The GIST of a stromal tumor

Abstract

Kim TW, Lee H, Kang Y-K, Choe MS, Ry M-H, Chang HM, Kim JS, Yook JH, Kim BS, Lee JS (Departments of Medicine, Diagnostic Pathology, and Surgery, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea). Prognostic significance of c-kit mutation in localized gastrointestinal stromal tumors. Clin Cancer Res 2004;10:3076–3081.Once grouped with leiomyomas, leiomyosarcomas, and other mesenchymal tumors of the gastrointestinal tract, gastrointestinal stromal tumors (GIST) are now considered distinct, arising from the Interstitial Cells of Cajal, and exhibit a variable clinical course. The vast majority of GISTs express the c-kit proto-oncogene protein, and gain-of-function mutations in this gene result in ligand-independent activation of c-kit tyrosine activity and subsequent oncogenesis. Because the clinical behavior of GISTs has been difficult to predict, a number of groups have sought to determine risk factors for aggressive tumors and thus poor prognosis. A National Institutes of Health (NIH)-sponsored GIST workshop in 2001 characterized risk of aggressive behavior (“very low”, “low”, “intermediate”, and “high”) based on tumor size and mitotic count, with tumor size >5 cm and mitoses > 5/50 high power fields (HPF) being indicators of high risk (Human Pathology 2002;33:459–465). Others have looked at the presence of mutations in the c-kit proto-oncogene (specifically exon 11, which encodes the juxtamembrane region of c-kit) as an independent factor for poor prognosis. Kim and colleagues report the results of a retrospective analysis of the prognostic significance of c-kit mutations located mainly in the stomach and small bowel in patients who had undergone curative resection for localized GIST. Using immunohistochemical staining for CD117, CD34, desmin, SMA, and S-100, these investigators identified 86 cases of true GIST treated at a single medical center between 1990 and 2001. The selected tumors were analyzed for histologic features and mitotic counts, and the patients were then stratified into groups based on the 2001 NIH consensus characterization of risk. Genomic DNA was extracted from each tumor, and polymerase chain reaction amplification of exon 11 was performed in each case. Samples negative for mutations in exon 11 mutations subsequently underwent analysis for exons 9, 13, and 17, which have been implicated in a minority of GISTs. Sixty-four of the 86 GISTs demonstrated mutations in c-kit. Of these, 61 had mutations in exon 11 and 3 had mutations in exon 9. No mutations in the other analyzed exons were found. The overall mutation rate of 71% was consistent with previous studies. Tumor size, histologic phenotype (spindle, epithelioid, mixed), cellularity (sparse, moderate, dense), tumor location, and mitotic count were compared in mutation-negative and mutation-positive lesions. Higher mitotic counts and higher cellularity were the only parameters that were significantly associated with the presence of c-kit mutations (P = .005 and .011, respectively). Mean number of mitoses per 50 HPF was 16.5 for mutation-positive lesions versus 4.9 in the negative tumors. Dense cellularity was found in 67% of positive lesions versus 41% of negative lesions. Average tumor size of c-kit mutation-positive lesions was greater (7.0 cm vs 6.4 cm), but this was not statistically significant. The authors also assessed 5-year relapse-free survival (RFS) using the Kaplan-Meier method and multivariate analysis. Median follow-up was 35.7 months for patients without relapse. The 5-year RFS trended toward being lower in patients with c-kit mutations than those without mutations (21% versus 60%, P = .098). Tumor size greater than 5 cm, mitotic count greater than 5/50 HPFs, epithelioid/mixed morphology, and nongastric GISTs were found to have significantly lower 5-year RFS. No difference between relapse-free survival rates of specific types of exon 11 mutations (ie, missense vs all other kinds) was found. Multivariate analyses led the authors to conclude that the presence of c-kit mutation (odds ratio [OR], 5.57; P = .023), mitoses >5/50 HPF (OR, 3.03; P = .018), and size >5 cm (OR, 4.23; P = .015) were independent indicators of a poorer 5-year RFS.CommentIt is estimated that about 5000–6000 new GISTs are identified in the United States yearly (Human Pathol 2002;33:459–465). As the utilization of endoscopy to investigate unrelated gastrointestinal symptoms increases, the number of small, incidentally discovered submucosal or subepithelial masses likely will grow. The diagnosis and management of GISTs poses a unique diagnostic challenge to the gastroenterologist. Tumors that are symptomatic, large, or felt to harbor malignancy require surgical excision. There also has been recent interest in the use of imantinib mesylate in individuals with malignant GISTs. Consequently, methods for determining the malignant nature or potential of these tumors are needed and must be evaluated for their performance characteristics. With widespread availability of the test, increasing numbers of patients with incidentally found lesions are undergoing endoscopic ultrasonography (EUS). EUS can accurately determine the exact size of a lesion, as well as its precise layer of origin. GISTs are hypoechoic and most commonly arise from the 4th sonographic layer—the muscularis propria. Whether endosonographic features alone can aid in the differentiation of high- and low-risk tumors is less certain. Chak and colleagues found that presence of irregular borders, cystic spaces, and echogenic foci within tumors were independent risk factors associated with malignancy; however, interobserver agreement in interpretation of these features was poor to fair, thus calling into question whether these results can be applied to general practice (Gastrointest Endosc 1997;45:468–473). EUS-guided FNA of suspected GISTs has been used widely in attempts to obtain a diagnosis. A single study by Ando and colleagues compared EUS-FNA to subsequently resected specimens. Using the surgical specimens as the gold standard for malignancy or benignity, they found that EUS-FNA had an accuracy of 91.3% and specificity of 100% (Gastrointest Endosc 2002;55:37–43). Sensitivity, however, was only 66.7%, with 2 malignant lesions missed at FNA. This study also found that the Ki-67 labeling index, an immunohistochemical stain for mitotic activity, predicted malignancy in FNA specimens. The largest drawback of EUS evaluation is the typical inability to obtain sufficient tissue for histology and immunohistochemical staining, likely contributing to its poor sensitivity. Ongoing improvements in FNA/biopsy capabilities and echoendoscopes may allow for greater tissue yield during EUS in the not too distant future. Our understanding of the clinical behavior and malignant potential of GISTs has improved with a broader appreciation of their origin and biology. Up to 30% of GISTs may harbor malignancy (Hum Pathol 1999;30:1213–1220). The ambiguity in defining risk underscores the fact that virtually any GIST can behave in a malignant fashion and illustrates the dire need for clear prognostic markers. The NIH consensus group expressed hesitation in terming any GIST as definitively “benign.” In contrast, they stratified risk for aggressive behavior in groups ranging from “very low” to “high.” Historically, risk stratification of GISTs has been based solely on data gleaned from resected surgical specimens, and tumor size and mitotic count have been the only cited independent risk factors for malignancy. Thus, the 3-cm GIST found incidentally at upper endoscopy would be classified as “low” risk by the endoscopist. The malignant potential of such lesions, however, is uncertain at best, and there has been a case report of a 3.5-cm mass growing to 25 cm in just 2 years (Gastrointest Endosc 2003;58:616–618). This uncertainty has led to the search for alternate markers of malignancy. Previous studies have shown that exon 11 mutations occur preferentially in malignant GISTs (50% vs 5% in benign lesions), and that their presence confers a poorer prognosis (Am J Pathol 1999;154:53–60, Cancer Res 1999;59:4297–4300). In contrast, a more recent analysis found exon 11 mutations in 77% of subcentimeter tumors—lesions that traditionally have been thought to be benign (Am J Pathol 2002;160:1567–1572). Yet another group found somatic C-Kit mutations in 44 of 48 GISTs (10 benign, 10 borderline, and 28 malignant), including 10 of 10 benign GISTs, underscoring the possibility that C-Kit activation is not a marker for malignancy but instead a ubiquitous oncogenic pathway in these tumors. This leaves open the possibility that inhibitors of C-Kit kinase function could play a role in tumor management (Cancer Res 2001;61:8118–8121). It is possible that the varying results may be related to different methods used to detect mutations and different types of tissue analyzed. Research in this arena is of tremendous significance. The selective tyrosine kinase inhibitor imatinib mesylate has been shown to decrease tumor load by 50%–96% in metastatic GISTs (N Engl J Med 2001;347:472–480). A recent study found that tumors with exon 11 mutations responded better to imatinib therapy (J Clin Oncol 2003;21:4342–4349).The present study attempts to clarify the controversy by using mutational analysis of localized GISTs. The authors conclude that that the presence of c-kit mutations in GISTs is associated with higher mitotic counts and imparts a poorer prognosis. In the study, 6 of 8 incidentally found GISTs less than 2 cm in size had c-kit mutations. When the analysis included these tumors, the presence of c-kit mutation was still associated with poor RFS. Several points are noteworthy. Tumor size was not significantly different between negative and positive lesions. The average tumor size in the mutation-negative group was 6.4 cm compared with 7 cm in the mutation-positive group, and the average number of mitoses was 4.9 versus 16.5 per 50 HPF. If we then believe in the role of c-kit in predicting the likelihood of malignancy, it raises questions about the traditional practice of using size in selecting patients for surgery. Finally, the relatively short follow-up period precluded comprehensive analysis of overall survival as it relates to prognostic factors. The present study, however, does reinforce the notion that mutations in c-kit, high mitotic count, and larger tumor size are risk factors for poor prognosis in patients with localized GISTs and predict reduced RFS after perceived curative resection.Gastroenterologists often face the dilemma of small lesions in asymptomatic patients. The main challenge with GISTs is to distinguish lesions that need surgery from those that need periodic surveillance. What lies over the horizon that will help clarify the management of these lesions? Ongoing improvements in FNA/biopsy capabilities and echoendoscopes may result in greater tissue yield during EUS in the not-too-distant future. This would enable accurate preoperative assessment of risk of malignancy by histologic evaluation, immunohistochemical staining, and mutational analysis. One hopes that better outcomes will be achieved with use of molecular markers to improve patient selection for surgical resection and drugs such as imantinib mesylate in the adjuvant and neoadjuvant setting. Kim TW, Lee H, Kang Y-K, Choe MS, Ry M-H, Chang HM, Kim JS, Yook JH, Kim BS, Lee JS (Departments of Medicine, Diagnostic Pathology, and Surgery, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea). Prognostic significance of c-kit mutation in localized gastrointestinal stromal tumors. Clin Cancer Res 2004;10:3076–3081. Once grouped with leiomyomas, leiomyosarcomas, and other mesenchymal tumors of the gastrointestinal tract, gastrointestinal stromal tumors (GIST) are now considered distinct, arising from the Interstitial Cells of Cajal, and exhibit a variable clinical course. The vast majority of GISTs express the c-kit proto-oncogene protein, and gain-of-function mutations in this gene result in ligand-independent activation of c-kit tyrosine activity and subsequent oncogenesis. Because the clinical behavior of GISTs has been difficult to predict, a number of groups have sought to determine risk factors for aggressive tumors and thus poor prognosis. A National Institutes of Health (NIH)-sponsored GIST workshop in 2001 characterized risk of aggressive behavior (“very low”, “low”, “intermediate”, and “high”) based on tumor size and mitotic count, with tumor size >5 cm and mitoses > 5/50 high power fields (HPF) being indicators of high risk (Human Pathology 2002;33:459–465). Others have looked at the presence of mutations in the c-kit proto-oncogene (specifically exon 11, which encodes the juxtamembrane region of c-kit) as an independent factor for poor prognosis. Kim and colleagues report the results of a retrospective analysis of the prognostic significance of c-kit mutations located mainly in the stomach and small bowel in patients who had undergone curative resection for localized GIST. Using immunohistochemical staining for CD117, CD34, desmin, SMA, and S-100, these investigators identified 86 cases of true GIST treated at a single medical center between 1990 and 2001. The selected tumors were analyzed for histologic features and mitotic counts, and the patients were then stratified into groups based on the 2001 NIH consensus characterization of risk. Genomic DNA was extracted from each tumor, and polymerase chain reaction amplification of exon 11 was performed in each case. Samples negative for mutations in exon 11 mutations subsequently underwent analysis for exons 9, 13, and 17, which have been implicated in a minority of GISTs. Sixty-four of the 86 GISTs demonstrated mutations in c-kit. Of these, 61 had mutations in exon 11 and 3 had mutations in exon 9. No mutations in the other analyzed exons were found. The overall mutation rate of 71% was consistent with previous studies. Tumor size, histologic phenotype (spindle, epithelioid, mixed), cellularity (sparse, moderate, dense), tumor location, and mitotic count were compared in mutation-negative and mutation-positive lesions. Higher mitotic counts and higher cellularity were the only parameters that were significantly associated with the presence of c-kit mutations (P = .005 and .011, respectively). Mean number of mitoses per 50 HPF was 16.5 for mutation-positive lesions versus 4.9 in the negative tumors. Dense cellularity was found in 67% of positive lesions versus 41% of negative lesions. Average tumor size of c-kit mutation-positive lesions was greater (7.0 cm vs 6.4 cm), but this was not statistically significant. The authors also assessed 5-year relapse-free survival (RFS) using the Kaplan-Meier method and multivariate analysis. Median follow-up was 35.7 months for patients without relapse. The 5-year RFS trended toward being lower in patients with c-kit mutations than those without mutations (21% versus 60%, P = .098). Tumor size greater than 5 cm, mitotic count greater than 5/50 HPFs, epithelioid/mixed morphology, and nongastric GISTs were found to have significantly lower 5-year RFS. No difference between relapse-free survival rates of specific types of exon 11 mutations (ie, missense vs all other kinds) was found. Multivariate analyses led the authors to conclude that the presence of c-kit mutation (odds ratio [OR], 5.57; P = .023), mitoses >5/50 HPF (OR, 3.03; P = .018), and size >5 cm (OR, 4.23; P = .015) were independent indicators of a poorer 5-year RFS. CommentIt is estimated that about 5000–6000 new GISTs are identified in the United States yearly (Human Pathol 2002;33:459–465). As the utilization of endoscopy to investigate unrelated gastrointestinal symptoms increases, the number of small, incidentally discovered submucosal or subepithelial masses likely will grow. The diagnosis and management of GISTs poses a unique diagnostic challenge to the gastroenterologist. Tumors that are symptomatic, large, or felt to harbor malignancy require surgical excision. There also has been recent interest in the use of imantinib mesylate in individuals with malignant GISTs. Consequently, methods for determining the malignant nature or potential of these tumors are needed and must be evaluated for their performance characteristics. With widespread availability of the test, increasing numbers of patients with incidentally found lesions are undergoing endoscopic ultrasonography (EUS). EUS can accurately determine the exact size of a lesion, as well as its precise layer of origin. GISTs are hypoechoic and most commonly arise from the 4th sonographic layer—the muscularis propria. Whether endosonographic features alone can aid in the differentiation of high- and low-risk tumors is less certain. Chak and colleagues found that presence of irregular borders, cystic spaces, and echogenic foci within tumors were independent risk factors associated with malignancy; however, interobserver agreement in interpretation of these features was poor to fair, thus calling into question whether these results can be applied to general practice (Gastrointest Endosc 1997;45:468–473). EUS-guided FNA of suspected GISTs has been used widely in attempts to obtain a diagnosis. A single study by Ando and colleagues compared EUS-FNA to subsequently resected specimens. Using the surgical specimens as the gold standard for malignancy or benignity, they found that EUS-FNA had an accuracy of 91.3% and specificity of 100% (Gastrointest Endosc 2002;55:37–43). Sensitivity, however, was only 66.7%, with 2 malignant lesions missed at FNA. This study also found that the Ki-67 labeling index, an immunohistochemical stain for mitotic activity, predicted malignancy in FNA specimens. The largest drawback of EUS evaluation is the typical inability to obtain sufficient tissue for histology and immunohistochemical staining, likely contributing to its poor sensitivity. Ongoing improvements in FNA/biopsy capabilities and echoendoscopes may allow for greater tissue yield during EUS in the not too distant future. Our understanding of the clinical behavior and malignant potential of GISTs has improved with a broader appreciation of their origin and biology. Up to 30% of GISTs may harbor malignancy (Hum Pathol 1999;30:1213–1220). The ambiguity in defining risk underscores the fact that virtually any GIST can behave in a malignant fashion and illustrates the dire need for clear prognostic markers. The NIH consensus group expressed hesitation in terming any GIST as definitively “benign.” In contrast, they stratified risk for aggressive behavior in groups ranging from “very low” to “high.” Historically, risk stratification of GISTs has been based solely on data gleaned from resected surgical specimens, and tumor size and mitotic count have been the only cited independent risk factors for malignancy. Thus, the 3-cm GIST found incidentally at upper endoscopy would be classified as “low” risk by the endoscopist. The malignant potential of such lesions, however, is uncertain at best, and there has been a case report of a 3.5-cm mass growing to 25 cm in just 2 years (Gastrointest Endosc 2003;58:616–618). This uncertainty has led to the search for alternate markers of malignancy. Previous studies have shown that exon 11 mutations occur preferentially in malignant GISTs (50% vs 5% in benign lesions), and that their presence confers a poorer prognosis (Am J Pathol 1999;154:53–60, Cancer Res 1999;59:4297–4300). In contrast, a more recent analysis found exon 11 mutations in 77% of subcentimeter tumors—lesions that traditionally have been thought to be benign (Am J Pathol 2002;160:1567–1572). Yet another group found somatic C-Kit mutations in 44 of 48 GISTs (10 benign, 10 borderline, and 28 malignant), including 10 of 10 benign GISTs, underscoring the possibility that C-Kit activation is not a marker for malignancy but instead a ubiquitous oncogenic pathway in these tumors. This leaves open the possibility that inhibitors of C-Kit kinase function could play a role in tumor management (Cancer Res 2001;61:8118–8121). It is possible that the varying results may be related to different methods used to detect mutations and different types of tissue analyzed. Research in this arena is of tremendous significance. The selective tyrosine kinase inhibitor imatinib mesylate has been shown to decrease tumor load by 50%–96% in metastatic GISTs (N Engl J Med 2001;347:472–480). A recent study found that tumors with exon 11 mutations responded better to imatinib therapy (J Clin Oncol 2003;21:4342–4349).The present study attempts to clarify the controversy by using mutational analysis of localized GISTs. The authors conclude that that the presence of c-kit mutations in GISTs is associated with higher mitotic counts and imparts a poorer prognosis. In the study, 6 of 8 incidentally found GISTs less than 2 cm in size had c-kit mutations. When the analysis included these tumors, the presence of c-kit mutation was still associated with poor RFS. Several points are noteworthy. Tumor size was not significantly different between negative and positive lesions. The average tumor size in the mutation-negative group was 6.4 cm compared with 7 cm in the mutation-positive group, and the average number of mitoses was 4.9 versus 16.5 per 50 HPF. If we then believe in the role of c-kit in predicting the likelihood of malignancy, it raises questions about the traditional practice of using size in selecting patients for surgery. Finally, the relatively short follow-up period precluded comprehensive analysis of overall survival as it relates to prognostic factors. The present study, however, does reinforce the notion that mutations in c-kit, high mitotic count, and larger tumor size are risk factors for poor prognosis in patients with localized GISTs and predict reduced RFS after perceived curative resection.Gastroenterologists often face the dilemma of small lesions in asymptomatic patients. The main challenge with GISTs is to distinguish lesions that need surgery from those that need periodic surveillance. What lies over the horizon that will help clarify the management of these lesions? Ongoing improvements in FNA/biopsy capabilities and echoendoscopes may result in greater tissue yield during EUS in the not-too-distant future. This would enable accurate preoperative assessment of risk of malignancy by histologic evaluation, immunohistochemical staining, and mutational analysis. One hopes that better outcomes will be achieved with use of molecular markers to improve patient selection for surgical resection and drugs such as imantinib mesylate in the adjuvant and neoadjuvant setting. It is estimated that about 5000–6000 new GISTs are identified in the United States yearly (Human Pathol 2002;33:459–465). As the utilization of endoscopy to investigate unrelated gastrointestinal symptoms increases, the number of small, incidentally discovered submucosal or subepithelial masses likely will grow. The diagnosis and management of GISTs poses a unique diagnostic challenge to the gastroenterologist. Tumors that are symptomatic, large, or felt to harbor malignancy require surgical excision. There also has been recent interest in the use of imantinib mesylate in individuals with malignant GISTs. Consequently, methods for determining the malignant nature or potential of these tumors are needed and must be evaluated for their performance characteristics. With widespread availability of the test, increasing numbers of patients with incidentally found lesions are undergoing endoscopic ultrasonography (EUS). EUS can accurately determine the exact size of a lesion, as well as its precise layer of origin. GISTs are hypoechoic and most commonly arise from the 4th sonographic layer—the muscularis propria. Whether endosonographic features alone can aid in the differentiation of high- and low-risk tumors is less certain. Chak and colleagues found that presence of irregular borders, cystic spaces, and echogenic foci within tumors were independent risk factors associated with malignancy; however, interobserver agreement in interpretation of these features was poor to fair, thus calling into question whether these results can be applied to general practice (Gastrointest Endosc 1997;45:468–473). EUS-guided FNA of suspected GISTs has been used widely in attempts to obtain a diagnosis. A single study by Ando and colleagues compared EUS-FNA to subsequently resected specimens. Using the surgical specimens as the gold standard for malignancy or benignity, they found that EUS-FNA had an accuracy of 91.3% and specificity of 100% (Gastrointest Endosc 2002;55:37–43). Sensitivity, however, was only 66.7%, with 2 malignant lesions missed at FNA. This study also found that the Ki-67 labeling index, an immunohistochemical stain for mitotic activity, predicted malignancy in FNA specimens. The largest drawback of EUS evaluation is the typical inability to obtain sufficient tissue for histology and immunohistochemical staining, likely contributing to its poor sensitivity. Ongoing improvements in FNA/biopsy capabilities and echoendoscopes may allow for greater tissue yield during EUS in the not too distant future. Our understanding of the clinical behavior and malignant potential of GISTs has improved with a broader appreciation of their origin and biology. Up to 30% of GISTs may harbor malignancy (Hum Pathol 1999;30:1213–1220). The ambiguity in defining risk underscores the fact that virtually any GIST can behave in a malignant fashion and illustrates the dire need for clear prognostic markers. The NIH consensus group expressed hesitation in terming any GIST as definitively “benign.” In contrast, they stratified risk for aggressive behavior in groups ranging from “very low” to “high.” Historically, risk stratification of GISTs has been based solely on data gleaned from resected surgical specimens, and tumor size and mitotic count have been the only cited independent risk factors for malignancy. Thus, the 3-cm GIST found incidentally at upper endoscopy would be classified as “low” risk by the endoscopist. The malignant potential of such lesions, however, is uncertain at best, and there has been a case report of a 3.5-cm mass growing to 25 cm in just 2 years (Gastrointest Endosc 2003;58:616–618). This uncertainty has led to the search for alternate markers of malignancy. Previous studies have shown that exon 11 mutations occur preferentially in malignant GISTs (50% vs 5% in benign lesions), and that their presence confers a poorer prognosis (Am J Pathol 1999;154:53–60, Cancer Res 1999;59:4297–4300). In contrast, a more recent analysis found exon 11 mutations in 77% of subcentimeter tumors—lesions that traditionally have been thought to be benign (Am J Pathol 2002;160:1567–1572). Yet another group found somatic C-Kit mutations in 44 of 48 GISTs (10 benign, 10 borderline, and 28 malignant), including 10 of 10 benign GISTs, underscoring the possibility that C-Kit activation is not a marker for malignancy but instead a ubiquitous oncogenic pathway in these tumors. This leaves open the possibility that inhibitors of C-Kit kinase function could play a role in tumor management (Cancer Res 2001;61:8118–8121). It is possible that the varying results may be related to different methods used to detect mutations and different types of tissue analyzed. Research in this arena is of tremendous significance. The selective tyrosine kinase inhibitor imatinib mesylate has been shown to decrease tumor load by 50%–96% in metastatic GISTs (N Engl J Med 2001;347:472–480). A recent study found that tumors with exon 11 mutations responded better to imatinib therapy (J Clin Oncol 2003;21:4342–4349). The present study attempts to clarify the controversy by using mutational analysis of localized GISTs. The authors conclude that that the presence of c-kit mutations in GISTs is associated with higher mitotic counts and imparts a poorer prognosis. In the study, 6 of 8 incidentally found GISTs less than 2 cm in size had c-kit mutations. When the analysis included these tumors, the presence of c-kit mutation was still associated with poor RFS. Several points are noteworthy. Tumor size was not significantly different between negative and positive lesions. The average tumor size in the mutation-negative group was 6.4 cm compared with 7 cm in the mutation-positive group, and the average number of mitoses was 4.9 versus 16.5 per 50 HPF. If we then believe in the role of c-kit in predicting the likelihood of malignancy, it raises questions about the traditional practice of using size in selecting patients for surgery. Finally, the relatively short follow-up period precluded comprehensive analysis of overall survival as it relates to prognostic factors. The present study, however, does reinforce the notion that mutations in c-kit, high mitotic count, and larger tumor size are risk factors for poor prognosis in patients with localized GISTs and predict reduced RFS after perceived curative resection. Gastroenterologists often face the dilemma of small lesions in asymptomatic patients. The main challenge with GISTs is to distinguish lesions that need surgery from those that need periodic surveillance. What lies over the horizon that will help clarify the management of these lesions? Ongoing improvements in FNA/biopsy capabilities and echoendoscopes may result in greater tissue yield during EUS in the not-too-distant future. This would enable accurate preoperative assessment of risk of malignancy by histologic evaluation, immunohistochemical staining, and mutational analysis. One hopes that better outcomes will be achieved with use of molecular markers to improve patient selection for surgical resection and drugs such as imantinib mesylate in the adjuvant and neoadjuvant setting. ReplyGastroenterologyVol. 128Issue 7PreviewAs imatinib mesylate has been proven effective in treating metastatic or recurrent gastrointestinal stromal tumors (GISTs), its role in postoperative adjuvant setting has become a hot topic, and several phase III clinical trials are currently underway worldwide to address this issue. Generally speaking, the best candidates for postoperative adjuvant treatment would be those whose tumors have not only a high risk of relapse but also a high likelihood of responding to the treatment. Full-Text PDF