Deletions Affecting Codons 557-558 of the c-KIT Gene Indicate a Poor Prognosis in Patients With Completely Resected Gastrointestinal Stromal
http://www.100md.com
《临床肿瘤学》
the Hospital Universitario Son Dureta, Palma de Mallorca
Instituto Valenciano de Oncología
Universidad de Valencia, Valencia
Instituto Catalán de Oncología
Hospital Clinic Barcelona, Barcelona
Hospital de León, León
Hospital Miguel Servet, Hospital Clinico Lozano Blesa, Zaragoza
Hospital Marqués de Valdecilla, Santander
Hospital Virgen del Camino, Pamplona
Consorcio Cruz Roja, Llobregat
Hospital de Basurto, Bilbao
Hospital Clínico Madrid, Madrid
Hospital San Millán, Logro?o
Consorcio Sanitario de Terrassa, Hospital Central de Asturias, Oviedo, Spain
ABSTRACT
PURPOSE: To explore the prognostic value of mutations in c-KIT and PDGFR- genes with respect to relapse-free survival (RFS) in patients with gastrointestinal stromal tumors (GIST). We have investigated the prognostic relevance of the type and position of the mutations, in addition to other clinicopathologic factors, in a large series of patients with GIST.
METHODS: For this study, 162 patients were selected according to the following criteria: completely resected tumors with negative margins attended between 1994 and 2001; no metastasis at diagnosis; tumor larger than 2 cm, c-KIT–positive immunostaining; and no other primary tumors.
RESULTS: The median follow-up was 42 months for patients free of recurrence. Mutations were detected in 96 tumors (60%): 82 cases involving c-KIT and 14 cases involving PDFGR-. Univariate analysis demonstrated the following as poor prognostic factors for RFS: tumors larger than 10 cm (P < .0001); mitotic count higher than 10 mitoses per 50 high-power fields (P < .0001); high risk index (P < .0001); intestinal GIST location (P = .0041); high cellularity (P < .0001); tumor necrosis (P < .0001); deletions affecting exon 11 (P = .0007); and deletions affecting codons 557 to 558 (P < .0001). After the multivariate analysis, only the high risk index (relative risk [RR], 12.36), high cellularity (RR, 3.97), and deletions affecting codons 557 to 558 of c-KIT (RR, 2.57) corresponded to independent prognostic factors for RFS in GIST patients.
CONCLUSION: Deletions affecting codons 557 to 558 are relevant for the prognosis of RFS in GIST patients. This critical genetic alteration should be considered to be a new prognostic stratification variable for randomized trials exploring imatinib mesylate in the adjuvant setting in GIST patients.
INTRODUCTION
Gastrointestinal stromal tumors (GIST) are the most frequent mesenchymal tumors of the digestive tract. Interest in these neoplasms has been renewed since Hirota et al1 shed light on the molecular basis of their oncogenesis. The diagnosis of GIST is based on histomorphological features and, in most cases, histochemical immunoreactivity to the tyrosine kinase receptor of c-KIT (CD117).2-4 The cells implicated in the oncogenic process in GIST are primitive cells that are related to interstitial cells of Cajal.5 Mutations in the c-KIT proto-oncogene, which encodes a type III tyrosine kinase receptor (KIT), lead to constitutive activation of the protein.6 Although the majority of mutations in c-KIT have been detected within the juxtamembrane region encoded by exon 11, they also have been found within exon 9 and, more rarely, in exons 13 and 17. Additionally, in tumors in which mutations in c-KIT have not been identified, oncogenic mutations have been reported in the gene encoding the platelet-derived growth factor receptor alpha (PDGFR-).7 Several studies of prognostic factors in mesenchymal tumors of the gastrointestinal tract have been published; however, most of these studies did not take into account the current definition of GIST.8-11 Other recent studies analyzing clinical and pathologic variables as well as mutations in exon 11 of c-KIT as prognostic factors have also been carried out.12-18 Nevertheless, data exploring the prognostic relevance of such factors on relapse-free survival (RFS) in completely resected patients is scarce and controversial.15-18
The life expectancy of metastatic GIST patients has dramatically changed through the appearance of imatinib mesylate (IM), a drug that inhibits the tyrosine kinase activity in the context of this disease.19 It therefore seems important, considering GIST as a model for target-based molecular therapy, to continue exploring the prognostic factors beyond size and mitotic count relevant to this type of tumor. Hence, we have analyzed the molecular and clinicopathologic prognostic factors in a large group of patients with GIST who submitted to complete surgical resection. The size of our sample enabled us to examine the position and type of mutation as prognostic factors for RFS in patients with GIST. We feel that these data will help better explain the results in future randomized trials with IM as an adjuvant treatment for GIST including a control, nontreated group.
PATIENTS AND METHODS
Patients
We assessed the information from patients with mesenchymal tumors of the gastrointestinal tract that were included in the Spanish Group for Sarcoma Research (GEIS) registry and diagnosed from January 1994 to December 2001 in 30 of the 53 hospitals that are members of GEIS in Spain. The study was approved by the institutional review board of GEIS. The clinical staging at diagnosis took into consideration computed tomography/magnetic resonance imaging or abdominal echography data. The inclusion criteria for subsequent biologic analysis were: tumors larger than 2 cm; c-KIT–positive tumors; absence of other primary tumors; adequate surgery defined as complete gross removal, usually by segmental resection without tumor rupture; and the absence of intra-abdominal implants in the operative report. From November 2002 to March 2003, clinical data were updated and checked on a queries-based task between the central reviewer and the medical oncologists from the referral centers. Two pathologists with expertise in analyzing sarcomas (A.L.-B. and R.R.) performed independent pathologic reviews of the paraffin-embedded tissue blocks. To additionally meet the inclusion criteria, in all patients the surgical margins had to be negative and the diagnosis of GIST confirmed histologically with complete agreement of the panel of pathologists.
Age at diagnosis and sex were the demographic characteristics that were taken into consideration for each patient. The clinical presentation, if symptomatic, was registered for all patients along with the date of the first symptom. Tumor location was divided into stomach, small intestine, esophagus, rectum, colon, and omentum. The largest diameter of the primary tumor was determined after surgery in each case and measured by the pathologist. The diagnostic methods, defined as the first diagnostic procedure, were registered as either core biopsy or incisional or excisional biopsy. Postoperative surveillance varied between the different centers. The status (alive with or without disease, dead with or without disease) and recurrence data (date and type of treatments and additional treatments) were obtained by follow-up of the patients.
Pathology
Paraffin sections of formalin-fixed tissue (3 μm) were used for conventional hematoxylin and eosin staining and immunohistochemistry. The immunohistochemical analysis involved the StreptABComplex method coupled to heat-induced epitope retrieval, and the following antibodies were used: CD117, c-KIT (A-4502, polyclonal, 1:400 dilution; DAKO, Copenhagen, Denmark), SMA (N-1584, monoclonal, 1:2 dilution; DAKO), desmin (N-1526, monoclonal, 1:2 dilution; DAKO), S-100 protein (N-1573, polyclonal, 1:2 dilution; DAKO), and PDGFR- (SC-338, polyclonal, 1:200 dilution; Santa Cruz Biotechnology, Santa Cruz, CA). Both pathologists involved in the study reviewed all cases separately, and when disagreement arose, a consensus was reached after a review meeting. The pathologists did not have access to clinical information or macroscopic data regarding the tumors. The histological variables analyzed for each tumor were as follows: mitotic count per 50 high-power fields (HPF), necrosis (presence or absence), pleomorphism, cellularity20 (high or moderate-paucicellular), predominant cell type (spindle, epithelioid, or mixed), immunoreactivity of c-KIT and PDGFR- antisera (strong, moderate, or weak), and pattern of c-KIT immunostaining (diffuse or perinuclear).
DNA Sequencing
DNA was isolated from 3- to 5-μm sections of fixed and paraffin-embedded tissue. After deparaffinization, the tumor tissue was resuspended in lysis solution (0.5% sodium dodecyl sulfate, 0.5 mg/mL proteinase K, 10 mmol/L Tris-HCl, pH 8, 0.15 M NaCl, and 5 mmol/L EDTA) and incubated overnight at 55°C. DNA was extracted with phenol/chloroform/isoamyl alcohol (25:24:1), precipitated with ethanol, and dissolved in 10 to 35 μL of ultrapure water. Intronic polymerase chain reaction (PCR) primers were used to amplify exons 9, 11, 13,21 and 17 of c-KIT22 and exons 12 and 18 of PDGFRA-.7 PCR was performed in a reaction volume of 50 μL containing 1 μL of DNA, 50 mmol/L KCl, 10 mmol/L Tris-HCl, pH 8.3, 2 mmol/L MgCl2, 0.2 mmol/L each dNTP, 0.2 μmol/L for each primer, and 2 units of AmpliTaq Gold (Perkin Elmer, Norwalk, CT). PCR was carried out in a DNA Thermal Cycler 9600 (Perkin Elmer) after preheating the samples at 95°C for 10 minutes. The DNA was amplified over 40 cycles of: 1 minute of denaturation at 94°C; 1.5 minutes of annealing at 56°C for c-KIT primers or 2 minutes at 65°C for PDGFR- primers; and 1 minute of extension at 72°C with an additional final extension step of 10 minutes. Ten microliters of the PCR products were visualized in ethidium-bromide–stained 2% UltraPure agarose gels (Life Technologies, Paisley, Scotland) and photographed. Negative controls were included in every set of amplifications. Bidirectional sequencing with specific primers was performed on an ABI 310 sequencer using the BigDye Terminator v1.1 kit (Applied Biosystems, Inc, Foster City, CA).
Statistical Methods
We considered recurrence as the most suitable event for statistical analysis, because overall survival could be biased by the introduction of IM as a treatment of patients with metastatic GIST. RFS was measured from the date of surgery and was estimated according to the Kaplan-Meier method.23 The factors assessed at the time of diagnosis and evaluated as possible prognostic indicators for RFS were: age, sex, tumor location (gastric v intestinal v other), tumor size (2 to 5 v > 5 to 10 v > 10 cm), number of mitoses per 50 HPF (0 to 5 v 6 to 10 v > 10), cellularity (high v moderate-paucicellular), risk of GIST4 (low risk v intermediate risk v high risk), presence of tumor necrosis, nuclear pleomorphism, presence of a mutation in exon 11 of c-KIT, type of mutation (missense and deletion), position of mutation in exon 11 (from codons 550 to 591), and the presence of a mutation in the PDGFR gene.
Univariate analysis was performed by using the log-rank test.24 Factors that were potentially predictive of RFS by univariate analysis, as well as potential confounding variables, were entered into a multivariate analysis using the Cox proportional-hazards model.25 All P values reported are two tailed, and statistical significance was defined at P < .05. All statistical analyses were performed by using Statistical Package for the Social Sciences (SPSS) for Windows version 10.0 (SPSS Inc, Chicago, IL).
RESULTS
The files of 555 patients were considered, and of these patients, 357 were reviewed as both paraffin blocks and medical information were available. According to the inclusion criteria, a subset of 162 patients was selected for this study, and the reasons for exclusion of the remaining patients are shown in Figure 1.
The median age for the group of 162 patients, 82 males and 80 females, was 63 years (range, 24 to 91 years). Of these patients, 84% had symptoms at diagnosis, of which symptomatic anemia and abdominal pain were the most frequent. The median tumor size was 6 cm (range, 2 to 26 cm), and there were 93 gastric tumors, 63 small intestinal and six with another primary site (one esophagus, three colorectal, and two omentum; Table 1). Patients with gastric GIST presented more frequently with symptomatic anemia, and conversely, patients with small bowel GIST suffered abdominal pain as the predominant clinical presentation. The majority of cases were diagnosed after surgery (93%), and a preoperative biopsy was performed in only 12 (7%) patients.
In most tumors the mitotic rate was low: 59% had less than six mitoses per 50 HPF, 12% had six to 10 mitoses per 50 HPF, and 29% had more than 10 mitoses per 50 HPF. This distribution was similar for patients with gastric and small bowel GIST. Spindle cell type was present in 71.4% of cases and was more predominant in small bowel GIST than gastric GIST. In contrast, epithelioid and mixed cell types were more frequent in gastric tumors. High cellularity was found in 38 patients and was not associated with a specific location.
Regarding immunohistochemistry, strong, diffuse, and cytoplasmic expression of c-KIT (CD117) was observed in 82% of the tumors. A Golgi (dot-like accentuation) and a mixed pattern of immunostaining were seen in 10 and 16 tumors, respectively, mostly in patients with gastric GIST. The tumors also expressed smooth muscle actin (24%), desmin (2.4%), S-100 protein (3%), and PDGFR- (85%).
Mutations were detected in 96 tumors (60%), of which 77 mutations were in exon 11 of c-KIT. Of such exon 11 mutations, 41 were deletions (53%), 23 were missense mutations (30%), and 11 were in-frame duplications (14%; Fig 2). Of the 41 patients with deletions in exon 11, five also carried a missense mutation, as did one of the 11 with in-frame duplications (Table 2). Each type of mutation had a tendency to cluster. In-frame deletions more frequently involved codons 550 to 572, whereas in-frame duplications were more predominant in codons 573 to 591. Deletions simultaneously involving codons 557 and 558 were detected in 19 of the cases involving a deletion in exon 11 (46%). In contrast, 17 of the 30 missense mutations (57%) affected codons 559 to 560. In this subgroup of patients, no phenotypic association was detected, but only two patients suffered a recurrence (V559D and V560D). With respect to the duplications, we found 11 tumors (6.7%) carrying internal tandem duplications at the 3' end of the c-KIT juxtamembrane domain (internal tandem duplications in the 3' c-KIT-JM), of which six were females (all spindle histology) and five were males (three epithelioid and two spindle histology). Of these 11 tumors, eight originated in the stomach and three originated in the small bowel. Despite their relatively low mitotic rate and a reported favorable evolution,26 there were three recurrences among these patients (two epithelioid and one spindle). We found two tumors with nonsense mutations within exon 11, one of which (codon 557) involved a similar mutation to that described in a patient with B-cell lymphoma of the brain27 and was discovered along with a missense mutation involving codon V560D. To the best of our knowledge, the second nonsense mutation involving codon W582 has not been described previously.
Mutations in exon 9 were found in four patients, and in three of them, this mutation represented the insertion of GCC TAT that resulted in the duplication of amino acid residues Ala502 and Tyr503. The other patient incorporated a point mutation E490G with a different histology (small-cell variant). Two of the four patients with mutations in exon 9 relapsed. Only one patient had mutation in exon 13 (K642E), and the associated tumor was epithelioid, located in the small bowel and with 30 mitoses per 50 HPF. Six months after radical surgery, this patient's tumor recurred in the liver. No mutations were found in exon 17.
Mutations in PDGFR- were detected in 14 cases (9%) and were mutually exclusive of c-KIT mutations. Of these, seven mutations were found in exon 12, five of which involved deletions of codons 554 to 555. The remaining seven mutations were observed in exon 18, most often involving a missense mutation (six cases) and three of them affected codon 842 (D842V).
RFS Analysis
In January 2004, the median follow-up for patients free of recurrence was 42 months. There were 42 relapses with the following primary locations: stomach (15 cases), small bowel (23 cases), colorectal (three cases), and omentum (one case). The 5-year RFS for the 162 patients was 68% (± 9%). All but one of the recurrences were metastatic.
Clinicopathologic Variables
Age, sex, and clinical presentation had no influence on RFS for the whole group or for the different tumor locations. With regard to size, the 5-year RFS for the different risk categories was 82% ± 12% (0 to 5 cm), 79% ± 11% (> 5 to 10 cm), and 42% ± 17% (> 10 cm). There was a significant difference in the RFS between tumors larger or smaller than 10 cm (P < .0001). When the mitotic count was considered, we found a significantly higher RFS rate in patients with tumors having less than five mitoses per 50 HPF than in those with more than 10 mitoses per 50 HPF (P < .0001). Analyzing the index of risk as a prognostic factor,4 we found 5-year RFS rates of 94% ± 6% (low risk), 85% ± 15% (intermediate risk), and 44% ± 14% (high risk; P < .0001; Fig 3). The RFS for patients with intestinal GIST was significantly lower than that of patients with gastric GIST, with 5-year RFS rates of 55% (± 14%) and 76% (± 10%), respectively (P = .0041). High cellularity was a significant predictor of RFS, with recurrence in 25 of 38 tumors (65.7%). The 5-year RFS rates for high and moderate-paucicellular tumors were 23% (± 16%) and 81% (± 7%), respectively (P < .0001; Fig 4). Tumor necrosis was detected in 31 cases, of which recurrence occurred in 19 patients (5-year RFS rate, 33% ± 18%); in contrast, 23 recurrences were detected in the 125 tumors without necrosis (5-year RFS rate, 76% ± 10%; P < .0001).
Mutation Variables
Tumors with mutations in c-KIT recurred more frequently (5-year RFS, 57% ± 13%) than tumors without such mutations (5-year RFS, 80% ± 11%; P = .0084). We assessed whether the type of mutation within exon 11 might be related to the deleterious RFS. Patients carrying missense mutations in exon 11 had a similar RFS to the remaining patients. Nevertheless, patients with in-frame deletions in exon 11 had a significantly lower 5-year RFS rate (76% ± 9% v 44% ± 18%; P = .0007; Fig 5). We noticed that there was a relatively large number of relapses (11 of 19) among patients with deletions involving codons 557 to 558 of exon 11 (Fig 6), as well as a worse clinical outcome in a univariate analysis of the results. In other words, tumors with in-frame deletions involving both codons 557 to 558 provoked a significant reduction in the 5-year RFS (23% ± 2% v 74% ± 9%; P < .0001; Fig 7). When only patients with mutations in exon 11 were considered, this difference still remained significant (P = .0027). If either codon was analyzed separately, we found a similar deleterious RFS for each codon but not for the rest of exon 11 codons studied: codon 558 5-year RFS rate, 19% ± 2% v 76% ± 9% (P < .0001); codon 557 5-year RFS rate, 25% ± 2% v 75% ± 9% (P < .0001).
Multivariate Analysis
Only those variables showing a statistically significant relationship to RFS in the univariate analysis were entered in the Cox's proportional-hazard model. Therefore, tumor size, mitotic count, index of risk, location of GIST, cellularity, tumor necrosis, presence of a mutation in c-KIT, and the type or position of the mutation in c-KIT (exon 11 deletion mutation type and exon 11 deletion involving codons 557 to 558) were examined in the multivariate model with a forward stepwise method. The following factors were significantly associated with recurrence: high risk index, presence of high cellularity, and deletions in codons 557 to 558 within exon 11 (Table 3). When mitotic count and tumor size were entered in the Cox model instead of the risk index, small bowel location and mitotic count were also independent prognostic factors.
If the deletion in codons 557 to 558 of exon 11 was not included in the model, the type of c-KIT deletion mutation was an independent prognostic factor for RFS. It can be deduced that the prognostic relevance of deletion-type mutations in c-KIT is due to the presence of deletions involving codons 557 to 558.
DISCUSSION
This study demonstrates the prognostic relevance of the type and position of deletions in the c-KIT gene (ie, involving codons 557 and 558) in addition to the risk index and cellularity in patients with localized GIST who submitted to complete resection.
Our data regarding the impact of classical prognostic factors (size and mitotic count) on RFS are consistent with the results of other studies. Tumor size has been recognized as an independent prognostic marker by some authors9,10,17 but not by others.14,16,28 However, the largest series of patients reported10 did not take into account the current definition of GIST. Tumor size dropped out after entering mitotic count into the model, probably because there is a strong correlation between both factors.29 Otherwise, the risk index seemed to be an independent prognostic factor but only for the high-risk category, because the intermediate-risk category could not be considered independent as having prognostic value.
Cellularity has been reported in some studies as a prognostic variable in GIST but only in univariate analysis.20,30 Although our data revealed high cellularity as having independent prognostic value, we treat this result with some caution, because this parameter could be observer dependent.
Several studies have reported that when the c-KIT gene harbors mutations, GIST has a worse prognosis12-15,17 regardless of the percentage of patients with mutations, which ranged from 37% to 57%. In four separate series that included 30 to 113 patients,15-18 the relationship of c-KIT mutations with RFS was explored in patients with GIST who submitted to surgery and had negative margins. The influence of the type of mutation (ie, missense or deletion) was analyzed in two studies.16,17 In contrast to our results, missense mutations had a more favorable RFS in a series of 30 patients.16 However, in the other study of 86 patients, although the RFS rate was worse for tumors having mutations in c-KIT, no relationship was found between RFS rates and the type of mutation in exon 11.17
As observed previously, we found that codons 557 and 558 were those most often involved in deletions within exon 11.14,15 However, the poor RFS of patients with deletions involving codons 557 to 558 was particularly striking when compared with patients with other mutations in exon 11, with patients with any c-KIT mutation, or with the whole series. Multivariate analysis confirmed the independent prognostic value of this finding, which also was detected recently in a study of 55 patients.18
In a study of 39 patients with primary GIST, the tumors were divided as benign, borderline, or malignant according to their mitotic count and size.31 Interestingly, relapses were suffered in 10 of 11 patients with tumors larger than 2 cm and with deletions involving codons 557 to 558. In contrast, in four tumors in which the deletion involved other codons, a relapse was observed in only one patient. Other investigators detected deletions involving codons 557 to 558 in 67% of tumors.14 Nevertheless, they concluded that the strongest correlation with malignancy was seen in cases involving codons 563 to 569, because an aggressive behavior was always associated with these patients' tumors (n = 8). According to these data, deletions involving codon 558 arose in 37 cases, and among these cases, only in 10 did tumors not recur. Therefore 72% of patients with deletions involving codon 558 developed metastasis, a figure that is not too different from the 58% observed in the patients with deletions of codons 557 to 558 analyzed here. However, we did not observe recurrence in patients whose tumors involved deletion of codons 563 to 569, which may be explained by the exclusion here of patients with metastasis at diagnosis.
Two facts seem to support our findings that abnormalities involving codons 557 to 558 within exon 11 confer a significantly higher risk of recurrence in patients with GIST. On one hand, Trp557 has been identified among juxtamembrane residues with inhibitory roles in the control of receptor kinase activity. On the other hand, mutant forms of c-KIT generated by substitution of proline for Lys558 led to a high degree of constitutive receptor phosphorylation.32
Our findings are relevant for studies exploring a possible role for IM in the adjuvant setting. In forthcoming randomized trials, which include a control, nontreated group, the failure to stratify patients according to the presence of critical deletions may introduce a bias that would influence therapeutic results.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
Joan Carles (Hospital del Mar); Joaquín Bellmunt (Hospital Vall d'Hebrón); Auxiliadora Gómez (Hospital Reina Sofía); Andrés Meana (Hospital General de Alicante); Isabel Sevilla (Hospital Clínico de Málaga); Inmaculada Maestu (Hospital Virgen de los Lirios); Vicente Valentí (Hospital Sant Joan de Reus); Carmen Bala?à (Hospital Germans Trias I Pujol); José Luis Carrasco (Hospital Xeral-Cíes); Ricardo Cubedo (Hospital Puerta de Hierro); Javier Cassinello (Hospital de Guadalajara); Fernando Molano (Hospital Insular Gran Canaria); Jordi Rubió (ICO Girona); Isabel Bover (Hospital Son Llàtzer); Fina Cruz (Hospital Universitario de Canarias); and Antonio López-Pousa (Hospital Sant Pau) also contributed to this work. We thank the pathologists associated with GEIS for providing us with the tissues. Finally, thanks to Aina Rifà and Celi Rodriguez for data management and Mark Sefton for careful revision of the English language of the manuscript.
NOTES
Supported partially by a grant from Novartis Pharmaceuticals and the Cancer Genomics Network C03/10 of the Spanish Ministry of Health. These funds covered the expenses of the immunohistochemical and mutational analyses.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Hirota S, Isozaki K, Moriyama Y, et al: Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science 279:577-580, 1998
Miettinen M, Lasota J: Gastrointestinal stromal tumors: Definition, clinical, histological, immunohistochemical, and molecular genetic features and differential diagnosis. Virchows Arch 438:1-12, 2001
Sarlomo-Rikala M, Kovatich AJ, Barusevicius A, et al: CD117: A sensitive marker for gastrointestinal stromal tumors that is more specific than CD34. Mod Pathol 11:728-734, 1998
Fletcher CD, Berman JJ, Corless C, et al: Diagnosis of gastrointestinal stromal tumors: A consensus approach. Hum Pathol 33:459-465, 2002
Kindblom LG, Remotti HE, Aldenborg F, et al: Gastrointestinal pacemaker cell tumor (GIPACT): Gastrointestinal stromal tumors show phenotypic characteristics of the interstitial cells of Cajal. Am J Pathol 152:1259-1269, 1998
Heinrich MC, Rubin BP, Longley BJ, et al: Biology and genetic aspects of gastrointestinal stromal tumors: KIT activation and cytogenetic alterations. Hum Pathol 33:484-495, 2002
Heinrich MC, Corless CL, Duensing A, et al: PDGFRA activating mutations in gastrointestinal stromal tumors. Science 299:708-710, 2003
Somerhausen NdS, Fletcher CD: Gastrointestinal stromal tumors: An update. Sarcoma 2:133-141, 1998
DeMatteo R, Lewis J, Leung D, et al: Two hundred gastrointestinal stromal tumors: Recurrence patterns and prognostic factors for survival. Ann Surg 231:51-58, 2000
Emory T, Sobin L, Lukes L, et al: Prognosis of gastrointestinal smooth muscle (stromal) tumors: Dependence on anatomic site. Am J Surg Pathol 23:82-87, 1999
Howe J, Karnell L, Scott-Conner C: Small bowel sarcoma: Analysis of survival from the national cancer data base. Ann Surg Oncol 6:496-508, 2001
Lasota J, Jasinski M, Sarlomo-Rikala M, et al: Mutations in exon 11 of c-kit occur preferentially in malignant versus benign gastrointestinal stromal tumors and do not occur in leiomyomas or leiomyosarcomas. Am J Pathol 154:53-60, 1999
Ernst S, Hubbs A, Przygodzki R, et al: KIT mutation portends poor prognosis in gastrointestinal stromal/smooth muscle tumors. Lab Invest 78:1633-1636, 1998
Emile J, Théou N, Tabone S, et al: Clinicopathologic, phenotypic, and genotypic characteristics of gastrointestinal mesenchymal tumors. Clin Gastroenterol Hepatol 2:597-605, 2004
Taniguchi M, Nishida T, Hirota S, et al: Effect of c-kit mutation on prognosis of gastrointestinal stromal tumors. Cancer Res 59:4297-4300, 1999
Singer S, Rubin B, Lux M, et al: Prognostic value of KIT mutation type, mitotic activity, and histologic subtype in gastrointestinal stromal tumors. J Clin Oncol 20:3898-3905, 2002
Kim T, Lee H, Kang Y, et al: Prognostic significance of c-kit mutation in localized gastrointestinal stromal tumors. Clin Cancer Res 10:3076-3081, 2004
Wardelmann E, Losen I, Hans V, et al: Deletion of Trp-557 and Lys-558 in the juxtamembrane domain of the c-kit protooncogene is associated with metastatic behavior of gastrointestinal stromal tumors. Int J Cancer 106:887-895, 2003
Joensuu H, Roberts P, Sarlomo-Rikala M, et al: Effect of the tyrosine kinase inhibitor STI571 in a patient with a metastatic gastrointestinal stromal tumor. N Engl J Med 344:1052-1056, 2001
Reith J, Goldblum J, Lyles R, et al: Extragastrointestinal (Soft tissue) Stromal tumors: An analysis of 48 cases with emphasis on histologic predictors of outcome. Mod Pathol 13:577-585, 2000
Lasota J, Wozniak A, Sarlomo-Rikala M, et al: Mutations in exons 9 and 13 of KIT gene are rare events in gastrointestinal stromal tumors. A study of 200 cases. Am J Pathol 157:1091-1095, 2000
Corless C, McGreevey L, Haley A, et al: KIT mutations are common in incidental gastrointestinal stromal tumors one centimeter or less in size. Am J Pathol 160:1567-1572, 2002
Kaplan E, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958
Mantel N: Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 50:163-170, 1966
Cox D: Regression models and life-tables. J R Stat Soc [Ser A] 34:187-220, 1972
Lasota J, Dansonka-Mieszkowska A, Stachura T, et al: Gastrointestinal stromal tumors with internal tandem duplications in 3' end of KIT juxtamembrane domain occur predominantly in stomach and generally seem to have a favorable course. Mod Pathol 16:1257-1264, 2003
Kuwahara Y, Hirata A, Miwa H, et al: Epstein-Barr virus associated B-cell lymphoma of brain developing in myelodysplastic syndrome with c-kit mutation (Try-557 –> stop). Am J Hematol 65:234-238, 2000
Aparicio T, Boige V, Sabourin J, et al: Prognostic factors after surgery of primary resectable gastrointestinal stromal tumours. Eur J Surg Oncol 30:1098-1103, 2004
Lin S, Huang M, Zeng C, et al: Clinical manifestations and prognostic factors in patients with gastrointestinal stromal tumors. World J Gastroenterol 9:2809-2812, 2003
Fujijmoto Y, Nakanishi Y, Yoshimura K, et al: Clinicopathologic study of primary malignant gastrointestinal stromal tumor of the stomach, with special reference to prognostic factors: Analysis of results in 140 surgically resected patients. Gastric Cancer 6:39-48, 2003
Schneider-Stock R, Boltze C, Lasota J, et al: High prognostic value of p16INK4 alterations in gastrointestinal stromal tumors. J Clin Oncol 21:1688-1697, 2003
Ma Y, Cunningham M, Wang X, et al: Inhibition of spontaneous receptor phosphorylation by residues in a putative alpha-helix in the KIT intracellular juxtamembrane regionJ Biol Chem 274:13399-13402, 1999(Javier Martín, Andrés Pov)
Instituto Valenciano de Oncología
Universidad de Valencia, Valencia
Instituto Catalán de Oncología
Hospital Clinic Barcelona, Barcelona
Hospital de León, León
Hospital Miguel Servet, Hospital Clinico Lozano Blesa, Zaragoza
Hospital Marqués de Valdecilla, Santander
Hospital Virgen del Camino, Pamplona
Consorcio Cruz Roja, Llobregat
Hospital de Basurto, Bilbao
Hospital Clínico Madrid, Madrid
Hospital San Millán, Logro?o
Consorcio Sanitario de Terrassa, Hospital Central de Asturias, Oviedo, Spain
ABSTRACT
PURPOSE: To explore the prognostic value of mutations in c-KIT and PDGFR- genes with respect to relapse-free survival (RFS) in patients with gastrointestinal stromal tumors (GIST). We have investigated the prognostic relevance of the type and position of the mutations, in addition to other clinicopathologic factors, in a large series of patients with GIST.
METHODS: For this study, 162 patients were selected according to the following criteria: completely resected tumors with negative margins attended between 1994 and 2001; no metastasis at diagnosis; tumor larger than 2 cm, c-KIT–positive immunostaining; and no other primary tumors.
RESULTS: The median follow-up was 42 months for patients free of recurrence. Mutations were detected in 96 tumors (60%): 82 cases involving c-KIT and 14 cases involving PDFGR-. Univariate analysis demonstrated the following as poor prognostic factors for RFS: tumors larger than 10 cm (P < .0001); mitotic count higher than 10 mitoses per 50 high-power fields (P < .0001); high risk index (P < .0001); intestinal GIST location (P = .0041); high cellularity (P < .0001); tumor necrosis (P < .0001); deletions affecting exon 11 (P = .0007); and deletions affecting codons 557 to 558 (P < .0001). After the multivariate analysis, only the high risk index (relative risk [RR], 12.36), high cellularity (RR, 3.97), and deletions affecting codons 557 to 558 of c-KIT (RR, 2.57) corresponded to independent prognostic factors for RFS in GIST patients.
CONCLUSION: Deletions affecting codons 557 to 558 are relevant for the prognosis of RFS in GIST patients. This critical genetic alteration should be considered to be a new prognostic stratification variable for randomized trials exploring imatinib mesylate in the adjuvant setting in GIST patients.
INTRODUCTION
Gastrointestinal stromal tumors (GIST) are the most frequent mesenchymal tumors of the digestive tract. Interest in these neoplasms has been renewed since Hirota et al1 shed light on the molecular basis of their oncogenesis. The diagnosis of GIST is based on histomorphological features and, in most cases, histochemical immunoreactivity to the tyrosine kinase receptor of c-KIT (CD117).2-4 The cells implicated in the oncogenic process in GIST are primitive cells that are related to interstitial cells of Cajal.5 Mutations in the c-KIT proto-oncogene, which encodes a type III tyrosine kinase receptor (KIT), lead to constitutive activation of the protein.6 Although the majority of mutations in c-KIT have been detected within the juxtamembrane region encoded by exon 11, they also have been found within exon 9 and, more rarely, in exons 13 and 17. Additionally, in tumors in which mutations in c-KIT have not been identified, oncogenic mutations have been reported in the gene encoding the platelet-derived growth factor receptor alpha (PDGFR-).7 Several studies of prognostic factors in mesenchymal tumors of the gastrointestinal tract have been published; however, most of these studies did not take into account the current definition of GIST.8-11 Other recent studies analyzing clinical and pathologic variables as well as mutations in exon 11 of c-KIT as prognostic factors have also been carried out.12-18 Nevertheless, data exploring the prognostic relevance of such factors on relapse-free survival (RFS) in completely resected patients is scarce and controversial.15-18
The life expectancy of metastatic GIST patients has dramatically changed through the appearance of imatinib mesylate (IM), a drug that inhibits the tyrosine kinase activity in the context of this disease.19 It therefore seems important, considering GIST as a model for target-based molecular therapy, to continue exploring the prognostic factors beyond size and mitotic count relevant to this type of tumor. Hence, we have analyzed the molecular and clinicopathologic prognostic factors in a large group of patients with GIST who submitted to complete surgical resection. The size of our sample enabled us to examine the position and type of mutation as prognostic factors for RFS in patients with GIST. We feel that these data will help better explain the results in future randomized trials with IM as an adjuvant treatment for GIST including a control, nontreated group.
PATIENTS AND METHODS
Patients
We assessed the information from patients with mesenchymal tumors of the gastrointestinal tract that were included in the Spanish Group for Sarcoma Research (GEIS) registry and diagnosed from January 1994 to December 2001 in 30 of the 53 hospitals that are members of GEIS in Spain. The study was approved by the institutional review board of GEIS. The clinical staging at diagnosis took into consideration computed tomography/magnetic resonance imaging or abdominal echography data. The inclusion criteria for subsequent biologic analysis were: tumors larger than 2 cm; c-KIT–positive tumors; absence of other primary tumors; adequate surgery defined as complete gross removal, usually by segmental resection without tumor rupture; and the absence of intra-abdominal implants in the operative report. From November 2002 to March 2003, clinical data were updated and checked on a queries-based task between the central reviewer and the medical oncologists from the referral centers. Two pathologists with expertise in analyzing sarcomas (A.L.-B. and R.R.) performed independent pathologic reviews of the paraffin-embedded tissue blocks. To additionally meet the inclusion criteria, in all patients the surgical margins had to be negative and the diagnosis of GIST confirmed histologically with complete agreement of the panel of pathologists.
Age at diagnosis and sex were the demographic characteristics that were taken into consideration for each patient. The clinical presentation, if symptomatic, was registered for all patients along with the date of the first symptom. Tumor location was divided into stomach, small intestine, esophagus, rectum, colon, and omentum. The largest diameter of the primary tumor was determined after surgery in each case and measured by the pathologist. The diagnostic methods, defined as the first diagnostic procedure, were registered as either core biopsy or incisional or excisional biopsy. Postoperative surveillance varied between the different centers. The status (alive with or without disease, dead with or without disease) and recurrence data (date and type of treatments and additional treatments) were obtained by follow-up of the patients.
Pathology
Paraffin sections of formalin-fixed tissue (3 μm) were used for conventional hematoxylin and eosin staining and immunohistochemistry. The immunohistochemical analysis involved the StreptABComplex method coupled to heat-induced epitope retrieval, and the following antibodies were used: CD117, c-KIT (A-4502, polyclonal, 1:400 dilution; DAKO, Copenhagen, Denmark), SMA (N-1584, monoclonal, 1:2 dilution; DAKO), desmin (N-1526, monoclonal, 1:2 dilution; DAKO), S-100 protein (N-1573, polyclonal, 1:2 dilution; DAKO), and PDGFR- (SC-338, polyclonal, 1:200 dilution; Santa Cruz Biotechnology, Santa Cruz, CA). Both pathologists involved in the study reviewed all cases separately, and when disagreement arose, a consensus was reached after a review meeting. The pathologists did not have access to clinical information or macroscopic data regarding the tumors. The histological variables analyzed for each tumor were as follows: mitotic count per 50 high-power fields (HPF), necrosis (presence or absence), pleomorphism, cellularity20 (high or moderate-paucicellular), predominant cell type (spindle, epithelioid, or mixed), immunoreactivity of c-KIT and PDGFR- antisera (strong, moderate, or weak), and pattern of c-KIT immunostaining (diffuse or perinuclear).
DNA Sequencing
DNA was isolated from 3- to 5-μm sections of fixed and paraffin-embedded tissue. After deparaffinization, the tumor tissue was resuspended in lysis solution (0.5% sodium dodecyl sulfate, 0.5 mg/mL proteinase K, 10 mmol/L Tris-HCl, pH 8, 0.15 M NaCl, and 5 mmol/L EDTA) and incubated overnight at 55°C. DNA was extracted with phenol/chloroform/isoamyl alcohol (25:24:1), precipitated with ethanol, and dissolved in 10 to 35 μL of ultrapure water. Intronic polymerase chain reaction (PCR) primers were used to amplify exons 9, 11, 13,21 and 17 of c-KIT22 and exons 12 and 18 of PDGFRA-.7 PCR was performed in a reaction volume of 50 μL containing 1 μL of DNA, 50 mmol/L KCl, 10 mmol/L Tris-HCl, pH 8.3, 2 mmol/L MgCl2, 0.2 mmol/L each dNTP, 0.2 μmol/L for each primer, and 2 units of AmpliTaq Gold (Perkin Elmer, Norwalk, CT). PCR was carried out in a DNA Thermal Cycler 9600 (Perkin Elmer) after preheating the samples at 95°C for 10 minutes. The DNA was amplified over 40 cycles of: 1 minute of denaturation at 94°C; 1.5 minutes of annealing at 56°C for c-KIT primers or 2 minutes at 65°C for PDGFR- primers; and 1 minute of extension at 72°C with an additional final extension step of 10 minutes. Ten microliters of the PCR products were visualized in ethidium-bromide–stained 2% UltraPure agarose gels (Life Technologies, Paisley, Scotland) and photographed. Negative controls were included in every set of amplifications. Bidirectional sequencing with specific primers was performed on an ABI 310 sequencer using the BigDye Terminator v1.1 kit (Applied Biosystems, Inc, Foster City, CA).
Statistical Methods
We considered recurrence as the most suitable event for statistical analysis, because overall survival could be biased by the introduction of IM as a treatment of patients with metastatic GIST. RFS was measured from the date of surgery and was estimated according to the Kaplan-Meier method.23 The factors assessed at the time of diagnosis and evaluated as possible prognostic indicators for RFS were: age, sex, tumor location (gastric v intestinal v other), tumor size (2 to 5 v > 5 to 10 v > 10 cm), number of mitoses per 50 HPF (0 to 5 v 6 to 10 v > 10), cellularity (high v moderate-paucicellular), risk of GIST4 (low risk v intermediate risk v high risk), presence of tumor necrosis, nuclear pleomorphism, presence of a mutation in exon 11 of c-KIT, type of mutation (missense and deletion), position of mutation in exon 11 (from codons 550 to 591), and the presence of a mutation in the PDGFR gene.
Univariate analysis was performed by using the log-rank test.24 Factors that were potentially predictive of RFS by univariate analysis, as well as potential confounding variables, were entered into a multivariate analysis using the Cox proportional-hazards model.25 All P values reported are two tailed, and statistical significance was defined at P < .05. All statistical analyses were performed by using Statistical Package for the Social Sciences (SPSS) for Windows version 10.0 (SPSS Inc, Chicago, IL).
RESULTS
The files of 555 patients were considered, and of these patients, 357 were reviewed as both paraffin blocks and medical information were available. According to the inclusion criteria, a subset of 162 patients was selected for this study, and the reasons for exclusion of the remaining patients are shown in Figure 1.
The median age for the group of 162 patients, 82 males and 80 females, was 63 years (range, 24 to 91 years). Of these patients, 84% had symptoms at diagnosis, of which symptomatic anemia and abdominal pain were the most frequent. The median tumor size was 6 cm (range, 2 to 26 cm), and there were 93 gastric tumors, 63 small intestinal and six with another primary site (one esophagus, three colorectal, and two omentum; Table 1). Patients with gastric GIST presented more frequently with symptomatic anemia, and conversely, patients with small bowel GIST suffered abdominal pain as the predominant clinical presentation. The majority of cases were diagnosed after surgery (93%), and a preoperative biopsy was performed in only 12 (7%) patients.
In most tumors the mitotic rate was low: 59% had less than six mitoses per 50 HPF, 12% had six to 10 mitoses per 50 HPF, and 29% had more than 10 mitoses per 50 HPF. This distribution was similar for patients with gastric and small bowel GIST. Spindle cell type was present in 71.4% of cases and was more predominant in small bowel GIST than gastric GIST. In contrast, epithelioid and mixed cell types were more frequent in gastric tumors. High cellularity was found in 38 patients and was not associated with a specific location.
Regarding immunohistochemistry, strong, diffuse, and cytoplasmic expression of c-KIT (CD117) was observed in 82% of the tumors. A Golgi (dot-like accentuation) and a mixed pattern of immunostaining were seen in 10 and 16 tumors, respectively, mostly in patients with gastric GIST. The tumors also expressed smooth muscle actin (24%), desmin (2.4%), S-100 protein (3%), and PDGFR- (85%).
Mutations were detected in 96 tumors (60%), of which 77 mutations were in exon 11 of c-KIT. Of such exon 11 mutations, 41 were deletions (53%), 23 were missense mutations (30%), and 11 were in-frame duplications (14%; Fig 2). Of the 41 patients with deletions in exon 11, five also carried a missense mutation, as did one of the 11 with in-frame duplications (Table 2). Each type of mutation had a tendency to cluster. In-frame deletions more frequently involved codons 550 to 572, whereas in-frame duplications were more predominant in codons 573 to 591. Deletions simultaneously involving codons 557 and 558 were detected in 19 of the cases involving a deletion in exon 11 (46%). In contrast, 17 of the 30 missense mutations (57%) affected codons 559 to 560. In this subgroup of patients, no phenotypic association was detected, but only two patients suffered a recurrence (V559D and V560D). With respect to the duplications, we found 11 tumors (6.7%) carrying internal tandem duplications at the 3' end of the c-KIT juxtamembrane domain (internal tandem duplications in the 3' c-KIT-JM), of which six were females (all spindle histology) and five were males (three epithelioid and two spindle histology). Of these 11 tumors, eight originated in the stomach and three originated in the small bowel. Despite their relatively low mitotic rate and a reported favorable evolution,26 there were three recurrences among these patients (two epithelioid and one spindle). We found two tumors with nonsense mutations within exon 11, one of which (codon 557) involved a similar mutation to that described in a patient with B-cell lymphoma of the brain27 and was discovered along with a missense mutation involving codon V560D. To the best of our knowledge, the second nonsense mutation involving codon W582 has not been described previously.
Mutations in exon 9 were found in four patients, and in three of them, this mutation represented the insertion of GCC TAT that resulted in the duplication of amino acid residues Ala502 and Tyr503. The other patient incorporated a point mutation E490G with a different histology (small-cell variant). Two of the four patients with mutations in exon 9 relapsed. Only one patient had mutation in exon 13 (K642E), and the associated tumor was epithelioid, located in the small bowel and with 30 mitoses per 50 HPF. Six months after radical surgery, this patient's tumor recurred in the liver. No mutations were found in exon 17.
Mutations in PDGFR- were detected in 14 cases (9%) and were mutually exclusive of c-KIT mutations. Of these, seven mutations were found in exon 12, five of which involved deletions of codons 554 to 555. The remaining seven mutations were observed in exon 18, most often involving a missense mutation (six cases) and three of them affected codon 842 (D842V).
RFS Analysis
In January 2004, the median follow-up for patients free of recurrence was 42 months. There were 42 relapses with the following primary locations: stomach (15 cases), small bowel (23 cases), colorectal (three cases), and omentum (one case). The 5-year RFS for the 162 patients was 68% (± 9%). All but one of the recurrences were metastatic.
Clinicopathologic Variables
Age, sex, and clinical presentation had no influence on RFS for the whole group or for the different tumor locations. With regard to size, the 5-year RFS for the different risk categories was 82% ± 12% (0 to 5 cm), 79% ± 11% (> 5 to 10 cm), and 42% ± 17% (> 10 cm). There was a significant difference in the RFS between tumors larger or smaller than 10 cm (P < .0001). When the mitotic count was considered, we found a significantly higher RFS rate in patients with tumors having less than five mitoses per 50 HPF than in those with more than 10 mitoses per 50 HPF (P < .0001). Analyzing the index of risk as a prognostic factor,4 we found 5-year RFS rates of 94% ± 6% (low risk), 85% ± 15% (intermediate risk), and 44% ± 14% (high risk; P < .0001; Fig 3). The RFS for patients with intestinal GIST was significantly lower than that of patients with gastric GIST, with 5-year RFS rates of 55% (± 14%) and 76% (± 10%), respectively (P = .0041). High cellularity was a significant predictor of RFS, with recurrence in 25 of 38 tumors (65.7%). The 5-year RFS rates for high and moderate-paucicellular tumors were 23% (± 16%) and 81% (± 7%), respectively (P < .0001; Fig 4). Tumor necrosis was detected in 31 cases, of which recurrence occurred in 19 patients (5-year RFS rate, 33% ± 18%); in contrast, 23 recurrences were detected in the 125 tumors without necrosis (5-year RFS rate, 76% ± 10%; P < .0001).
Mutation Variables
Tumors with mutations in c-KIT recurred more frequently (5-year RFS, 57% ± 13%) than tumors without such mutations (5-year RFS, 80% ± 11%; P = .0084). We assessed whether the type of mutation within exon 11 might be related to the deleterious RFS. Patients carrying missense mutations in exon 11 had a similar RFS to the remaining patients. Nevertheless, patients with in-frame deletions in exon 11 had a significantly lower 5-year RFS rate (76% ± 9% v 44% ± 18%; P = .0007; Fig 5). We noticed that there was a relatively large number of relapses (11 of 19) among patients with deletions involving codons 557 to 558 of exon 11 (Fig 6), as well as a worse clinical outcome in a univariate analysis of the results. In other words, tumors with in-frame deletions involving both codons 557 to 558 provoked a significant reduction in the 5-year RFS (23% ± 2% v 74% ± 9%; P < .0001; Fig 7). When only patients with mutations in exon 11 were considered, this difference still remained significant (P = .0027). If either codon was analyzed separately, we found a similar deleterious RFS for each codon but not for the rest of exon 11 codons studied: codon 558 5-year RFS rate, 19% ± 2% v 76% ± 9% (P < .0001); codon 557 5-year RFS rate, 25% ± 2% v 75% ± 9% (P < .0001).
Multivariate Analysis
Only those variables showing a statistically significant relationship to RFS in the univariate analysis were entered in the Cox's proportional-hazard model. Therefore, tumor size, mitotic count, index of risk, location of GIST, cellularity, tumor necrosis, presence of a mutation in c-KIT, and the type or position of the mutation in c-KIT (exon 11 deletion mutation type and exon 11 deletion involving codons 557 to 558) were examined in the multivariate model with a forward stepwise method. The following factors were significantly associated with recurrence: high risk index, presence of high cellularity, and deletions in codons 557 to 558 within exon 11 (Table 3). When mitotic count and tumor size were entered in the Cox model instead of the risk index, small bowel location and mitotic count were also independent prognostic factors.
If the deletion in codons 557 to 558 of exon 11 was not included in the model, the type of c-KIT deletion mutation was an independent prognostic factor for RFS. It can be deduced that the prognostic relevance of deletion-type mutations in c-KIT is due to the presence of deletions involving codons 557 to 558.
DISCUSSION
This study demonstrates the prognostic relevance of the type and position of deletions in the c-KIT gene (ie, involving codons 557 and 558) in addition to the risk index and cellularity in patients with localized GIST who submitted to complete resection.
Our data regarding the impact of classical prognostic factors (size and mitotic count) on RFS are consistent with the results of other studies. Tumor size has been recognized as an independent prognostic marker by some authors9,10,17 but not by others.14,16,28 However, the largest series of patients reported10 did not take into account the current definition of GIST. Tumor size dropped out after entering mitotic count into the model, probably because there is a strong correlation between both factors.29 Otherwise, the risk index seemed to be an independent prognostic factor but only for the high-risk category, because the intermediate-risk category could not be considered independent as having prognostic value.
Cellularity has been reported in some studies as a prognostic variable in GIST but only in univariate analysis.20,30 Although our data revealed high cellularity as having independent prognostic value, we treat this result with some caution, because this parameter could be observer dependent.
Several studies have reported that when the c-KIT gene harbors mutations, GIST has a worse prognosis12-15,17 regardless of the percentage of patients with mutations, which ranged from 37% to 57%. In four separate series that included 30 to 113 patients,15-18 the relationship of c-KIT mutations with RFS was explored in patients with GIST who submitted to surgery and had negative margins. The influence of the type of mutation (ie, missense or deletion) was analyzed in two studies.16,17 In contrast to our results, missense mutations had a more favorable RFS in a series of 30 patients.16 However, in the other study of 86 patients, although the RFS rate was worse for tumors having mutations in c-KIT, no relationship was found between RFS rates and the type of mutation in exon 11.17
As observed previously, we found that codons 557 and 558 were those most often involved in deletions within exon 11.14,15 However, the poor RFS of patients with deletions involving codons 557 to 558 was particularly striking when compared with patients with other mutations in exon 11, with patients with any c-KIT mutation, or with the whole series. Multivariate analysis confirmed the independent prognostic value of this finding, which also was detected recently in a study of 55 patients.18
In a study of 39 patients with primary GIST, the tumors were divided as benign, borderline, or malignant according to their mitotic count and size.31 Interestingly, relapses were suffered in 10 of 11 patients with tumors larger than 2 cm and with deletions involving codons 557 to 558. In contrast, in four tumors in which the deletion involved other codons, a relapse was observed in only one patient. Other investigators detected deletions involving codons 557 to 558 in 67% of tumors.14 Nevertheless, they concluded that the strongest correlation with malignancy was seen in cases involving codons 563 to 569, because an aggressive behavior was always associated with these patients' tumors (n = 8). According to these data, deletions involving codon 558 arose in 37 cases, and among these cases, only in 10 did tumors not recur. Therefore 72% of patients with deletions involving codon 558 developed metastasis, a figure that is not too different from the 58% observed in the patients with deletions of codons 557 to 558 analyzed here. However, we did not observe recurrence in patients whose tumors involved deletion of codons 563 to 569, which may be explained by the exclusion here of patients with metastasis at diagnosis.
Two facts seem to support our findings that abnormalities involving codons 557 to 558 within exon 11 confer a significantly higher risk of recurrence in patients with GIST. On one hand, Trp557 has been identified among juxtamembrane residues with inhibitory roles in the control of receptor kinase activity. On the other hand, mutant forms of c-KIT generated by substitution of proline for Lys558 led to a high degree of constitutive receptor phosphorylation.32
Our findings are relevant for studies exploring a possible role for IM in the adjuvant setting. In forthcoming randomized trials, which include a control, nontreated group, the failure to stratify patients according to the presence of critical deletions may introduce a bias that would influence therapeutic results.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
Joan Carles (Hospital del Mar); Joaquín Bellmunt (Hospital Vall d'Hebrón); Auxiliadora Gómez (Hospital Reina Sofía); Andrés Meana (Hospital General de Alicante); Isabel Sevilla (Hospital Clínico de Málaga); Inmaculada Maestu (Hospital Virgen de los Lirios); Vicente Valentí (Hospital Sant Joan de Reus); Carmen Bala?à (Hospital Germans Trias I Pujol); José Luis Carrasco (Hospital Xeral-Cíes); Ricardo Cubedo (Hospital Puerta de Hierro); Javier Cassinello (Hospital de Guadalajara); Fernando Molano (Hospital Insular Gran Canaria); Jordi Rubió (ICO Girona); Isabel Bover (Hospital Son Llàtzer); Fina Cruz (Hospital Universitario de Canarias); and Antonio López-Pousa (Hospital Sant Pau) also contributed to this work. We thank the pathologists associated with GEIS for providing us with the tissues. Finally, thanks to Aina Rifà and Celi Rodriguez for data management and Mark Sefton for careful revision of the English language of the manuscript.
NOTES
Supported partially by a grant from Novartis Pharmaceuticals and the Cancer Genomics Network C03/10 of the Spanish Ministry of Health. These funds covered the expenses of the immunohistochemical and mutational analyses.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Hirota S, Isozaki K, Moriyama Y, et al: Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science 279:577-580, 1998
Miettinen M, Lasota J: Gastrointestinal stromal tumors: Definition, clinical, histological, immunohistochemical, and molecular genetic features and differential diagnosis. Virchows Arch 438:1-12, 2001
Sarlomo-Rikala M, Kovatich AJ, Barusevicius A, et al: CD117: A sensitive marker for gastrointestinal stromal tumors that is more specific than CD34. Mod Pathol 11:728-734, 1998
Fletcher CD, Berman JJ, Corless C, et al: Diagnosis of gastrointestinal stromal tumors: A consensus approach. Hum Pathol 33:459-465, 2002
Kindblom LG, Remotti HE, Aldenborg F, et al: Gastrointestinal pacemaker cell tumor (GIPACT): Gastrointestinal stromal tumors show phenotypic characteristics of the interstitial cells of Cajal. Am J Pathol 152:1259-1269, 1998
Heinrich MC, Rubin BP, Longley BJ, et al: Biology and genetic aspects of gastrointestinal stromal tumors: KIT activation and cytogenetic alterations. Hum Pathol 33:484-495, 2002
Heinrich MC, Corless CL, Duensing A, et al: PDGFRA activating mutations in gastrointestinal stromal tumors. Science 299:708-710, 2003
Somerhausen NdS, Fletcher CD: Gastrointestinal stromal tumors: An update. Sarcoma 2:133-141, 1998
DeMatteo R, Lewis J, Leung D, et al: Two hundred gastrointestinal stromal tumors: Recurrence patterns and prognostic factors for survival. Ann Surg 231:51-58, 2000
Emory T, Sobin L, Lukes L, et al: Prognosis of gastrointestinal smooth muscle (stromal) tumors: Dependence on anatomic site. Am J Surg Pathol 23:82-87, 1999
Howe J, Karnell L, Scott-Conner C: Small bowel sarcoma: Analysis of survival from the national cancer data base. Ann Surg Oncol 6:496-508, 2001
Lasota J, Jasinski M, Sarlomo-Rikala M, et al: Mutations in exon 11 of c-kit occur preferentially in malignant versus benign gastrointestinal stromal tumors and do not occur in leiomyomas or leiomyosarcomas. Am J Pathol 154:53-60, 1999
Ernst S, Hubbs A, Przygodzki R, et al: KIT mutation portends poor prognosis in gastrointestinal stromal/smooth muscle tumors. Lab Invest 78:1633-1636, 1998
Emile J, Théou N, Tabone S, et al: Clinicopathologic, phenotypic, and genotypic characteristics of gastrointestinal mesenchymal tumors. Clin Gastroenterol Hepatol 2:597-605, 2004
Taniguchi M, Nishida T, Hirota S, et al: Effect of c-kit mutation on prognosis of gastrointestinal stromal tumors. Cancer Res 59:4297-4300, 1999
Singer S, Rubin B, Lux M, et al: Prognostic value of KIT mutation type, mitotic activity, and histologic subtype in gastrointestinal stromal tumors. J Clin Oncol 20:3898-3905, 2002
Kim T, Lee H, Kang Y, et al: Prognostic significance of c-kit mutation in localized gastrointestinal stromal tumors. Clin Cancer Res 10:3076-3081, 2004
Wardelmann E, Losen I, Hans V, et al: Deletion of Trp-557 and Lys-558 in the juxtamembrane domain of the c-kit protooncogene is associated with metastatic behavior of gastrointestinal stromal tumors. Int J Cancer 106:887-895, 2003
Joensuu H, Roberts P, Sarlomo-Rikala M, et al: Effect of the tyrosine kinase inhibitor STI571 in a patient with a metastatic gastrointestinal stromal tumor. N Engl J Med 344:1052-1056, 2001
Reith J, Goldblum J, Lyles R, et al: Extragastrointestinal (Soft tissue) Stromal tumors: An analysis of 48 cases with emphasis on histologic predictors of outcome. Mod Pathol 13:577-585, 2000
Lasota J, Wozniak A, Sarlomo-Rikala M, et al: Mutations in exons 9 and 13 of KIT gene are rare events in gastrointestinal stromal tumors. A study of 200 cases. Am J Pathol 157:1091-1095, 2000
Corless C, McGreevey L, Haley A, et al: KIT mutations are common in incidental gastrointestinal stromal tumors one centimeter or less in size. Am J Pathol 160:1567-1572, 2002
Kaplan E, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958
Mantel N: Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 50:163-170, 1966
Cox D: Regression models and life-tables. J R Stat Soc [Ser A] 34:187-220, 1972
Lasota J, Dansonka-Mieszkowska A, Stachura T, et al: Gastrointestinal stromal tumors with internal tandem duplications in 3' end of KIT juxtamembrane domain occur predominantly in stomach and generally seem to have a favorable course. Mod Pathol 16:1257-1264, 2003
Kuwahara Y, Hirata A, Miwa H, et al: Epstein-Barr virus associated B-cell lymphoma of brain developing in myelodysplastic syndrome with c-kit mutation (Try-557 –> stop). Am J Hematol 65:234-238, 2000
Aparicio T, Boige V, Sabourin J, et al: Prognostic factors after surgery of primary resectable gastrointestinal stromal tumours. Eur J Surg Oncol 30:1098-1103, 2004
Lin S, Huang M, Zeng C, et al: Clinical manifestations and prognostic factors in patients with gastrointestinal stromal tumors. World J Gastroenterol 9:2809-2812, 2003
Fujijmoto Y, Nakanishi Y, Yoshimura K, et al: Clinicopathologic study of primary malignant gastrointestinal stromal tumor of the stomach, with special reference to prognostic factors: Analysis of results in 140 surgically resected patients. Gastric Cancer 6:39-48, 2003
Schneider-Stock R, Boltze C, Lasota J, et al: High prognostic value of p16INK4 alterations in gastrointestinal stromal tumors. J Clin Oncol 21:1688-1697, 2003
Ma Y, Cunningham M, Wang X, et al: Inhibition of spontaneous receptor phosphorylation by residues in a putative alpha-helix in the KIT intracellular juxtamembrane regionJ Biol Chem 274:13399-13402, 1999(Javier Martín, Andrés Pov)