Molecular Profile of Gastrointestinal Stromal Tumors in Sixty-Eight Patients from a Single Swiss Institution

Gastrointestinal stromal tumor (GIST) is the most common mesenchymal tumor of the gastrointestinal tract [1, 2]. The annual incidence of clinically relevant GIST is around 10–15 per million per year [3]. However, a much higher number of small GIST with a diameter <1 cm (“microGIST”) can be found after thorough screening of the gastrointestinal tract [4]. GIST are located most commonly in the stomach (60%) or small intestine (30%) but can occur in any other part of the gastrointestinal tract, including the colon and the esophagus [1, 3]. The median age at GIST diagnosis is 65 years, with a similar prevalence between men and women [3]. Most GIST develop sporadically. However, a small percentage of around 5% can be associated with genetic syndromes like neurofibromatosis type 1, Carney’s triad, or Carney-Stratakis syndrome [2].

A pathological diagnosis of GIST is made by use of histological and immunohistochemical analysis [5]. GIST can be divided into the following 3 different histological subtypes: spindle cell, epitheloid cell, and mixed [5]. Immunohistochemistry shows positivity for CD117 and/or DOG-1 in about 95% of cases, which can be used to separate GIST from tumors with a similar morphology like leiomyomas or neurofibromas [5]. The analysis of GIST includes measurement of mitosis per 5 mm², which is used for risk stratification in combination with tumor size and tumor site [6]. Mutational analysis of GIST is another key element in the diagnostic work-up of GIST [7]. First, because mutational analysis has predictive value for molecular-targeted therapy and, second, because some genotypes have a distinct natural history. Therefore, current guidelines tend to recommend that every GIST be tested for mutations [7]. KIT is the most commonly mutated gene (70–80%), followed by PDGFRA (5–10%) [2]. Both mutations are mutually exclusive and play a central role in GIST pathogenesis. They encode proteins belonging to the receptor tyrosine kinase class III family and, when activated, they promote proliferation while apoptosis is inhibited [2]. KIT mutations in GIST are clustered in different exons and appear most frequently in exons 11 and 9, whereas mutations in exons 8, 13, and 17 are rare (1–2%) [2, 8]. PDGFRA mutations usually occur in gastric GIST and most commonly affect exon 18 (rarely exons 12 and 14) [2, 8]. GIST with no mutation in KIT and PDGFRA have been described to carry mutations in other genes such as BRAF, NF1, PIK3CA, RAS, FGFR1, or the SDH complex [8, 9]. More recently, fusions such as ETV6-NTRK3 were also described in the absence of genetic aberrations in the above listed genes [8]. Due to more extensive molecular analysis, the percentage of so called “wild-type GIST,” with no genetic aberration, is diminishing. In fact, we are beginning to understand in more detail the spectrum of genetic changes beyond KIT and PDGFRA mutations. The aim of our retrospective study was therefore to assemble the molecular alterations in a cohort of GIST (n = 68), analyzed at our Institute of Pathology in Basel, Switzerland, within a 5-year period. We examined the mutational status of KIT and PDGFRA in all of the cases. Furthermore, we tested additional cancer genes in more than half of the cases by use of targeted next-generation sequencing (NGS).

We searched our database at the Institute of Pathology of the University Hospital Basel for GIST, which were analyzed by either Sanger sequencing or NGS between January 2014 and 2019. We retrieved 73 cases in total, 3 of which had to be excluded from our study due to a lack of patient consent; 2 referral cases were excluded because of incomplete analysis at our institute. 97% (n = 66) were formalin-fixed and paraffin-embedded tissue samples and 3% (n = 2) were cytological samples.

All cases were reviewed by 3 pathologists (S.H., M.S.M., and K.M.) and classified based on their morphology into the histological subtypes: spindle cell, epitheloid or mixed [5]. Every case was positive for CD117 and/or DOG1 on immunohistochemistry. Primary GIST were classified to their pT-stage according to the UICC TNM staging system (8th edition) and the risk for recurrence was determined after Miettinen [6].

For DNA isolation, tissue blocks were cut to sections of 4 µm thickness, which were placed on a glass slide. Afterwards, the area of interest was marked by a pathologist using an H&E stained tissue section as a guide. The selected area was macrodissected by scratching and DNA was isolated using the Maxwell DNA purification kit according to the manufacturer’s protocol. The DNA concentration was quantified using the Qubit® dsDNA HS Assay Kit (ThermoFisher Scientific).

For Sanger sequencing, polymerase chain reaction amplification and direct sequencing of KIT exons 9, 11, 13, and 17 and PDGFRA exons 12, 14, and 18 were performed using methods previously described [10]. Thereafter, the amplicons were analyzed by capillary electrophoresis and laser-induced fluorescence detection with an ABI PRISM 3130 Genetic Analyzer (ThermoFisher Scientific) according to the manufacturer’s protocol.

NGS was conducted using the Ion Torrent^TM Oncomine^TM Focus Assay or the Ion AmpliSeq Cancer Hotspot Panel v2 (both from ThermoFisher Scientific). The Oncomine Focus Assay uses 269 amplicons to cover hotspot mutations and copy number variance in 35 genes involved in cancer. Hotspot mutations of the -following genes are included: AKT1, ALK, AR, BRAF, CDK4, CTNNB1, DDR2, EGFR, ERBB2, ERBB3, ERBB4, ESR1, FGFR2, FGFR3, GNA11, GNAQ, HRAS, IDH1, IDH2, JAK1, JAK2, JAK3, KIT ( exons 8, 9, 11, 13, and 17), KRAS, MAP2K1, MAP2K2, MET, MTOR, NRAS, PDGFRA ( exons 12, 14, and 18), PIK3CA, RAF1, RET, ROS1, and SMO. The Ion AmpliSeq Cancer Hotspot Panel v2 uses 207 amplicons to cover hotspot mutations in 50 cancer genes including: ABL1, EGFR, GNAS, KRAS, PTPN11, AKT1, ERBB2, GNAQ, MET, RB1, ALK, ERBB4, HNF1A, MLH1, RET, APC, EZH2, HRAS, MPL, SMAD4, ATM, FBXW7, IDH1, NOTCH1, SMARCB1, BRAF, FGFR1, JAK2, NPM1, SMO, CDH1, FGFR2, JAK3, NRAS, SRC, CDKN2A, FGFR3, IDH2, PDGFRA ( exons 12, 14, 15, 18), STK11, CSF1R, FLT3, KDR, PIK3CA, TP53, CTNNB1, GNA11, KIT ( exons 2, 9, 10, 11, 13, 13, 14, 15, 17, and 18), PTEN, and VHL. Twenty-four cases were analyzed with the Oncomine Focus Assay and 15 cases were analyzed with the Ion AmpliSeq Cancer Hotspot Panel v2.

The Ion AmpliSeq Library Kit version 2.0 (ThermoFisher Scientific) was used to prepare the libraries from 10 ng of DNA. Ion Ampliseq HiFi Master Mix (ThermoFisher Scientific) was used to prepare the amplicons that were digested with FUPA reagent and tagged with IonCode Barcode Adapters. Finally, the amplified products were purified using the Agencourt AMPure XP PCR purification system (Beckman Coulter, USA). The Ion Library Equalizer Kit method was used to normalize the library concentration at ∼100 pM. Finally, equal volumes of normalized DNA library were combined and amplified on Ion Sphere particles (ISP; ThermoFisher Scientific) by emulsion PCR using the Ion PI HiQ OT2 200 Kit (both from ThermoFisher Scientific). Quality control was performed using the Ion Sphere Quality Control Kit (ThermoFisher Scientific) to ensure that 10–30% of template-positive ISP were generated in the emulsion PCR. The template-positive Ion PI ISP were loaded onto an Ion PI Chip and sequenced on an Ion S5^TM XLSequencer (ThermoFisher Scientific) with the Ion PI HiQ Sequencing 200 Kit (ThermoFisher Scientific) according to the manufacturer’s instructions.

For NGS data analysis, the Ion Reporter Software within Torrent Suite Software (ThermoFisher Scientific) was used. Detected sequence variants were evaluated for their pathogenicity based on previous literature, databases (COSMIC, ClinVar, The Jackson Laboratory Clinical Knowledgebase [CKB], dbSNP), reports and in silico functional prediction (MetaLR score) [11]. Mutations were visualized with the Complex heatmap package in R [12]. Mutations classified as benign were not reported. For statistical analysis, the χ² test and, in the case of low patient numbers, Fisher’s exact test were applied using R. All tests were hypothesis generated and we did not correct for multiple testing.

We included sequencing data from 68 GIST, which were analyzed consecutively at our institute over a 5-year period from the beginning of 2014 to the beginning of 2019. The median age of the GIST patients was 65 years (±14 years) and there were 40 males and 28 females (Table 1). The analysis was performed on 66 formalin-fixed and paraffin-embedded histological tissue samples and 2 cytological specimens stained with Papanicolaou. Our study comprised 60 primary GIST and 8 metastasized GIST. Most primary GIST were located in the stomach (n = 31; 52%) or in the small intestine (jejunum/ileum, n = 22; 35%). The remaining were found in the rectum (n = 4; 7 %), the duodenum (n = 2; 4%), and the esophagus (n = 1; 2%). The average primary tumor diameter was 74 mm (±47 mm). Metastatic sites were either the liver (n = 7) or the lung (n = 1). A summary of the clinical data, including pT-stage and risk of recurrence, can be found in Table 1.

Before 2016, mutational analysis was performed by Sanger sequencing (n = 29; 43%); thereafter, it was done by NGS (n = 39; 57%). Mutations were found in the KIT, PDGFRA, BRAF, or PIK3CA gene (Fig. 1; Tables 2, 3). Of the 60 primary GIST, 47 (78%) carried 1 or 2 KIT mutations, 9 (15%) had a PDGFRA mutation, 2 (3%) had a BRAF mutation, and 1 (2%) had a PIK3CA mutation. One case (2%) did not show any mutation. All metastatic GIST (8 cases) harbored exclusively KIT mutations.

Of the KIT mutated primary GIST (n = 47), 45 cases (96%) had a single mutation, whereas 2 cases (4%) showed a double KIT mutation. Single mutations included missense mutations (37%), deletions (37%), indels (22%), and duplications (4%). KIT mutations were located in exon 11 in 43 (88%) cases, in exon 9 in 4 (8%) cases, and in exon 17 in 2 (4%) cases. The first case (No. 16), with 2 concomitant pathogenic KIT mutations, carried both mutations right next to each other in exon 11 (p.V559A and p.V560D). The second case (No. 42) was an imatinib-treated GIST and carried a mutation in exon 11 (p.W557_K558del) in combination with one in exon 17 (p.D820V), which typically evolves under imatinib therapy [13]. KIT mutations in the metastatic GIST (n = 8) were located mainly in exon 11 (n = 7; 88%) and in 1 case in exon 9 (n = 1; 12%). They included deletions (37%), indels (37%), missense mutations (13%), and duplications (13%).

Of the 9 PDGFRA-mutated primary GIST, 8 (89%) carried the well-known p.D842V point mutation in exon 18 and 1 (11%) harbored a p.W559_V561delinsF in exon 12.

Four cases did not harbor a KIT or PDGFRA mutation. Two of them (No. 32 and 46) showed an activating BRAF mutation. One carried the well-known BRAF mutation p.V600E; the other one harbored a deletion in exon 12 (p.N486_P490del), which has not been described in GIST yet. Additionally, we found 1 case (No. 37) harboring an activating PIK3CA mutation (p.M1043I) in exon 9, which has not been reported in GIST either. Finally, 1 case did not show any mutation after analysis by NGS. Moreover, the expression of SDHB was retained by immunohistochemistry. Therefore, this case was considered wild-type.

Next, we classified all GIST into the 3 different known histological subtypes [5]. 47 (69%) showed spindle cell morphology, 10 (14%) showed epithelioid morphology, and 11 (17%) showed a mixed epithelioid-spindle cell morphology. Examples of the different morphologies are shown in online supplementary Figure 1 (see www.karger.com/doi/10.1159/000505407 for all online suppl. material). Immunohistochemical analysis (online suppl. Fig. 1) revealed that 64 (94%) out of 68 cases showed positivity for CD117 and 47 (98%) out of 48 were positive for DOG-1, whereas, the 4 cases negative for CD117 were positive for DOG-1. The correlation of morphology with molecular results revealed that GIST harboring a KIT mutation were associated (p = 0.03, χ² test, online suppl. Fig. 2) with a spindle cell morphology (n = 42; 76%) and less frequently with a mixed epitheloid-spindle cell morphology (n = 6; 11%) or an epitheloid cell morphology (n = 7; 13%). In contrast, PDGFRA mutated GIST were associated (p = 0.0003, Fisher’s exact test, online suppl. Fig. 3) with an epitheloid cell morphology (n = 4; 44%) or a mixed epitheloid-spindle cell morphology (n = 4; 44%), whereas 1 case (12%) showed a spindle cell morphology. Both cases with a BRAF mutation and the wild-type case showed a spindle cell morphology, whereas the single case with a PIK3CA mutation had an epithelioid morphology.

We reviewed the mutational profile of a cohort of 68 patients diagnosed with GIST. Although other studies from Europe have been published, this is the first from northern Switzerland. We found a mutational rate for KIT of 78% and one for PDGFRA of 15% in primary GIST. Population-based studies from other European regions such as Norway, Greece, and Poland showed KIT mutations in 75, 69, and 63% of cases and PDGFRA in 10, 13, and 18% of cases, respectively [14-16]. Therefore, our cohort revealed slightly increased frequencies of KIT mutations and similar frequencies of PDGFRA mutations. The slightly higher frequency of KIT mutation may be explained by the fact that our cohort consists mainly of patients at an advanced disease stage who were amenable to tyrosine kinase receptor inhibitor therapy. It is well-known that KIT-mutated GIST generally have worse clinical outcomes compared to PDGFRA-mutated GIST [17]. Therefore, KIT-mutated GIST may be overrepresented in our study. In fact, a study with advanced GIST showed a mutation rate of 83% for KIT [13]. In line with this fact, all metastasized GIST in our study exclusively showed KIT mutations. Our analysis further revealed that the majority of KIT mutations were located in exon 11 (88%), followed by mutations in exon 9 (8%), which is in accordance with previous studies [8, 15]. Almost all of our cases showed only 1 mutation, except for 2 cases harboring concomitant KIT mutations. Concomitant KIT mutations are rare and often encountered after TKI therapy [8]. Indeed, 1 of our patients with concomitant KIT mutations was treated with imatinib and showed a mutation in exon 17 (p.D820V), which has been described to evolve under therapy [13]. Our second case was a therapy-naive GIST with a double mutation (with each located right to each other) in exon 11 (p.V559A and p.V560D). As described before, most GIST reveal double primary mutations in the same exon of the same gene, indicating that they originate from a single mutagenic event [18-20].

The PDGFRA mutations discovered in our cohort were similar to those found in earlier studies and consisted almost exclusively of the imatinib resistance-associated mutation p.D842V in exon 18. Only 1 case with an additional PDGFRA mutation was found, i.e., a p.W559_V561delinsF in exon 12. In accordance with earlier studies, we found very few cases with a BRAF mutation. All of them were devoid of KIT or PDGFRA mutations. BRAF kinase is a downstream protein of the tyrosine kinase-signaling cascade of both KIT and PDGFRA. In our study, we identified 1 case (No. 46) with a BRAF V600E mutation in exon 15. In addition, we found 1 case (No. 32) with a p.N486_P490del in exon 12. This mutation has not been described in GIST yet, but it has been reported in ovarian cancer and in histiocytic disease [21, 22]. We also found a case with a single PIK3CA mutation (p.M1043I) in exon 9. Similarly, this mutation has not been described in GIST yet, but it has been found in other tumors, including endometrial carcinoma, thyroid carcinoma, and glioblastoma [23-26]. Single PIK3CA mutations in untreated GIST are rare [27]. In contrast, concurrent PIK3CA mutations with exon 11 KIT mutations have been demonstrated more frequently [3, 28]. Finally, 1 case remained without any mutation. This case was a primary GIST from a 68-year-old male patient located in the small intestine (ileum). DOG1 and CD117 were positive and the tumor harbored a spindle cell morphology. Analysis by NGS using the Ion AmpliSeq Cancer Hotspot Panel, which covers hotspot mutations in KIT, PDGFRA, PIK3CA, BRAF, RAS, and FGFR1 did not show any aberration. Moreover, immunohistochemistry for SDHB did not show any loss, indicating that it was a sporadic tumor. Therefore, we considered this cases as wild-type, although we cannot exclude that other genes such as NF1 are mutated or a translocation for ETV6-NTRK3 is present, as these aberrations are not covered by the NGS panel used.

In conclusion, most mutations in our cohort occurred in the genes KIT and PDGFRA with a similar frequency compared to other studies. However, we also report previously undescribed mutations in the downstream molecules of KIT, i.e., BRAF and PIK3CA, expanding the molecular landscape of GIST. Our study therefore encourages mutational analysis beyond KIT and PDGFRA in case these genes are devoid of mutations.

We would like to thank Sibylle Tschumi, Tanja Dietsche, Valeria Perrina, and Linda Schweizer for their excellent technical assistance. We also would like to thank Sarah Niederberger for her outstanding support.

This study was approved by the Ethics Commission of Northern Switzerland (EKNZ; study ID: 2019-00776).

M.S.M. has received speaker’s honoraria from ThermoFisher. Otherwise, the authors have no conflicts of interest to declare.

Swiss Cancer Research Foundation Grant KFS-4168-02-2017 to M.S.M. The sponsor of the study did not have any role in the study design, or collection, analysis, and interpretation of date.

S.H. and M.S.M. conceived and designed this project. M.B. analyzed the Sanger and NGS data and reported the mutations. S.H., K.M., P.K., E.S., G.C., S.H., L.T., S.D., L.M.T., and M.S.M. performed the histological analysis and interpreted the sequencing data. D.J., M.S.M., and N.M.-S. performed the data analysis. S.H. and M.S.M. wrote this paper. All of the authors revised and agreed to the content of this paper.

M.B. and M.S.M. contributed equally to this work.