in c-kit or alternative PDGFRA proto-oncogenes. 2-7 These mutant genes confer constitutive kinase activity

A modification of NIH consensus criteria to better distinguish the highly lethal subset of primary localized gastrointestinal stromal tumors: A subdivision of the original high-risk group on the basis of outcome Hsuan-Ying Huang, MD, a Chien-Feng Li, MD, b Wen-Wei Huang, MD, c Tsung-Hui Hu, MD, d Ching-Nan Lin, MD, b Yih-Huei Uen, MD, e Ching-Yeh Hsiung, MD, f and David Lu, MD, g Tainan, Taiwan Background. By reappraising the National Institutes of Health (NIH) consensus criteria, we worked on establishing a modified scheme to identify highly lethal gastrointestinal stromal tumors (GISTs), which have an imperative demand for sequencing analysis to assess the suitability of an adjuvant imatinib trial. Methods. Clinicopathologic features, including NIH and modified schemes, were retrospectively analyzed for 289 patients with localized GISTs. We combined the very low/low-risk GISTs into a single risk level I group ( 5 cm, 5/50 high power fields [HPFs]) and redesignated the intermediate-risk GISTs as risk level II (either 5 cm, 6 to 10/50 HPFs or 5 to 10 cm, 5/50 HPFs). The GISTs of risk level IV group were 5 cmand 10/50 HPF, with the rest of high-risk GISTs defined as risk level III. Results. The cumulative 5-year rate of disease-specific survival (DSS) for all 289 patients was 82%, and the DSS rates for patients with GISTs classified as risk levels I to IV were 100%, 96%, 67%, and 25% at 5 years, respectively. The prognostic differences were striking between the risk level II and III groups (P.0001) and between the risk level III and IV groups (P.0002). The higher risk level of our scheme represented the strongest independent adverse factor (risk ratio [RR] 11.299 for risk level III; RR 33.815 for risk level IV; P.0001), followed by mixed/epithelioid histology (RR 2.837, P.003) and older age ( 70 years, RR 1.955, P.044). Conclusions. Remarkable prognostic heterogeneity exists in the high-risk category of the NIH scheme, which is not as effective as the modified criteria in identifying highly lethal GISTs that we classified as risk level IV. (Surgery 2007;141:748-56.) From the Departments of Pathology a and Radiation Oncology f and Division of Gastroenterology, d Department of Internal Medicine, Chang Gung Memorial Hospital-Kaohsiung Medical Center, Chang Gung University College of Medicine, Kaohsiung; Departments of Pathology b and Surgery, e Chi-Mei Foundation Medical Center, Tainan; Department of Family Medicine, c College of Medicine, National Cheng Kung University, Tainan; and Department of Pathology, g Chi-Mei Foundation Hospital, Liouying Campus, Tainan, Taiwan Presented in part at the 95th annual meeting of United States and Canadian Academy of Pathology, Atlanta, Georgia, 2006. Supported in part by grants from National Science Council, Taiwan (NSC94-2320-B-182A-005) and Chi-Mei Medical Center (CMFHR9359). Accepted for publication January 7, 2007. Reprint requests: David Lu, MD, Department of Pathology, Chi- Mei Foundation Hospital, Liouying Campus, 201, Taikang Village, Liouying Township, Tainan County, Taiwan. E-mail: angelo.p@ yahoo.com.tw. Gastrointestinal stromal tumors (GISTs) constitute the most common primary mesenchymal tumors of the digestive tract and characteristically express CD117 in most cases. 1-3 Central to the tumorigenesis of GISTs are gain-of-function mutations in c-kit or alternative PDGFRA proto-oncogenes. 2-7 These mutant genes confer constitutive kinase activity 0039-6060/$ - see front matter 2007 Mosby, Inc. All rights reserved. doi:10.1016/j.surg.2007.01.024 748 SURGERY

Surgery Huang et al 749 Volume 141, Number 6 to drive tumor development, which is the rationale for applying imatinib mesylate (a selective tyrosine kinase inhibitor) for targeted therapy. 7-10 Although most GISTs initially manifest as localized resectable lesions, disease relapse is not unlikely even if tumors are completely excised in primary surgery. 1,2,11-13 The success of imatinib mesylate in treating metastatic and/or unresectable GISTs that harbor certain types of c-kit or PDGFRA mutations has prompted the use of postoperative adjuvant therapy for aggressive primary localized tumors. 14-16 However, the potential risk of unscreened adjuvant therapy is acquired resistance to imatinib after a short-term, initial benefit, which was recently shown to occur via secondary mutations in c-kit kinase domain 17-19 or, less commonly, amplification of c-kit or c-kit/pdgfra gene. 19 From the perspectives of cost-effectiveness and drug safety, it is therefore highly desirable to better delineate which GISTs are truly highly lethal with an imperative requirement of sequencing analysis for assessing the suitability of an adjuvant imatinib trial. Despite new insights into the pathogenesis of GISTs, they have been considered as prognostically treacherous with a wide clinical spectrum ranging from indolent to highly aggressive behavior. 1-3,11,20 Up to 30% of newly diagnosed GISTs have features connoting a high malignant potential, whereas a minority of small and mitotically inactive GISTs may still behave aggressively. 1-3,11,20 Based on the combination of mitotic rate and tumor size, the National Institutes of Health (NIH) recently proposed a consensus scheme to stratify patients with GISTs into 4 risk categories: very low-risk, low-risk, intermediate-risk, and high-risk groups, irrespective of tumor sites. 2 Accordingly, GISTs with both size 5 cm and mitotic count 5/50 high power fields (HPFs) are regarded as at high risk for clinical aggressiveness, together with any tumor featuring either large size ( 10 cm) or high mitotic activity ( 10/50 HPFs). 2 However, Miettinen et al 5 more recently reported their new risk scheme by stratifying GISTs into more (1 to 6b) prognostic groups on the basis of more than 1700 patients with long-term follow-up. Few, if any, studies have evaluated GISTs of a sufficiently large sample size to examine the prognostic efficacy of the NIH consensus system. 21-23 As for high-risk GISTs, we have noted in our practice that tumors with high mitotic activity have appeared to behave worse than their large but mitotically inactive counterparts behave. This observation is in agreement with the large series studied by Miettinen et al, 5 who stated that tumor-specific mortality was relatively low for tumors 10 cm with mitotic activity 5/50 HPFs (group 3b in their classification). However, they did not specifically address whether small but mitotically active tumors ( 5 cm, 10/50 HPFs) would behave differently. If the previously stated prognostic heterogeneity among high-risk GISTs were confirmed, it would call for modification of the current NIH scheme. By reappraising the clinicopathologic findings related to outcomes in 289 GISTs treated in the preimatinib era, the aims of this study were to (1) assess the potential limitations in individual risk category for refining the criteria of the current NIH scheme, and (2) to define the highly aggressive subset with a top priority for sequencing analysis to guide adjuvant targeted therapy. MATERIALS AND METHODS Case selection. We retrieved 339 consecutively resected gastrointestinal mesenchymal tumors between 1984 and 2002 from the archives of 2 tertiary medical centers in Southern Taiwan. After histologic review, 32 tumors were reclassified as entities simulating GISTs and were excluded from the study. A total of 307 histologically typical GISTs were verified by strong reactions to c-kit (1:300; DAKO, Carpenteria, Calif) (Fig 1, A) and/or positive PKC- staining (1:100, 27; BD Pharmingen, San Diego, Calif) for a minor subset with no or weak c-kit expression. 24 Institutional review boards of both centers approved retrospective clinical data collection. By directly contacting patients themselves or families, follow-up information was available in 289 patients, who did not receive imatinib therapy until last appointment and formed the cohort of statistical analyses. All patients with primary GISTs resected after the year 2000 were scheduled to visit the clinics at 3-month intervals. Abdominal computed tomography (CT) was performed within 3 months after primary surgery for high-risk patients or when there were clinical indications. Assessment of clinicopathological variables and proposed modified scheme. Clinical factors recorded for statistical analyses included sex, age, tumor size, and location (gastric vs intestinal). The morphologic features analyzed were histologic type (spindle, epithelioid, or mixed), cellularity (low, moderate, or high), nuclear pleomorphism (mild, moderate, and prominent), mitotic count/50 HPFs, and microscopic margin status. The criteria to determine histologic type have been reported previously. 2,25 Cellularity was semiquantified as low ( 25% cells, 75% stroma), moderate (cells vs stroma between 25% and 75%), or high ( 75% cells, 25% stroma). Prominent nuclear pleomorphism was defined as a marked variation in

750 Huang et al Surgery June 2007 Fig 1. According to our modified criteria, representative GISTs with increasing mitoses and cellularity were verified by c-kit staining (A) and classified as risk level I (B, C), risk level II (D), risk level III (E), and risk level IV (F), respectively. Table I. Criteria of the original vs modified NIH schemes to define aggressiveness of GISTs Original NIH scheme Modified scheme Very low-risk Risk level I 2 cm, 5/50 HPF 5 cm, 5/50 HPF Low-risk 2-5 cm, 5/50 HPF Intermediate-risk Risk level II 5 cm, 6-10/50 HP 5 cm, 6-10/50 HPF 5-10 cm, 5/50 HPF 5-10 cm, 5/50 HPF High-risk Risk level III 5 cm, 5/50 HPF 5 cm, 10/50 HPF 10 cm, any mitosis 5-10 cm, 6-10/50 HPF any size, 10/50 HPF 10 cm, 5/50 HPF Risk level IV 5 cm, 10/50 HPF NIH, National Institutes of Health; GIST, gastrointestinal stromal tumor; HPF, high power field. nuclear size and shape with the presence of tumor giant cells. Using Olympus BX40 microscope (field area, 0.237 mm 2 [Olympus, Tokyo, Japan]), mitotic activity was assessed by reporting the number of mitotic figures observed in the most mitotically active area for 50 wide-filed HPFs ( 40). Margin status was considered positive when gross residual disease was found at operation or when tumor cells were within 1 mm from the inked end on microscopic examination. As shown in Table I, the risk category of each GIST was first determined according to NIH consensus criteria proposed by Fletcher et al. 2 The statistical powers of combining or dividing various risk categories of the NIH system in predicting patient survival were accordingly examined, by which we proposed a modified scheme with better prognostic efficacy. In brief, the very low-risk and low-risk GISTs were combined into a single risk level I group, and the intermediate-risk GISTs were retained and redesignated as risk level II. The highly lethal risk level IV group referred to GISTs with size 5 cm and mitoses 10/50 HPFs, whereas the rest of the original high-risk GISTs comprised the risk level III group. Therefore, this new scheme consists of 4 risk levels to predict clinical aggressiveness of GISTs (Table I). Statistical analyses. Statistical analyses were performed using the SPSS 10 software package (SPSS Corp, Chicago, Ill). Survival data were calculated from the date of primary surgery to the date of death or last review. For survival analyses, the endpoint evaluated was disease-specific survival (DSS). All time-to-event endpoints according to various clinicopathologic factors, including various combinations or divisions of NIH risk categories, were plotted by the Kaplan-Meier method and compared by log-rank tests. In principle, those significant parameters in univariate analysis (P.05) were entered into multivariate Cox regression model to determine the risk ratio (RR) and independence of prognostic impact in a stepwise backward fashion. Nevertheless, as component factors of both the original and modified schemes, tumor size and mitotic count were not analyzed for multivariate comparisons. Furthermore, we only adopted our modified criteria rather than the original NIH consensus scheme in Cox regression analysis because it proved more effective in distinguishing prognostic

Surgery Huang et al 751 Volume 141, Number 6 differences among GISTs (see Results for details). For all analyses, two-sided tests of significance were used with P.05 considered statistically significant. RESULTS Clinicopathologic findings. Salient clinicopathologic features are summarized in Table II. The median age for the entire cohort of 289 patients, 140 males and 149 females, was 61 years (range, 23 to 87 years), with 79 patients (27.3%) aged 70 years. There were 177 gastric and 112 intestinal tumors, including 91 and 21 located in the small bowel and rectum, respectively. The tumor size ranged from 0.6 to 30 cm (median, 5.5 cm), with 130 (45.0%), 104 (36.0%), and 55 (19.0%) cases being 5 cm, 5 to 10 cm, and 10 cm, respectively. Histologically, 224 (77.5%) tumors were pure spindle type and 65 tumors (22.5%) showed at least focal epithelioid histology. Hypercellularity was observed in 54 patients (18.7%), whereas prominent nuclear pleomorphism was present in 23 patients (8.0%). Mitotic counts revealed a wide variation from 0 to 184 (median, 2) per 50 HPF and were 0 to 5/50 HPF in 208 patients (72.0%), 6 to 10/50 HPF in 37 (12.8%), and 10/50 HPF in 44 (15.2%). Resection margins were positive in 54 patients (18.9%), including large, ruptured tumors with gross residual disease and those treated by wedge resection with microscopically unclear margins. According to the original NIH consensus criteria, 16 (5.5%), 92 (31.8%), 88 (30.4%), and 93 (32.2%) of GISTs were classified into very low-risk groups, low-risk groups, intermediate-risk groups, and high-risk groups, respectively. By using our modified scheme, there were 108 (37.4%) GISTs designated as risk level I (Fig 1, B, C); 88 (30.4%) as risk level II (Fig 1, D); 61 (21.1%) as risk level III (Fig 1, E); and 32 (11.1%) as risk level IV (Fig 1, F). Follow-up and patient outcomes. Clinical follow-up was available in all 289 patients with primary localized GISTs; follow-up interval ranged from 6 to 231 months (median, 41 months). At last follow-up, 49 patients (16.8%) experienced tumor relapses with development of local recurrences in 24 and metastases in 33 to the liver (n 23), peritoneum (n 7), lymph node (n 1), and bone (n 1), respectively. Relapse events occurred in 3 (3.3%), 9 (10.2%), 11 (18.0%), 26 (81.3%) of 108 cases classified as risk level I, 88 cases as risk level II, 61 cases as risk level III, and 32 cases as risk level IV, respectively. No patients with GISTs of risk level I died of disease, while 3.4% (3/88) of risk level II cases, 16.4% (10/61) of risk level III cases, and 78.1% (25/32) of risk level IV cases died of their GISTs. Table II. Clinicopathologic features of 289 primary localized GISTs Parameters No. of cases (%) Sex Male 140 (48.4%) Female 149 (51.6%) Age (y)(mean: 60; median: 61; range: 23-87) 40 23 (8.0%) 40-49 36 (12.5%) 50-59 71 (24.6%) 60-69 80 (27.7%) 70-79 61 (21.1%) 80 18 (6.2%) Location Gastric 177 (61.2%) Intestinal 112 (38.8%) Tumor size (cm)(mean: 6.4, median: 5.5, range: 0.6-30) 5 cm 130 (45.0%) 5 to 10 cm 104 (36.0%) 10 cm 55 (19.0%) Histologic type Spindle 224 (77.5%) Epithelioid and mixed 65 (22.5%) Cellularity Low 46 (15.9%) Moderate 189 (65.4%) High 54 (18.7%) Nuclear pleomorphism No/mild 266 (92.0%) Prominent 23 (8.0%) Mitotic rate (per 50 HPF) (mean: 8.79, median: 2, range: 0-184) 5 208 (72.0%) 5 but 10 37 (12.8%) 10 44 (15.2%) Margin status Negative 232 (81.1%) Positive 54 (18.9%) Unknown 3 Risk categories Original NIH scheme Very low 16 (5.5%) Low 92 (31.8%) Intermediate 88 (30.4%) High 93 (32.2%) Modified scheme Risk level I 108 (37.3%) Risk level II 88 (30.4%) Risk level III 61 (21.1%) Risk level IV 32 (11.1%) NIH, National Institutes of Health; GIST, gastrointestinal stromal tumor; HPF, high power field.

752 Huang et al Surgery June 2007 Table III. Univariate survival analyses for diseasespecific survival Parameters No. patients No. events P values Age (y).0091* 70 210 21 70 79 17 Sex.1703 Female 149 16 Male 140 22 Location.1587 Gastric 177 19 Intestinal 112 19 Size.0001* 5 cm 130 2 5to10cm 104 14 10 cm 55 22 Margin.4663 Positive 54 5 Negative 232 33 Histologic type.0001* Spindle 224 18 Epithelioid and mixed 65 20 Cellularity.0012* Low/moderate 235 23 High 54 15 Nuclear pleomorphism.5861 Mild/moderate 266 34 Prominent 23 4 Mitotic count/50 HPFs.0001* 5 208 10 5 but 10 37 7 10 44 21 NIH consensus scheme.0001* Very low/low-risk 108 0 Intermediate-risk 88 3 High-risk 93 35 Modified scheme.0001* Risk level I 108 0 Risk level II 88 3 Risk level III 61 16 Risk level IV 32 19 NIH, National Institutes of Health; HPF, high power field. *Statistically significant. Univariate survival analyses. Correlations between clinicopathologic parameters and patient survival are presented in Table III and Figs 2 and 3; the cumulative 5-year rate of DSS for all 289 patients was 82%. The following variables were found significantly associated with poor DSS: larger tumor size (P.0001, Fig 2, A), higher mitotic count (P.0001, Fig 2, B), presence of epithelioid histology (P.0001, Fig 2, C), hypercellularity (P.0012), and older age ( 70 years, P.0091, Fig 2, D). However, sex, tumor location, marginal status, and prominent nuclear pleomorphism were not related to patient survival. By virtue of no difference in outcome, very low-risk and low-risk GISTs of the NIH system were grouped together (ie, risk level I of the modified scheme). On univariate survival analyses, we found that higher risk categories defined by both schemes (P.0001) were significantly predictive of adverse outcomes (Fig 3, A, B). In subset analyses, we found that high mitotic activity ( 10/50 HPFs, P.0127), rather than large tumor size ( 10 cm, P.2467), is significantly predictive of inferior DSS for NIH-defined high-risk GISTs. This finding indicated that the original high-risk category might represent a group of prognostically heterogeneous tumors, which showed a cumulative DSS rate of 51% at 5 years. However, by using both tumor size 5 cm and mitoses 10/50 HPFs as the threshold to dichotomize this high-risk group, we substantiated that GISTs of risk level IV were more highly lethal than GISTs of risk level III. The 5-year rate of DSS was 25.4% for risk level IV vs. 67.0% for risk level III (P.0002, Fig 3, B). According to our modified scheme, the prognostic discrimination in DSS between risk level II and risk level III cases still remained very robust (P.0001, Fig 3, B). The findings confirmed that significant prognostic heterogeneity exists in high-risk GISTs of NIH consensus criteria, which is not as effective as our modified scheme. Therefore, only the modified scheme and other significant factors, except for tumor size and mitoses, were entered for multivariate analyses. Multivariate survival analyses. Given no diseasespecific death event in patients with risk level I GISTs, they were combined with patients with risk level II GISTs to form a joint group for multivariate comparisons (Table IV). As for GIST-related DSS, higher risk levels of modified NIH scheme represented the strongest negative factor (P.0001). Of note, the level III and level IV categories identified patients at 11.299-fold and 33.815-fold increased risks in comparison to the joint level I/level II group, substantiating that our modified scheme is highly representative of biologic aggressiveness of GISTs. In addition, presence of epithelioid histology (P.003, RR 2.837) and age 70 years (P.044, RR 1.955) were also independently associated with a poor outcome, whereas hypercellularity lost statistical significance. DISCUSSION Metastatic or intraabdominally disseminated GISTs have been a virtually incurable disease until the elucidation of the pathogenetic role of c-kit

Surgery Huang et al 753 Volume 141, Number 6 Fig 2. Kaplan-Meier plots to predict disease-specific survival according to tumor size (A), mitotic count (B), histologic type (C), and age at presentation (D). Fig 3. A comparison of prognostic efficacy between the original (A) and modified (B) schemes in relation to disease-specific survival. mutations. Imatinib mesylate is a potent and specific inhibitor of various receptor tyrosine kinases, including c-kit receptor. 3,4,6,7,10 The dramatic results of imatinib therapy in metastatic or unresectable GISTs have shifted the focus of research in localized GISTs from prediction of aggressiveness to proper selection of patients for adjuvant therapy. 14-16 However, adjuvant imatinib therapy should be applied with great caution because of the potential risk of acquired drug resistance and its enormous cost. 15,17-19,26 These concerns are conceivable because a considerable proportion, ranging from 24% to 45%, of GISTs were classified as high-risk based on the data of various studies using the original NIH consensus scheme. 21-23,27,28 Because of the concerns regarding drug safety and cost-effectiveness, sequencing analysis to assess the potentiality of adjuvant targeted therapy might be comparatively more imperative for a subset of highrisk GISTs that are proven to be highly lethal. To improve risk-based stratification for proper adjuvant imatinib therapy, we retrospectively reappraised the prognostic value of original NIH consensus scheme compared to our modified system, in combination with other critical clinicopathologic features, for 289 primary localized GISTs. The 5-year DSS rate of the 289 GISTs from various sites used in our study was 82% and reached a plateau

754 Huang et al Surgery June 2007 Table IV. Results of Cox multivariate analyses in relation to disease-specific survival Variable Category RR 95% CI P value Modified scheme Risk levels I and II 1.00 -.0001 Risk level III 11.299 3.697-34.531 Risk level IV 33.815 10.945-104.472 Histologic type Spindle 1.00 -.003* Epithelioid and mixed 2.837 1.438-5.596 Age (y) 70 1.00 -.044* 70 1.955 1.019-3.750 Cellularity Low/moderate 1.00 -.768 High 0.901 0.451-1.801 RR, Relative risk; CI, confidence interval. *Statistically significant. thereafter, which is similar to the results demonstrated in other published reports. 5,12,21-23,25,28 In survival analyses, we substantiated that our modified scheme was more effective than the original NIH criteria, given its stronger discrimination in identifying highly lethal GISTs at the aggressive end of the clinical spectrum. Furthermore, the modified scheme also represented the strongest independent adverse factor of DSS, followed by histologic subtype and age at presentation. With regard to most mesenchymal malignancies, the status of microscopic margin of primary tumors is only associated with recurrence-free survival but not predictive of disease-related mortality. 29 This also holds true for primary localized GISTs, because positive margins, albeit not independent, were previously found significantly related to decreased relapse-free survival on univariate analysis. 25 For primary localized GISTs undergoing complete gross resection, DeMatteo et al 12 found the status of microscopic margin did not affect DSS. Despite no significant prognostic impact of margin status in our series, it should be noted that a retrospective study might contain a potential bias regarding how to clearly distinguish grossly positive from microscopically positive margins for a subset of GISTs. The prognostic impact of histologic subtype on GISTs has not yet been fully elucidated and has remained a matter of debate in terms of its independence of statistical significance. 5,20,22,25,30 Several studies demonstrated that pure spindle histology represented a favorable prognosticator for various survival endpoints at the univariate level alone, 22,30 whereas the presence of epithelioid histology in the current series was independently predictive of inferior DSS. Our finding was more concordant with the report of Singer et al 25 that identified mixed spindle/epithelioid histology, together with high mitotic activity, as one of the two most important independent adverse factors for recurrence-free survival. Few groups have suggested that age at presentation is a significant prognosticator of patient survival in GISTs. 11,31 In our analysis, we found that age 70 years independently correlated with DSS, with an approximately 2-fold increased risk. This finding was in keeping with a Japanese series by Hasegawa et al, 31 who reported that age 60 years independently correlated with inferior overall survival. We consider that the prognostic effect of age is less likely related to ethnic difference, because a large US series also addressed its strong, independent prognostic value. 11 Furthermore, a recent population-based study by a Sweden group demonstrated the additive adverse effect of increasing age in GIST mortality, regardless of risk scores (Fig 4 in their article). 28 More likely, the influence of an older age was mainly attributed to the bias that more conservative therapeutic strategy was usually preferred in these patients should tumor relapses occur. The original NIH consensus scheme by Fletcher et al 2 did not consider the effect of tumor site, 2 although other investigators previously demonstrated a seemingly more aggressive course in intestinal lesions. 11,32,33 This contention appears to be an area of divergent viewpoints. Based on the extremely large cohort of over 2600 GISTs, Miettinen et al 5,34 recently asserted that the tumor-related mortality of small intestinal GISTs was 2 times higher than gastric GISTs. Although other researchers including ourselves could not reproduce this finding, it must be noted that none of these series enrolled more than 300 cases, 23,28,35 including one population-based study. 28 Our finding that the original high-risk group comprised GISTs with remarkable prognostic heterogeneity was of great consequence in prediction of clinical behavior and patient-tailored therapeutic strategy. This limita-

Surgery Huang et al 755 Volume 141, Number 6 tion was at least in part ascribed to improper lumping of large, mitotically inactive GISTs ( 10 cm, 5 mitotes/50 HPF) within the high-risk category, as already noted by Miettinen et al. 5 Despite mitotic rate being well recognized as an important prognostic factor of GISTs, 1,2,11,20 it remains less clear how many and what cutoff points of mitotic rate should be applied to establish an effective risk scheme. In this study, only high mitotic activity ( 10 mitoses/50 HPF), rather than large size, significantly correlated with DSS for patients with high-risk GISTs. This finding suggests the insufficiency of using only 5 mitoses/50 HPF as the highest cutoff to risk-stratify GISTs, which has been used in some prior studies. 1,20 This contention was supported by Singer et al, 25 who found an even higher cutoff of mitotic rate (ie, 15 mitoses/30 HPF) as the strongest independent prognosticator and no statistical significance of tumor size on multivariate analysis after adjusting for mitotic activity and histologic subtype. By using our modified criteria, both tumor relapses and mortality significantly escalated with the increment of risk levels, thus substantiating that our proposal to subdivide high-risk GISTs into risk level III and risk level IV groups more accurately predicted clinical outcomes. Only 3 cases (2.8%, 1 gastric and 2 small intestinal) among 108 GISTs of risk level I experienced tumor relapses, whereas 10.2%, 18.0%, and 81.3% of risk level II, risk level III, and risk level IV cases, respectively, did. These results reinforce our finding that our modified scheme is highly representative of biologic aggressiveness and tumor progression of GISTs. In summary, our proposed modified criteria can more effectively distinguish prognostic heterogeneity in high-risk GISTs of the original system. It is extremely crucial to accurately identify individual patients with GISTs who are at exceptionally high risk because this consideration has become particularly relevant with the advent of imatinib mesylate. More specifically, a few phase III prospective randomized trials of adjuvant imatinib therapy (eg, ACOSOG Z9001 and EORTC 62024) are currently ongoing for primary localized GISTs of high and moderate risk. 36 These therapeutic approaches may provide significant survival benefits for GISTs classified into different risk groups similar to those used for unresectable GISTs. At least in this and other studies, 5 large, mitotically inactive GISTs do not necessarily follow a very aggressive clinical course. Moreover, large tumor size has been proven to be independently associated with late resistance to imatinib therapy and occurrence of skin toxicity in treating advanced and/or metastatic GISTs. 37,38 Therefore, we recommend that GISTs with tumor size 5 cm and mitoses 10/50 HPFs be specified as the risk level IV category; these GISTs are highly lethal with a top priority for genetic analysis to assess the suitability of postoperative adjuvant imatinib therapy. REFERENCES 1. Miettinen M, El-Rifai W, L HLS, Lasota J. Evaluation of malignancy and prognosis of gastrointestinal stromal tumors: a review. Hum Pathol 2002;33:478-83. 2. Fletcher CD, Berman JJ, Corless C, Gorstein F, Lasota J, Longley BJ, et al. Diagnosis of gastrointestinal stromal tumors: a consensus approach. Hum Pathol 2002;33:459-65. 3. Corless CL, Fletcher JA, Heinrich MC. Biology of gastrointestinal stromal tumors. J Clin Oncol 2004;22:3813-25. 4. Martin J, Poveda A, Llombart-Bosch A, Ramos R, Lopez- Guerrero JA, Garcia del Muro J, et al. Deletions affecting codons 557-558 of the c-kit gene indicate a poor prognosis in patients with completely resected gastrointestinal stromal tumors: a study by the Spanish Group for Sarcoma Research (GEIS). J Clin Oncol 2005;23:6190-8. 5. Miettinen M, Sobin LH, Lasota J. Gastrointestinal stromal tumors of the stomach: a clinicopathologic, immunohistochemical, and molecular genetic study of 1765 cases with long-term follow-up. Am J Surg Pathol 2005;29:52-68. 6. Hirota S, Ohashi A, Nishida T, Isozaki K, Kinoshita K, Shinomura Y, et al. Gain-of-function mutations of plateletderived growth factor receptor alpha gene in gastrointestinal stromal tumors. Gastroenterology 2003;125:660-7. 7. Corless CL, Schroeder A, Griffith D, Town A, McGreevey L, Harrell P, et al. PDGFRA mutations in gastrointestinal stromal tumors: frequency, spectrum and in vitro sensitivity to imatinib. J Clin Oncol 2005;23:5357-64. 8. Heinrich MC, Blanke CD, Druker BJ, Corless CL. Inhibition of KIT tyrosine kinase activity: a novel molecular approach to the treatment of KIT-positive malignancies. J Clin Oncol 2002;20:1692-703. 9. Heinrich MC, Corless CL, Demetri GD, Blanke CD, von Mehren M, Joensuu H, et al. Kinase mutations and imatinib response in patients with metastatic gastrointestinal stromal tumor. J Clin Oncol 2003;21:4342-9. 10. Dagher R, Cohen M, Williams G, Rothmann M, Gobburu J, Robbie G, et al. Approval summary: imatinib mesylate in the treatment of metastatic and/or unresectable malignant gastrointestinal stromal tumors. Clin Cancer Res 2002;8: 3034-8. 11. Emory TS, Sobin LH, Lukes L, Lee DH, O Leary TJ. Prognosis of gastrointestinal smooth-muscle (stromal) tumors: dependence on anatomic site. Am J Surg Pathol 1999; 23:82-7. 12. DeMatteo RP, Lewis JJ, Leung D, Mudan SS, Woodruff JM, Brennan MF. Two hundred gastrointestinal stromal tumors: recurrence patterns and prognostic factors for survival. Ann Surg 2000;231:51-8. 13. Li CF, Chuang SS, Lu CL, Lin CN. Gastrointestinal stromal tumor (GIST) in southern Taiwan: a clinicopathologic study of 93 resected cases. Pathol Res Pract 2005;201:1-9. 14. Benjamin RS, Blanke CD, Blay JY, Bonvalot S, Eisenberg B. Management of gastrointestinal stromal tumors in the imatinib era: selected case studies. Oncologist 2006;11:9-20. 15. Blay JY, Bonvalot S, Casali P, Choi H, Debiec-Richter M, Dei Tos AP, et al. Consensus meeting for the management of gastrointestinal stromal tumors. Report of the GIST Con-

756 Huang et al Surgery June 2007 sensus Conference of 20-21 March 2004, under the auspices of ESMO. Ann Oncol 2005;16:566-78. 16. van der Zwan SM, DeMatteo RP. Gastrointestinal stromal tumor: 5 years later. Cancer 2005;104:1781-8. 17. Antonescu CR, Besmer P, Guo T, Arkun K, Hom G, Koryotowski B, et al. Acquired resistance to imatinib in gastrointestinal stromal tumor occurs through secondary gene mutation. Clin Cancer Res 2005;11:4182-90. 18. Chen LL, Trent JC, Wu EF, Fuller GN, Ramdas L, Zhang W, et al. A missense mutation in KIT kinase domain 1 correlates with imatinib resistance in gastrointestinal stromal tumors. Cancer Res 2004;64:5913-9. 19. Debiec-Rychter M, Cools J, Dumez H, Sciot R, Stul M, Mentens N, et al. Mechanisms of resistance to imatinib mesylate in gastrointestinal stromal tumors and activity of the PKC412 inhibitor against imatinib-resistant mutants. Gastroenterology 2005;128:270-9. 20. Trupiano JK, Stewart RE, Misick C, Appelman HD, Goldblum JR. Gastric stromal tumors: a clinicopathologic study of 77 cases with correlation of features with nonaggressive and aggressive clinical behaviors. Am J Surg Pathol 2002; 26:705-14. 21. Kim TW, Lee H, Kang YK, Choe MS, Ryu MH, Chang HM, et al. Prognostic significance of c-kit mutation in localized gastrointestinal stromal tumors. Clin Cancer Res 2004;10: 3076-81. 22. Feakins RM. The expression of p53 and bcl-2 in gastrointestinal stromal tumours is associated with anatomical site, and p53 expression is associated with grade and clinical outcome. Histopathology 2005;46:270-9. 23. Nakamura N, Yamamoto H, Yao T, Oda Y, Nishiyama K, Imamura M, et al. Prognostic significance of expressions of cell-cycle regulatory proteins in gastrointestinal stromal tumor and the relevance of the risk grade. Hum Pathol 2005; 36:828-37. 24. Blay P, Astudillo A, Buesa JM, Campo E, Abad M, Garcia- Garcia J, et al. Protein kinase C theta is highly expressed in gastrointestinal stromal tumors but not in other mesenchymal neoplasias. Clin Cancer Res 2004;10(12 Pt 1):4089-95. 25. Singer S, Rubin BP, Lux ML, Chen CJ, Demetri GD, Fletcher CD, et al. Prognostic value of KIT mutation type, mitotic activity, and histologic subtype in gastrointestinal stromal tumors. J Clin Oncol 2002;20:3898-905. 26. Van Glabbeke M, Verweij J, Casali PG, Le Cesne A, Hohenberger P, Ray-Coquard I, et al. Initial and late resistance to imatinib in advanced gastrointestinal stromal tumors are predicted by different prognostic factors: a European Organisation for Research and Treatment of Cancer-Italian Sarcoma Group-Australasian Gastrointestinal Trials Group study. J Clin Oncol 2005;23:5795-804. 27. Goettsch WG, Bos SD, Breekveldt-Postma N, Casparie M, Herings RM, Hogendoorn PC. Incidence of gastrointestinal stromal tumours is underestimated: results of a nation-wide study. Eur J Cancer 2005;41:2868-72. 28. Nilsson B, Bumming P, Meis-Kindblom JM, Oden A, Dortok A, Gustavsson B, et al. Gastrointestinal stromal tumors: the incidence, prevalence, clinical course, and prognostication in the preimatinib mesylate era a population-based study in western Sweden. Cancer 2005;103:821-9. 29. Gronchi A, Casali PG, Mariani L, Miceli R, Fiore M, Lo Vullo S, et al. Status of surgical margins and prognosis in adult soft tissue sarcomas of the extremities: a series of patients treated at a single institution. J Clin Oncol 2005;23:96-104. 30. Cooper PN, Quirke P, Hardy GJ, Dixon MF. A flow cytometric, clinical, and histological study of stromal neoplasms of the gastrointestinal tract. Am J Surg Pathol 1992;16:163-70. 31. Hasegawa T, Matsuno Y, Shimoda T, Hirohashi S. Gastrointestinal stromal tumor: consistent CD117 immunostaining for diagnosis, and prognostic classification based on tumor size and MIB-1 grade. Hum Pathol 2002;33:669-76. 32. Antonescu CR, Sommer G, Sarran L, Tschernyavsky SJ, Riedel E, Woodruff JM, et al. Association of KIT exon 9 mutations with nongastric primary site and aggressive behavior: KIT mutation analysis and clinical correlates of 120 gastrointestinal stromal tumors. Clin Cancer Res 2003; 9:3329-37. 33. Lasota J, Kopczynski J, Sarlomo-Rikala M, Schneider-Stock R, Stachura T, Kordek R, et al. KIT 1530ins6 mutation defines a subset of predominantly malignant gastrointestinal stromal tumors of intestinal origin. Hum Pathol 2003; 34:1306-12. 34. Miettinen M, Makhlouf H, Sobin LH, Lasota J. Gastrointestinal stromal tumors of the jejunum and ileum: a clinicopathologic, immunohistochemical, and molecular genetic study of 906 cases before imatinib with long-term follow-up. Am J Surg Pathol 2006;30:477-89. 35. Schneider-Stock R, Boltze C, Lasota J, Peters B, Corless CL, Ruemmele P, et al. Loss of p16 protein defines high-risk patients with gastrointestinal stromal tumors: a tissue microarray study. Clin Cancer Res 2005;11(2 Pt 1):638-45. 36. van der Zwan SM, DeMatteo RP. Gastrointestinal stromal tumors: 5 years later. Cancer 2005;104:1781-8. 37. Van Glabbeke M, Verweij J, Casali PG, Le Cesne A, Hohenberger P, Ray-Coquard I, et al. Initial and late resistance to imatinib in advanced gastrointestinal stromal tumors are predicted by different prognostic factors: a European Organisation for Research and Treatment of Cancer-Italian Sarcoma Group-Australasian Gastrointestinal Trials Group study. J Clin Oncol 2005;23:5795-804. 38. Van Glabbeke M, Verweij J, Casali PG, Simens J, Le Cesne A, Reichardt P, et al. Predicting toxicities for patients with advanced gastrointestinal stromal tumours treated with imatinib: a study of the European Organisation for Research and Treatment of Cancer, the Italian Sarcoma Group, and the Australasian Gastro-Intestinal Trials Group (EORTC- ISG-AGITG). Eur J Cancer 2006;42:2277-85.

本文献由 学霸图书馆 - 文献云下载 收集自网络, 仅供学习交流使用 学霸图书馆 (www.xuebalib.com) 是一个 整合众多图书馆数据库资源, 提供一站式文献检索和下载服务 的 24 小时在线不限 IP 图书馆 图书馆致力于便利 促进学习与科研, 提供最强文献下载服务 图书馆导航 : 图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具