ULTRASONOGRAPHIC FEATURES OF CANINE GASTROINTESTINAL STROMAL TUMORS COMPARED TO OTHER GASTROINTESTINAL SPINDLE CELL TUMORS JOSHUA HOBBS, JAMES SUTHERLAND-SMITH, DOMINIQUE PENNINCK, SAMUEL JENNINGS, LISA BARBER , BRUCE BARTON
Canine gastrointestinal stromal tumors (GISTs) are a recent subtype of gastrointestinal spindle cell tumor recognized with the increasing use of immunohistochemistry. To our knowledge, no imaging features have been described in immunistochemically confirmed canine GISTs. The objective of this retrospective, crosssectional study was to describe ultrasonographic features of canine GISTs compared with other spindle cell tumors. Thirty-seven dogs with an ultrasonographically visible gastrointestinal mass and a histopathologic diagnosis of spindle cell neoplasia were examined. Immunohistochemistry staining was performed for retrieved tissue samples to further differentiate the tumor type and each sample was interpreted by a single veterinary pathologist. Ultrasonographic features recorded examined included mass echogenicity, homogeneity, presence of cavitation, layer of origin, bowel wall symmetry, and loss of wall layering, location, size, vascularity, and evidence of perforation or ulceration. Tumor types included 19 GISTs, eight leiomyosarcomas, six leiomyomas, and four nonspecified sarcomas. Gastrointestinal stromal tumors were significantly more likely to be associated (P < 0.03) with abdominal effusion than other tumor types. There was overlap between the anatomical locations of all tumors types with the exception of the cecum where all eight tumors identified were GISTs. Besides location, there were no unique ultrasound features of GISTs that would allow distinction from other gastrointestinal spindle cell tumors. Similar to previous studies, GISTs appeared to be the most common spindle cell tumor associated with the cecum in our sample of dogs. The high frequency of abdominal effusion C 2015 American with GIST’s was of unknown etiology could possibly have been due to septic peritonitis. College of Veterinary Radiology. Key words: cecum, canine, gastrointestinal stromal tumor, ultrasound.
Introduction
G
arise from the interstitial cells of Cajal that are considered the pacemaker cells of the gastrointestinal tract controlling peristalsis and intestinal motility. Gastrointestinal stromal tumors may be morphologically indistinguishable from smooth muscle tumors or other spindle cell tumors with routine histologic staining. Immunohistochemistry is required in order to differentiate these neoplasm types, with GISTs being characterized by the expression of the protein CD117 (KIT). In contrast, other spindle cell tumors lack expression of this protein.5,6,9 Retrospective pathological studies have suggested that canine GISTs are primarily associated with the colon and the cecum, whereas leiomyomas and leiomyosarcomas are more often associated with the stomach; the opposite distribution pattern has been reported in humans.10–12 In humans, ultrasonographic features of GISTs include a large extramural tumor arising from the muscularis layer, that rarely affects the mucosal layer, and often contain regions of cavitation.12 In veterinary medicine, ultrasonographic characteristics of smooth muscle tumors include eccentric
approximately 3% of all reported canine neoplasms.1 Historically, gastrointestinal smooth muscle tumors such as leiomyomas and leiomyosarcomas accounted for approximately 20–30% of intestinal tumors and were thought of as the most common intestinal spindle cell tumors.1–5 With the advent of immunohistochemistry, many previously diagnosed leiomyosarcomas have been reclassified as gastrointestinal stromal tumors (GISTs). This reclassification may affect the true incidence of smooth muscle tumors.6–9 Gastrointestinal stromal tumors are thought to principally ASTROINTESTINAL TUMORS COMPRISE
From the Tufts Cummings School of Veterinary Medicine, North Grafton, MA 01536 (Hobbs, Sutherland-Smith, Penninck, Jennings, Barber); and University of Massachusetts, Worcester, MA 01655 (Barton). Supported by a grant from Universal Imaging, Bedford Hills, NY. Abstract presented at 2013 ACVR, Savannah, GA. Address correspondence and reprint requests to James SutherlandSmith, at the above address. E-mail: [email protected] Received May 17, 2014; accepted for publication December 22, 2014. doi: 10.1111/vru.12253
Vet Radiol Ultrasound, Vol. 56, No. 4, 2015, pp 432–438.
432
VOL. 56, NO. 4
ULTRASOUND FEATURES OF CANINE GIST
masses with cavitations but these were described prior to our understanding of GISTs.5 Thus, the ultrasonographic characteristics of confirmed GISTs have yet to be described. Dogs with GISTs have been reported to have a longer survival time in comparison to leiomyosarcomas.8 Therefore the ability to differentiate between these tumor types could aid in providing a prognosis prior to invasive surgical intervention. Additionally, it is important for ultrasonographers to be aware of emerging tumor types in order to provide a prioritized and thorough differential list. The purposes of this retrospective cross sectional study were to describe the ultrasonographic features of GISTs and determine whether ultrasound features could differentiate GISTs from other spindle cell tumors (non-GISTs). Based on our clinical experience and the existing literature, our hypothesis was that other than location, no ultrasonographic feature would differentiate GISTs from non-GISTs.
Material and Methods Patient Group Inclusion criteria consisted of gastrointestinal or abdominal masses identified surgically or at autopsy as originating from the gastrointestinal tract of dogs with a diagnosis of a leiomyoma, leiomyosarcoma, spindle cell sarcoma (fibrosarcoma, undifferentiated sarcoma) or GIST. Additional inclusion criteria included the presence of adequate stored ultrasound images or video, and stored histologic slides or paraffin-embedded tissues.
Imaging Studies The ultrasound images of the gastrointestinal masses were evaluated by at least one of two board-certified veterinary radiologists (J.S.S. and D.P.) and a veterinary radiology resident (J.H.) using a consensus opinion approach. The readers were not aware of the diagnosis at the time of evaluation. The gastrointestinal masses were evaluated for size, number, location within the gastrointestinal tract, bowel wall layering, and mass relationship to the gastrointestinal lumen, layer of origin when apparent, internal echogenicity, echotexture, presence of cavitation, pattern of vascularization, and evidence of ulceration or perforation. Vascularization was assessed using power or color flow Doppler. The videos, images and ultrasound reports were also reviewed for presence of abdominal effusion, or signs of metastasis such as hepatic or splenic nodules/masses, lymphadenomegaly, or mesenteric nodules. Both transverse and sagittal plane images were used to evaluate the masses. The largest dimension of the mass was recorded. The location of the mass was identified as either stomach, duodenum, jejunum, ileum, cecum, or colon. The layer of origin of the mass was based on evidence of specific wall layer enlargement
433
FIG. 1. Sagittal sonogram of cavitated regions (double headed white arrow) within a cecal mass confirmed as GIST. The lumen is seen as a hyperechoic line (white arrow) dorsal to the hypoechoic region. The white arrowheads depict the dorsal and ventral margins of the mass.
at the junction between the mass and the normal tissue. Layers were recorded as mucosal, submucosal, muscularis, or undetermined. Circumferential masses were defined as masses that symmetrically surrounded the bowel lumen and caused transmural loss of wall layering. Asymmetric circumferential masses were defined as masses that caused transmural loss of wall layering but was larger on one side of the bowel wall.13 Eccentric masses were defined as masses that caused transmural loss of wall layering one side of the bowel wall with retained visible layering on the opposite wall. Internal echogenicity of the mass was characterized as poor (anechoic or almost anechoic), moderate (similar echogenicity to spleen), or high (greater echogenicity to spleen). The internal echotexture of the mass was subjectively assessed using approximate criteria that homogenous tissue echotexture had uniformity over at least 90% of the mass, while inhomogeneous masses had uniformity levels below this threshold. Masses were evaluated for anechoic or hypoechoic regions (Fig. 1) referred to collectively as cavitations. When available, the presence of vascularization of the mass was assessed using color Doppler and was subjectively categorized as absent (0), poor (1–3 vessels per image field), moderate (3–5 vessels per image field), and marked (>5 per image field). Perforation was defined as free gas within the abdomen with or without concurrent hyperechoic foci or line extending from the mass lumen to the serosal surface. Free gas was defined as nondependent hyperechoic foci within the peritoneal cavity that displayed distal reverberation artifact. Ulceration was defined as gas dissecting the wall of the mass but not free within the abdomen. Maximal short axis dimensions were used to assess lymph node size with enlargement considered greater than 8 mm.14 The ultrasound machine and the frequency of the transducer used to obtain the images were recorded. Abdominal and thoracic radiographs were reviewed when available for the presence of an abdominal mass, abdominal effusion, or pulmonary nodules.
434
HOBBS ET AL.
Histology A single board-certified veterinary anatomic pathologist (S.J.) reviewed all the histology slides without knowledge of the previous diagnosis. Tumors were included in the study if the neoplastic cells were spindle shaped with H&E staining. The immunohistochemical (IHC) stains reviewed were directed against CD 117 (KIT) and smooth muscle actin (SMA). Immunohistochemical staining was not repeated on cases if this had been performed previously for the clinical diagnosis and the pathologist deemed the quality adequate. The SMA expression was considered positive if there was moderate to strong staining in at least 10% of the neoplastic cells, although typically expression was observed in greater than 50% of the neoplastic cells when staining was present. The KIT expression was considered positive if there was strong expression in greater than 10% of the neoplastic cells or weak to moderate expression in greater than 25% of the neoplastic cells. A GIST was diagnosed if the neoplastic cells expressed KIT regardless of SMA expression. Leiomyoma and leiomyosarcoma were diagnosed if the neoplastic cells expressed SMA but not KIT. Leiomyosarcoma was differentiated from leiomyoma by the presence of criteria of malignancy, such as cellular or nuclear pleomorphism, elevated mitotic activity, invasion, hemorrhage, or tumor necrosis.9 Nonspecified sarcoma was diagnosed if the neoplastic cells failed to express KIT or SMA.
Statistical Analysis All statistical tests were selected and performed by a statistician (B.B.), using statistical software (SAS/STAT software, SAS Institute, Cary, NC). Categorical ultrasonographic characteristics of GISTs were compared to those for non-GISTs (leiomyomas, leiomyosarcomas, and nonspecified sarcomas) using contingency table analysis with likelihood ratio chi-square tests used to determine the significance of any observed differences, including the comparison of ultrasonographic location to actual tumor location. For continuous characteristics (such as maximum size of mass), because of small sample size, we used a nonparametric test (Wilcoxon signed rank) to test differences in central tendency between GIST and non-GIST tumors without assuming a normal distribution. For the descriptive table of results we presented the frequencies and percentages for GIST, leiomyoma, leiomyosarcoma, and nonspecified sarcoma. The critical level for statistical significance was set at a P value
Canine gastrointestinal stromal tumors (GISTs) are a recent subtype of gastrointestinal spindle cell tumor recognized with the increasing use of immunohistochemistry. To our knowledge, no imaging features have been described in immunistochemically confirmed canine GISTs. The objective of this retrospective, crosssectional study was to describe ultrasonographic features of canine GISTs compared with other spindle cell tumors. Thirty-seven dogs with an ultrasonographically visible gastrointestinal mass and a histopathologic diagnosis of spindle cell neoplasia were examined. Immunohistochemistry staining was performed for retrieved tissue samples to further differentiate the tumor type and each sample was interpreted by a single veterinary pathologist. Ultrasonographic features recorded examined included mass echogenicity, homogeneity, presence of cavitation, layer of origin, bowel wall symmetry, and loss of wall layering, location, size, vascularity, and evidence of perforation or ulceration. Tumor types included 19 GISTs, eight leiomyosarcomas, six leiomyomas, and four nonspecified sarcomas. Gastrointestinal stromal tumors were significantly more likely to be associated (P < 0.03) with abdominal effusion than other tumor types. There was overlap between the anatomical locations of all tumors types with the exception of the cecum where all eight tumors identified were GISTs. Besides location, there were no unique ultrasound features of GISTs that would allow distinction from other gastrointestinal spindle cell tumors. Similar to previous studies, GISTs appeared to be the most common spindle cell tumor associated with the cecum in our sample of dogs. The high frequency of abdominal effusion C 2015 American with GIST’s was of unknown etiology could possibly have been due to septic peritonitis. College of Veterinary Radiology. Key words: cecum, canine, gastrointestinal stromal tumor, ultrasound.
Introduction
G
arise from the interstitial cells of Cajal that are considered the pacemaker cells of the gastrointestinal tract controlling peristalsis and intestinal motility. Gastrointestinal stromal tumors may be morphologically indistinguishable from smooth muscle tumors or other spindle cell tumors with routine histologic staining. Immunohistochemistry is required in order to differentiate these neoplasm types, with GISTs being characterized by the expression of the protein CD117 (KIT). In contrast, other spindle cell tumors lack expression of this protein.5,6,9 Retrospective pathological studies have suggested that canine GISTs are primarily associated with the colon and the cecum, whereas leiomyomas and leiomyosarcomas are more often associated with the stomach; the opposite distribution pattern has been reported in humans.10–12 In humans, ultrasonographic features of GISTs include a large extramural tumor arising from the muscularis layer, that rarely affects the mucosal layer, and often contain regions of cavitation.12 In veterinary medicine, ultrasonographic characteristics of smooth muscle tumors include eccentric
approximately 3% of all reported canine neoplasms.1 Historically, gastrointestinal smooth muscle tumors such as leiomyomas and leiomyosarcomas accounted for approximately 20–30% of intestinal tumors and were thought of as the most common intestinal spindle cell tumors.1–5 With the advent of immunohistochemistry, many previously diagnosed leiomyosarcomas have been reclassified as gastrointestinal stromal tumors (GISTs). This reclassification may affect the true incidence of smooth muscle tumors.6–9 Gastrointestinal stromal tumors are thought to principally ASTROINTESTINAL TUMORS COMPRISE
From the Tufts Cummings School of Veterinary Medicine, North Grafton, MA 01536 (Hobbs, Sutherland-Smith, Penninck, Jennings, Barber); and University of Massachusetts, Worcester, MA 01655 (Barton). Supported by a grant from Universal Imaging, Bedford Hills, NY. Abstract presented at 2013 ACVR, Savannah, GA. Address correspondence and reprint requests to James SutherlandSmith, at the above address. E-mail: [email protected] Received May 17, 2014; accepted for publication December 22, 2014. doi: 10.1111/vru.12253
Vet Radiol Ultrasound, Vol. 56, No. 4, 2015, pp 432–438.
432
VOL. 56, NO. 4
ULTRASOUND FEATURES OF CANINE GIST
masses with cavitations but these were described prior to our understanding of GISTs.5 Thus, the ultrasonographic characteristics of confirmed GISTs have yet to be described. Dogs with GISTs have been reported to have a longer survival time in comparison to leiomyosarcomas.8 Therefore the ability to differentiate between these tumor types could aid in providing a prognosis prior to invasive surgical intervention. Additionally, it is important for ultrasonographers to be aware of emerging tumor types in order to provide a prioritized and thorough differential list. The purposes of this retrospective cross sectional study were to describe the ultrasonographic features of GISTs and determine whether ultrasound features could differentiate GISTs from other spindle cell tumors (non-GISTs). Based on our clinical experience and the existing literature, our hypothesis was that other than location, no ultrasonographic feature would differentiate GISTs from non-GISTs.
Material and Methods Patient Group Inclusion criteria consisted of gastrointestinal or abdominal masses identified surgically or at autopsy as originating from the gastrointestinal tract of dogs with a diagnosis of a leiomyoma, leiomyosarcoma, spindle cell sarcoma (fibrosarcoma, undifferentiated sarcoma) or GIST. Additional inclusion criteria included the presence of adequate stored ultrasound images or video, and stored histologic slides or paraffin-embedded tissues.
Imaging Studies The ultrasound images of the gastrointestinal masses were evaluated by at least one of two board-certified veterinary radiologists (J.S.S. and D.P.) and a veterinary radiology resident (J.H.) using a consensus opinion approach. The readers were not aware of the diagnosis at the time of evaluation. The gastrointestinal masses were evaluated for size, number, location within the gastrointestinal tract, bowel wall layering, and mass relationship to the gastrointestinal lumen, layer of origin when apparent, internal echogenicity, echotexture, presence of cavitation, pattern of vascularization, and evidence of ulceration or perforation. Vascularization was assessed using power or color flow Doppler. The videos, images and ultrasound reports were also reviewed for presence of abdominal effusion, or signs of metastasis such as hepatic or splenic nodules/masses, lymphadenomegaly, or mesenteric nodules. Both transverse and sagittal plane images were used to evaluate the masses. The largest dimension of the mass was recorded. The location of the mass was identified as either stomach, duodenum, jejunum, ileum, cecum, or colon. The layer of origin of the mass was based on evidence of specific wall layer enlargement
433
FIG. 1. Sagittal sonogram of cavitated regions (double headed white arrow) within a cecal mass confirmed as GIST. The lumen is seen as a hyperechoic line (white arrow) dorsal to the hypoechoic region. The white arrowheads depict the dorsal and ventral margins of the mass.
at the junction between the mass and the normal tissue. Layers were recorded as mucosal, submucosal, muscularis, or undetermined. Circumferential masses were defined as masses that symmetrically surrounded the bowel lumen and caused transmural loss of wall layering. Asymmetric circumferential masses were defined as masses that caused transmural loss of wall layering but was larger on one side of the bowel wall.13 Eccentric masses were defined as masses that caused transmural loss of wall layering one side of the bowel wall with retained visible layering on the opposite wall. Internal echogenicity of the mass was characterized as poor (anechoic or almost anechoic), moderate (similar echogenicity to spleen), or high (greater echogenicity to spleen). The internal echotexture of the mass was subjectively assessed using approximate criteria that homogenous tissue echotexture had uniformity over at least 90% of the mass, while inhomogeneous masses had uniformity levels below this threshold. Masses were evaluated for anechoic or hypoechoic regions (Fig. 1) referred to collectively as cavitations. When available, the presence of vascularization of the mass was assessed using color Doppler and was subjectively categorized as absent (0), poor (1–3 vessels per image field), moderate (3–5 vessels per image field), and marked (>5 per image field). Perforation was defined as free gas within the abdomen with or without concurrent hyperechoic foci or line extending from the mass lumen to the serosal surface. Free gas was defined as nondependent hyperechoic foci within the peritoneal cavity that displayed distal reverberation artifact. Ulceration was defined as gas dissecting the wall of the mass but not free within the abdomen. Maximal short axis dimensions were used to assess lymph node size with enlargement considered greater than 8 mm.14 The ultrasound machine and the frequency of the transducer used to obtain the images were recorded. Abdominal and thoracic radiographs were reviewed when available for the presence of an abdominal mass, abdominal effusion, or pulmonary nodules.
434
HOBBS ET AL.
Histology A single board-certified veterinary anatomic pathologist (S.J.) reviewed all the histology slides without knowledge of the previous diagnosis. Tumors were included in the study if the neoplastic cells were spindle shaped with H&E staining. The immunohistochemical (IHC) stains reviewed were directed against CD 117 (KIT) and smooth muscle actin (SMA). Immunohistochemical staining was not repeated on cases if this had been performed previously for the clinical diagnosis and the pathologist deemed the quality adequate. The SMA expression was considered positive if there was moderate to strong staining in at least 10% of the neoplastic cells, although typically expression was observed in greater than 50% of the neoplastic cells when staining was present. The KIT expression was considered positive if there was strong expression in greater than 10% of the neoplastic cells or weak to moderate expression in greater than 25% of the neoplastic cells. A GIST was diagnosed if the neoplastic cells expressed KIT regardless of SMA expression. Leiomyoma and leiomyosarcoma were diagnosed if the neoplastic cells expressed SMA but not KIT. Leiomyosarcoma was differentiated from leiomyoma by the presence of criteria of malignancy, such as cellular or nuclear pleomorphism, elevated mitotic activity, invasion, hemorrhage, or tumor necrosis.9 Nonspecified sarcoma was diagnosed if the neoplastic cells failed to express KIT or SMA.
Statistical Analysis All statistical tests were selected and performed by a statistician (B.B.), using statistical software (SAS/STAT software, SAS Institute, Cary, NC). Categorical ultrasonographic characteristics of GISTs were compared to those for non-GISTs (leiomyomas, leiomyosarcomas, and nonspecified sarcomas) using contingency table analysis with likelihood ratio chi-square tests used to determine the significance of any observed differences, including the comparison of ultrasonographic location to actual tumor location. For continuous characteristics (such as maximum size of mass), because of small sample size, we used a nonparametric test (Wilcoxon signed rank) to test differences in central tendency between GIST and non-GIST tumors without assuming a normal distribution. For the descriptive table of results we presented the frequencies and percentages for GIST, leiomyoma, leiomyosarcoma, and nonspecified sarcoma. The critical level for statistical significance was set at a P value
