CYTOLOGIC-PATHOLOGIC CORRELATION
Diagnosis of Gastrointestinal Stromal Tumor by Endoscopic Ultrasound-Guided Fine Needle Aspiration Biopsy – A Potential Pitfall Kai Fu, MD, PhD, Mohamad A. Eloubeidi, MD, MHS, Nirag C. Jhala, MD, Darshana Jhala, MD, David C. Chhieng, MD, and Isam-Eldin A. Eltoum, MD Endoscopic ultrasound-guided fine needle aspiration biopsy (EUS-FNA) is considered to be a reliable and accurate method for the evaluation of submucosal lesions in the gastrointestinal tract. Herein, we report our experience with the diagnosis of 10 cases of gastrointestinal stromal tumor (GIST) using EUS-FNA. The materials obtained from the EUS-FNA were stained with the rapid Romanowsky or the Papanicolaou method for cytologic examination. The subsequent surgical resection specimens were submitted for histopathologic examination. Immunoperoxidase stains were performed on the cell blocks and/or representative histologic sections of the tumor using commercially available antibodies against c-kit (CD117), CD34, S-100, and smooth muscle actin. Of the 10 cases studied, there were five men and five women with an average age of 62 years (range, 38 to 87 years). Five tumors were located in the stomach, and five in the duodenum. Tumor size ranged from 3.5 to 16.2 cm. Immediate on-site evaluation and cytologic diagnoses were given in eight cases (80%) with an average of three passes. The diagnoses were confirmed by strong and diffuse tumor cell c-kit immunoreactivity in the cell blocks. However, the final diagnoses of two other cases (20%) were not established until surgical resections were obtained. Retrospectively, reviews of cytologic smears of both cases demonstrated rare cohesive sheets or clusters of spindle cells with cigar-shaped nuclei. These observations were initially misinterpreted as benign fibrous tissue and/or fragments of smooth muscle of the gastrointestinal wall such as one might encounter in a routine transgastric or transduodenal EUS-FNA. The current study showed that when combining cytologic and immunocytochemical studies, EUS-FNA is accurate and efficient in the diagnosis of GIST. It exemplified the importance of considering GIST in the differential diagnosis of gastrointestinal lesions and also demonstrated the potential pitfalls of EUS-FNA evaluation of submucosal lesions in the gastrointestinal tract. Ann Diagn Pathol 6: 294-301, 2002. Copyright 2002, Elsevier Science (USA). All rights reserved. Index Words: Endoscopic ultrasound guided fine needle aspiration, gastrointestinal stromal tumor, cytologic diagnosis, immunocytochemistry
G
ASTROINTESTINAL stromal tumors (GISTs) are the most common mesenchymal tumors of the gastrointestinal (GI) tract. These tumors usually do not involve the mucosa but commonly originate in the wall of the GI tract1; therefore, the
From the Departments of Pathology and Gastroenterology and Hepatology, The University of Alabama at Birmingham, AL. Address reprint requests to Isam-Eldin A. Eltoum, MD, Department of Pathology, The University of Alabama at Birmingham, KB627, Kracke Bldg, 619 S 19th St, Birmingham, AL 35233. Copyright 2002, Elsevier Science (USA). All rights reserved. 1092-9134/02/0605-0005$35.00/0 doi:10.1053/adpa.2002.35741
294
opportunity to obtain adequate materials for the histologic diagnosis via endoscopic forcep biopsy is limited. The diagnosis of these tumors is generally based on surgical specimens, laparoscopic biopsies, or occasionally percutaneous ultrasound-guided fine needle aspirations.2-6 Endoscopic ultrasound (EUS) is not only capable of characterizing lesions of the GI tract and adjacent structures but it is also capable of guiding the fine needle aspiration (FNA) under real-time ultrasound using a through-the-scope needle aspiration system.7,8 Recent studies have shown that EUS combined with FNA cytology appears to be of great value in the evaluation of intramural lesions
Annals of Diagnostic Pathology, Vol 6, No 5 (October), 2002: pp 294-301
Diagnosis of GIST by EUS-FNA
295
Figure 1. (Top) Using a radial scanning echoendoscope, an irregular hypoechoic mass with a central cyst is identified in the wall of the duodenum. (Bottom) Real time, ultrasound-guided FNA biopsy of the submucosal mass is performed. The echogenic needle is visualized within the tumor.
of the GI tract,9-11 especially GIST.3,12-14 Herein, we report our experience of using EUS-FNA in the diagnosis of 10 cases of GIST at our institution. Materials and Methods Patients Ten patients with a submucosal lesion of the upper GI tract who had undergone EUS-FNA analysis were included
in this study (Fig 1). Mean age was 62 years (range, 38 to 87 years). There were five men and five women. Four patients presented with abdominal pain, two with GI bleeding, one with obstructive jaundice, while the other three were asymptomatic. As shown in Table 1, these tumors arose from the stomach (five cases), or duodenum (five cases), respectively. There were no esophageal GIST encountered. Tumors ranged in size from 3.5 to 16.2 cm in dimension. The tumors that arose from the stomach were significantly
296
Fu et al Table 1. Clinical Information of Ten Cases of Gastrointestinal Stromal Tumor
Case No.
Age
Sex
Site
1 2 3 4 5 6 7 8 9 10
84 53 87 48 81 51 62 56 65 38
F M F M M F F F M M
Duodenum Stomach Duodenum Duodenum Duodenum Stomach Stomach Stomach Stomach Duodenum
Presentation
Size (cm)
Asymptomatic 4.6 ⫻ 3.6 Gl bleeding 12.0 ⫻ 11.0 Abdominal pain 4.1 ⫻ 2.5 Asymptomatic 3.5 ⫻ 3.5 Asymptomatic 4.0 ⫻ 3.1 Abdominal pain 14.0 ⫻ 11.1 Gl bleeding 6.7 ⫻ 4.7 Abdominal pain 16.2 ⫻ 10.0 Abdominal pain 5.6 ⫻ 5.2 Jaundice 7.4 ⫻ 5.2
larger than the ones from the duodenum. While five gastric cases presented with abdominal pain or GI bleeding, three of five tumors arising from the duodenum were asymptomatic and found incidentally by image studies for other causes including cirrhosis (one case) or colon cancer staging (two cases).
Methods EUS-FNA was performed under endosonographic guidance (UC-30P, Olympus America, Melville, NY) using a 22-gauge needle (Wilson Cook; Winston Salem, NC). The materials obtained from the aspiration were spread onto the glass sides and divided between air-dried smears stained for immediate interpretation with rapid Romanowsky (Diff-Quik, Baxter; McGaw Park, IL) and alcohol-fixed smears for staining by the Papanicolaou staining method. The needle was then rinsed with Hank’s balanced salt solution and cell blocks were prepared from the deposit after centrifugation. The surgically removed specimen was fixed in 10% neutral buffered formalin and embedded in paraffin. The sections submitted for histopathologic examination were stained with hematoxylin-eosin. Immunoperoxidase stains were subsequently performed on the cell block and/or representative histologic sections of the tumor using commercially available antibodies against c-kit (CD117), CD34, S-100, and smooth muscle actin. Appropriate positive and negative controls were used.
Results
FNA Diagnosis
Surgical Resection
GIST (c-kit ⫹) Nondiagnostic Nondiagnostic GIST (c-kit ⫹) GIST (c-kit ⫹) GIST (c-kit ⫹) GIST (c-kit ⫹) GIST (c-kit ⫹) GIST (c-kit ⫹) GIST (c-kit ⫹)
None Malignant GIST Spindle cell tumor None None None GIST (c-kit ⫹) Malignant GIST None GIST (c-kit ⫹)
Comment
History of colon cancer C-kit negative, SMA positive Cirrhosis History of colon cancer
History of Breast cancer
ment of nuclei. The cells had varying amounts of delicate and lightly stained cytoplasm. The cell membranes were inconspicuous. Many cells contained long cytoplasmic processes, especially in the periphery of the cell clusters. The tumor cells had elongated, cigar-shaped, blunted-end nuclei with relatively smooth contour. Occasionally, a small indentation was noted in the middle of the nuclei along the longitudinal axis (Figs 2B and 2C). The chromatin was finely granular and evenly distributed. Significant atypia or mitoses were absent in all cases studied. No evidence of necrosis was noted. The immediate on-site evaluation and cytologic diagnoses of GIST were made on these cases. Immunoperoxidase stains were subsequently performed on the cell blocks. In all cases, the tumor cells exhibited strong and diffuse c-kit immunoreactivity (Fig 2D), which confirmed the cytologic diagnoses. The cytology smears from case nos. 2 (Fig 3A) and 3 (Fig 4A) demonstrated scant cellularity with rare cohesive sheets or clusters of spindle cells with cigar-shaped nuclei. These findings were initially interpreted as benign fibrous tissue and/or fragments of smooth muscle of the GI tract; therefore, both cases were interpreted as nondiagnostic in the cytologic examinations. The cell blocks in both cases were inadequate for further studies.
Cytologic Findings The FNA smears from eight cases were moderately cellular and composed of a monotonous neoplastic population of spindle cells arranged in flat cohesive sheets, three-dimensional clusters, or as single cells in a bloody background (Fig 2A). The tumor cells formed fascicles with a parallel arrange-
Histologic and Immunohistochemical Findings Patient no. 2 subsequently underwent surgical excision. A 12-cm tumor that originated from the gastric wall with a central cystic degeneration and hemorrhage was resected. The histologic sections demonstrated a malignant GIST with moderate
Diagnosis of GIST by EUS-FNA
297
Figure 2. EUS-FNA smears of case no. 1. (A) Moderately cellular clusters of spindle cells are evident (Diff-Quik stain). (B and C) The spindle cells are bland looking with elongated, cigar-shaped pale nuclei, fine chromatin, and inconspicuous nucleoli. The cytoplasm show taping at both ends (Diff-Quik stain). (D) Diffuse and strong immunoreactivity of tumor cells for c-kit (immunoperoxidase stain on cell block).
cytologic atypia and brisk mitotic activities (Fig 3B). Immunoperoxidase stains were performed and the tumor cells were strongly immunoreactive for c-kit and CD34, but negative for S-100 and smooth muscle actin. Patient no. 3 subsequently underwent exploratory laparotomy, which revealed a 4-cm mass that originated from the duodenum and adhered to the aortic wall. Because of the patient’s overall frail condition the mass was not resected, but a needle biopsy was obtained. The biopsy specimen showed a mass lesion composed of spindle cells with elongated nuclei and scant cytoplasm (Fig 4B). The tumor cells had mild cytologic atypia. There were 1 to 2 mitoses/10 high power field. Immunoperoxidase stains were then performed and the tumor
cells showed strong smooth muscle actin positivity, but were negative for c-kit, CD34, and S-100 protein. This tumor was diagnosed as a smooth muscle tumor with undetermined malignant potential. Retrospectively, the cytologic smears of these two cases were re-examined and compared to the histology of the surgical specimen. The cytologic features of those rare spindle cell fragments were identical to those in histologic sections (Figs 3 and 4). However, these fragments appeared to be slightly more cellular and the nuclei of the cells more crowded than those seen in benign fibrous tissue or smooth muscle of the GI tract. Three other patients also underwent surgical excision. The histologic sections presented GIST with low malignant potential (case nos. 7 and 10)
298
Fu et al
Figure 3. EUS-FNA smear of case no. 2. (A) A single cluster of spindle cell with elongated nuclei, fine chromatin, and inconspicuous nucleoli is evident (Diff-Quik stain). (B) Gastric resection specimen showing spindle cell tumor with mitotic active tumor cells forming interlacing fascicles (hematoxylin-eosin stain).
and a malignant GIST with moderate cytologic atypia and brisk mitotic activities (case no. 8), respectively. Immunoperoxidase stains were performed, the tumor cells were strongly immunoreactive for CD117 and CD34, but negative for S-100 and smooth muscle actin. Discussion Most GI mesenchymal tumors have been labeled leiomyomas, leiomyosarcomas, or Schwann cell tumors in the past.1 Immunohistochemical and ultrastructural studies in the early 1990s showed that the majority of these tumors lack specific differentiation features of classical smooth muscle tumors or Schwann cell tumors.15 Thus, the term “gastrointestinal stromal tumor” (GIST) has been intro-
duced to acknowledge the lack of evidence of muscular or nerve sheath differentiating features and the variable phenotypic profile in this family of tumors. These tumors span a wide clinical spectrum from benign incidentally detected nodules to frankly malignant tumors, and are generally resistant to chemotherapy and radiation treatment. Until very recently surgery has been the primary treatment modality for GISTs. In the past several years, it has been recognized that activation of KIT signaling, frequently occurring via mutations in the c-kit gene, is the most important factor in the pathogenesis of GISTs. These findings have significant clinical impact and provided potential targets for therapeutic intervention of GISTs.16 This progress has
Diagnosis of GIST by EUS-FNA
299
Figure 4. EUS-FNA smear of case no. 3. (A) A sheet of spindle cells with elongated nuclei and inconspicuous nucleoli is evident (Papanicolaou stain). (B) Biopsy of the periduodenal mass showing spindle cell tumor with relatively bland looking tumor cells forming interlacing fascicles (hematoxylin-eosin stain).
made it necessary to accurately define and diagnose these tumors preoperatively. In the current study, we showed that GISTs could be diagnosed preoperatively on EUS-FNA cytologic specimen combined with immunocytochemistry studies. Eight of 10 cases were correctly diagnosed on site with subsequent confirmation by immunocytochemical studies. Clinically, the most important differential diagnoses of the GIST include the epithelial neoplasms or malignant lymphomas, because the treatment of choice is generally different.1,15,17 The significance of this distinction is exemplified in case nos. 1 and 5. In both cases, a metastatic adenocarcinoma to the lymph node was the clinical and radiographic impression. The differential diagnosis of GIST with other mesenchy-
mal tumors of the abdomen (ie, leiomyomas, leiomyosarcomas, or Schwann cell tumors) may not be possible on routine stained cytologic smears. However, the immunocytochemical findings with positivity for c-kit and/or CD34 add valuable information for the correct diagnosis.4,18 In the current study, the diagnoses of GIST in eight cases were further supported by strong and diffuse c-kit immunoreactivity in the tumor cells. The diagnoses of the other two cases, however, were not established until surgical resections were obtained. In retrospect, the cytologic smears from these two false-negative cases demonstrated rare clusters of spindle-shaped cells arranged in fascicles. In both cases, sufficient materials were lacking in the smears for establishing a confident diagnosis
300
Fu et al
or in the cellblocks for further studies. Therefore, it is crucial to have an experienced cytologist to evaluate the sufficiency of the aspirate on-site when EUS-FNA is performed. Both cases contain rare cohesive sheets of spindle cells with cigar-shaped nuclei, which were initially interpreted as benign fibrous tissue or smooth muscle of the GI wall. As we know, the aspirate material of EUS-FNA is usually obtained through the wall of the GI tract. Clusters of stromal cells from the normal intramural structures, which sometimes manifest as spindle cells groups, can complicate the cytologic diagnosis of GI tumors, especially when the cellularity of the aspirate is scant. Without high suspicion, a cytologist may easily confuse the rare clusters of blandlooking spindle tumor cells with the normal stromal structures of the GI tract. Although the diagnosis of GISTs can be made with a certain degree of confidence by using the FNA findings, predictions about the biological features of the tumor are difficult. The cytomorphologic characterization of malignant GISTs was quite limited in previous reports.5,19 In the current study, although the histology sections of the both malignant GIST cases (2 and 8) exhibited cytologic atypia and high mitotic activity, the cytologic smears show a relative bland population of spindled cells without apparent evidence of malignancy. Therefore, predictions about the potential aggressiveness of the GIST are best reserved for histologic examination or, better yet, the patient’s ultimate outcome. Interestingly, Ando et al14 has recently shown that the presence of mitotic cells and the Ki-67 labeling index are significant predictive factors for malignant GIST. Therefore, in our opinion, further studies to evaluate the clinical use of Ki-67 immunostaining may be warranted. The presence of c-kit mutations may portend poor prognosis clinically in GISTs.20-25 A recent study by Li et al3 showed the use of detection of c-kit mutation by polymerase chain reaction in the FNA specimen.3 Although the presence of c-kit mutation in GISTs was strongly predictive of malignant behavior, they found that some malignant GISTs did not show c-kit mutations. However, their study is limited by amplifying only a segment of c-kit exon 11, as we know that cases of malignant GISTs with mutations in either exon 9 or 13, but not in exon 11, have been reported.26 In summary, the current study showed demonstrated that a combination of cytologic and immu-
nocytochemical studies makes EUS-FNA an accurate and efficient method in the diagnosis of GIST of the upper GI tract. It exemplified the importance of considering GIST in the differential diagnosis of GI lesions. It also catalogued potential pitfalls of EUS-FNA evaluation of submucosal lesions in the GI tract. References 1. Miettinen M, Sarlomo-Rikala M, Lasota J: Gastrointestinal stromal tumors: Recent advances in understanding of their biology. Hum Pathol 1999;30:1213-1220 2. Seidal T, Edvardsson H: Diagnosis of gastrointestinal stromal tumor by fine-needle aspiration biopsy: a cytological and immunocytochemical study. Diagn Cytopathol 2000;23:397-401 3. Li SQ, O’Leary TJ, Sobin LH, et al: Analysis of KIT mutation and protein expression in fine needle aspirates of gastrointestinal stromal/smooth muscle tumors. Acta Cytol 2000;44: 981-986 4. Cheuk W, Lee KC, Chan JK: c-kit immunocytochemical staining in the cytologic diagnosis of metastatic gastrointestinal stromal tumor. A report of two cases. Acta Cytol 2000;44:679-685 5. Dodd LG, Nelson RC, Mooney EE, et al: Fine-needle aspiration of gastrointestinal stromal tumors. Am J Clin Pathol 1998;109:439-443 6. Isimbaldi G, Santangelo M, Cenacchi G, et al: Gastrointestinal autonomic nerve tumor (plexosarcoma): Report of a case with fine needle aspiration biopsy and histologic, immunocytochemical and ultrastructural study. Acta Cytol 1998;42:11891194 7. Erickson RA, Sayage-Rabie L, Avots-Avotins A: Clinical utility of endoscopic ultrasound-guided fine needle aspiration. Acta Cytol 1997;41:1647-1653 8. Gress FG, Hawes RH, Savides TJ, et al: Endoscopic ultrasound-guided fine-needle aspiration biopsy using linear array and radial scanning endosonography. Gastrointest Endosc 1997; 45:243-250 9. Williams DB, Sahai AV, Aabakken L, et al: Endoscopic ultrasound guided fine needle aspiration biopsy: A large single centre experience. Gut 1999;44:720-726 10. Bentz JS, Kochman ML, Faigel DO, et al: Endoscopic ultrasound-guided real-time fine-needle aspiration: Clinicopathologic features of 60 patients. Diagn Cytopathol 1998;18:98-109 11. Frazee RC, Singh H, Erickson RA: Endoscopic ultrasound for peripancreatic masses. Am J Surg 1997;174:596-598 12. Boggino HE, Fernandez MP, Logrono R: Cytomorphology of gastrointestinal stromal tumor: Diagnostic role of aspiration cytology, core biopsy, and immunochemistry. Diagn Cytopathol 2000;23:156-160 13. Gu M, Ghafari S, Nguyen PT, et al: Cytologic diagnosis of gastrointestinal stromal tumors of the stomach by endoscopic ultrasound-guided fine-needle aspiration biopsy. Diagn Cytopathol 2001;25:343-350 14. Ando N, Hidemi G, Yasumasa N, et al: The diagnosis of GI tumors with EUS-guided fine needle aspiration with immunohistochemical analysis. Gastrointest Endosc 2002;55:37-43 15. Berman J, O’Leary TJ: Gastrointestinal stromal tumor workshop. Hum Pathol 2001;32:578-582 16. Chan JK: Mesenchymal tumors of the gastrointestinal
Diagnosis of GIST by EUS-FNA tract: a paradise for acronyms (STUMP, GIST, GANT, and now GIPACT), implication of c-kit in genesis, and yet another of the many emerging roles of the interstitial cell of Cajal in the pathogenesis of gastrointestinal diseases? Adv Anat Pathol 1999; 6:19-40 17. Miettinen M, Sarlomo-Rikala M, Lasota J: Gastrointestinal stromal tumours. Ann Chir Gynaecol 1998;87:278-281 18. Sarlomo-Rikala M, Kovatich AJ, Barusevicius A, et al: CD117: A sensitive marker for gastrointestinal stromal tumors that is more specific than CD34. Mod Pathol 1998;11:728-734 19. Tao LC, Davidson DD: Aspiration biopsy cytology of smooth muscle tumors. A cytologic approach to the differentiation between leiomyosarcoma and leiomyoma. Acta Cytol 1993; 37:300-308 20. Sakurai S, Fukasawa T, Chong JM, et al: C-kit gene abnormalities in gastrointestinal stromal tumors (tumors of interstitial cells of Cajal). Jpn J Cancer Res 1999;90:1321-1328 21. Taniguchi M, Nishida T, Hirota S, et al: Effect of c-kit
301
mutation on prognosis of gastrointestinal stromal tumors. Cancer Res 1999;59:4297-4300 22. Seidal T, Edvardsson H: Expression of c-kit (CD117) and Ki67 provides information about the possible cell of origin and clinical course of gastrointestinal stromal tumours. Histopathology 1999;34:416-424 23. Lasota J, Jasinski M, Sarlomo-Rikala M, et al: Mutations in exon 11 of c-Kit occur preferentially in malignant versus benign gastrointestinal stromal tumors and do not occur in leiomyomas or leiomyosarcomas. Am J Pathol 1999;154:53-60 24. Nakahara M, Isozaki K, Hirota S, et al: A novel gain-offunction mutation of c-kit gene in gastrointestinal stromal tumors. Gastroenterology 1998;115:1090-1095 25. Ernst SI, Hubbs AE, Przygodzki RM, et al: KIT mutation portends poor prognosis in gastrointestinal stromal/smooth muscle tumors. Lab Invest 1998;78:1633-1636 26. Lux ML, Rubin BP, Biase TL, et al: KIT extracellular and kinase domain mutations in gastrointestinal stromal tumors. Am J Pathol 2000;156:791-795