Fax: (011) 48-22-643-9791
Clinical utility of the new American Joint Committee on Cancer staging system for gastrointestinal stromal tumors†
Current overall survival after primary tumor resection
Article first published online: 31 MAR 2011
Copyright © 2011 American Cancer Society
Volume 117, Issue 21, pages 4916–4924, 1 November 2011
How to Cite
Rutkowski, P., Wozniak, A., Dębiec-Rychter, M., Kąkol, M., Dziewirski, W., Zdzienicki, M., Ptaszynski, K., Jurkowska, M., Limon, J. and Siedlecki, J. A. (2011), Clinical utility of the new American Joint Committee on Cancer staging system for gastrointestinal stromal tumors. Cancer, 117: 4916–4924. doi: 10.1002/cncr.26079
Presented as an oral presentation at the 16th Annual Meeting of the Connective Tissue Oncology Society; Paris, France; November 11-13, 2010.
- Issue published online: 19 OCT 2011
- Article first published online: 31 MAR 2011
- Manuscript Accepted: 3 FEB 2011
- Manuscript Received: 21 DEC 2010
- gastrointestinal stromal tumor;
The objectives of the current study were to assess the reliability of the new revision of the American Joint Committee on Cancer (AJCC) staging system for gastrointestinal stromal tumors (GISTs) based on the National Comprehensive Cancer Network-Armed Forces Institute of Pathology risk classification and to analyze the factors that influence after resection for primary GISTs in 2 AJCC groups: patients with GISTs originating from the stomach and omentum (G-GISTs) and patients with other primary GISTs located mainly in the small bowel (nongastric GISTs [NG-GISTs]).
The authors prospectively analyzed a group of 640 patients with primary, CD117-positive GISTs who underwent surgery with curative intention (R0/R1 resection), including 340 G-GISTs (55.5%) and 300 NG-GISTs (44.5%). Factors were explored that had an effect on disease-free survival time (DFS), which was calculated from the date of radical operation to the date of recurrence or last follow-up. The median follow-up was 39 months.
Compared with NG-GISTs, G-GISTs were characterized by a significantly lower median size (5.3 cm and 8.5 cm, respectively; P < .0001) and lower mitotic activity (median, 3 in 50 high-power fields [HPF] vs 5 in 50 HPF; P < .0001), and they were diagnosed in older patients (median age, 62 years vs 57 years; P = .002). The most commonly detected mutations in G-GIST were those located in KIT exon 11 (60.5%) and platelet-derived growth factor receptor alpha (PDGFRA) exon 18 (19%) versus KIT exons 11 and 9 in NG-GISTs (72% and 17.4%, respectively). The prognosis of patients who had G-GISTs was significantly better compared that of patients who had NG-GISTs, with 5-year DFS rates of 69% (median, 83 months) versus 43% (median, 33 months), respectively (P < .00001). The most significant prognostic factors that correlated with shorter DFS in both G-GISTs and NG-GISTs were primary tumor size >5 cm and >10 cm (P < .0001) and mitotic index >5 in 50 HPF and >10 in 50 HPF (P < .0001). The 5-year DFS rates in G-GISTs according to AJCC stage categories were as follows: 96% for stage IA tumors, 92% for stage IB tumors, 51% for II tumors, 22% for stage IIIA tumors, and 22% for stage IIIB tumors (P < .0001). The 5-year DFS rates in NG-GISTs according to AJCC categories were as follows: 92% for stage I tumors, 66% for stage II tumors, 28% for IIIA tumors, and 16% for IIIB tumors (P < .0001). The high prognostic significance of the AJCC classification also was confirmed for overall survival data, including the impact of therapy with tyrosine kinase inhibitors.
The reliability of AJCC risk classification after resection of primary GIST was confirmed for DFS and overall survival. Patients with primary G-GISTs had a better prognosis than patients with NG-GISTs. In both groups, primary tumor size and mitotic activity were the most important prognostic factors in terms of DFS. Cancer 2011;. © 2011 American Cancer Society.
Gastrointestinal stromal tumors (GISTs) are the most common intra-abdominal mesenchymal neoplasms and may occur throughout the entire gastrointestinal tract.1-3 The majority of GISTs are associated with activating, somatic, mutually exclusive mutations of 2 genes, KIT and platelet-derived growth factor receptor alpha (PDGFRA), which are the early oncogenic events during GIST development. They result in overexpression and activation of the KIT and PDGFR oncogenes, which belong to the subclass III family of receptor tyrosine kinases (RTKs). RTKs are transmembrane proteins involved in signal transduction and regulation of cellular growth, differentiation, and angiogenesis.4-9
GISTs are morphologically and clinically heterogeneous, ranging from clinically benign to malignant, and their biologic behavior is difficult to predict. Radical surgery is the treatment of choice in primary resectable tumors, but virtually all GISTs are associated with a risk of recurrence. Approximately 40% to 50% of patients who undergo potentially curative resection develop recurrent or metastatic disease, mainly in the liver or intraperitoneally.2, 3, 10-15 The prognosis for patients with recurrent/unresectable/metastatic disease has changed dramatically during last decade because of introduction into clinical practice of targeted therapy with tyrosine kinase inhibitors (TKIs) (imatinib, sunitinib), which target the KIT and PDGFRA receptors.3, 16-18 Moreover, 1 trial of imatinib adjuvant therapy after resection of primary GIST demonstrated a significant reduction in the risk of recurrence from 17% to 2% at 1 year, although it still had no significant impact on overall survival (OS).19 Nevertheless, imatinib was registered as adjuvant therapy for patients after resection of primary GIST at significant risk of relapse, although no definite guidance was provided regarding the optimal duration of treatment or patient selection. This drug registration implies the use of a reliable, common system for the assessment of recurrence risk and OS after resection of a primary tumor. A consensus conference held at the National Institutes of Health in 2001 provided both an evidence-based definition and a first practical scheme for the assessment of risk in the clinical course of this disease.20 The risk categorization was based on evaluation of the size and mitotic rate (evaluated per 50 high-power fields [HPF]) of tumors. It was supported by analyses of some series of patients, but reliability for the intermediate-risk group was criticized.21, 22 Miettinen and Lasota from the Armed Forces Institute of Pathology created classifications for risk assessment of GISTs in different locations that were approved for common use by the National Comprehensive Cancer Network in 2007.23, 24 In particular, that system reflects the finding that gastric GISTs have a much lower rate of aggressive behavior compared with jejunal and ileal GISTs of comparable size and/or mitotic rate.25, 26 The final step was to create the separate new staging system for GIST by the American Joint Committee on Cancer (AJCC), which was implemented for public use on January 1, 2010.27, 28 The utility of this system has never been confirmed on a prospective, contemporary group of patients in terms of either disease free-survival (DFS) or OS. The objectives of the current study were to evaluate the new AJCC staging system for GIST based on a large, prospectively collected group of patients after primary tumor resection and to identify prognostic factors for primary GISTs of different primary sites.
MATERIALS AND METHODS
We analyzed a group of 640 patients with CD117-immunopositive GISTs who underwent surgery with curative intent for a primary resectable tumor without detectable metastases. All patients were included in the prospectively collected database of the Polish Clinical GIST Registry (http://gist.coi.waw.pl accessed December 1, 2010), which was started in 2001, without further data selection. Clinicopathologic data were supplemented by a review of all available medical and histopathologic records from the referring hospitals. Imatinib was not received as adjuvant treatment by any of these patients. All patients provided informed consent for therapy, and the survivors were carefully followed for a median of 39 months (range, 6-192 months). Follow-up information was obtained during regular outpatient visits or by telephone with the patient and/or the referring physician. Postoperative follow-up consisted of physical examination and routine imaging investigations (computed tomography scans of the abdominal cavity and pelvis, chest x-rays). Routinely, in high-risk/intermediate-risk groups, thorough examinations were recommended every 3 to 4 months during the first 2 years, every 4 months in third year, and every 6 months thereafter. In the low-risk/very-low-risk groups, patients followed every 6 to 12 months according to Polish national guidelines.29 Patients who developed recurrent disease routinely received TKIs (mainly imatinib in the first-line setting).
Mutational analysis of KIT and PDGFRA was performed in 276 randomly selected patients (43%) based on DNA that was isolated from paraffin-embedded or fresh-frozen tumor tissue, as described previously.30 KIT exons 9, 11, 13, and 17 and PDGFRA exons 12, 14, and 18 were amplified by polymerase chain reaction and prescreened using denaturing high-performance liquid chromatography (WAVE System; Transgenomic Limited, Glasgow, United Kingdom). Then, all mutations were confirmed by sequencing. Analyses of clinicopathologic data and molecular results were approved by the local bioethical committee according to good clinical practice guidelines.
All statistical analyses were performed using the R statistical program (version 2.10.1; R Development Core Team, Vienna, Austria; http://www.R-project.org accessed November 20, 2010) and Statistica software (version 6.1; Statsoft, Tulsa, Okla). Contingency tables were analyzed by using the chi-square test for non-normal distribution of parameters. For comparison of 2 groups with non-normal, nonparametric distribution, the Mann-Whitney U test was applied.
For survival analyses (DFS and OS), Kaplan-Meier estimates were used with the log-rank tests for bivariate comparisons. We calculated DFS from the date of primary tumor resection to the date of either recurrence or last follow-up.
In addition, for AJCC stage groups and for different genotypes, we calculated OS from the date primary tumor resection to the date of either death from disease progression/therapy complications or last follow-up. All deaths from other causes were recorded as censored.
DFS was assessed with respect to the following variables: demographic data (aged ≤45 years or >45 years at diagnosis), sex, tumor size (≤5 cm vs >5-10 cm vs >10 cm), mitotic rate (0-5 per 50 HPF vs 6-10 per 50 HPF vs >10 per 50 HPF), histologic subtype (spindle-cell vs epithelial and mixed-cell), primary tumor location (gastric/omental [G]-GIST vs nongastric [NG]-GIST, including the small intestine, colon/rectum, mesentery, esophagus, and peritoneum, according to AJCC anatomic stage), molecular findings (exon 11 KIT mutations [ie, deletions including codons 557 and/or 558, point mutations/insertions, and other deletions]; exon 9 KIT mutations and other KIT mutations; exon 18 PDGFRA aspartic acid to valine substitution at codon 842 [p.D842V] substitutions and other PDGFRA mutations; and wild-type tumors), and AJCC stage (for the entire group and separately for the G-GIST and NG-GIST groups27, 28) (Table 1).
|IA||T1 or T2||N0||M0||Low|
|Any T||Any N||M1||Any|
|I||T1 or T2||N0||M0||Low|
|Any T||Any N||M1||Any|
In multivariate analysis of the factors associated with DFS after radical surgery for primary GIST, we used a Cox proportional hazards model with a stepwise model-building procedure and included all covariates that were significant at the 20% level in bivariate analysis. We did not include mutational data in the multivariate analysis, because those data were obtained in only 43% of patients and will be the subject of a separate analysis. Then, 2-way interactions were considered in the model. Differences were considered statistically significant if P values were < .05.
The study cohort consisted of 640 patients, including 295 men (46.1%) and 345 women (53.9%), and the median patient age at diagnosis was 60 years (range, 9-89 years). All patients who had primary, inoperable and/or metastatic disease at presentation and all patients who received treatment with adjuvant imatinib after undergoing primary tumor resection were excluded from the current analyses. Most primary tumors were located in the stomach (52.8%), followed by the small intestine (34.4%). According to AJCC staging, 42.6% of patients had stage I tumors, 18.4% of patients had stage II tumors, 12.5% of patients had stage IIIA tumors, and 26.5% of patients had stage IIIB tumors. Detailed clinicopathologic data for the entire group of patients with primary GISTs are provided in Table 2.
|No. of Patients (%)|
|All patients||640 (100)||340 (53)||300 (47)|
|Median [range]||60 [9-89]||62 [9-89]||57 [22-85]|
|<45||95 (14.8)||40 (11.7)||55 (18.3)||.02|
|>45||545 (85.2)||300 (88.3)||245 (81.7)|
|Women||345 (53.9)||188 (56.3)||157 (52.3)|
|Men||295 (46.1)||152 (44.7)||143 (47.7)|
|Primary tumor site|
|Stomach||338 (52.8)||338 (99.4)|
|Duodenum||33 (5.1)||33 (11)|
|Small bowel/jejunum and ileum||220 (34.4)||220 (73.3)|
|Large bowel/rectum||23 (3.6)||23 (7.7)|
|Omentum||2 (0.3)||2 (0.6)|
|Other; eg, mesentery, esophagus, intraperitoneally NOS||24 (3.8)||24 (8)|
|Primary tumor size, cm||<.0001|
|Median [range], cm||6.5 [0.5-40]||5.3 [0.7-40]||8.5 [2-35]|
|≤5||237 (37.8)||168 (49.9)||69 (23.8)|
|>5-10||237 (37.8)||111 (32.9)||126 (43.3)|
|>10||153 (24.4)||58 (17.2)||95 (32.7)|
|Mitotic count per 50 HPF||<.0001|
|Median [range]||4 [0-100]||3 [0-100]||5 [0-100]|
|≤5||357 (60.4)||215 (73.9)||142 (51.6)|
|6-10||81 (13.7)||23 (10.9)||58 (21.1)|
|>10||153 (25.9)||78 (6.5)||75 (27.3)|
|I||255 (42.6)||203 (61.5)||52 (19.3)|
|II||110 (18.4)||61 (18.5)||49 (18.2)|
|IIIA||75 (12.5)||33 (10)||42 (15.6)|
|IIIB||159 (26.5)||33 (10)||126 (46.9)|
|KIT exon 11 del involving 557-558||70 (25.4)||28 (19.5)||42 (31.8)|
|KIT exon 11 PM/INS||62 (22.5)||38 (26.4)||24 (18.2)|
|KIT exon 11 other del||42 (15.2)||19 (13.2)||23 (17.4)|
|KIT other||8 (2.9)||2 (1.4)||6 (4.6)|
|KIT exon 9||23 (8.3)||0 (0)||23 (17.4)|
|PDGFRA exon 18 p.D842V||21 (7.6)||21 (14.5)||0 (0)|
|PDGFRA other||16 (5.8)||16 (11.1)||0 (0)|
|Wild-type||34 (12.3)||20 (13.9)||14 (10.6)|
We observed significant differences between both analyzed groups of patients. The median size of G-GISTs compared with NG-GISTs was lower (5.3 cm and 8.5 cm, respectively; P < .0001), and G-GISTs were characterized by lower mitotic activity (median, 3 per 50 HPF vs 5 per 50 HPF; P < .0001). Patients in the G-GIST group were older than patients in the NG-GIST group (median age, 62 years vs 57 years; P = .002).
The most common detected mutations were KIT exon 11 mutations (60.5%) and PDGFRA exon 18 mutations (19%) in G-GIST and KIT exons 11 and 9 mutations in NG-GIST (72% and 17.4%, respectively; P < .0001). Moreover, among patients with G-GISTs who had exon 11 KIT mutations, the most frequent were point mutations/insertions (26.4%). In the NG-GIST group, the most frequent mutations were exon 11 KIT deletions with involvement of codons 557 and/or 558 (31.8%).
We detected 249 disease recurrences (39%) during follow-up. The median time to disease recurrence was 20 months (range, 2-168 months). The median DFS duration after resection of primary GIST was 57 months, and the estimated 5-year DFS rate was 50.2% (95% confidence interval [CI], 45.2%-55.7%). The prognosis for patients who had G-GISTs was significantly better compared with that for patients who had NG-GISTs, with 5-year DFS rates of 67.5% (95% CI, 61%-74.7%; median, 86 months) versus 36.5% (95% CI, 30.3%-43.9%; median, 33 months; P < .0001) (Fig. 1).
The categorization of risk groups according to AJCC staging categories revealed an excellent correlation with DFS. For the entire group, DFS according to risk group indicated that the 5-year estimated DFS rate was 94.4% (95% CI, 89.9%-99.3%) for patients with stage I tumors, 60.6% (95% CI, 49.7%-73.9%) for patients with stage II tumors, 25.4% (95% CI, 15.6%-41.4%) for patients with stage IIIA tumors, and 18% (95% CI, 12.1%-26.7%) for patients with stage IIIB tumors. The statistical difference between these rates was highly significant (P < .0001). The 5-year DFS rate for patients with AJCC stage IA, IB, II, IIIA, and IIIB tumors in the G-GIST group were 96%, 92%, 51%, 22%, and 22%, respectively (P < .0001) (Fig. 2); whereas, in the NG-GIST group, the rates for AJJC stage I, II, IIIA and IIIB tumors were 92%, 66%, 28%, and 16%, respectively (P < .0001) (Fig. 3).
In univariate analysis, the following factors had a negative impact on DFS in the G-GIST group: being in the stage II, IIIA or IIIB risk group (P < .00001); a mitotic index >5 per 50 HPF and >10 per 50 HPF (P < .00001); primary tumor size >5 cm and >10 cm (P < .00001); male gender (P = .03). In the NG-GIST group, the following factors had a negative influence on DFS: stage II, IIIA, or IIIB tumors (P < .00001); a mitotic index >5 per 50 HPF (P < .0001); primary tumor size >5 cm (P < .0001); and epithelioid or mixed cell pathologic subtype (P = .05).
In the subgroup of 276 patients who had data on tumor genotype available, we observed differences in terms of DFS according to mutational status, with the worst prognosis for patients who had exon 11 KIT deletions involving codons 557 and/or 558 and exon 9 KIT mutants. The 3-year DFS rates according to mutational status were 39% (95% CI, 25.7%-56.2%) for exon 11 KIT deletions with codons 557 and/or 558 involvement, 51% (95% CI, 37.6%-74.2%) for exon 11 KIT point mutations/insertions, 50% (95% CI, 38.3%-74.0%) for other exon 11 deletions, 86% (95%CI, 50.4%-91.8%) for other KIT mutations, 43% (95% CI, 25.6%-68%) for exon 9 KIT mutations, 69% (range, 50.55%-79.3%) for PDGFRA exon 18 p.D842V substitutions, 91% (95% CI, 68%-95.4%) for other PDGFRA mutations, and 46% (95% CI, 38%-65.3%) for wild-type tumors (P = .02) (Fig. 4).
According to multivariate analysis, 2 independent factors had a negative impact on DFS in both the G-GIST group and the NG-GIST group: a mitotic index >5 per 50 HPF (G-GIST: P = .000001; hazard ratio [HR], 11.2; 95% CI, 5.9-21.1; NG-GIST: P = .0000001; HR, 3.8; 95% CI, 2.5-5.8) and a primary tumor size >5 cm (G-GIST: P = .00001; HR, 3.6; 95% CI, 1.8-7.3; NG-GIST: P = .0000001; HR, 6.0; 95% CI, 2.97-12.2). We did not include mutational status in the multivariate analysis because we had data only for 276 patients, and this will be the subject of a separate analysis.
At the time of the current analysis, 88 patients (13.8%) had died. The estimated 5-year OS rate was 83.6% (95% CI, 79.6%-87.8%), and the median OS had not been reached (patients routinely received TKIs, mainly imatinib, in the first-line setting).
The AJCC staging categories were correlated with OS and 5-year OS rates according to stage groups as follows: 99.1% (95% CI, 97.4%-100%) for stage I tumors, 88.2% (95% CI, 80%-97.2%) for stage II tumors, 70.2% (95% CI, 58.3%-84.6%) for stage IIIA tumors, and 74.4% (95% CI, 66.3%-83.5%) for stage IIIB tumors (Fig. 5). We did not observe any significant differences in OS according to mutational status (P = .07).
The reliable assessment of cancer according to disease stage is important in terms of evaluating patient prognosis, choosing the correct follow-up schedule, and discussing adjuvant therapy options with the patient. The presented analysis demonstrated for the first time in the large series of patients the reliability of the new AJCC staging system/risk classification after resection of primary GIST in terms of DFS and OS. Moreover, we observed that the risk of recurrence in patients with stage III tumors was very high (>75% of patients had a recurrence within 5 years), and these patients clearly should be considered for adjuvant imatinib therapy. We also confirmed the assumption that patients who have GISTs originating from the bowel have a poorer prognosis compared with patients who have GISTs localized in the stomach.11, 23, 31 In both groups, primary tumor size and mitotic activity were the most important, reproducible prognostic factors in terms of DFS, but there were striking differences in the primary characteristics of these groups of tumors. G-GISTs generally were smaller than the intestinal tumors, which may be related to their easier diagnosis using standard endoscopy or endoscopic ultrasound techniques. However, there also are intrinsic differences in the mitotic index and distribution of mutational status that cannot be explained only by the delayed finding of a primary tumor outside the stomach. The mitotic index demonstrated a striking influence on patient outcome. The problem of the impact of primary tumor mutational status is more complicated. It has been suggested that there are significant associations between specific mutations and tumor aggressiveness/recurrences. In particular, patients with KIT exon 11 deletions involving codons 557 and/or 558 and those with KIT exon 9 mutations have disease with proliferating potential characterized by a poorer prognosis than patients with KIT exon 11 substitutions or insertions or with PDGFRA mutations.32-34 We also confirmed these associations in univariate analysis. We did not perform a multivariate analysis for mutational status, because data on mutational status were available only for approximately 40% of the analyzed patients and will be the focus of a separate analysis. The suggestion of prognostic significance for specific genotypes must be confirmed in larger series of patients, because even patients who have very small and low-risk tumors may present with KIT mutations, which are believed to be more aggressive (like KIT exon 11 557 and/or 558 deletions). However, we have confirmed that KIT exon 9 mutations occur mostly in tumors of the bowel and demonstrate the most aggressive behavior.32 Conversely, PDGFRA mutations characterize GIST tumors with less aggressive biology.33 It is interesting to note changes in the significance of these mutations after the introduction of TKI therapy for recurrent tumors, which likely is the reason for the observed loss of differences between different genotypes in terms of OS. It has been established that KIT exon 11 mutants respond the best to imatinib therapy, and long-term outcomes in such patients, even after disease recurrences, are relatively good. However, patients who have the PDGFRA p.D842V mutation, although it is associated less commonly with disease recurrence, have a poorer prognosis after relapse because of intrinsic resistance to imatinib and sunitinib therapy.9, 30
The important message from the results of our study is the analysis of current OS data from patients with resectable GIST. To the best of our knowledge, there are no contemporary data on the OS of patients after resection of primary GIST in the TKI era. Currently, these patients receive a TKI (mainly imatinib in the first-line setting) after they develop a recurrence, which has significantly changed their final outcome. The outcome of patients with metastatic GIST before the introduction of TKI therapy was dismal, as reported previously, with a median OS of 19 months.35 Historic data from patients with primary, resectable GISTs before the advent of TKI therapy era indicated a 5-year survival rate of approximately 50%,10, 36-38 which is about 30% lower than the rate reported in current data. One study on a series of 186 patients who reached 5-year DFS over a period >25 years mixed together a contemporary series of patients (who could be offered imatinib therapy) with cases from the 1980s.31 The only phase 2 study on imatinib in advanced GIST that presented OS data, in which the median OS approached 5 years, also comprised a heterogeneous group of patients and took into account patients who developed recurrent disease after primary surgery and patients who initially had unresectable/metastatic tumors.39 Not surprisingly, the highest impact of TKI therapy on OS after disease recurrence was observed in the high-risk group, in which a 5-year DFS rate of approximately 25% translated into a 5-year OS rate that approached >70%. We have demonstrated an excellent long-term OS rate of almost 100% in patients with stage I disease, confirming that adjuvant therapy should not be recommended for these patients. Our data also indicated that, even without adjuvant therapy, the intermediate-risk patients (stage II) had a survival rate of 88%. This is important in the context of the results from adjuvant therapy, because it implies that it is difficult to demonstrate an increase in OS from adjuvant therapy in patients who have intermediate-risk disease.
To summarize, the current study is the first to our knowledge demonstrating the utility of the new AJCC staging system in GIST in terms of DFS and OS on the basis of a prospectively collected, large series of patients from a tumor registry. Patients who have GISTs originating in the bowel have a greater risk of recurrence than patients who have gastric GISTs, but tumor size and mitotic index universally are the most important prognostic factors. Surgery is the standard treatment for primary GIST; however, surgical resection often is not curative, particularly in-high risk patients, implying the necessity of considering adjuvant imatinib therapy for patients with stage III disease. To the best of our knowledge, this also is the only analysis to include final outcomes in patients with resectable GIST in the era of TKI therapy, and the results indicate that the emergence of KIT/PDGFR TKIs have substantially altered the prognosis of patients with recurrent disease.
We thank all specialists who cooperated in the Polish Clinical Gastrointestinal Stromal Tumor Registry and D. Rabczenko for statistical advice.
This study was supported by the Polish Ministry of Science (grant N N402 2092 35 to J. Limon, A. Woźniak, and P. Rutkowski).
CONFLICT OF INTEREST DISCLOSURES
P. Rutkowski and A. Woźniak have received honoraria and travel grants from Novartis, and P. Rutkowski has received honoraria from Pfizer and was a member of the Advisory Board for Novartis.
- 24NCCN Task Force report: management of patients with gastrointestinal stromal tumor (GIST)—update of the NCCN Clinical Practice Guidelines. J Natl Compr Canc Netw. 2007l; 5( suppl 2): S1-S29., , , et al.
- 27Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. AJCC Staging Manual. 7th ed. New York: Springer; 2009.
- 28Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. AJCC Cancer Staging Handbook. 7th ed. New York: Springer; 2010.
- 29Recommendations for diagnostics and therapy of gastrointestinal stromal tumors (GIST) in 2008 [article in Polish]. Nowotwory. 2008; 58: 537-549., , , et al.
- 30Use of c-KIT/PDGFRA mutational analysis to predict the clinical response to imatinib in patients with advanced gastrointestinal stromal tumours entered on phase I and II studies of the EORTC Soft Tissue and Bone Sarcoma Group. Eur J Cancer. 2004; 40: 689-695., , , et al.