Main

Gastrointestinal stromal tumors (GISTs) are potentially malignant mesenchymal tumors of the gastrointestinal tract. KIT or — less often — PDGFRA (platelet-derived growth factor receptor alpha) mutations leading to a ligand-independent autoactivation of one of two tyrosine kinases are the oncogenetic drivers of these tumors.1, 2

In GIST, primary KIT mutations do not occur arbitrarily but only in the hot spot regions, which are exons 9, 11, 13, and 17. These encode an extracellular domain, the juxtamembrane domain, and the kinase domains 1 and 2, respectively. Corresponding affected exons of the closely related PDGFRA gene are 12, 14, and 18.3, 4, 5 Activating mutations in the remaining exons of both the genes have not been described so far in sporadic tumors. It is now widely accepted that GISTs are characterized by a close genotype–phenotype correlation in terms of location within the gastrointestinal tract, morphology, clinical behavior, and response to treatment with imatinib, which is a potent tyrosine kinase inhibitor. Thus, the mutational status has a high clinical impact as: (1) it can confirm the diagnosis of GIST in selected cases that are hard to diagnose; (2) it has a prognostic value; and (3) it is predictive for the response to molecular-targeted therapies, eg, with imatinib. Concerning the prognostic value, specific mutations in KIT exon 11 were shown to be associated with a higher risk of metastatic disease,6 whereas most PDGFRA-mutated GISTs show a less aggressive behavior.7, 8 In intestinal GISTs, KIT exon 9 mutations go along with aggressive behavior.9 In the context of targeted treatment, it has been recognized that the most common PDGFRA mutation (p.D842V), which nearly exclusively occurs in GISTs of the stomach, leads to resistance against imatinib.10 The detection of KIT exon 9 mutations in metastatic GISTs warrants doubling of the daily dose of imatinib to 800 mg/day in order to increase the response rate to treatment.11, 12 Thus, mutational testing of GIST is recommended in the diagnostic work-up of all GIST specimens, eg, by NCCN guidelines13 and the ESMO Clinical Practice Guidelines,14 at least in GISTs with intermediate or high risk of recurrence.

However, about 10–15% of GISTs carry wild-type sequences in all known hot spots of KIT and PDGFRA (wt-GISTs).15 In most cases, these wt-GISTs are known to have no or a poor response to tyrosine kinase inhibition. It now becomes apparent that wt-GISTs constitute an obviously heterogeneous group of different cancer subtypes. A subset of these tumors seem to be driven by other amplifications or oncogenic mutations of genes such as in IGF1R, BRAF, or PIK3CA.16, 17 Another subset of wt-GISTs arises in the context of hereditary syndromes. In the autosomal-dominant disorder, neurofibromatosis type 1 multiple wt-GISTs can occur,18, 19 arising predominantly in the small bowel. Patients with Carney–Stratakis syndrome suffer from gastric wt-GIST and paragangliomas.20 This syndrome is caused by germline mutations in the succinate dehydrogenase (SDH) subunits B, C, or D, leading to dysfunction of complex II of the electron transport chain.20, 21 However, it was shown that a subset of patients outside the Carney–Stratakis syndrome reveals mutations in SDH subtype B or C, pointing toward a pathogenetic role of SDH dysregulation in wt-GISTs.22 Paediatric GISTs, typically present in the stomach of young girls, display an epithelioid phenotype and can be multifocal. KIT and PDGFRA mutations are mostly absent23 and a subgroup of them is again attributable to succinate hydrogenase deficiency. Thus, the group of wt-GISTs is heterogeneous and further molecular characterization is needed to define subgroups, which might lead to a more specific therapy for these patients.

As stated earlier, wt-GISTs are currently defined by having wild-type sequences in exons 9, 11, 13, and 17 of the KIT gene and exons 12, 14, and 18 of the PDGFRA gene. Until now, the analysis of additional exons is not recommended in the routine molecular diagnostics of GISTs. However, we have previously published a report on a KIT exon 8 germline mutation,24 which was associated with systemic mastocytosis and familial GISTs.

The aim of this study was to investigate whether KIT exon 8 mutations also occur in sporadic GISTs. We therefore screened a cohort of 145 wt-GISTs for somatic mutations in KIT exon 8.

Materials and methods

Cases and Immunohistochemistry

A total of 145 wt-GISTs were extracted from our consultation files. All cases were wild type for KIT exons 9, 11, 13, and 17 and PDGFRA exons 12, 14, and 18. Baseline characteristics of the cases are displayed in Tables 1 and 2. The diagnosis of GISTs was made in all the presented cases by a senior pathologist specialized in soft tissue pathology (H-US, EW or RB). Immunohistochemistry for CD34 (clone QBEnd10, DAKO, 1:100), KIT (A4502, DAKO, 1:100), PDGFRA (SC-338, Santa Cruz, 1:50), DOG1 (Spring Bioscience, Fremont, CA, USA; 1:100) and Ki67 (Mib1, DAKO, 1:1000) was carried out as described earlier.25

Table 1 Clinicopathological features of the wt-GIST cohort (n=145)
Full size table
Table 2 Summary of two gastrointestinal stromal tumors with sporadic KIT exon 8 mutations
Full size table

Mutational Analysis

For genomic DNA-extraction, the lesional areas were marked on a haematoxylin and eosin-stained slide by a senior pathologist (H-US, EW, or RB). Subsequently, the tissue was scraped from unstained slides and DNA was extracted using standard methods as described previously.26 KIT exons 9, 11, 13, and 17 as well as PDGFRA exons 12, 14, and 18 DNA were amplified by PCR with specific primers and sequenced as described previously.26, 27 KIT exon 8 DNA was amplified by PCR with the following specific primers: Forward: 5′-GAAGTGAATGTTGCTGAGG-3′; and Reverse: 5′-GTGAATTGCAGTCCTTCC-3′. PCR and cycle sequencing were repeated twice.

Fluorescence In-Situ Hybridization (FISH) Analysis

FISH was performed on interphase nuclei of 4-μm sections of FFPE tissue. Locus-specific probe was derived from the BAC clone RP11-74L18 (imaGenes GmbH, Berlin, Germany) containing the full-length KIT gene using a Nick Translation Kit (Abbott Molecular, Germany) and spectrum orange-labelled dUTP (Abbott Molecular, Germany). KIT probe was used together with the chromosome 4 centromeric enumeration probe (CEP4, Abbott Molecular, Germany).

Deparaffinized slides were incubated in 0.2 M HCl for 20 min, followed by washing and 80 °C heat pretreatment in pretreatment solution (Abbott Molecular, Germany) for 30 min and further washing. Tissue was digested for 1.5 h with protease, slides were washed again, and fixed in 4% buffered formalin. All pretreatment steps were carried out in a prehybdridization processor (VP2000, Abbott Molecular). After washing, dehydration, and air drying, an appropriate amount of fluorochrome-labeled probe mixture was applied to the centre of the tissue section. Tissue DNA and probe mixture were co-denatured at 85 °C for 10 min, cooled down to 37 °C, and hybridized overnight in a Hybrite Hybridizer (Abbott Molecular, Germany). After stringent washing and dehydration, sections were counterstained using fluorescence mounting media containing de-amidino-phenyl-indole (DAPI, Zytovision GmbH, Germany). FISH slides were evaluated with a fluorescence microscope (DM 5500, Leica, Germany) using appropriate filter sets. Orange (KIT) and green (CEP4) signals were counted in 60 non-overlapping tumor cell nuclei from the three tumor areas.

Results

Clinicopathological Features

A total of 145 patients with proven wild-type sequences in all known mutational hot spots of KIT and PDGFRA were included in our study. Detailed clinicopathological data are provided in Table 1.

A comparison of these data with established clinicopathological features of GISTs (including wt-GISTs) demonstrates that our wt-GIST cases display a more or less representative cohort. For instance, small bowel GISTs (n=80) outnumbered GISTs of the stomach (n=53) and of the oesophagus and colon (n=1 each). Most cases displayed a spindled or mixed cell morphology (n=125/139) and only 14 cases revealed a pure epithelioid phenotype. In all, 12 patients were known to fulfill clinical criteria of neurofibromatosis type 1 and one patient presented with a Carney dyade. The tumor size was known in 121 cases and ranged from 0.1 to 35.0 cm in diameter (mean: 5.4 cm, median: 4.5 cm). Mitotic count was assessable in 136 cases and ranged from 0 to 90 (mean: 4.8; median: 1), with most cases having a low mitotic count. Therefore, according to the NCCN-AFIP criteria,28 only 43 cases were classified as being at ‘intermediate’ or ‘high’ risk for progressive disesase among 120 classifiable cases. Four other cases presented as liver metastasis. The overall patients’ mean age was 57.4 years (median: 60 years, range: 8–98 years; n=143), including five pediatric patients. Clinicopathological data were not available for all the patients, because most samples had been evaluated in the setting of a referral center with limited access to external clinical parameters.

Mutational Analysis

Two primary GISTs (cases 1 and 2) with an identical KIT exon 8 mutation (c.1255_1257delGAC) leading to the deletion of an aspartatic acid on residue 419 (p.D419del) were detected, representing 1.4% of all the cases investigated. The mutation was also found in a peritoneal metastasis of case 1, which was sequenced independently. No activating mutations in KIT exon 8 could be identified in the remaining 143 wild-type cases. However, silent mutations were found in KIT exon 17 (p.I798I, n=7, 4.8%) and PDGFRA exon 18 (p.V824V, n=24, 16.6%). Based on evaluation of the mutational status of 1351 well-characterized GIST samples from the Bonn–Cologne GIST registry, the overall frequency of KIT exon 8 mutations was 0.15%. In our referral center cohort, KIT exon 9 mutations were detected in 126 cases (9.33%), exon 11 mutations in 810 cases (59.96%), exon 13 mutations in 25 cases (1.85%), and exon 17 mutations in 23 cases (1.70%). PDGFRA mutations in exon 12 were found in 25 cases (1.85%), exon 14 mutations in 8 cases (0.59%), and exon 18 mutations in 189 cases (13.99%) (Figure 1).

Figure 1
figure 1

Mutational status of 1351 GIST cases in the Bonn–Cologne GIST registry (n=1351), KIT mutations were found in 72.99% and PDGFRA mutations in 16.43% of cases. The remaining 10.58% were classified as wt-GISTs. The overall frequency of KIT exon 8 mutations is 0.15%.

Full size image

Clinicopathological Characteristics of KIT Exon 8 Mutant GISTs

One GIST with a sporadic KIT exon 8 mutation (p.D419del) was found in a 53-year-old male patient. The tumor originated in the small bowel and measured 5.4 cm in diameter. The lesion showed a biphasic spindled-epithelioid histological appearance. Large tumor nodules consisted of epithelioid tumor cells with eosinophilic to clear cytoplasm, which were aggregated in an alveolar fashion. Multinucleated giant cells were not observed. High proliferative activity with numerous mitoses was noted and tumor necrosis was seen. The mitotic count was 13/50 HPF in this area. Furthermore, a second spindle cell pattern was noted. In this area, the mitotic count was 4/50 HPF (Figure 2, Table 2). According to the NCCN-AFIP criteria,28 the patient was classified as being at high risk of progressive disease. The patient developped multiple peritoneal metastases 29 months after surgical removal of the primary tumor. Mutational analysis of the metastasic lesions revealed the same deletion in KIT exon 8. Notably, a heterozygous mutation was detected in the primary tumor, whereas a homozygous pattern was detected in the metastatic disease. After 34 months, the patient was lost to follow up with progressive metastatic disease. As the patient was diagnosed in the pre-imatinib era, no adjuvant treatment was administered. Instead, the patient was treated with adriamycin and ifosfamide for his metastatic disease.

Figure 2
figure 2

Histopathological and molecular features of a GIST with KIT exon 8 mutation. In case 1, epithelioid tumor cells with eosinophilic to clear cytoplasm were aggregated in an alveolar fashion ((a), × 200, H&E). A second tumor cell component showed a spindeled cell appearance and was situated in close proximity to the epithelioid areas ((b,c), × 200, H&E). Abundant necrosis was noted ((d), × 200, H&E). Metastatic tissue revealed epithelioid tumor cells ((e,f), × 200, H&E). Strong expression of the KIT receptor ((g), × 200, KIT). The p.D419del mutation occurred heterozygously in the primary tumor ((h), top) and homozygously in the metastasis ((h), bottom). FISH analysis showed polysomy by presenting multiple orange KIT signals as well as multiple green CEP4 signals in both the primary tumor and metastasis (i).

Full size image

In case 2, the same heterozygous mutational subtype (p.D419del) was found in a GIST arising in the proximal jejunum. The 57-year-old female patient followed, however, a benign clinical course under adjuvant treatment (400 mg imatinib/day). The tumor measured 10 cm in size and a spindle cell phenotype of the entire tumor was observed (Figure 3, Table 2). One mitosis was found in 50 HPF. According to the NCCN-AFIP criteria,28 the patient was classified as being at moderate risk of progressive disease. Twenty-four months after removal of the GIST, there is no evidence of recurrence, while the patient is currently under adjuvant treatment with imatinib (400 mg/day).

Figure 3
figure 3

Histopathological and molecular features of a GIST with KIT exon 8 mutation. In case 2, a spindled cell morphology was noted ((a), × 200, H&E). Immunohistochemistry revealed strong positivity for DOG1 (b, × 200) and KIT (c). Sanger sequencing showed a heterozygous deletion in KIT exon 8 (p.D419del) (d).

Full size image

FISH Analysis

In case 1, the primary tumor showed the deletion in exon 8 of the KIT gene in a heterozygous pattern, whereas a homozygous status was detected in the relapse. In order to further elucidate the development of this homozygous pattern, we performed FISH analysis of the KIT gene locus in primary tumor and in the relapse. Spindle cell and epithelioid areas of the primary tumor did not differ in terms of KIT and centromer 4 copy numbers. Both the primary lesion and metastasis tumor cells showed polysomy by presenting multiple orange KIT signals as well as multiple green CEP4 signals (Figure 2). Thus, FISH analysis revealed neither a monosomy of chromosome 4 nor a mono-allelic deletion of the KIT gene locus as the cause for the homozygous mutational pattern.

Discussion

We have previously reported familial GISTs caused by germline mutation in KIT exon 8.24 This led us to the hypothesis that KIT exon 8 mutations might also represent driver mutations in a subset of sporadic GISTs. We therefore screened a well-defined cohort of 145 wt-GISTs for mutations in KIT exon 8.

On the molecular level, GISTs are characterized by mutations in the KIT or PDGFRA genes, both encoding for closely related tyrosine kinase proteins, which consist of regulatory and katalytic domains. The regulatory domain consists of an extracellular ligand-binding domain (EC) and a juxtamembrane domain (JM). Main parts of the katalytic domain are the two tyrosine kinase domains (TK1 and TK2), which are separated by a kinase insert.3 Each exon of KIT and PDGFRA corresponds to a specific molecular structure of the receptor. In sporadic GISTs, primary KIT mutations were hitherto found in EC and JM domains (exons 9 and 11, respectively) as well as in both kinase domains (exons 13 and 17); corresponding affected exons of PDGFRA are 12 (JM), 14 (TK1) and 18 (TK2).3 Beside GISTs, PDGFRA translocations or mutations are associated with various human neoplasms such as myeloproliferative diseases associated with hypereosinophilia29 or inflammatory fibroid polyps.25, 30 KIT mutations have also been reported in other malignancies, such as melanomas or seminomas.31 Some of these tumors share the same mutational hot spots and affected exons with GISTs.

KIT exon 8 mutations are located in the EC domain of the receptor and are frequently found in hematopoietic disorders, such as acute myeloid leukemia and/or mastocytosis.32, 33 Several mutational subtypes have been described that mostly affect codons 417–421 (Table 3). Up to now, KIT exon 8 mutations have not been systematically investigated in sporadic GISTs, and we report, for the first time, two cases with somatic mutations. In our cohort of wt-GISTs, which obviously appears to be representative, the frequency was 1.4%. However, they represent very rare mutational events, contributing to <0.2% of all GISTs.

Table 3 Reported KIT exon 8 mutations
Full size table

In both our cases, we observed the same deletion (p.D419del), which was also shown to be associated with pediatric-onset cutaneous mastocytosis (urticaria pigmentosa) and acute myeloid leukemia (Table 3). The first case differed substantially from the second case in terms of morphology and clinical behavior: it had a peculiar biphasic and partially alveolar epithelioid growth pattern and followed an aggressive clinical course. A heterozygous mutation was found in the primary tumor, shifting to a homozygous mutation in the metastatic disease, which might reflect tumor progression. FISH analysis showed polysomy with more than three copies both of centromer 4 and the KIT locus in all the analyzed tumor cells. This finding rules out mono-allelic deletion of the wild-type KIT locus and is consistent with a duplication of the KIT mutant allele. This mechanism was previously described for KIT exon 11-mutated tumors by Lasota et al.34 They showed that a loss of heterozygosity was attributed to the loss of the wild-type allele with subsequent gain of one or multiple copies of the mutated one. As in our study, the phenomenon was associated with a malignant course of disease.

Our second case was an intermediate risk tumor of spindled subtype, which did not progress under adjuvant tyrosine kinase inhibition with imatinib. The 67-year-old female patient suffered from a small bowel tumor but is well with no evidence of disease 24 months after her surgery.

Exactly the same type of mutation was once reported as p.D419del germline mutation in familial GISTs. We could identify a kindred with both familiar GISTs and mastocytosis.24 These familial cases were typical low-risk GISTs with a spindled morphological subtype. By in vitro experiments, we could previously show that this particular mutation results in constitutively activated KIT, which can be inhibited by imatinib.24 In our present cases, no clue for neither a familiar background nor an associated mastocytosis was found, based on information from the patients’ files and oral consultation with their physicians.

We conclude that sporadic KIT exon 8 mutations in GISTs represent rare events. A small proportion (about 1–2%) of GISTs being formerly reported as ‘wild type’ might, in fact, carry a mutation in KIT exon 8. Only one mutational subtype (p.D419del) was identified so far that was shown to be activating and imatinib sensitive in vitro. We therefore propose that the screening for KIT exon 8 mutations should be implemented in the routine molecular diagnostics of GIST. In addition, patients with proven KIT exon 8 mutations and significant risk for tumor progression should be treated with imatinib.