| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Department of Radiology (G.J.C.B., M.B., E.C.M.), Sarcoma Unit (O.A.M., J.M.T., I.R.J.), and Department of Histopathology (C.F.), Royal Marsden Hospital, Fulham Rd, London SW3 6JJ, England. From the 2001 RSNA scientific assembly. Received November 25, 2001; revision requested February 5, 2002; revision received April 26; accepted May 24. Address correspondence to E.C.M. (e-mail: eleanor.moskovic@rmh.nthames.nhs.uk).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: The medical records of all patients at our institution with a histologic diagnosis of GIST were reviewed. Two radiologists with knowledge of the diagnosis reviewed the radiologic findings by means of consensus. Sixty-seven patients underwent computed tomography, and scans of the primary tumor were available in 38 patients.
RESULTS: One hundred sixteen patients with malignant GISTs were identified (76 men and 40 women; mean age, 54.6 years ± 13.5 [SD]). The primary tumor locations in descending order of frequency were the small bowel (n = 49), stomach (n = 43), colon (n = 7), rectum (n = 6), other (n = 3), and not specified (n = 8). Mean primary tumor size was 13 cm ± 6. Tumors were typically well defined (31 of 36 [86%]), with a heterogeneous rim of soft tissue with lower signal intensity than that of the contrast material–enhanced liver. Central fluid attenuation was seen in 24 of 36 (67%) patients. Metastases were seen in 23 of 38 (61%) patients at presentation and in 53 of 61 (87%) patients during follow-up. Spread was usually to the liver or peritoneum. Visceral obstruction rarely occurred, even in the presence of extensive peritoneal metastatic disease. Ascites was an unusual finding.
CONCLUSION: Malignant GISTs are typically large, well-circumscribed, heterogeneous, centrally necrotic tumors that arise in the wall of the small bowel or stomach. They rarely obstruct viscera, despite their large size and propensity to metastasize to the liver and peritoneum.
© RSNA, 2003
Index terms: Gastrointestinal stromal tumor (GIST), 70.322 • Gastrointestinal tract, CT, 70.12114 • Gastrointestinal tract, neoplasms, 70.322 • Sarcoma, 70.322
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Most GISTs (70%–80%) are benign. There is, however, a continuum from benign to malignant that can be predicted, although not absolutely, according to tumor size and mitotic frequency. The mitotic frequency is a measure of the number of mitoses per high-power microscopic field (8). In addition, tumor behavior varies according to the anatomic site of origin. Benign behavior is particularly common in gastric tumors, where benign tumors outnumber malignant ones by a ratio of 3–5:1. Benign tumors are often discovered incidentally at surgery and are completely excised (8). The increasing use of computed tomography (CT) and endoscopy of the upper gastrointestinal tract is also a means for the detection of asymptomatic tumors.
GISTs can originate anywhere along the gastrointestinal tract or beyond it in the mesentery or omentum (8–10). They typically arise in the bowel wall, usually from or between the muscularis propria and muscularis mucosa (9). Articles on the investigation of large series of GISTs have been published in the surgical and pathologic literature, but little emphasis has been placed on the specific radiologic appearances of these tumors. The purpose of the present study was to investigate and describe the anatomic distribution, imaging features, and pattern of metastatic spread of malignant GISTs.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The GIST database at our institution was reviewed. All patient records in the database showed a histologically confirmed diagnosis, some with the use of immunohistochemistry since its introduction. Body CT scans were available for review in 67 patients. The earliest available CT study for review was from a referring hospital and was acquired in 1982. The date range of CT studies reviewed from our institution was April 1993 to August 2001. The mean number of scans per patient was five (range, 2–17). Chest images were available in 60 patients (CT and radiographic, n = 39; CT alone, n = 12; radiographic alone, n = 9). Images of the brain were available in five patients: CT scans in four and magnetic resonance images in one.
Abdominopelvic CT scans acquired at our institution since the installation of a spiral CT scanner in 1996 (HiSpeed CT/I; GE Medical Systems, Milwaukee, Wis) are typically acquired after oral administration of water (500 mL) and intravenous administration of contrast material (100 mL; flow rate, 3 mL/sec) in the portal venous phase with a section thickness of 10 mm and a pitch of 1.5. A number of patients were referred from other institutions with initial CT scans acquired with some variation in technique from our own, predominantly with respect to the use of iodinated orally administered contrast material.
Clinical data were reviewed by three investigators (G.J.C.B. and M.B. separately but in consensus with O.A.M.). The same investigators recorded the site of the primary tumor by means of pathologic reports and surgical notes when possible, and when these were noncontributory, from clinical records and preoperative CT scans if available. When preoperative CT scans were used, the anatomic location was determined by means of consensus (G.J.C.B., E.C.M.).
Two radiologists (G.J.C.B., E.C.M.) retrospectively reviewed the hardcopy CT scans by means of consensus. Findings evaluated for the 38 primary tumors included tumor diameter and whether the tumor was predominantly extra- or endoluminal, defined as whether more or less than half of the tumor lay outside the bowel lumen. The tumor margin was categorized as well defined (ie, a smooth or lobular contour without surface projections), irregular (ie, with surface projections), or clearly invasive (ie, when soft tissue of similar attenuation to the tumor breached an adjacent organ). Vascular encasement was deemed present if there was no interposed fat attenuation and more than 180° contact between tumor and vessel wall. The presence of intratumoral fluid, gas, or calcific attenuation was recorded with reference to, for example, the gallbladder or urinary bladder, gas in the bowel, and bones. The homogeneity of the soft-tissue elements of the tumor (ie, completely uniform or not and tumor attenuation compared with that of the liver) were visually assessed. Evidence of bowel, biliary, or renal obstruction and the presence of ascites were noted. The location of metastases and the number and attenuation of hepatic metastases in those who underwent CT at presentation were recorded.
Follow-up analysis included review of all postoperative or postpresentation CT scans. For patients with both pre- and postoperative scans, local relapse was recorded if tumor recurred at the resection site. For all patients with follow-up images, the time to detection and location of disease relapse was recorded. Ascites and visceral obstruction were searched for on all follow-up CT scans. The imaging features of hepatic metastases that developed during follow-up were recorded from the scan on which they were first apparent.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
Of the 15 patients who underwent CT and had no metastases at presentation, 12 had follow-up images for review. Eleven of these patients developed metastases at a mean of 19 months (range, 4–62 months). The location of metastases was as follows: liver and peritoneum in six patients, liver alone in three, peritoneum alone in one, and peritoneum, retroperitoneum, and pleura in one. The remaining patient was disease free at 24 months of follow-up.
Eight of 19 patients (42%) in whom CT scans were available before and after resection of the primary tumor developed local recurrence, which was usually detected at the time of metastatic involvement.
Sixty-one patients underwent follow-up CT (range of follow-up from date of diagnosis, 1–385 months; mean, 41 months). Fifty-three patients had or eventually developed metastases. In 27 patients, a single site was involved, namely, the liver in 18 and the peritoneum in nine. Most of the remaining patients (23 of 26) had two sites of metastatic disease—once again, usually to the liver and peritoneum. Other locations for metastatic disease included the lungs, pleura, retroperitoneum, bone, and subcutaneous tissues in relation to a laparotomy scar (Fig 5; Table 2). None of the five patients who underwent brain imaging had evidence of intracerebral metastases. At follow-up, fluid-fluid levels were seen in hepatic metastases in four patients. The liver deposits were of low attenuation compared with that of the normal surrounding parenchyma in all but one patient with hyperattenuating metastases at portal venous phase CT. Seven patients developed ascites. Visceral obstruction was distinctly unusual, with just one case of small-bowel obstruction and one case of unilateral renal obstruction at follow-up CT. There were no cases of biliary obstruction, despite 14 patients with peritoneal tumor invading the porta hepatis (Fig 6).
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In the present study, the mean age at presentation was over 50 years, in line with that in other series (3,9,11,12). Male predominance is a feature in most previous studies (3,9,11–13) but is usually less marked than the nearly 2:1 ratio in our cohort.
The stomach and small bowel consistently account for the vast majority of cases of GIST, usually with a division that favors the stomach (3,9,11,14,15). Multiplicity of primary tumors is rare (9,12,16). However, tumors arising outside the bowel wall constitute up to 10% of cases (3). The small bowel rather than the stomach was the most common primary site in our series. A possible explanation for this finding could be the misclassification of a number of mesenteric and omental tumors as being of small-bowel origin, a theory supported by the large proportion of extraluminal tumors and the relative paucity of extraintestinal tumors in our study. As a tertiary cancer care center, our referral pattern is biased toward the more aggressive end of the GIST spectrum. This may provide a second explanation for our higher rate of small-bowel primary tumors over gastric ones. Tumors of small-bowel origin tend to have more aggressive behavior and thus a worse prognosis than that of tumors originating in other gastrointestinal sites (8,14). Such tumors are likely to accumulate in the specialist cancer care unit. In addition to anatomic site, there are a number of other prognosticators. A poor prognosis is also conferred by advanced patient age, large tumor size (>5 cm), irresectability, metastases at presentation, tumor necrosis, and a high mitotic index (3,11,14,15,17). Mitotically inactive tumors can metastasize, however, which indicates that a low mitotic count does not necessarily rule out malignant behavior (8).
Features at clinical presentation are dependent on tumor size. Small tumors are usually an incidental surgical finding, while large tumors are usually symptomatic. Symptoms include gastrointestinal hemorrhage, often with an acute episode, abdominal pain, a mass, weight loss, nausea, and vomiting. Small-bowel obstruction is reported in up to 30% of clinical series but accounts for less than 10% of presentations in most reports (9,11,15,18–20). There was a single case of small-bowel obstruction on follow-up CT scans in our series. Some patients may have experienced small-bowel obstruction that was managed at their local hospital. Furthermore, a large number of the tumors in our series were predominantly extraluminal and would therefore be less likely to result in obstruction.
Biliary and renal obstruction is not mentioned in the above clinical series (9,11,15,18–20), except in the context of duodenal GISTs compromising the common bile duct (11). This concurs with our findings, which also show such obstructions to be rare, even in advanced metastatic disease. Specific tumor characteristics offer an explanation for these findings, since GISTs at gross pathologic evaluation are soft, fleshy, lobulated masses with a heterogeneous internal appearance and areas of necrosis and cystic degeneration (21). Even when the tumor infiltrates surrounding structures, it does so with a smooth, broad pushing border rather than with insinuating fascicles, which once again correlates well with the radiologic patterns observed previously (21). The rarity of ascites suggests that these tumors do not incite a local inflammatory reaction. The occurrence of ascites at follow-up is likely, due at least in part to treatment.
The predominance of large primary tumor size and central necrosis reflects the bias toward malignant GISTs in our series (12). Tumors found incidentally, which have a better prognosis, have a mean diameter of 1.5 cm (9). Although extraluminal expansion is the more common method of primary tumor growth, the striking difference in number of extra- versus endoluminal tumors in our series is indicative of a group of patients presenting with tumors that have a tendency toward late manifestation (19). Endoluminal tumors will result in symptoms sooner, with a greater expectation for curative surgery (19). Our referral pattern also accounts for the high rate of metastases at presentation. However, the incidence of metastases at presentation in the largest clinical series (3) of malignant GISTs approached 50%.
The development of local recurrence and metastases at follow-up is a feature of this disease (3,12,17). The distribution of metastases is predictable, with the liver and peritoneum dominating (3,12). The liver is the most common metastatic site at both presentation and disease relapse (3). Metastases to bone and the lung have been previously described, but they are distinctly uncommon, as was the case in our patients (3,15). The incidence of lymph node metastases is more controversial. We found no cases of enlarged lymph nodes according to CT criteria. A number of other investigators (9,18,20) have also not reported metastatic disease to the lymph nodes. However, pathologic reports indicate lymph node metastases can occur but with insufficient frequency to warrant routine lymphadenectomy (3). Some of the peritoneal deposits in our series may have been mesenteric nodal metastases.
There is agreement that complete surgical excision of the primary tumor offers the best chance of cure (3,9,18). In the absence of metastatic disease, complete resection is usually achievable, since the tumor is typically limited by the serosa of the organ, and when invasive, does so with a pushing rather than infiltrative border, allowing en bloc resection with clear margins (3,18,21). However, the high rate of local and distant recurrence indicates the need for effective nonsurgical treatment. Until recently, chemotherapeutic intervention has offered only disappointing results (3,22,23). A promising new drug, STI571 (imatinib, Glivec; Novartis Pharma, Basel, Switzerland), a tyrosine kinase inhibitor, is currently undergoing clinical trial following encouraging initial reports (24).
The diagnosis of malignant GIST can be suggested on CT scans with the presence of a large well-circumscribed tumor arising from the stomach or small bowel that is usually predominantly extraluminal and has a heterogeneously enhancing soft-tissue rim surrounding a necrotic center. Metastases, if present, will be to the liver or peritoneum. Lymph node enlargement is not a feature. The differentiation from other primary gastrointestinal malignancies can often be made on the basis of these specific findings, since lymphomas tend to cause circumferential mural thickening with homogeneous enhancement and/or lymph node enlargement. Carcinoids are mainly found in the terminal ileum and excite a desmoplastic reaction, while carcinomas are likely to demonstrate local infiltration and visceral obstruction, especially if large. Metastases to the bowel are typically multifocal, unlike GISTs, and often come with an appropriate history of primary malignancy (25). The diagnoses that are more difficult to differentiate both radiologically and pathologically are those of other soft-tissue tumors, including leiomyosarcoma, intraabdominal fibromatosis of the bowel wall, and malignant tumors of nerve sheath and vascular origin (21,25,26). However, transabdominal biopsy is not recommended in potentially resectable cases because of the risk of tumor seeding, which presumably accounted for our two cases of metastasis to scar tissue.
We accept a number of limitations of this study. Most notably, this was a retrospective unblinded review of cases collected over a number of years at a specialist center that carries with it a number of biases. In addition, we were unable to collate a complete set of imaging studies. Scans reviewed included those from other institutions in which different imaging techniques were used, and we did not have access to soft-copy scans, with the added benefits of electronic measurement and attenuation assessment. However, many of these problems are simply unavoidable because of the rarity of this type of tumor.
In conclusion, despite overlap of the radiologic appearances of GISTs and other sarcomatous tumors, the authors believe that if the above imaging features and/or tumor behaviors are recognized, there should be a high level of suspicion for GIST.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, E.C.M.; study concepts, E.C.M., J.M.T., C.F., I.R.J.; study design, G.J.C.B., M.B., O.A.M.; literature research, G.J.C.B., M.B.; clinical studies, G.J.C.B., M.B., E.C.M.; data acquisition, G.J.C.B., O.A.M., M.B.; data analysis/interpretation, G.J.C.B., E.C.M., M.B., O.A.M.; manuscript preparation, G.J.C.B., M.B., O.A.M.; manuscript definition of intellectual content, E.C.M., C.F., I.R.J., J.M.T.; manuscript editing, M.B., O.A.M.; manuscript revision/review, E.C.M., I.R.J., J.M.T., C.F.; manuscript final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Wronski, W. Cebulski, M. Slodkowski, and I. W. Krasnodebski Gastrointestinal Stromal Tumors: Ultrasonographic Spectrum of the Disease J. Ultrasound Med., July 1, 2009; 28(7): 941 - 948. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Kuzmich, D. C. Howlett, A. Andi, D. Shah, and T. Kuzmich Transabdominal Sonography in Assessment of the Bowel in Adults Am. J. Roentgenol., January 1, 2009; 192(1): 197 - 212. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S ULUSAN, Z KOC, and F KAYASELCUK Gastrointestinal stromal tumours: CT findings Br. J. Radiol., August 1, 2008; 81(968): 618 - 623. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Perrone, G. Serafini, A. Vitali, F. Lacelli, L. Sconfienza, and L. E. Derchi Gastrointestinal Stromal Tumor Metastatic to the Scrotum J. Ultrasound Med., June 1, 2008; 27(6): 961 - 964. [Full Text] [PDF]
|
|
|
|
|
|
F. Berton, G. Gola, and S. R. Wilson Perspective on the Role of Transrectal and Transvaginal Sonography of Tumors of the Rectum and Anal Canal Am. J. Roentgenol., June 1, 2008; 190(6): 1495 - 1504. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Wu, X. Zhu, and Y. Ding Diagnosis and Treatment of Gastrointestinal Stromal Tumors of the Stomach: Report of 28 Cases. Ann. Clin. Lab. Sci., January 1, 2007; 37(1): 15 - 21. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Khaja, R. Kilani, W. Jacobson, and A. K. Hiett Gastrointestinal Stromal Tumor Presenting as a Mass on Pelvic Sonography J. Ultrasound Med., January 1, 2007; 26(1): 117 - 120. [Full Text] [PDF]
|
|
|
|
 |
|
 X. Hong, H. Choi, E. M. Loyer, R. S. Benjamin, J. C. Trent, and C. Charnsangavej Gastrointestinal Stromal Tumor: Role of CT in Diagnosis and in Response Evaluation and Surveillance after Treatment with Imatinib. RadioGraphics, March 1, 2006; 26(2): 481 - 495. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. R. Warakaulle and F. Gleeson MDCT Appearance of Gastrointestinal Stromal Tumors After Therapy with Imatinib Mesylate Am. J. Roentgenol., February 1, 2006; 186(2): 510 - 515. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Clark, C. Fisher, I. Judson, and J. M. Thomas Soft-Tissue Sarcomas in Adults N. Engl. J. Med., August 18, 2005; 353(7): 701 - 711. [Full Text] [PDF]
|
|
|
|
 |
|
 U. De Giorgi and J. Verweij Imatinib and gastrointestinal stromal tumors: Where do we go from here? Mol. Cancer Ther., March 1, 2005; 4(3): 495 - 501. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. D. Levy, A. M. Quiles, M. Miettinen, and L. H. Sobin Gastrointestinal Schwannomas: CT Features with Clinicopathologic Correlation Am. J. Roentgenol., March 1, 2005; 184(3): 797 - 802. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Sandrasegaran, A. Rajesh, J. Rydberg, D. A. Rushing, F. M. Akisik, and J. D. Henley Gastrointestinal Stromal Tumors: Clinical, Radiologic, and Pathologic Features Am. J. Roentgenol., March 1, 2005; 184(3): 803 - 811. [Full Text] [PDF]
|
|
|
|
|
|
A. D. Levy, N. Patel, R. M. Abbott, N. Dow, M. Miettinen, and L. H. Sobin Gastrointestinal Stromal Tumors in Patients with Neurofibromatosis: Imaging Features with Clinicopathologic Correlation Am. J. Roentgenol., December 1, 2004; 183(6): 1629 - 1636. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Sinha, R. Verma, and A. Kong Mesenteric Gastrointestinal Stromal Tumor in a Patient with Neurofibromatosis Am. J. Roentgenol., December 1, 2004; 183(6): 1844 - 1846. [Full Text] [PDF]
|
|
|
|
|
|
H.-C. Kim, J. M. Lee, K. W. Kim, S. H. Park, S. H. Kim, J. Y. Lee, J. K. Han, and B. I. Choi Gastrointestinal Stromal Tumors of the Stomach: CT Findings and Prediction of Malignancy Am. J. Roentgenol., October 1, 2004; 183(4): 893 - 898. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H.-C. Kim, J. M. Lee, K. R. Son, S. H. Kim, K. H. Lee, K. W. Kim, M. Lee, J. K. Han, and B. I. Choi Gastrointestinal Stromal Tumors of the Duodenum: CT and Barium Study Findings Am. J. Roentgenol., August 1, 2004; 183(2): 415 - 419. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H.-C. Kim, J. M. Lee, S. H. Kim, K. W. Kim, M. Lee, Y. J. Kim, J. K. Han, and B. I. Choi Primary Gastrointestinal Stromal Tumors in the Omentum and Mesentery: CT Findings and Pathologic Correlations Am. J. Roentgenol., June 1, 2004; 182(6): 1463 - 1467. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. L. Eisenberg and I. Judson Surgery and Imatinib in the Management of GIST: Emerging Approaches to Adjuvant and Neoadjuvant Therapy Ann. Surg. Oncol., May 1, 2004; 11(5): 465 - 475. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Gayed, T. Vu, R. Iyer, M. Johnson, H. Macapinlac, N. Swanston, and D. Podoloff The Role of 18F-FDG PET in Staging and Early Prediction of Response to Therapy of Recurrent Gastrointestinal Stromal Tumors J. Nucl. Med., January 1, 2004; 45(1): 17 - 21. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Goyal, R E Mansel, and S Goyal Gastrointestinal stromal tumour in an inguinal hernial sac: an unusual presentation Postgrad. Med. J., December 1, 2003; 79(938): 707 - 708. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
