- Email: [email protected]
c-kit mutations in gastrointestinal stromal tumours
Pathology (2002 ) 34, pp. 315– 319
ANATOMICAL
PATHOLOGY
c-kit mutations in gastrointestinal stromal tumours ADRIENNE L. MOREY, G. DAVID WANIGESEKERA*, NICHOLAS J. HAWKINS† AND ROBYN L. WARD* Departments of Anatomical Pathology and *Medical Oncology, St Vincent’s Hospital; †School of Pathology, University of New South Wales, Sydney, NSW, Australia
Summary Aims: Gastrointestinal stromal tumours (GISTs), once assumed to be of smooth muscle origin, generally express CD117 and CD34, similar to the interstitial cells of Cajal. Assessment of malignant potential in GISTs is problematic, especially on small biopsies. Some recent data indicate that mutations in the juxtamembrane domain (exon 11) of the c-kit (CD117) proto-oncogene may be associated with a worse prognosis. In this study, the frequency of c-kit exon 11 mutations has been determined in a series of 18 gut stromal tumours. Methods: Immunophenotype was assessed by immunoperoxidase stains for smooth muscle actin, desmin, S100, CD34 and CD117, and each tumour classified as being of low, uncertain (intermediate) or high malignant potential based on standard histological criteria. DNA from each tumour was extracted from fresh (n = 5) or formalin-fixed, paraffin-embedded (n = 13) tissues using the direct lysis method. Exon 11 was amplified by PCR and sequencing of both sense and antisense strands was performed on two occasions using an ABI 377 sequencer. Results: Mutations in exon 11 were detected in three of 14 confirmed GISTs, two being point mutations at codon 560 and one a 3-bp deletion resulting in the in-frame deletion of glutamine at codon 561. All three tumours were of high or intermediate malignant potential histologically. Three other ‘high risk’ primary GISTs and a metastatic GIST deposit were negative for exon 11 mutations. Conclusions: Data on this relatively small cohort of Australian patients indicate that c-kit exon 11 mutation analysis does not correlate well with histological assessment of malignant potential, and cannot be regarded as a reliable objective marker for poor prognosis in GISTs. Key words: Gastrointestinal stromal tumour, GIST, KIT, CD117, mutation, immunohistochemistry. Received 12 December 2001, revised 1 March, accepted 4 March 2002
INTRODUCTION Gastrointestinal stromal tumours ( GISTs) are now recognised as the most common mesenchymal tumour in the gastrointestinal tract, but have long presented problems to pathologists in terms of prognosis and classification.1 Once assumed to be of smooth muscle origin, they are now known to express an immunophenotype similar to the interstitial cells of Cajal ( gastrointestinal pacemaker cells), being generally positive for CD117 ( c-kit) and CD34. 2–4
Assessment of malignant potential in these spindle cell tumours using standard histological criteria of size, mitotic rate and tumour necrosis does not always accurately predict clinical outcome.5 Recently, mutations in the c-kit gene have been proposed as an important prognostic marker.6 The c-kit proto-oncogene, located on chromosome 4q11– 21, codes for a transmembrane tyrosine kinase receptor with structural similarity to the platelet-derived growth factor ( PDGF) receptor.7 It is thought to play an important role in the development of a range of cells including haemopoietic cells, germ cells, melanocytes and mast cells, as well as the interstitial cells of Cajal.8 Activating mutations in the kinase domain ( exon 17) of c-kit are found in the majority of adult mast cell tumours.9 Hirota and colleagues first demonstrated ‘gain of function’ mutations in exon 11 of the c-kit gene in five out of six sporadic GISTs2 as well as a germline c-kit mutation in a family with multiple GISTs.10 Transfection of the mutant c-kit cDNA induced malignant transformation in murine lymphoid cell cultures, implying that c-kit mutations play a role in tumour development. Since then, several other groups have demonstrated mutations in the ‘juxtamembrane’ domain ( exon 11) of the c-kit gene in a subset of GISTs.6,11–16 Mutations in the juxtamembrane domain are believed to cause constitutive activation of the KIT receptor by permitting dimerisation without binding of the receptor ligand, stem cell factor.2 Most reported mutations have been clustered around codons 550–560, and have included in-frame deletions ( 3–48 bp), point mutations and occasional insertions. Small numbers of mutations have been found in the more distal part of the transmembrane domain14 and other domains.16 The proportion of GISTs demonstrating such mutations has varied considerably between studies, and their prognostic significance is still controversial. One Japanese study, for example, found exon 11 mutations in 71 of 124 GISTs ( 57% ), the presence of mutations correlating with ‘malignant’ histological features and poor prognosis.6 A second large Japanese study detected the presence of such mutations in 31% of tumours, but failed to demonstrate a significant association with high histological grade and poor clinical outcome.14 It has been suggested that racial origin and age may be factors in determining location and frequency of mutations.14 The identification of an unambiguous marker predicting aggressive biological behaviour in GISTs is all the more relevant now that promising results are being obtained with a novel drug, STI571 ( Glivec, Novartis) which competitively inhibits the KIT tyrosine kinase.17
ISSN 0031–3025 printed/ISSN 1465– 3931 online/02/040315 – 05 © 2002 Royal College of Pathologists of Australasia DOI:10.1080/00313020220147122
316
Pathology ( 2002 ), 34, August
MOREY et al.
TABLE 1
Clinicopathological, immunohistochemical and genetic characteristics of tumours in this study
No.
Sex
Age
Site
Diameter ( mm)
/50HPF*
Necrosis
Risk
CD117†
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
F F F M M F F M M F M M M M F M M M
60 90 76 73 71 78 42 72 34 76 63 67 41 58 70 27 27 30
Duodenum Jejunum Caecum Jejunum Stomach Stomach Stomach Rectum Liver met Stomach Stomach Stomach Duodenum Stomach Jejunum Jejunum Jejunum Jejunum
100 35 60 13 95 270 110 36 55 110 45 50 47 160 9 30 45 110
1 2 66 0 1 8 110 34 1 4 3 2 55 25 0 125 130 0
N N N N N Y N N Y N N N N N N Y Y N
Intemediate Low High Low Intermediate High High Intermediate [Met] Intermediate Low Intermediate High High Low Intermediate† Intermediate† Low
+ +
Mitoses
– + + + + + + + + + + + +
– – –
CD34†
SMA†
Desmin†
S100†
Exon 11
–
+ /– + /–
– – – – – – – – – – – – –
– – – – – – – – – – – – – – –
Point mutation Wild Wild Wild Wild Wild Wild Wild Wild Wild Wild 3 bp deletion Wild Point mutation Wild Wild Wild Wild
+ /– + /– + + /– + + +
– + + + + /– + +
– – –
– + /–
– – + /–
– – – – – + /–
– – + /– + /– +
+ /–
– – – +
+ +
–
* Mitoses/50 HPF represents the number of mitoses per 50 high power fields using an Olympus BX50 microscope ( Olympus, Australia) and ´40 objective ( field diameter, 0.5 mm). † Generalised positive and negative immunohistochemical staining are represented by + and –, respectively; + /– represents focal or patchy staining. ‡ Tumours 16 and 17 were from the same patient. Although nominally of uncertain malignant potential due to small size, the extremely high mitotic rate suggests probable malignancy. Abbreviation: Met, metastasis.
We have examined a series of 18 mesenchymal tumours of the gastrointestinal tract from the archives of St. Vincent’s Hospital, Sydney ( n = 14) and from Prince of Wales Hospital, Sydney ( n = 4) to determine whether there is an association between c-kit exon 11 mutations and ‘classical’ criteria for high risk of malignancy in this patient sample.
MATERIALS AND METHODS Immunohistochemistry Most of the cases had been reported prior to the availability of CD117 antibody, and had been variously diagnosed as GIST, stromal tumour, leiomyoma, leiomyosarcoma or poorly differentiated sarcoma. One tumour was a metastatic spindle cell deposit in the liver following resection of a duodenal primary tumour. Tumour characteristics were confirmed on 4-mm H&E-stained sections prior to mutation analysis ( Table 1) and assessment made of malignant potential based on standard criteria ( size > 5 cm and mitoses > 5/50 high power fields ( HPF) = high risk; size > 5 cm or mitoses > 5/50 HPF = intermediate risk; size < 5 cm and mitoses < 5/50 HPF = low risk).18 Tumours 16 and 17 were discrete lesions from the same patient. Paraffin sections ( 4-mm) were stained with antibodies to CD34 ( QBEND/10, 1:100; BioGenex, USA), smooth muscle actin ( asm-1, 1:100; Novocastra, UK), desmin ( 33, 1:100; BioGenex), S100 ( Z311, 1:1000; Dako, Denmark) and CD117 ( c-kit) ( A4502, 1:300; Dako) using a Dako autostainer with either proteinase K digestion ( S100 ) or pressurecooker antigen retrieval in citrate buffer, pH 6.0, for 4 min ( other antibodies ). Detection was via the Vector ABC Elite detection system ( PK6100, 1:100 for 10 min; Vector Laboratories, USA) with DAB as the chromogen. For the purposes of this study, GISTs were defined as tumours exhibiting CD117 ( c-kit) reactivity.
DNA sequencing DNA were extracted from paraffin-embedded sections or fresh tissues ( Cases 6, 8, 9, 16, 17) as previously described.19 Exon 11 of the c-kit gene was amplified by PCR using 0.27 pmol of each primer ( forward:
59-GATCTATTTTTCCCTTTCTC– 39; and reverse: 59-AGCCCCTGTTTCATACTGAC– 39)6 in a buffer containing 2 mM MgCl2 , 0.2 mM dNTPs and using 0.25 U of Tth Plus DNA polymerase according to the manufacturer ’s instructions ( Fisher Biotec, Australia). The mixture was cycled under the following conditions: 94°C for 2 min; 30 cycles of 94°C for 30 s, 48°C for 30 s, 72°C for 1 min followed by a final extension step of 72°C for 10 min. PCR amplicons were electrophoresed on an 8% polyacrylamide gel and visualised using ethidium bromide staining and UV illumination. The exon 11 amplicon was excised with a clean blade and the DNA extracted from the gel by incubation in a buffer of 1 mM EDTA, pH 8, 0.5 M ammonium acetate and 0.1% SDS for 24 hours at 37°C. The DNA was precipitated using ethanol. Amplicons were sequenced using the BigDye Terminator Kit ( Applied Biosystems, USA) and a model 377 DNA sequencer ( Applied Biosystems). Each sample was sequenced once in the 59 direction and at least twice in the 39 direction using the primers above. Nucleotide sequencing was performed without knowledge of the pathological and immunohistochemical results.
RESULTS Tumours 1, 2 and 4–15 were classified as definite GISTs, showing strong expression of c-kit (CD117 ), with most also expressing CD34 (Fig. 1). Five of these GISTs showed some focal staining for smooth muscle actin and one showed focal staining for desmin; none expressed S100. Four were of high ‘histological’ malignant potential. Interestingly, the proven GIST metastasis ( Tumour 9) showed a very low mitotic rate (original duodenal tumour not available for comparison). Tumours 16 and 17 had morphology typical of GISTs but lacked expression of CD34 and CD117 as well as desmin; they showed labelling for S100 suggesting probable nerve sheath origin ( metastatic melanoma having previously been excluded). Tumour 3 was a highly pleiomorphic caecal sarcoma which lacked CD117 expression and showed only focal staining for CD34; the true histiogenesis of this lesion has not been determined. Tumour 18 was classified as a
c-kit MUTATIONS IN GISTs
317
Fig. 1 Typical GIST ( Case 14 ) showing ( A) spindle cell morpholog y, H&E stain, ( B) positive immunoperoxidase staining for CD34 and ( C) CD117, but ( D) negative staining for smooth muscle actin ( original magnification, ´40).
benign oesophageal leiomyoma based on lack of pleiomorphism, a low mitotic rate, strong staining for smooth muscle actin and desmin, and absence of tumour cell staining for CD34 and CD117. Exon 11 of the c-kit gene was PCR amplified from genomic DNA extracted from 18 GIST tumours and one section of normal gastric mucosa. The normal tissue showed a nucleotide sequence identical to the published wild-type sequence ( Genbank Acc No: HSU63834). Three of 18 tumour samples had a mutation in exon 11 ( Fig. 2).
Tumours 1 and 14 showed the same point mutation, namely a T® A transition at nucleotide position 75693 resulting in a valine to aspartic acid substitution at codon 560 ( Fig. 2). Tumour 12 revealed a 3-bp deletion from 75694 to 75696 resulting in the in-frame deletion of glutamine at codon 561. The remaining 15 tumour samples were identical to wildtype sequence. The relationship between c-kit mutation status at exon 11 and pathological grade is shown in Table 2. It is apparent that while tumours showing mutations were from intermediate or high risk groups, the majority of tumours in these higher groups showed no mutations in c-kit exon 11. TABLE 2 Relationship between the c-kit mutation status at exon 11 and pathological grade
Fig. 2 Nucleotide and predicted amino acid sequence of c-kit exon 11 in normal and tumour DNA. Tumours 1 and 14 exhibited a point mutation at codon 560, while tumour 12 showed an in-frame 3-bp deletion which resulted in loss of the glutamine residue ( E) at codon 561.
Risk group
Mutation present
No mutation
% Mutant
High Intermediate Low Total
1 2 0 3
4 3 4 11
22 40 0 21
There was no association between the presence of c-kit mutations and prognostic group, based on pathological criteria ( P = 0.73, Fisher’s exact test, two-sided). Only CD117-positive tumours were included. The metastatic lesion ( Tumour 9) was classified as ‘high risk’ for the purpose of this analysis.
318
Pathology ( 2002 ), 34, August
MOREY et al.
DISCUSSION Mutations in exon 11 of the c-kit gene were detected in only three of 14 (21%) immunohistochemically confirmed GISTs, one being ‘high risk’ and two being of ‘intermediate’ malignant potential by standard histological criteria. Three other ‘high-risk’ primary GISTs and a hepatic GIST metastasis were negative for exon 11 mutations. Data on this small Australian sample indicate that a substantial proportion of KIT-positive GISTs develop ( and may metastasise ) without evidence of a mutation in the juxtamembrane domain of c-kit. One previous North American study of exon 11 mutations in GISTs found a similar low mutation rate of 21%,12 whereas several other studies have found rates between 42 and 85%.6,11,13,15,16 The mutations detected have varied in site and nature; the most common have been in-frame deletions in the proximal part of exon 11 ( which spans codons 550–592); however, point mutations in the same region are also frequently found. Small numbers of cases have demonstrated homozygous mutations.11,14 Reliance on PCR-based gel-shift assays alone for detecting mutations is inadequate, as evidenced by one study which found a 40% increase in mutations upon sequencing apparently ‘wildtype’ bands.13 All mutations appear to be ‘gain of function’ mutations; presumably the reliance of interstitial cells of Cajal on KIT for their maintenance means that any cell harbouring a ‘loss of function’ mutation in c-kit is selected against in tumourigenesis. While most studies have limited their analysis to exon 11 of the c-kit gene, one Japanese study of 124 cases also examined exon 17 ( the kinase domain) and found no mutations there.6 A recent North American study has found small numbers of mutations in exons 9 and 13.16 The majority of studies to date have concluded that mutations tend to occur in more biologically aggressive GISTs,6,11,13 and it has been suggested that variation in mutation rates may reflect the proportion of ‘malignant’ tumours analysed in each series.6 The authors of one Japanese study which detected no statistically significant relationship between c-kit mutation and survival, tumour size, proliferative activity or telomerase activity suggested that their discrepant results might be partially explained by a relatively high proportion of ‘distal’ exon 11 mutations, which were all found in elderly female patients and appeared to be associated with a better prognosis.14 However, there are examples in most previous studies, as well as the present study, of mutation-negative GISTs with the capacity for metastatic spread. Other factors must obviously play a role in tumour development and progression. In the present study, loss of CD34 staining did not correlate with any prognostic grouping. Such an association has been reported in a series of mesenteric/omental GISTs,20 as well as in five of six lethal GISTs,4 but the role of CD34 in tumourigenesis has not been extensively studied. It is possible that mutations involving the CD34 gene are also important in the development of GISTs; alternatively, it may be that particularly aggressive tumours can develop from a KIT-positive/CD34-negative subgroup of interstitial cells of Cajal.4 Cytogenetic analyses have demonstrated frequent losses of chromosomes 14 and 22 in GISTs,21 and one study has demonstrated loss of heterozygosity at chromosome 1p36 in 24 of 80 GISTs,22 suggesting possible loss of a tumour suppressor gene in this region may also be involved in tumourigenesis.
Malignant GISTs respond poorly to radiotherapy and standard chemotheraputic agents and, until recently, there has been no effective treatment for advanced metastatic disease. A recent report has described promising results with a ‘designer’ drug, ST1571 ( Glivec, Novartis) in a patient with metastatic GIST.17 The drug was developed to selectively inhibit the enzymatic action of the tyrosine kinase produced by the BCR-ABL translocation in chronic myeloid leukaemia.23 It acts by blocking the ATP ‘pocket’,23 and is also active against several other tyrosine kinases, including the PDGF receptor and c-kit tyrosine kinases. The GIST from the single reported patient treated with this drug had a 15-bp deletion mutation in exon 11 of the c-kit gene,17 and it was suggested that the drug achieved its effect by inhibiting the constitutively active mutant c-kit tyrosine kinase. Therefore, the possibility exists that this drug may be inappropriate (or less effective) therapy for patients with GISTs lacking evidence of a c-kit mutation ( and associated constitutive KIT activation), and in whom other factors such as CD34 mutation and loss of a tumour suppressor gene may play a significant role. Australian patients are currently being recruited for further trials of this drug, with entry criteria based on confirmed CD117 expression by immunohistochemistry. Until data are available on whether patients with GISTs lacking c-kit mutations are responsive to the drug, the outlook should perhaps be cautious in view of the relatively small percentage of GISTs exhibiting c-kit exon 11 mutations in this series of 18 tumours from Australian patients. ACKNOWLEDGEMENTS Drs J. Turner, A. Field and S. Rainer, Anatomical Pathology, St Vincent’s Hospital, and Dr R. Crouch, Anatomical Pathology, Prince of Wales Hospital, are acknowledged for initial diagnostic work and thanked for access to archival material. Address for correspondence: Dr A. Morey, Department of Anatomical Pathology, St Vincent’s Hospital, Darlinghurst, NSW 2010, Australia. E-mail: [email protected] u
References 1. Rubin BP, Fletcher JA, Fletcher CDM. Molecular insights into the histogenesis and pathogenesis of gastrointestinal stromal tumors. Int J Surg Pathol 2000; 8: 5–10. 2. Hirota S, Isozaki K, Moriyama Y, et al. Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science 1998; 279: 577– 80. 3. Kindblom, LG, Remotti HF, Aldenborg F, Meis-Kinblom JM. Gastrointestinal pacemaker cell tumour ( GIPACT). Gastrointestinal stromal tumours show phenotypic characteristics of the interstitial cells of Cajal. Am J Pathol 1998; 152: 1259–69. 4. Sircar K, Hewlett BR, Huizinga JD, Chorneyko K, Berezin I, Riddell RH. Interstitial cells of Cajal as precursors for gastrointestinal stromal tumours. Am J Surg Pathol 1999; 23: 377– 89. 5. Emory TS, Sobin LH, Lukes L, Lee DH, O’Leary TJ. Prognosis of gastrointestinal smooth-muscle ( stromal ) tumours. Am J Surg Pathol 1999; 23; 82–7. 6. Taniguchi M, Nishida T, Hirota S, et al. Effect of c-kit mutation on prognosis of gastrointestinal stromal tumors. Cancer Res 1999; 59: 4297– 300. 7. Yarden Y, Ullrich A. Growth factor receptor tyrosine kinases. Annu Rev Biochem 1988; 57: 443–8. 8. Maeda II, Yamagata A, Nishikawa S, et al. Requirement of C-kit for development of intestinal pacemaker system. Development 1992; 116: 369–75. 9. Hartmann K, Henz BM. Mastocytosis: recent advances in defining the disease. Br J Dermatol 2001; 144: 682–95. 10. Nishida T, Hirota S, Taniguchi M, et al. Familial gastrointestinal stromal tumours with germline mutation of the KIT gene. Nat Genet 1998; 19: 323–4.
c-kit MUTATIONS IN GISTs
11. Ernst SI, Habbs AE, Przygodski RM, et al. KIT mutation portends poor prognosis in gastrointestinal stromal/smooth muscle tumours. Lab Invest 1998; 78: 1633– 6. 12. Moskaluk CA, Tian Q, Marshall CR, et al. Mutations of c-kit JM domain are found in a minority of human gastrointestinal stromal tumours. Oncogene 1999; 18: 1897–902. 13. Lasota J, Jasinski M, Sarlomo-Rikala M, Miettinen M. Mutations in exon 11 of c-kit occur preferentially in malignant versus benign gastrointestinal tumors and do not occur in leiomyomas or leiomyosarcomas. Am J Pathol 1999; 154: 53– 60. 14. Sakurai S, Fukasawa T, Chong JM, Tanaka A, Fukayama M. C-kit gene abnormalities in gastrointestinal stromal tumors ( tumors of interstitial cells of Cajal). Jpn J Cancer Res 1999; 90: 1321– 8. 15. Miettinen M, Sarlomo-Rikala M, Sobin LH, Lasota J. Esophageal stromal tumors: a clinicopathologic, immunohistochemical, and molecular genetic study of 17 cases and comparison with esophagea l leiomyomas and leiomyosarcomas. Am J Surg Pathol 2000; 24: 211– 22. 16. Lux ML, Rubin BP, Biase TL, et al. KIT extracellular and kinase domain mutations in gastrointestinal stromal tumors. Am J Pathol 2000; 156: 791– 5.
319
17. Joensuu H, Roberts PJ, Sarlomo-Rikala M, et al. Effect of the tyrosine kinase inhibitor STI571 in a patient with a metastatic gastrointestinal stromal tumor. New Engl J Med 2001; 344: 1052–6. 18. Franquemont DW. Differentiation and risk assessment of gastrointestinal stromal tumours. Am J Clin Pathol 1995; 103: 41–7. 19. Ward RL, Hawkins NJ, O’Grady R, et al. REMS-PCR – a novel assay for the detection of K-ras mutations in clinical samples. Am J Pathol 1998; 153; 373– 9. 20. Miettinen M, Monihan JM, Sarlomo-Rikala M, et al. Gastrointestinal stromal tumours/smooth muscle tumours ( GISTs) primary in the omentum and mesentry. Am J Surg Pathol 1999; 23: 1109–18. 21. Marci V, Casorzo L, Sarotto I, Dogliani N, Milazzo MG, Risio M. Gastrointestinal stromal tumour, uncommitted type, with monosomies 14 and 22 as the only chromosomal abnormalities. Cancer Genet Cytogenet 1998; 102: 135–8. 22. O’Leary T, Ernst S, Przygodzki R, Emory T, Sobin L. Loss of heterozygosity at 1p36 predicts poor prognosis in gastrointestinal stromal/smooth muscle tumors. Lab Invest 1999; 79: 1461–7. 23. Druker BJ, Lydon NB. Lessons learned from the development of an abl tyrosine kinase inhibitor for chronic myelogenous leukemia. J Clin Invest 2000; 105: 3–7.
ANATOMICAL
PATHOLOGY
c-kit mutations in gastrointestinal stromal tumours ADRIENNE L. MOREY, G. DAVID WANIGESEKERA*, NICHOLAS J. HAWKINS† AND ROBYN L. WARD* Departments of Anatomical Pathology and *Medical Oncology, St Vincent’s Hospital; †School of Pathology, University of New South Wales, Sydney, NSW, Australia
Summary Aims: Gastrointestinal stromal tumours (GISTs), once assumed to be of smooth muscle origin, generally express CD117 and CD34, similar to the interstitial cells of Cajal. Assessment of malignant potential in GISTs is problematic, especially on small biopsies. Some recent data indicate that mutations in the juxtamembrane domain (exon 11) of the c-kit (CD117) proto-oncogene may be associated with a worse prognosis. In this study, the frequency of c-kit exon 11 mutations has been determined in a series of 18 gut stromal tumours. Methods: Immunophenotype was assessed by immunoperoxidase stains for smooth muscle actin, desmin, S100, CD34 and CD117, and each tumour classified as being of low, uncertain (intermediate) or high malignant potential based on standard histological criteria. DNA from each tumour was extracted from fresh (n = 5) or formalin-fixed, paraffin-embedded (n = 13) tissues using the direct lysis method. Exon 11 was amplified by PCR and sequencing of both sense and antisense strands was performed on two occasions using an ABI 377 sequencer. Results: Mutations in exon 11 were detected in three of 14 confirmed GISTs, two being point mutations at codon 560 and one a 3-bp deletion resulting in the in-frame deletion of glutamine at codon 561. All three tumours were of high or intermediate malignant potential histologically. Three other ‘high risk’ primary GISTs and a metastatic GIST deposit were negative for exon 11 mutations. Conclusions: Data on this relatively small cohort of Australian patients indicate that c-kit exon 11 mutation analysis does not correlate well with histological assessment of malignant potential, and cannot be regarded as a reliable objective marker for poor prognosis in GISTs. Key words: Gastrointestinal stromal tumour, GIST, KIT, CD117, mutation, immunohistochemistry. Received 12 December 2001, revised 1 March, accepted 4 March 2002
INTRODUCTION Gastrointestinal stromal tumours ( GISTs) are now recognised as the most common mesenchymal tumour in the gastrointestinal tract, but have long presented problems to pathologists in terms of prognosis and classification.1 Once assumed to be of smooth muscle origin, they are now known to express an immunophenotype similar to the interstitial cells of Cajal ( gastrointestinal pacemaker cells), being generally positive for CD117 ( c-kit) and CD34. 2–4
Assessment of malignant potential in these spindle cell tumours using standard histological criteria of size, mitotic rate and tumour necrosis does not always accurately predict clinical outcome.5 Recently, mutations in the c-kit gene have been proposed as an important prognostic marker.6 The c-kit proto-oncogene, located on chromosome 4q11– 21, codes for a transmembrane tyrosine kinase receptor with structural similarity to the platelet-derived growth factor ( PDGF) receptor.7 It is thought to play an important role in the development of a range of cells including haemopoietic cells, germ cells, melanocytes and mast cells, as well as the interstitial cells of Cajal.8 Activating mutations in the kinase domain ( exon 17) of c-kit are found in the majority of adult mast cell tumours.9 Hirota and colleagues first demonstrated ‘gain of function’ mutations in exon 11 of the c-kit gene in five out of six sporadic GISTs2 as well as a germline c-kit mutation in a family with multiple GISTs.10 Transfection of the mutant c-kit cDNA induced malignant transformation in murine lymphoid cell cultures, implying that c-kit mutations play a role in tumour development. Since then, several other groups have demonstrated mutations in the ‘juxtamembrane’ domain ( exon 11) of the c-kit gene in a subset of GISTs.6,11–16 Mutations in the juxtamembrane domain are believed to cause constitutive activation of the KIT receptor by permitting dimerisation without binding of the receptor ligand, stem cell factor.2 Most reported mutations have been clustered around codons 550–560, and have included in-frame deletions ( 3–48 bp), point mutations and occasional insertions. Small numbers of mutations have been found in the more distal part of the transmembrane domain14 and other domains.16 The proportion of GISTs demonstrating such mutations has varied considerably between studies, and their prognostic significance is still controversial. One Japanese study, for example, found exon 11 mutations in 71 of 124 GISTs ( 57% ), the presence of mutations correlating with ‘malignant’ histological features and poor prognosis.6 A second large Japanese study detected the presence of such mutations in 31% of tumours, but failed to demonstrate a significant association with high histological grade and poor clinical outcome.14 It has been suggested that racial origin and age may be factors in determining location and frequency of mutations.14 The identification of an unambiguous marker predicting aggressive biological behaviour in GISTs is all the more relevant now that promising results are being obtained with a novel drug, STI571 ( Glivec, Novartis) which competitively inhibits the KIT tyrosine kinase.17
ISSN 0031–3025 printed/ISSN 1465– 3931 online/02/040315 – 05 © 2002 Royal College of Pathologists of Australasia DOI:10.1080/00313020220147122
316
Pathology ( 2002 ), 34, August
MOREY et al.
TABLE 1
Clinicopathological, immunohistochemical and genetic characteristics of tumours in this study
No.
Sex
Age
Site
Diameter ( mm)
/50HPF*
Necrosis
Risk
CD117†
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
F F F M M F F M M F M M M M F M M M
60 90 76 73 71 78 42 72 34 76 63 67 41 58 70 27 27 30
Duodenum Jejunum Caecum Jejunum Stomach Stomach Stomach Rectum Liver met Stomach Stomach Stomach Duodenum Stomach Jejunum Jejunum Jejunum Jejunum
100 35 60 13 95 270 110 36 55 110 45 50 47 160 9 30 45 110
1 2 66 0 1 8 110 34 1 4 3 2 55 25 0 125 130 0
N N N N N Y N N Y N N N N N N Y Y N
Intemediate Low High Low Intermediate High High Intermediate [Met] Intermediate Low Intermediate High High Low Intermediate† Intermediate† Low
+ +
Mitoses
– + + + + + + + + + + + +
– – –
CD34†
SMA†
Desmin†
S100†
Exon 11
–
+ /– + /–
– – – – – – – – – – – – –
– – – – – – – – – – – – – – –
Point mutation Wild Wild Wild Wild Wild Wild Wild Wild Wild Wild 3 bp deletion Wild Point mutation Wild Wild Wild Wild
+ /– + /– + + /– + + +
– + + + + /– + +
– – –
– + /–
– – + /–
– – – – – + /–
– – + /– + /– +
+ /–
– – – +
+ +
–
* Mitoses/50 HPF represents the number of mitoses per 50 high power fields using an Olympus BX50 microscope ( Olympus, Australia) and ´40 objective ( field diameter, 0.5 mm). † Generalised positive and negative immunohistochemical staining are represented by + and –, respectively; + /– represents focal or patchy staining. ‡ Tumours 16 and 17 were from the same patient. Although nominally of uncertain malignant potential due to small size, the extremely high mitotic rate suggests probable malignancy. Abbreviation: Met, metastasis.
We have examined a series of 18 mesenchymal tumours of the gastrointestinal tract from the archives of St. Vincent’s Hospital, Sydney ( n = 14) and from Prince of Wales Hospital, Sydney ( n = 4) to determine whether there is an association between c-kit exon 11 mutations and ‘classical’ criteria for high risk of malignancy in this patient sample.
MATERIALS AND METHODS Immunohistochemistry Most of the cases had been reported prior to the availability of CD117 antibody, and had been variously diagnosed as GIST, stromal tumour, leiomyoma, leiomyosarcoma or poorly differentiated sarcoma. One tumour was a metastatic spindle cell deposit in the liver following resection of a duodenal primary tumour. Tumour characteristics were confirmed on 4-mm H&E-stained sections prior to mutation analysis ( Table 1) and assessment made of malignant potential based on standard criteria ( size > 5 cm and mitoses > 5/50 high power fields ( HPF) = high risk; size > 5 cm or mitoses > 5/50 HPF = intermediate risk; size < 5 cm and mitoses < 5/50 HPF = low risk).18 Tumours 16 and 17 were discrete lesions from the same patient. Paraffin sections ( 4-mm) were stained with antibodies to CD34 ( QBEND/10, 1:100; BioGenex, USA), smooth muscle actin ( asm-1, 1:100; Novocastra, UK), desmin ( 33, 1:100; BioGenex), S100 ( Z311, 1:1000; Dako, Denmark) and CD117 ( c-kit) ( A4502, 1:300; Dako) using a Dako autostainer with either proteinase K digestion ( S100 ) or pressurecooker antigen retrieval in citrate buffer, pH 6.0, for 4 min ( other antibodies ). Detection was via the Vector ABC Elite detection system ( PK6100, 1:100 for 10 min; Vector Laboratories, USA) with DAB as the chromogen. For the purposes of this study, GISTs were defined as tumours exhibiting CD117 ( c-kit) reactivity.
DNA sequencing DNA were extracted from paraffin-embedded sections or fresh tissues ( Cases 6, 8, 9, 16, 17) as previously described.19 Exon 11 of the c-kit gene was amplified by PCR using 0.27 pmol of each primer ( forward:
59-GATCTATTTTTCCCTTTCTC– 39; and reverse: 59-AGCCCCTGTTTCATACTGAC– 39)6 in a buffer containing 2 mM MgCl2 , 0.2 mM dNTPs and using 0.25 U of Tth Plus DNA polymerase according to the manufacturer ’s instructions ( Fisher Biotec, Australia). The mixture was cycled under the following conditions: 94°C for 2 min; 30 cycles of 94°C for 30 s, 48°C for 30 s, 72°C for 1 min followed by a final extension step of 72°C for 10 min. PCR amplicons were electrophoresed on an 8% polyacrylamide gel and visualised using ethidium bromide staining and UV illumination. The exon 11 amplicon was excised with a clean blade and the DNA extracted from the gel by incubation in a buffer of 1 mM EDTA, pH 8, 0.5 M ammonium acetate and 0.1% SDS for 24 hours at 37°C. The DNA was precipitated using ethanol. Amplicons were sequenced using the BigDye Terminator Kit ( Applied Biosystems, USA) and a model 377 DNA sequencer ( Applied Biosystems). Each sample was sequenced once in the 59 direction and at least twice in the 39 direction using the primers above. Nucleotide sequencing was performed without knowledge of the pathological and immunohistochemical results.
RESULTS Tumours 1, 2 and 4–15 were classified as definite GISTs, showing strong expression of c-kit (CD117 ), with most also expressing CD34 (Fig. 1). Five of these GISTs showed some focal staining for smooth muscle actin and one showed focal staining for desmin; none expressed S100. Four were of high ‘histological’ malignant potential. Interestingly, the proven GIST metastasis ( Tumour 9) showed a very low mitotic rate (original duodenal tumour not available for comparison). Tumours 16 and 17 had morphology typical of GISTs but lacked expression of CD34 and CD117 as well as desmin; they showed labelling for S100 suggesting probable nerve sheath origin ( metastatic melanoma having previously been excluded). Tumour 3 was a highly pleiomorphic caecal sarcoma which lacked CD117 expression and showed only focal staining for CD34; the true histiogenesis of this lesion has not been determined. Tumour 18 was classified as a
c-kit MUTATIONS IN GISTs
317
Fig. 1 Typical GIST ( Case 14 ) showing ( A) spindle cell morpholog y, H&E stain, ( B) positive immunoperoxidase staining for CD34 and ( C) CD117, but ( D) negative staining for smooth muscle actin ( original magnification, ´40).
benign oesophageal leiomyoma based on lack of pleiomorphism, a low mitotic rate, strong staining for smooth muscle actin and desmin, and absence of tumour cell staining for CD34 and CD117. Exon 11 of the c-kit gene was PCR amplified from genomic DNA extracted from 18 GIST tumours and one section of normal gastric mucosa. The normal tissue showed a nucleotide sequence identical to the published wild-type sequence ( Genbank Acc No: HSU63834). Three of 18 tumour samples had a mutation in exon 11 ( Fig. 2).
Tumours 1 and 14 showed the same point mutation, namely a T® A transition at nucleotide position 75693 resulting in a valine to aspartic acid substitution at codon 560 ( Fig. 2). Tumour 12 revealed a 3-bp deletion from 75694 to 75696 resulting in the in-frame deletion of glutamine at codon 561. The remaining 15 tumour samples were identical to wildtype sequence. The relationship between c-kit mutation status at exon 11 and pathological grade is shown in Table 2. It is apparent that while tumours showing mutations were from intermediate or high risk groups, the majority of tumours in these higher groups showed no mutations in c-kit exon 11. TABLE 2 Relationship between the c-kit mutation status at exon 11 and pathological grade
Fig. 2 Nucleotide and predicted amino acid sequence of c-kit exon 11 in normal and tumour DNA. Tumours 1 and 14 exhibited a point mutation at codon 560, while tumour 12 showed an in-frame 3-bp deletion which resulted in loss of the glutamine residue ( E) at codon 561.
Risk group
Mutation present
No mutation
% Mutant
High Intermediate Low Total
1 2 0 3
4 3 4 11
22 40 0 21
There was no association between the presence of c-kit mutations and prognostic group, based on pathological criteria ( P = 0.73, Fisher’s exact test, two-sided). Only CD117-positive tumours were included. The metastatic lesion ( Tumour 9) was classified as ‘high risk’ for the purpose of this analysis.
318
Pathology ( 2002 ), 34, August
MOREY et al.
DISCUSSION Mutations in exon 11 of the c-kit gene were detected in only three of 14 (21%) immunohistochemically confirmed GISTs, one being ‘high risk’ and two being of ‘intermediate’ malignant potential by standard histological criteria. Three other ‘high-risk’ primary GISTs and a hepatic GIST metastasis were negative for exon 11 mutations. Data on this small Australian sample indicate that a substantial proportion of KIT-positive GISTs develop ( and may metastasise ) without evidence of a mutation in the juxtamembrane domain of c-kit. One previous North American study of exon 11 mutations in GISTs found a similar low mutation rate of 21%,12 whereas several other studies have found rates between 42 and 85%.6,11,13,15,16 The mutations detected have varied in site and nature; the most common have been in-frame deletions in the proximal part of exon 11 ( which spans codons 550–592); however, point mutations in the same region are also frequently found. Small numbers of cases have demonstrated homozygous mutations.11,14 Reliance on PCR-based gel-shift assays alone for detecting mutations is inadequate, as evidenced by one study which found a 40% increase in mutations upon sequencing apparently ‘wildtype’ bands.13 All mutations appear to be ‘gain of function’ mutations; presumably the reliance of interstitial cells of Cajal on KIT for their maintenance means that any cell harbouring a ‘loss of function’ mutation in c-kit is selected against in tumourigenesis. While most studies have limited their analysis to exon 11 of the c-kit gene, one Japanese study of 124 cases also examined exon 17 ( the kinase domain) and found no mutations there.6 A recent North American study has found small numbers of mutations in exons 9 and 13.16 The majority of studies to date have concluded that mutations tend to occur in more biologically aggressive GISTs,6,11,13 and it has been suggested that variation in mutation rates may reflect the proportion of ‘malignant’ tumours analysed in each series.6 The authors of one Japanese study which detected no statistically significant relationship between c-kit mutation and survival, tumour size, proliferative activity or telomerase activity suggested that their discrepant results might be partially explained by a relatively high proportion of ‘distal’ exon 11 mutations, which were all found in elderly female patients and appeared to be associated with a better prognosis.14 However, there are examples in most previous studies, as well as the present study, of mutation-negative GISTs with the capacity for metastatic spread. Other factors must obviously play a role in tumour development and progression. In the present study, loss of CD34 staining did not correlate with any prognostic grouping. Such an association has been reported in a series of mesenteric/omental GISTs,20 as well as in five of six lethal GISTs,4 but the role of CD34 in tumourigenesis has not been extensively studied. It is possible that mutations involving the CD34 gene are also important in the development of GISTs; alternatively, it may be that particularly aggressive tumours can develop from a KIT-positive/CD34-negative subgroup of interstitial cells of Cajal.4 Cytogenetic analyses have demonstrated frequent losses of chromosomes 14 and 22 in GISTs,21 and one study has demonstrated loss of heterozygosity at chromosome 1p36 in 24 of 80 GISTs,22 suggesting possible loss of a tumour suppressor gene in this region may also be involved in tumourigenesis.
Malignant GISTs respond poorly to radiotherapy and standard chemotheraputic agents and, until recently, there has been no effective treatment for advanced metastatic disease. A recent report has described promising results with a ‘designer’ drug, ST1571 ( Glivec, Novartis) in a patient with metastatic GIST.17 The drug was developed to selectively inhibit the enzymatic action of the tyrosine kinase produced by the BCR-ABL translocation in chronic myeloid leukaemia.23 It acts by blocking the ATP ‘pocket’,23 and is also active against several other tyrosine kinases, including the PDGF receptor and c-kit tyrosine kinases. The GIST from the single reported patient treated with this drug had a 15-bp deletion mutation in exon 11 of the c-kit gene,17 and it was suggested that the drug achieved its effect by inhibiting the constitutively active mutant c-kit tyrosine kinase. Therefore, the possibility exists that this drug may be inappropriate (or less effective) therapy for patients with GISTs lacking evidence of a c-kit mutation ( and associated constitutive KIT activation), and in whom other factors such as CD34 mutation and loss of a tumour suppressor gene may play a significant role. Australian patients are currently being recruited for further trials of this drug, with entry criteria based on confirmed CD117 expression by immunohistochemistry. Until data are available on whether patients with GISTs lacking c-kit mutations are responsive to the drug, the outlook should perhaps be cautious in view of the relatively small percentage of GISTs exhibiting c-kit exon 11 mutations in this series of 18 tumours from Australian patients. ACKNOWLEDGEMENTS Drs J. Turner, A. Field and S. Rainer, Anatomical Pathology, St Vincent’s Hospital, and Dr R. Crouch, Anatomical Pathology, Prince of Wales Hospital, are acknowledged for initial diagnostic work and thanked for access to archival material. Address for correspondence: Dr A. Morey, Department of Anatomical Pathology, St Vincent’s Hospital, Darlinghurst, NSW 2010, Australia. E-mail: [email protected] u
References 1. Rubin BP, Fletcher JA, Fletcher CDM. Molecular insights into the histogenesis and pathogenesis of gastrointestinal stromal tumors. Int J Surg Pathol 2000; 8: 5–10. 2. Hirota S, Isozaki K, Moriyama Y, et al. Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science 1998; 279: 577– 80. 3. Kindblom, LG, Remotti HF, Aldenborg F, Meis-Kinblom JM. Gastrointestinal pacemaker cell tumour ( GIPACT). Gastrointestinal stromal tumours show phenotypic characteristics of the interstitial cells of Cajal. Am J Pathol 1998; 152: 1259–69. 4. Sircar K, Hewlett BR, Huizinga JD, Chorneyko K, Berezin I, Riddell RH. Interstitial cells of Cajal as precursors for gastrointestinal stromal tumours. Am J Surg Pathol 1999; 23: 377– 89. 5. Emory TS, Sobin LH, Lukes L, Lee DH, O’Leary TJ. Prognosis of gastrointestinal smooth-muscle ( stromal ) tumours. Am J Surg Pathol 1999; 23; 82–7. 6. Taniguchi M, Nishida T, Hirota S, et al. Effect of c-kit mutation on prognosis of gastrointestinal stromal tumors. Cancer Res 1999; 59: 4297– 300. 7. Yarden Y, Ullrich A. Growth factor receptor tyrosine kinases. Annu Rev Biochem 1988; 57: 443–8. 8. Maeda II, Yamagata A, Nishikawa S, et al. Requirement of C-kit for development of intestinal pacemaker system. Development 1992; 116: 369–75. 9. Hartmann K, Henz BM. Mastocytosis: recent advances in defining the disease. Br J Dermatol 2001; 144: 682–95. 10. Nishida T, Hirota S, Taniguchi M, et al. Familial gastrointestinal stromal tumours with germline mutation of the KIT gene. Nat Genet 1998; 19: 323–4.
c-kit MUTATIONS IN GISTs
11. Ernst SI, Habbs AE, Przygodski RM, et al. KIT mutation portends poor prognosis in gastrointestinal stromal/smooth muscle tumours. Lab Invest 1998; 78: 1633– 6. 12. Moskaluk CA, Tian Q, Marshall CR, et al. Mutations of c-kit JM domain are found in a minority of human gastrointestinal stromal tumours. Oncogene 1999; 18: 1897–902. 13. Lasota J, Jasinski M, Sarlomo-Rikala M, Miettinen M. Mutations in exon 11 of c-kit occur preferentially in malignant versus benign gastrointestinal tumors and do not occur in leiomyomas or leiomyosarcomas. Am J Pathol 1999; 154: 53– 60. 14. Sakurai S, Fukasawa T, Chong JM, Tanaka A, Fukayama M. C-kit gene abnormalities in gastrointestinal stromal tumors ( tumors of interstitial cells of Cajal). Jpn J Cancer Res 1999; 90: 1321– 8. 15. Miettinen M, Sarlomo-Rikala M, Sobin LH, Lasota J. Esophageal stromal tumors: a clinicopathologic, immunohistochemical, and molecular genetic study of 17 cases and comparison with esophagea l leiomyomas and leiomyosarcomas. Am J Surg Pathol 2000; 24: 211– 22. 16. Lux ML, Rubin BP, Biase TL, et al. KIT extracellular and kinase domain mutations in gastrointestinal stromal tumors. Am J Pathol 2000; 156: 791– 5.
319
17. Joensuu H, Roberts PJ, Sarlomo-Rikala M, et al. Effect of the tyrosine kinase inhibitor STI571 in a patient with a metastatic gastrointestinal stromal tumor. New Engl J Med 2001; 344: 1052–6. 18. Franquemont DW. Differentiation and risk assessment of gastrointestinal stromal tumours. Am J Clin Pathol 1995; 103: 41–7. 19. Ward RL, Hawkins NJ, O’Grady R, et al. REMS-PCR – a novel assay for the detection of K-ras mutations in clinical samples. Am J Pathol 1998; 153; 373– 9. 20. Miettinen M, Monihan JM, Sarlomo-Rikala M, et al. Gastrointestinal stromal tumours/smooth muscle tumours ( GISTs) primary in the omentum and mesentry. Am J Surg Pathol 1999; 23: 1109–18. 21. Marci V, Casorzo L, Sarotto I, Dogliani N, Milazzo MG, Risio M. Gastrointestinal stromal tumour, uncommitted type, with monosomies 14 and 22 as the only chromosomal abnormalities. Cancer Genet Cytogenet 1998; 102: 135–8. 22. O’Leary T, Ernst S, Przygodzki R, Emory T, Sobin L. Loss of heterozygosity at 1p36 predicts poor prognosis in gastrointestinal stromal/smooth muscle tumors. Lab Invest 1999; 79: 1461–7. 23. Druker BJ, Lydon NB. Lessons learned from the development of an abl tyrosine kinase inhibitor for chronic myelogenous leukemia. J Clin Invest 2000; 105: 3–7.