| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, Mayo Clinic and Mayo Foundation, 200 First St. S.W.,
Rochester, MN 55905.
2 Departmemt of Surgery, Mayo Clinic and Mayo Foundation, Rochester, MN
55905.
3 Department of Endocrinology, Mayo Clinic and Mayo Foundation, Rochester, MN
55905.
Received February 10, 2003;
accepted after revision April 1, 2003.
Presented at the annual meeting of the American Roentgen Ray Society, San
Diego, May 2003.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Prospective interpretations of multiphase helical CT scans were reviewed in 30 patients who had insulinomas resected over a 5-year period. CT scans were retrospectively reviewed to determine enhancement characteristics, tumor conspicuity in each phase of enhancement, and potential causes for false-negative findings.
RESULTS. Sixty-three percent (19/30) of tumors were identified on CT prospectively. An additional six tumors were visualized in retrospect, allowing characterization of 25 (83%) of 30 tumors. Most tumors were hyperdense on at least one phase (n = 19), three tumors were hypoattenuating, and three were isodense and pedunculated. Insulinomas were most conspicuous on the early phase in 15 patients and in the portal venous phase in three. All tumors that underwent pancreatic phase imaging were seen (13/13), whereas three of 18 arterial and six of 25 portal venous phase findings were inconclusive for tumor. In the six examinations with false-negative findings in which the tumor could be seen in retrospect, two tumors were isodense and pedunculated, three were in close proximity to vessels, and one had a cystic appearance.
CONCLUSION. Multiphasic CT has a moderate sensitivity in the detection of insulinomas. Most tumors are more conspicuous on the earlier phases of enhancement. The pancreatic phase may be more useful than the arterial phase. Potential sources of false-negative results include tumors adjacent to vessels, pedunculated morphology, or nonhyperattenuating lesions.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Other experts believe that preoperative imaging can be valuable by providing information regarding the type of surgery required. For example, insulinomas located near the surface of the gland may undergo simple enucleation; however, if the tumor is located deep in the gland, partial pancreatectomy may be required. Identification of a focal mass can help exclude nesidioblastosis as a cause of the hypoglycemia, which might require an extensive resection. Nesidioblastosis, or islet cell hyperplasia, is characterized by a proliferation of abnormal B cells throughout the pancreas. Finally, preoperative localization can reduce the operative palpation that is necessary [3, 5, 6]. Thus, if the sensitivity of noninvasive preoperative imaging improves, dedicated pancreatic imaging may be more widely used in preoperative evaluation algorithms.
Technical advances have led to an improvement in the quality of CT, and recent reports have suggested improved performance for the detection of islet cell tumors [7-11].
However, further studies are necessary to validate these promising results. In addition, disagreement exists on the appropriate CT protocol. With the development of helical CT and, more recently, multidetector scanners, we are now able to rapidly image the pancreas during multiple phases of contrast enhancement. Some investigators have shown improved conspicuity of islet cell tumors on the arterial phase of enhancement, whereas others have shown the portal venous phase to be more beneficial [7-9, 12]. The pancreatic phase of enhancement has been shown to provide maximal tumor conspicuity for adenocarcinoma of the pancreas, but the utility of this phase has not been assessed for the detection of insulinomas [13].
The aims of this study are to review our experience in the preoperative detection of insulinomas by using multiphase helical CT technology, determine the enhancement characteristics and relative conspicuity of insulinomas on the different phases of pancreatic enhancement, and analyze the cause of false-negative findings on CT.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CT
All CT scans were performed using helical scanners (General Electric
Medical Systems, Milwaukee, WI). Eighteen patients were scanned on
single-detector helical CT scanners (HiSpeed Advantage or CTi, General
Electric Medical Systems). Twelve patients were scanned on multidetector
scanners (LightSpeed or LightSpeed Plus, General Electric Medical Systems).
All CT was performed as dedicated multiphase examinations of the pancreas.
Twenty-three examinations were biphasic and seven were triphasic. Slice
thickness varied from 2.5 to 5 mm. Because of the lack of documentation,
contrast injection rates and scan-delay times could not be determined in all
cases. Our standard multidetector triphasic pancreatic protocol is performed
with a slice thickness of 1.25 mm through the pancreas at 20 sec after the
initiation of contrast injection (arterial phase) followed by 2.5-mm-thick
slices at 40 (pancreatic phase) and 65 sec (portal venous phase). Biphasic
examinations are performed with scan delays of 40 and 65 sec. One hundred
forty milliliters of contrast material is injected at a rate of 4mL/sec.
Table 1 summarizes the
techniques that were used for the examinations in this study.
|
Analysis
All preoperative imaging reports were reviewed to determine the prospective
sensitivity of each imaging technique. The radiologist interpreting the
examination was not blinded prospectively to any of the other imaging results.
These included sonography (n = 28), endoscopic sonography (n
= 5), angiography with arterial stimulation and venous sampling (n =
8), and intraoperative sonography (n = 16). CT scans were then
retrospectively reviewed by two of three radiologists in consensus. These
radiologists were aware of surgical reports, so that tumor presence and
location were correlated with imaging findings in each case. Radiologists
assessed for phases of enhancement acquired, tumor presence and location,
relative tumor conspicuity in each phase, tumor appearance (solid vs cystic,
intrapancreatic vs pedunculated), and enhancement characteristics.
Examinations were reviewed to determine the number of phases of enhancement that were performed. The phases were classified as arterial, pancreatic, or portal venous. The early phases, arterial and pancreatic, were differentiated on the basis of degree of enhancement of the superior mesenteric vein. The phase was considered to be arterial if there was opacification of the arteries without opacification of the superior mesenteric vein. The phase was considered pancreatic if there was opacification of the superior mesenteric vein. The last phase with opacification of the portal and hepatic veins was considered portal venous phase.
The conspicuity of the tumor on each phase was determined. Each phase was reviewed to determine if the tumor could be conclusively visualized. The phase in which the tumor had the greatest conspicuity was noted. Enhancement of the tumor on each phase was assessed and subjectively classified as hypo-, hyper-, or isoattenuating compared with the surrounding pancreas. Tumors were noted to have either homogeneous or heterogeneous internal attenuation.
The original prospective interpretations of the radiologists in our department were compared with our consensus retrospective review to determine potential causes of false-negative examinations.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Tumor Detection
The prospective sensitivity of the findings of 21 radiologists who
interpreted the multiphase CT examinations was 63% (19/30). Six additional
insulinomas were detected on CT by our retrospective consensus reading, which
also identified all tumors detected prospectively. A tiny 4-mm hyperenhancing
insulinoma was equivocally visualized, but a consensus could not be reached.
Assessment of imaging characteristics was subsequently performed.
Enhancement Characteristics
Most tumors were hyperenhancing compared with the pancreatic parenchyma on
at least one of the phases (n = 19), but three tumors were
hypoattenuating on all phases (Figs.
1 and
2A,
2B). Another three tumors were
isodense with the pancreatic parenchyma but could be visualized because of
pedunculated morphology (Figs.
3A, and
3B). Tumors were rated as
homogeneous on 56% (10/18), 85% (11/13), and 80% (20/25) of the arterial,
pancreatic, and portal venous phase images, respectively.
|
|
|
|
|
Tumors were most conspicuous on the early phase (arterial or pancreatic) in 15 patients and in the portal venous phase in three. In seven patients, there was no significant difference in tumor conspicuity among the different phases of enhancement.
Nineteen patients had biphasic examinations. The first phase was considered to be arterial in 12 patients and pancreatic in seven. In the 12 patients with arterial phase acquisitions, eight tumors were most conspicuous during the arterial phase (four of these were not conclusively seen during the portal venous phase), two were most conspicuous during the portal venous phase (one of these was not conclusively seen during the arterial phase), and two tumors were considered to be equally conspicuous. In the seven patients with pancreatic phase acquisitions, four tumors were considered most conspicuous during the pancreatic phase (one of these was not conclusively seen during the portal venous phase), one was most conspicuous during the portal venous phase (also seen on the pancreatic phase), and two were equally conspicuous.
Six patients underwent triphasic examinations. One tumor was most conspicuous on the arterial phase (seen on all phases), and two tumors were most conspicuous on the pancreatic phase (both tumors were not conclusively seen on the arterial phase) (Figs. 4A, and 4B). Three tumors were equally conspicuous on all phases. None of these six tumors was most conspicuous on the portal venous phase.
|
|
The arterial and portal venous phases were thought to be inconclusive for tumor presence in three of 18 and six of 25 cases, respectively. All tumors that underwent pancreatic phase imaging could be identified on the pancreatic phase (13/13).
False-Negative Examinations and Retrospective Tumor Detection
There were 11 examinations in which the prospective interpretation by a
radiologist in our department failed to identify an insulinoma, despite its
presence being documented biochemically. Our retrospective consensus
interpretation could definitely identify the tumor in six of these patients.
Three tumors not identified were in close proximity to a vessel and most
likely misinterpreted as being part of the vessel (Figs.
5A,
5B,
5C and
6A,
6B). These tumors were similar
to the adjacent vessel in size and shape. One tumor had a cystic appearance
and did not appear solid (Fig.
1). This tumor was hypoechoic on intraoperative sonography and was
located at a site of previous exploration and dissection. Perhaps the
hypoenhancement on CT was due to devascularization by the previous
exploration. Two tumors were noted to be pedunculated at surgery. Both of
these tumors were isodense on CT and could be seen in retrospect as focal
protuberant nodules along the pancreatic surface (Figs.
3A, and
3B). Because of the isodensity
to the normal pancreas, these were most likely interpreted as normal
nodularity of the gland.
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our review is one of the largest studies specifically addressing the detection and characterization of insulinomas using multiphasic helical CT technology. The radiologists at our institution could prospectively detect 63% of insulinomas, and 83% of all insulinomas in our study could be seen in retrospect on CT, allowing assessment of enhancement characteristics. Whereas most insulinomas exhibit hyperenhancement on at least one of the phases, some will be hypoenhancing or, rarely, cystic. Tumors may also be pedunculated. Three of the lesions that were missed prospectively had unusual appearances that may account for the initial errors in detection (Figs. 1, 2A, 2B, 3A, 3B) and for the lower prospective sensitivity seen in some of the smaller studies mentioned previously that reported more typical-appearing insulinomas.
Familiarity with the entire spectrum of these atypical appearances should help improve tumor detection. Tumors that lie close to enhancing vessels can potentially be missed and accounted for most of our false-negative interpretations that could be seen in retrospect. This result may be especially true when the tumor is the same size and shape as the adjacent vessel. Close scrutiny of all enhancing structures in the pancreas is critical to help differentiate tumor from vessel (Figs. 5A, 5B, 5C and 6A, 6B). The use of a workstation may be helpful by allowing the radiologist to scroll through the pancreas in axial stack mode and evaluate for contiguity of enhancing structures with vessels. Other investigators have shown coronal reformatted images to be useful in differentiating tumor from vessel. Chung et al. [11] detected five of seven islet cell tumors by reviewing axial images. However when these were combined with multi-planar reformatted images, all seven tumors were depicted.
The development of multidetector CT technology now allows high-resolution images to be obtained during multiple phases of enhancement. Most biphasic studies assessing insulinomas have been performed using arterial and portal venous phase imaging to exploit the hypervascular nature of these tumors. Both phases appear to be complementary. Some investigators have shown the utility of arterial phase images, whereas others have not found significant benefit. King et al. [7] found arterial phase images to be superior to portal venous phase images in a study of six patients with small insulinomas. All six tumors were detected in the arterial phase, and only four were seen during the portal venous phase. The four insulinomas detected on both phases were more conspicuous on the arterial phase.
However, in a retrospective review of 19 patients with 26 islet cell tumors (eight functioning, 11 nonfunctioning), Ichikawa et al. [12] found that the portal venous images identified slightly more lesions than those obtained in the arterial phase. This group of patients only had insulinoma, however. Van Hoe et al. [8] and Keogan et al. [9], respectively, found that only 55% (6/11) and 40% (2/5) of islet cell tumors were hyperenhancing on the arterial phase. Despite this finding, Van Hoe et al. considered both phases to be useful because tumor conspicuity was better in the arterial phase for two patients and in the portal venous phase for another two patients. Therefore, both phases are complementary because some lesions may be more conspicuous on one phase than on the other.
Our results would support the use of multiphasic imaging. Sixty percent (15/25) of tumors were thought to be most conspicuous during the early phases (arterial or pancreatic). Whereas many could be seen on the portal venous phase, this improved conspicuity on the early phases would very likely improve reviewer confidence in clinical practice. In addition, six tumors could not be definitely seen on the portal venous phase.
More recently, pancreatic phase imaging has been performed to detect hypoattenuating adenocarcinoma of the pancreas [13]. However, the utility of the pancreatic phase for insulinoma has not been assessed. Investigators have shown that a late arterial phase may be more helpful in showing hypervascular hepatocellular carcinomas in the liver than an early arterial phase, possibly by allowing slightly more time for the contrast material to stain the tumor [14]. The advantage of this phase may be seen with insulinomas as well. We found three of 18 tumors could not be seen on arterial phase images. However, all the tumors (n = 13) that were examined on a pancreatic phase acquisition could be seen on that phase (Fig. 4B). Further studies will be necessary to assess this issue.
Other investigators have shown good sensitivities with modalities such as MRI and endoscopic sonography. Using a 1.5-T magnet, Thoeni et al. [15] detected 85% of 20 small islet cell tumors, most of which were insulinomas. Catalano et al. [16] detected 92% of 26 insulinomas using a 0.5-T magnet. MRI is not routinely used in our practice unless there is a contraindication to CT. The results obtained with endoscopic sonography for the detection of pancreatic tumors have been good, with sensitivities of greater than 70% in most studies [17-21]. However, results vary depending on the site of the tumor and are operator-dependent. Areas with lower sensitivity include the pancreatic tail, duodenal wall, and pedunculated lesions [17, 19-22]. In our group of patients, endoscopic sonography was reserved for those patients with negative findings on transabdominal sonography or CT or to improve the confidence level in diagnosing questionable tumors. In this study cohort, endoscopic sonography identified four of five insulinomas. There were three patients with negative transabdominal and positive endoscopic sonographic examinations.
Several limitations to our study are noteworthy. Because of the clinical approach used at our institution, there was a selection bias against CT. Our clinicians usually perform transabdominal sonography as the initial examination. If findings on sonography are positive, many patients go directly to surgery without further imaging. For those patients with a negative sonogram, CT is usually subsequently performed. Arterial stimulation and venous sampling are usually reserved for patients with negative sonography and CT. During the time of this review, 53 insulinomas were resected at this institution. Forty-eight patients had sonographic imaging preoperatively. Sonography could identify 27 (56%) of these tumors. In 20 patients, sonography was performed without CT. For those patients undergoing sonography without CT, the mean tumor size was 1.8 cm, and the sensitivity of sonography was 75%. Therefore, the exclusion of these patients may have lowered the sensitivity of CT by selecting out more conspicuous tumors.
Our scanning technique was not standardized among patients. There was a variation in the number of phases and slice thickness acquired and empiric use of scan-delay times, and different radiologists (n = 21) performed the prospective interpretations. Therefore, the results are subject to interobserver variability. However, this variation mirrors our clinical practice because patients are scanned on different CT scanners with different capabilities and interpreted by a spectrum of radiologists. The relatively small number of triphasic examinations limits the ability to compare the sensitivity of the arterial to that of the pancreatic phase for insulinoma detection. Further study is needed in this area.
The retrospective review of the images was not performed in a blinded fashion, nor did a control population assist in estimating specificity. These were not intended aims of the study because we analyzed the localization and enhancement characteristics of insulinomas, given the discrepant reports in the imaging literature. If the images had been reviewed in a blinded fashion, the results for tumor conspicuity and conclusive visualization might have been different.
In summary, multidetector CT technology allows high-resolution multiphasic evaluation of the pancreas for the detection of insulinomas and has promising sensitivities. Meticulous technique and awareness of atypical appearances such as isoattenuating, hypoattenuating, and cystic tumors and careful scrutiny for tumors near vessels are critical to help obtain the highest possible sensitivities. Both pancreatic and arterial phase images are helpful in tumor detection; however, the pancreatic phase may replace the arterial phase in insulinoma protocols. Further studies will be necessary to confirm these preliminary results.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  K. M. Horton, R. H. Hruban, C. Yeo, and E. K. Fishman Multi-Detector Row CT of Pancreatic Islet Cell Tumors. RadioGraphics, March 1, 2006; 26(2): 453 - 464. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Takeshita, K. Kutomi, K. Takada, H. Kohtake, and S. Furui 3D Pancreatic Arteriography with MDCT During Intraarterial Infusion of Contrast Material in the Detection and Localization of Insulinomas Am. J. Roentgenol., March 1, 2005; 184(3): 852 - 854. [Full Text] [PDF]
|
|
|
|
 |
|
 P Manoharan and M B Sheridan Neoplasms of the pancreas Imaging, September 1, 2004; 16(4): 323 - 337. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
