Synchronous, Yet Genomically Distinct, GIST Offer New Insights Into Precise Targeting of Tumor Driver Mutations

INTRODUCTION

GI stromal tumor (GIST) is the most common sarcoma and mesenchymal tumor of the GI tract.¹ These tumors are often characterized according to their location and molecular aberrations. The majority of GIST occur in the stomach (62%)² and harbor constitutively activated mutant isoforms of either KIT (75%-85%) or platelet-derived growth factor receptor alpha (PDGFRA; approximately 5%), both of which are type III receptor tyrosine kinases.³ Other genetic abnormalities associated with GIST include mutations of RAS pathway components (KRAS [2%-5%]^4,5, BRAF [2%-4%],⁶ or NF1 [1.5%-10% depending on the study]⁷) and loss of function in the succinate dehydrogenase protein complex (5%-10%).^8,9 Rarer molecular alterations include FGFR1 and ETV6-NTRK3 gene fusions.⁶ Improved understanding of the molecular aberrancies underlying GIST has served as the foundation for the paradigm shift toward precision oncology, highlighting the importance of molecularly matching driver mutations with cognate agents. For example, the development of tyrosine kinase inhibitors such as imatinib has revolutionized the treatment of KIT mutant GIST and drastically improved survival for patients suffering from this malignancy.^8,10

With the advent of next-generation sequencing (NGS), precision treatment of patients with GIST according to tumor genotype has become essential in optimizing clinical outcomes.^11,12 Herein, we report the first case of a patient with two synchronous GISTs in the stomach and duodenal-jejunal flexure (DJF) that were driven by KIT V560E (exon 11) and BRAF V600E mutations, respectively. This case will highlight the importance of NGS in understanding GIST biology, response to tyrosine kinase inhibitors therapy, and mechanisms of GIST development.

CASE REPORT

A 57-year-old male patient presented in Spring 2017 with 6 weeks of right lower quadrant abdominal pain and fevers. His medical history was significant for gastroesophageal reflux disease, for which he took a proton pump inhibitor (omeprazole 20 mg daily), and for a right inguinal hernia repair in 2011. His family history was significant for breast cancer in both his mother and paternal aunt, as well as lung cancer in his father. The patient endorsed smoking cigars almost daily since the age of 19 years (approximate equivalent of 5 pack years). He also reported drinking two glasses of wine daily.

Initial evaluation of the patient’s abdominal pain via computed tomography (CT) abdomen and pelvis revealed a lobulated, well-circumscribed, exophytic mass in the left upper quadrant arising from the stomach. This measured 14.5 × 13.0 cm. Endoscopic ultrasound-guided biopsy of the mass revealed a spindle cell lesion without necrosis or mitoses. Immunohistochemical profiling was consistent with GIST, demonstrating strong positivity for DOG-1 and KIT (CD117), weak positivity for AE1/AE3, and negativity for desmin, melan-A, and S100.

Upon referral to our institution, restaging CT chest, abdomen, and pelvis about 6 weeks after the initial imaging demonstrated growth of the left upper quadrant mass, now measuring 16.7 × 13.2 cm. It was arising from the gastric wall with possible invasion of the spleen and left hemidiaphragm but without evidence of metastasis. The biopsied specimen was sent for an NGS panel, which revealed a KIT V560E (exon 11) gain-of-function mutation. Subsequently, the patient was started on neoadjuvant imatinib 400 mg daily, which he tolerated well with only intermittent periorbital edema and minimal diarrhea.

The patient underwent bimonthly CT chest, abdomen, and pelvis restaging until maximal response to imatinib was achieved at 12 months, with decrease of the size of the gastric mass to 9.4 × 7.3 cm and no evidence of metastatic disease. Following plateau tumor response to imatinib after 14 months, the patient underwent a planned en bloc resection of gastric GIST with partial sleeve gastrectomy. Because of ongoing splenic hilar involvement, a splenectomy was performed. The gastric tumor had evidence of significant treatment effect (< 5% viability; Fig 1) with R0 margins and a mitotic index of 1 per 5 mm². During the operation, a second lesion was identified in the proximal jejunum. This was a suspected metastasis and resected. No other peritoneal or hepatic disease was noted. Pathologic assessment of this lesion (0.9 cm) was also consistent with GIST, spindle cell type, but with 100% viability (Fig 1). The mitotic index was 0 per 5 mm². At that time, it was unclear whether the jejunal mass was an imatinib-resistant metastasis or a synchronous primary lesion.

FIG 1.

Histology of gastric and jejunal GISTs. The gastric tumor had evidence of treatment effect with < 5% viability, and mitotic index of 1 per 5 mm². The jejunal lesion was also consistent with GIST, spindle cell type, with 100% viability, and mitotic index of 0 per 5 mm². GIST, GI stromal tumor.

Molecular profiling of both lesions (Tempus, IL) demonstrated a 269× average depth of coverage with distinct genetic aberrations between the two tumors. Germline NGS was used as a control. The gastric GIST was driven by a KIT V560E exon 11 gain-of-function mutation. Other somatic mutations included copy-number losses of SMARCB1, CHEK2, EP300, LZTR1, NF2, and ZNRF3, and copy-number gains of MCL1 and PDPK1 (Table 1). By contrast, the jejunal GIST was driven by a distinct underlying BRAF V600E missense gain-of-function mutation. This tumor also harbored a frameshift loss-of-function somatic mutation of KMT2A P773fs (Table 1). Both the gastric and jejunal lesions were negative for expression of the programmed death ligand-1 immunotherapeutic marker and demonstrated microsatellite stability, as well as normal expression of the DNA mismatch repair proteins MLH1, PMS2, MSH2, and MSH6. The distinct molecular profiling of the jejunal lesion, as well as the lack of response to the neoadjuvant imatinib regimen, indicates that this was a synchronous primary GIST. With these findings combined, using the American Joint Committee on Cancer staging system for GIST, the gastric lesion was staged as IIIA (T3N0Mx) and the jejunal lesion was staged as IA (T1NxMx) rather than stage IV disease with a secondary imatinib-resistance mutation.^13,14

TABLE 1.

Molecular Profiles of Gastric and Jejunal GISTs

Further genomic analysis of the two tumors was performed to identify any distinct variant aberrations or dissimilar molecular signatures. Variant effect predictor analysis was performed to map the effects of sequenced variants on genes and transcripts in both lesions using the Ensembl Variant Effect Predictor.¹⁵ This revealed 20 variants within the gastric GIST and 14 within the jejunal GIST, with 25% and 21% of variants, respectively, corresponding to missense mutations (Table 2, Appendix Table A1). The gastric GIST was confirmed to possess a missense mutation in KIT, whereas the jejunal GIST possessed missense mutations in BRAF, MAFA, and SMC1A genes (Fig 2 and Table 2). Of the variants, only three were shared between the two tumors. Both lesions carried the same missense mutations in ZNF30 and CYP3A43, suggesting that they have common clonal ancestry (Table 3).

TABLE 2.

Variant Effect Predictor Consequence Analysis for Gastric and Jejunal GISTs

FIG 2.

Variant effect predictor analysis of the somatic mutations found in the two GISTs. Variant consequence was attributed based on ENSEMBL variant effect predictor annotation. The gastric GIST carried 25% missense variants, whereas the jejunal GIST carried 21% missense variants. GIST, GI stromal tumor.

TABLE 3.

Shared Mutations Between Gastric and Jejunal GISTs

Previous studies have demonstrated that multiple mutational processes generate characteristic mutational signatures in human cancers.¹⁶ Two of these processes correlate with chronologic age and have been termed clock-like mutational signatures.¹⁷ Analysis of mutational signatures for both tumors was performed to determine the set of operative mutational signatures in these cancers. No signatures other than the two clock-like signatures were found in the samples (Fig 3). The two tumors had similar proportions of mutations attributed to the two different clock-like signatures; 34% of mutations in the gastric GIST and 36% of mutations in the jejunal lesion were attributed to an endogenous process with spontaneous or enzymatic deamination of cytosine to thiamine (ie, SBS1). The remaining mutations were attributed to SBS5, a clock-like mutational signature with unknown etiology.

FIG 3.

Mutational signatures of the lesions revealed that all mutations in both samples can be explained by clock-like signatures. (A) Attribution of single base substitutions to mutational signatures. The bars show the percentage of mutations attributed to each signature in the two GIST samples. (B) Mutational profiles of clock-like mutational signatures SBS1 and SBS5. Mutational profiles of substitutions are shown using six subtypes: C>A, C>G, C>T, T>A, T>C, and T>G. Underneath each subtype are 16 bars reflecting the sequence contexts determined by the four possible bases immediately 5′ and 3′ to each mutated base. GIST, GI stromal tumor.

Per National Comprehensive Cancer Network guidelines, the patient resumed adjuvant imatinib with a plan to continue this regimen, given the high risk of recurrence associated with his high-risk, KIT exon 11-mutated gastric GIST.¹⁸ However, there remains debate and ongoing studies as to the ideal length of adjuvant therapy.^19,20 But, there was no evidence-based indication for adjuvant therapy directed for the BRAF-mutated GIST, so this was not offered. Surveillance CT chest, abdomen, and pelvis 21 months post-operation demonstrated no evidence of local recurrence or metastasis. Any recurrence in the future will require repeat biopsy and molecular profiling to determine whether the lesion is driven by KIT or BRAF mutation, as this will have profound treatment implications.

DISCUSSION

The routine use of NGS panels of cancer-related genes has provided both clinicians and scientists with the opportunity to gain a deeper understanding of actionable cancer genomics and to begin investigating personalized treatments based on tumor molecular profiling.²¹ The development of precise therapeutic regimens targeting specific molecular aberrations harbored by GIST exemplifies the paradigm of genomics-driven clinical care of patients with cancer. Here, we have described a patient with two synchronous, yet genomically distinct, primary GISTs. Although one case of synchronous, multicentric GIST with distinct KIT mutations (nongermline) has been previously described in the literature,²² this is the first report of a patient with synchronous KIT-mutant and BRAF-mutant GIST.

Many previous studies have demonstrated the clonal nature of metastatic cancer. In one study analyzing sequencing data from 76 treatment-naive metastases from 20 patients with breast, colorectal, endometrial, gastric, lung, pancreatic, and prostate cancers and melanoma, the majority of driver gene mutations were present in all metastases of each individual patient.²³ In a different study of metastatic solid tumor genomes, including whole-genome sequencing data for 2,520 pairs of tumor and normal tissue and surveying more than 70 million somatic variants, the mutations of metastases reflected those of the primary tumors with 96% of driver mutations being clonal with similar mutational landscape and driver genes to primary tumors.²⁴ By contrast, genomic analysis on the two lesions presented here revealed that not only do these lesions possess differing driver mutations, but they also have multiple differences in genomic variants, as well as evidence of distinct mutational profiles over time. These findings taken together suggest that these tumors may have had a common ancestor or it is also possible that there was a somatic mosaic event. Therefore, this case highlights the profound utility of molecular profiling in deciphering the presence of independent primary lesions versus metastatic lesions in patients with multiple tumors.

The genomic findings from this case carry important prognostic and therapeutic implications. Because of genetic profiling of this patient’s tumors, the patient was reclassified from stage IV disease to stage IIIA gastric GIST and stage IA jejunal GIST, which have predicted metastatic rates of 12% and 0%, respectively, based on AFIP-Miettinen risk assessment.^25,26 This is important prognostic information as the 5-year GIST-specific mortality in patients with localized GIST (5.6%) is significantly more favorable compared to patients with regionally advanced (34.0%) or metastatic (34.3%) GIST² and may influence future decisions regarding patient care. In addition, although this patient's KIT-mutant gastric GIST demonstrated significant treatment response to neoadjuvant imatinib with < 5% viability at the time of resection, his BRAF-mutated jejunal lesion exhibited 100% viability, which is in line with existing literature that suggest BRAF-mutant GIST have primary imatinib resistance.²⁷ This finding supports evidence that kinase inhibitors targeting BRAF may be more effective therapeutic options in this molecular GIST subset.²⁸

This case report also demonstrates the correlation between anatomic distribution and underlying molecular aberration of various GISTs. We have previously shown that 83% of patients with GIST located at the DJF (also known as the ligament of Treitz) harbored NF1 mutations.⁷ Moreover, in the same study, we discovered that 29% of these DJF tumors were also associated with a BRAF V600E mutation, which is the same alteration heralded by our patient’s proximal jejunal lesion near the DJF. Although the exact mechanism by which certain mutations lead to genomically distinct GIST arising at specific sites remains unknown, this case illustrates that anatomic location represents an important consideration as it may have genetic and, ultimately, therapeutic implications for patients with GIST.

This report also reinforces the notion that GIST can be a secondary cancer to itself. We previously reported that 17% of patients with GIST have increased rates of additional malignancies including other sarcomas, neuroendocrine tumors, non-Hodgkin lymphoma, and colorectal adenocarcinoma both before and after diagnosis of GIST.²⁹ However, our patient’s synchronous, yet genomically distinct, GIST serves as a cautionary tale that multifocal GISTs are not necessarily metastases. This phenomenon highlights the power of clinical-grade molecular testing as these lesions otherwise may have been assumed to harbor the same underlying aberration, leading to potential treatment refractoriness.

Finally, this case underscores the need to explore risk factors that contribute to the development of GIST. Currently, the incidence of GIST in the United States is 1 per 147,000³⁰; therefore, the risk of developing two independent GISTs is 1 per 26.63 billion individuals. Extremely rare cases, such as our patient, may offer insight into potential risk factors that can be mitigated. Although our patient did not have any mutational signatures that correlated with a specific carcinogenic signature,¹⁷ our patient had a history of chronic tobacco and alcohol use, which are risk factors for a multitude of malignancies. Interestingly, both lesions shared the same missense mutations in ZNF30 (zinc finger 30) and CYP3A43, suggesting these lesions at one point shared a common ancestor or that there was a somatic mosaic event. Additional research is warranted to further elucidate possible environmental factors that may result in increased cancer risk in patients with GIST.

In conclusion, this case report is the first to our knowledge that describes two synchronous and genetically distinct KIT-mutant and BRAF-mutant GISTs. This study exemplifies how tumor molecular profiling both improves our scientific understanding of the tumor ancestry of synchronous GIST and affords clinicians with the innovative ability to precisely deliver personalized cancer care by enabling specific targeting of independent tumors within each patient.

Appendix

TABLE A1.

Somatic Mutations Identified in Gastric GIST and Jejunal GISTs

AUTHOR CONTRIBUTIONS

Conception and design: Eleftherios A. Makris, Sudeep Banerjee, Adam Burgoyne, Jason K. Sicklick

Provision of study materials or patients: Adam Burgoyne

Collection and assembly of data: Eleftherios A. Makris, Jorge de la Torre, Vi Nguye, Mojgan Hosseini, Adam Burgoyne, Jason K. Sicklick

Data analysis and interpretation: Eleftherios A. Makris, Ashwyn K. Sharma, Erik N. Bergstrom, Xiaojun Xu, Vi Nguyen, Mojgan Hosseini, Adam Burgoyne, Olivier Harismendy, Ludmil B. Alexandrov, Jason K. Sicklick

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO’s conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).

Adam Burgoyne

Consulting or Advisory Role: Exelixis, Eisai, Deciphera, Genentech/Roche

Speakers' Bureau: Deciphera

Olivier Harismendy

Stock and Other Ownership Interests: Sanofi, Novartis

Expert Testimony: Personal Genome Diagnostics

Jason K. Sicklick

Stock and Other Ownership Interests: Personalis

Consulting or Advisory Role: Deciphera

Speakers' Bureau: QED Therapeutics, Foundation Medicine, Roche

Research Funding: Foundation Medicine

No other potential conflicts of interest were reported.