance imaging (MRI) has been extensively evaluated over the past decade.16 With improving technologies and experience with the technique, its sensitivity has been reported to be over 90%, with lowering of the lesion size detection threshold.17,18 This is particularly true for imaging using MR specific contrast agents.19 Most experience has been obtained with the iron oxide reticuloendothelial agents, particularly ferumoxides.20–22 While sensitive in liver lesion detection, this agent has been reported to cause untoward side effects, and requires a 30-minute infusion with a filter for use. More recently, the use of a hepatocellular MR manganese-based contrast agent, Mangafodipir Trisodium (Mn-DPDP, Teslascan; Nycomed Amersham Imaging, Little Chalfont, UK), has been reported. Its advantages are fewer side effects, ease of use, high sensitivity, and potential for improved lesion characterization.23–25 The use of this potentially useful agent is not currently widespread, and there are no reports to date of its use in the preoperative imaging setting. The goal of this study is to review our experience with this agent, compare it to CECT, determine its influence on surgical decision making, and perform a cost analysis with its use.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Diagnostic Imaging
MR examinations of the abdomen were acquired with a 1.5 T superconducting magnet (GE Signa System; GE Medical Systems, Milwaukee, WI) utilizing a standard body or phased-array torso receiver coil. Precontrast scans included T1 (TR 300–600, TE-minimum, 4 NEX) axial and conventional spin echo T2-weighted (TR 2000–3000 msec, TE 80/100 msec, 2 NEX) or FSE T2-weighted (TR 10000 msec, TE 78/152 eff msec, 8–16 ETL, 2–3 NEX) images. Images were obtained in axial and coronal planes (FO 32–40 cm2, 256 x 192 matrix, slice thickness 7–10 mm, 1–2 mm interslice gap).
Postcontrast MRI was accomplished in the following fashion: 3-D gradient echo T1 weighted (TR 6.1 msec, TE 1.1 msec) breath-hold images of the liver were obtained following intravenous administration of 0.l cc/kg Teslascan hepatobiliary MR contrast (Nycomed Amersham Imaging) in axial and coronal planes with 5-mm slice thickness (0.5–1 NEX, 512 x 128 matrix). This agent is selectively taken up by normal hepatocytes, shortening the T1 of liver and rendering normal hepatic tissue bright on T1-weighted imaging sequences. Fat saturation was utilized on 3-D gradient echo and FSE sequences. First order flow compensation and respiratory triggering were implemented on FSE images.
Abdominal CT imaging was accomplished within 30 days of MRI. Seven to ten mm thick axial CT scans were obtained with incremental or helical technique. All patients evaluated for surgical resection were scanned with and without iodinated intravenous contraSt. A minority of noncontrasted outside CT scans were not repeated when all clinically relevant information was obtained with the MRI.
Data Analysis
Patients undergoing both CECT and T-MRI were compared with respect to ability of the specific modality to detect and characterize focal liver lesions. The same reviewers read CT and MRI scans separately, results then were recorded and compared. The number of lesions detected by T-MRI were compared with CECT and, when available, intraoperative findings and IOUS. Matched pair analysis was carried out utilizing the paired t test, performed with JMP software, version 4 (SAS Institute, Cary, NC). A P value < .05 was considered significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Comparison of T-MRI, CECT, Intraoperative Findings, IOUS
All 54 patients were imaged with both CT and T-MRI (Table 1, Fig. 1). Most of these patients, and all of those evaluated for surgical resection, were imaged with CECT. These images were reviewed and compared with each other and, where appropriate, to the findings at the time of surgery. CT was able to detect a mean of 2.7 lesions per patient, significantly fewer than the 4.4 detected by T-MRI (P < .0001). The number of lesions detected by CT and T-MRI correlated in 22 patients; T-MRI detected more lesions in 27 patients and fewer lesions in 5 patients.
|
|
Of the 25 patients explored, 20 underwent IOUS and were analyzed separately. T-MRI detected a mean of 2.7 lesions per patient, and IOUS detected 3.2 lesions per patient (P < .02). IOUS detected a total of nine additional lesions in six patients, all of which were <1 cm. Of these 20 patients, 17 had their lesions resected and/or ablated. Pathological confirmation of lesions in the 17 patients was used as a gold standard to compare sensitivity and the positive predictive value (PPV) of lesion detection modalities. CECT had a 61% sensitivity and 96% PPV in lesion detection. The respective results for T-MRI were 83% and 97% and combined intraoperative evaluation (palpation and IOUS) 93% and 95%. Thus pathology confirmed 37% more lesions than CECT, but only 15% more lesions than T-MRI. The additional information obtained with IOUS influenced the surgeon’s decision not to resected/ablate the liver tumor(s) in only one patient. Two other patients had extrahepatic disease documented by biopsy. A fourth patient, with a pancreatic islet cell tumor, had metastases excluded by IOUS-guided biopsy of the suspicious lesion.
Influence of Preoperative T-MRI on Patient Management
Figures 2 and 3 depict the influence of preoperative T-MRI on patient management. T-MRI was obtained either to evaluate the extent of liver disease prior to considering patients for surgery (43 patients) or to evaluate the response of malignancy to chemotherapy (11 patients). Of the 43 patients evaluated by surgeons, T-MRI influenced the operative decision in 32 (74%): helping to dissuade surgical intervention in 14 patients (33%) and to confirm resectability in 18 (42%). T-MRI helped to avoid celiotomy in the 14 patients by demonstrating unresectable liver lesions not fully appreciated on CECT in 5 patients, characterizing the lesions as benign in 8 patients (focal nodular hyperplasia, hepatic adenoma, or hemangioma), and failing to demonstrate a lesion seen on CECT in 1 patient (false-positive CT finding). Follow-up of the latter nine patients confirmed lack of lesion progression.
|
|
Thus, of the 25 patients explored, 17 patients had their liver lesions resected or ablated, giving a resectability rate of 68%. Of the nine patients who were not resected, T-MRI failed to predict hepatic-specific resectability in only one, this patient having more extensive disease recognized intraoperatively. The patient with an islet cell tumor and focal lesions found on preoperative imaging had IOUS-guided biopsy failing to confirm a neoplasm and represents the only false-positive finding.
Cost Analysis
In our institution, the charge for a CECT is $1,470, and for a T-MRI is $1,420. If the patient undergoes a nontherapeutic exploration with a 2- to 3-day hospitalization, charges are in the order of $15,000. Should the patient undergo resection or ablation of their liver lesions, charges for a single-day ICU stay and a total of 7 to 8 days of hospitalization would be in the range of $40,000.
Given that all 43 patients considered for surgery were imaged with both modalities, total costs for these tests were approximately $122,120 or $61,060 more than the cost of undergoing a CECT alone. The cost savings for the 14 patients alone who were spared an unnecessary, nontherapeutic operation by T-MRI amounted to $210,000. Thus the total savings, in this analysis, are at least $148,940 or $2,758 per patient overall. Conversely, if T-MRI were to replace CECT altogether, total savings in patients evaluated for surgery would be $63,210 and could potentially improve savings to $3,928 per patient overall.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Transcutaneous ultrasound is noninvasive and inexpensive. It is operator dependent and, even in experienced hands, has an overall sensitivity of 53%, which decreases to 20% for lesions <1 cm.10 It has been replaced by CT as the most widely used initial modality for evaluating solid liver lesions. CT is cost effective, and its accuracy has improved with the advent of newer technologies such as helical CECT.9 CECT has a reported sensitivity of up to 91% for lesions larger than 1 cm, which decreases to 56% for lesions <1 cm.11 CT portography (CTAP) has been used to improve sensitivity in detection of liver tumors, particularly those <1 cm. Studies have shown that CTAP is able to detect 91%– 94% of all liver metastases and 71% of lesions <1 cm.14,26 Sensitivities are lower for primary malignant liver neoplasms, 72% overall, and, in one study, 25% for lesions <2 cm.13 The increased expense, invasiveness, and comorbidity, as well as a false-positive rate of up to 15%,12 have reduced the selection of CTAP as a preoperative examination.
MR imaging has assumed an increasingly important role in the evaluation of the liver over the past decade. Improved technology, experience with the technique, and the multiple pulse sequences now available are partly responsible for this.16 It is noninvasive, costs have decreased, and it is less expensive than CTAP. It is able to provide precise anatomic information, as well as visualization of the tumor’s relation to vascular structures and the biliary tree. Using the pulse sequences available, one is able to comprehensively examine the liver and provide better lesion characterization than available with CT technology and even replace traditional tests such as tagged red cell scanning in the diagnosis of hemangioma.8 Several liver-directed MR imaging contrast agents have been developed.19 Their evaluation requires consideration of both lesion detection and lesion characterization. These contrast agents fall into two major groups: reticuloendothelial-specific and hepatocyte-specific.
Superparamagnetic iron oxide (SPIO) particles, such as ferumoxides, are cleared from the blood through phagocytosis by the reticuloendothelial system. Kupffer cells in the liver are their primary target. Uptake of SPIO causes signal loss on T2-weighted images. Since most focal lesions do not contain Kupffer cells, SPIO agents act as positive contrast agents, with lesions not taking up SPIO maintaining their signal and surrounding normal liver darkening. This results in bright lesions and increases their conspicuity.
Several studies have evaluated SPIO-enhanced MRI for evaluating focal liver masses. Using biopsy, autopsy, or follow up as a standard, an early study found SPIO-enhanced MRI to have a 96% overall sensitivity and a 89% sensitivity for lesions <1 cm. By comparison, CECT and unenhanced MRI had overall sensitivities of 60% and 77%, respectively.20 Another study found SPIO-enhanced MRI to have a 99% sensitivity in the preoperative setting, but only 56% when compared with the gold standard of IOUS and pathology.17 A prospective multicenter study with 208 patients confirmed the findings from these smaller studies. Compared with unenhanced MRI and CECT, additional lesions were noted in 27% and 40% of patients, respectively.21 In addition, SPIO-enhanced MRI has been noted to be as least as accurate as CTAP for the detection of hepatic metastases and more specific in its ability to differentiate metastases from cysts.18 Given these advantages, a recent retrospective review analyzed the impact of SPIO-enhanced MRI in patient care. MRI was found to have a sensitivity of 86% when compared with the standard of IOUS and pathology. Importantly, MRI altered clinical management in 67% of patients, contributing to a net cost savings of $1,901 per patient.22 Although hypotension occurred 1%–2% of the time in early reports, 19 the current technique of infusion over 30 minutes through a filter has abrogated this problem. However, the 5%–15% incidence of other side effects such as lumbar back pain and flushing17,21 are problematic to patients.
The hepatocellular contrast agent, Mn-DPDP, undergoes specialized intracellular uptake by hepatocytes, a function lacking in other tissues such as metastases. On T1-weighted images, normal liver shows an increased signal intensity, which is lacking in metastases. This increases lesion-to-liver contrast and subsequent conspicuity of focal liver masses. Advantages of Mn-DPDP include an excellent safety profile, with few reports of serious side effects, and ease of use, being administered as a bolus injection without the need for infusion or filters.19 This makes its use more convenient and increases patient throughput. Importantly, as a hepatocellular agent, it is able to characterize lesions to a greater degree than SPIO agents. Well-differentiated hepatocellular carcinoma (HCC) and hepatic adenomas actually take up the contrast agent and appear brighter on the T1 images than metastases or poorly differentiated HCC.23,25 These latter lesions appear dark, surrounded by the postcontrast enhanced normal liver parenchyma. Sometimes an Mn-DPDP-enhanced rim is noted around the liver tumors, due to surrounding parenchymal compression, bile duct proliferation, and malignant infiltration.27 This rim, together with the marked improvement in lesion conspicuity, contributes to increased ease of lesion detection. Although not yet the subject of a study, it is our opinion that use of this agent could improve detection of lesions in nonspecialized centers, such as community hospitals, which have access to an appropriate MRI and where much patient care is delivered.
In this study, T-MRI detected significantly more lesions than CECT (4.4 vs. 2.7; P < .0001). A confounding issue with this assessment was the fact that many patients were referred for evaluation of liver lesions found on CT scan done at other institutions. These CT scans were not always repeated and there is some minor lack in uniformity of CECT technique. Another weakness with this study, and indeed with all studies that compare imaging modalities, is the lack of definitive identification of lesions seen radiologically. This would necessitate surgical exploration and pathological confirmation in all patients. Indeed, to identify all lesions would require liver explant and serial sectioning, something obviously not possible. Calculations of specificity, accuracy, and negative predictive value are not possible. Our results do confirm the increased sensitivity of T-MRI over CECT, but even further improved lesion detection with intraoperative evaluation.
Our study is the first, to our knowledge, to report on the use of T-MRI in the preoperative setting. Of the 43 patients referred for surgical consultation, T-MRI influenced management in 74%. Most importantly, it persuaded the surgeon not to operate on one-third of the patients, either because of more extensive liver disease than noted on CECT or because the T-MRI was able to characterize the lesion as benign. These benign lesions were confirmed with reimaging in 3 to 6 months, thus avoiding biopsy or surgery. In the 25 patients explored, T-MRI compared favorably with careful surgical palpation. Similar to previous studies, IOUS was found to be more sensitive than T-MRI, detecting additional lesions in 30% of patients.28,29 Despite this, the management of these patients was seldom altered intraoperatively because of unanticipated hepatic disease. Rather, most patients not resected were done so because of extrahepatic disease. All told, only one patient did not undergo resection/ablation because of more extensive isolated liver disease recognized during surgical palpation.
In today’s health care environment, the expense of an additional study as well as its impact on patient morbidity always needs evaluation. Avoidance of an unnecessary celiotomy in 14 patients represented a savings of $2,758 per patient for all patients imaged with T-MRI. Other obvious benefits in these patients, although not formally analyzed or measured, were the reduction in patient discomfort and the reduction in potentially costly complications associated with operative procedures. In a prospective study evaluating hepatic resection for metastatic colorectal carcinoma, Steele et al.30 noted that the complication rate for exploration alone was 5%. Taken together, T-MRI is an efficient and cost-effective imaging modality.
Several additional limitations are evident with this study and T-MRI in general. The current study’s limited numbers, retrospective nature, and subjective patient selection could introduce bias and requires validation in a prospective manner. Mn-DPDP T-MRI is an excellent modality for imaging the liver, as well as the biliary tree, because it excreted in the bile. It is less well suited, however, for imaging the remainder of the abdomen, important in the assessment and staging of locoregional disease and/or its recurrence. In this regard it does not replace CECT. Recently, Low et al.31 reported the use of state-of-the-art gadolinium-enhanced MR imaging in screening for extrahepatic tumor in patients with malignancy. Of 154 surgically confirmed extrahepatic tumor sites, MRI depicted 139 while helical CT only depicted 101, a statistical difference. This has prompted these authors to utilize MRI as the primary modality for detecting tumor within the abdomen. It is likely that both mangafodipir and gadolinium contrast could be used to accurately define the extent of liver and extrahepatic disease in the same patient. In fact, we have used this strategy in selected patients, proving its technical feasibility. Too few patients have been imaged in this manner to draw conclusions at this point in time.
Use of T-MRI preoperatively is promising, both in patient selection and treatment planning. It replaces the need for CTAP and, if combined with gadolinium enhanced MRI, may ultimately obviate the need for CECT. Patients explored for liver resection/ablation should still undergo IOUS prior to a final decision to proceed. Evidence from this and other studies indicate that newer imaging modalities alter the clinical management of patients significantly, improve cost effectiveness, and help to avoid unnecessary invasive diagnostic and therapeutic maneuvers. Further prospective comparative studies are required to more formally define a role for T-MRI and new state-of-the-art imaging modalities, such as 2-fluoro-2-deoxyglucose positron emission tomography (FDG-PET), in evaluation and treatment selection of patients with focal hepatic masses.
|
Footnotes |
|---|
Received for publication October 30, 2000. Accepted for publication April 24, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
