Gastrointestinal stromal tumors (GISTs) are rare neoplasms. Although they represent only 0.1% to 3% of all gastrointestinal malignancies,^1–4 they account for 80% of gastrointestinal mesenchymal neoplasms.⁵ Approximately 5000 to 6000 new cases are diagnosed per year in the United States, for an annual incidence of 14.5 per million and prevalence of 129 per million.⁶ In the last 15 years, the understanding and treatment of GIST has witnessed remarkable advances due to two key developments: (1) the identification of constitutively active signals (oncogenic mutation of the c-KIT and platelet-derived growth factor alpha [PDGFRA] gene-encoding receptor tyrosine kinases) and (2) the development of therapeutic agents that suppress tumor growth by specifically targeting and inhibiting these signals. These developments in the management of GIST illustrate the principle of translational therapeutics in oncology, confirming that specific inhibition of tumor-associated receptor tyrosine kinase activity is an effective cancer treatment. The advent of effective targeted medical therapy for GIST has increased the complexity of management and opened new dialogues regarding the need for integrated multimodality therapy. This chapter reviews the biology, treatment, and emerging clinical challenges of these mesenchymal neoplasms.

Historical Background

The term “GIST” was initially coined in 1983 by Mazur and Clark to describe intra-abdominal nonepithelial neoplasms which lacked the ultrastructural features of smooth muscle cells and the immunohistochemical characteristics of Schwann cells.⁷ GISTs typically exhibit heterogeneous histologic features. They are most commonly composed of long fascicles of spindle cells with pale to eosinophilic cytoplasm and rare nuclear pleomorphism, but may occasionally exhibit epithelioid characteristics, including sheets of round- to oval-shaped cells with abundant eosinophilic cytoplasm and nuclear atypia (Fig. 33-1). As such, they are typically classified as spindle cell type, epithelioid type, or mixed type. The majority of GISTs are of spindle cell appearance (70%), while epithelioid (20%) and mixed (10%) cell morphology are less common.

In 1995, Miettinen and colleagues discovered that 70% of GISTs were positive for CD34 by immunohistochemistry, a myeloid progenitor cell antigen also present in endothelial cells and fibroblasts.⁸ Based upon their histologic features, GISTs are believed to arise from the interstitial cells of Cajal, components of the intestinal autonomic nervous system that serve as intestinal pacemakers and also express CD34.⁹ Nonetheless, until the late 1990s, there were no objective criteria to classify GISTs. They were frequently misclassified as leiomyomas, leiomyoblastomas, leiomyosarcomas, Schwannomas, gastrointestinal autonomic nerve tumors, or other similar soft tissue histologies.¹⁰ Consequently, interpretation of clinical results for reports on “GISTs” published before 2000 is challenging.

Receptor Tyrosine Kinase Mutations

In a landmark publication in 1998, Hirota and colleagues reported two critical findings: (1) near-universal expression of the transmembrane receptor tyrosine kinase KIT in GISTs and (2) presence of gain-of-function mutations in the corresponding c-KIT proto-oncogene.¹¹ The KIT receptor is activated by binding its cytokine ligand, known as steel factor or stem cell factor,¹² which then causes receptor homodimerization, phosphorylation, and cellular proliferation. KIT plays a critical role in the development and maintenance of components of hematopoiesis, gametogenesis, and intestinal pacemaker cells.^13–15 Oncogenic KIT mutations have been identified as molecular drivers of neoplasms corresponding to these functions, including mast cell tumors, myelofibrosis, chronic myelogenous leukemia, germ cell tumors, and GIST.¹³ Mutated KIT remains constitutively active even in the absence of ligand binding and results in both unregulated cell growth and malignant transformation.¹¹

GISTs are identified by immunohistochemical staining for the CD117 antigen, part of the KIT receptor (Fig. 33-2). CD117 expression is characteristic of most GISTs, but not of other gastrointestinal smooth muscle tumors such as leiomyosarcoma, which are more likely to express high levels of desmin and smooth muscle actin.^13–16 Application of CD117 staining as a diagnostic criterion for GIST has heightened understanding of disease prevalence but is an imperfect isolated surrogate for GIST diagnosis. Some GISTs may stain strongly for KIT (CD117) by immunohistochemistry (KIT-positive) yet lack KIT mutations,¹³ while others that do not stain for KIT (KIT-negative) may nevertheless harbor KIT mutations.¹⁷

Over 85% of GISTs have activating KIT mutations (Fig. 33-3).¹³ These mutations commonly occur in exon 11 (in 57% to 71% of cases), exon 9 (10% to 18%), exon 13 (1% to 4%), and exon 17 (1% to 4%).^18–21 GISTs with KIT exon 9 mutations predominantly arise in the small intestine, and homozygous mutant GISTs are often associated with recurrent disease. Mutations in exon 11 may include deletions, insertions, single-base substitutions, and various combinations of these and are associated with variable rates of disease recurrence following complete resection.^22–25 Deletion mutations in exon 11 are an independent adverse prognostic factor, with worse prognosis than those with point mutations.^26–28 Deletions specifically involving codon 557 and 558 are considered mutational “hotspots” and are associated with more aggressive and often metastatic behavior.^29–30

Approximately 35% of neoplasms lacking KIT mutations have activating mutations in a gene encoding a related receptor tyrosine kinase, the platelet-derived growth factor receptor alpha (PDGFRA).^31–33 PDGFRA mutations have been identified in exon 12 (1% to 2% of GISTs), exon 18 (2% to 6%), and exon 14 (<1%).^31,34 KIT and PDGFRA mutations are mutually exclusive events in GIST pathogenesis. However, no differences in the activation of downstream signaling intermediates have been observed between PDGFRA-mutant and KIT-mutant tumors, suggesting that both pathways result in parallel oncogenic molecular signals. Substitution of valine (V) for aspartic acid (D) at codon 842 in exon 18 accounts for 70% of all PDGFRA mutations and is associated with imatinib resistance.³⁵ GISTs harboring a PDGFRA exon 18 D842V mutation have been shown to have a gastric location predilection, have a lower risk of recurrence than GISTs with KIT mutations, and tend to have a more indolent course.³⁶

Finally, wild-type (WT) GISTs exhibit no detectable KIT or PDGFRA mutations, and have alternate pathways for pathogenesis. Additional putative mutations have been identified to molecularly characterize this group of emerging GIST tumors. These include mutations in BRAF V600E exon 15 and insulin-like growth factor-1 receptor overexpression.^37,38 Mutations in succinate dehydrogenase (SDH) subunits and type 1 neurofibromatosis (NF1) genes have also been linked with GIST oncogenesis (see discussion below).³⁹

Incidence

Investigators have attempted to determine the true incidence of GIST using the Surveillance, Epidemiology, and End Results (SEER) database from the National Cancer Institute. However, these data are difficult to interpret since many GISTs were previously misclassified as other gastrointestinal mesenchymal neoplasms.⁴⁰ Although a near doubling of the incidence of all gastrointestinal mesenchymal tumors (over 80% were GIST) has been reported (0.17/100,000 in 1992 to 0.31/100,000 in 2002), this may be due to a combination of increased recognition, increased screening, and/or true increased incidence.⁴⁰ The annual incidence in the United States is estimated to be approximately 5000 new cases per year.⁴¹ European population-based studies identify annual incidence rates ranging from 11 to 14.5 cases per million population.^6,42

Age

The median age at diagnosis of GIST is 60 years (range 58-65 years).^2,6 There is a slight male predominance, and there is no significant racial or ethnic predilection. GIST does occur rarely in children (1.4%-2.6% of cases), often as a familial syndrome or as part of Carney’s triad (see below).^43,44 The clinical presentation is typically different in children and tends to present with multifocal epithelioid gastric GISTs, harbor wild-type KIT/PDGFRA genes, and have a higher incidence of lymph node metastases.³⁷

Hereditary GIST

The overwhelming majority of GISTs are sporadic. Nevertheless, families with germline KIT and PDGFRA mutations have been reported.^45–53 Individuals with GISTs secondary to familial germline KIT mutations are usually younger than those with sporadic GISTs, manifest multifocal disease at presentation, and rarely develop metastatic disease.⁵² This phenotype includes skin hyperpigmentation and diffuse hyperplasia of the intestinal myenteric plexus and is associated with a germline KIT exon 11 mutation.⁵⁴ Mutations in exons 8, 13, and 17 have also been identified at the germline level with constitutive activation resulting in variable phenotypes including dysphagia, urticarial pigmentosa, and macrocytosis.^55–57

Germline PDGFRA mutations are less common and have been reported to affect exon 12.⁵³ These individuals have congenitally enlarged hands, small intestinal polyps, fibroid tumors, and lipomas.

Von Recklinghausen neurofibromatosis type 1 (NF1) is the most common autosomal dominant inherited disorder and is characterized by a spectrum of clinical features including cutaneous neurofibromatosis, café-au-lait macules, axillary and inguinal freckling, ocular hamartomas, and benign and malignant intestinal tumors.⁵⁵ Approximately 7% of these individuals have multifocal GISTs, most commonly in the small intestine.^56–58 In addition to their NF1 mutations, these individuals express KIT and PDGFRA point mutations in 8% and 6% of GISTs, respectively.⁵⁹ Conversely, NF1 mutations have not been identified in non-NF1 individuals with sporadic GISTs.⁶⁰

Gastric GISTs are components of both Carney’s triad and Carney–Stratakis syndrome. Fewer than 100 cases of Carney’s triad, consisting of gastric GISTs, pulmonary chondromas, and extra-adrenal paragangliomas, have been reported.^44,61 Esophageal leiomyomas and adrenal cortical adenomas have recently also been added as elements of this syndrome.⁴⁴ Approximately 85% of cases occur in women and 80% are diagnosed before age of 30. Patients with Carney’s triad lack germline KIT or PDGFRA mutations; however, chromosome 1 deletions of 1q12 to 1q21 involving the SDH gene and the 1p region have been implicated in the pathogenesis of this syndrome. GISTs seen in these patients are typically multifocal and are more likely to manifest lymph node metastases.

The similarly eponymous Carney–Stratakis syndrome describes familial cases expressing the dyad of gastric GIST and paraganglioma.⁴⁶ Recently, inactivating germline mutations in several SDH subunits have been reported in Carney–Stratakis syndrome kindreds and are reflected as a characteristic clinical picture.⁶²

Primary GISTs commonly arise in the stomach (50% to 70%), small intestine (25% to 35%), colon and rectum (5 % to 10%), mesentery or omentum (7%), and esophagus (<5%).^10,63 Occasionally, GISTs may arise in the duodenal ampulla, appendix, gallbladder, and urinary bladder.^64–69

GISTs are generally found due to symptoms. In a large population-based study, 69% of tumors were symptomatic, 21% were discovered incidentally at surgery, and 10% were discovered at autopsy.⁶ Primary extragastrointestinal GISTs are rare (<10% of cases) and may be the sequela of a yet unrecognized primary gastrointestinal tumor. These GISTs usually present in the omentum or mesentery and have a more aggressive clinical course compared with similar-sized gastric counterparts.⁷⁰

GISTs are often highly vascular, soft, and friable, and bleeding is therefore a common complaint. Other common presenting symptoms may include abdominal pain, distension suggestive of obstruction, or a palpable mass. GISTs may cause life-threatening hemorrhage by erosion into the gastric or bowel lumen. Alternatively, tumor rupture may cause potentially catastrophic intraperitoneal bleeding and/or dissemination by peritoneal seeding. Intestinal obstruction may lead to perforation. Small tumors may remain asymptomatic and be incidentally detected on radiographic studies, endoscopy, or laparotomy.

Between 15% and 47% of patients with GIST have metastatic disease at diagnosis.^2,71 Common sites of metastasis include liver, peritoneum, and omentum; lymph node metastases are rare.⁵ Extra-abdominal metastases (lung, bone, subcutaneous tissues, and brain) are rare, observed in approximately 5% of patients.⁷²

Radiographic Studies

The initial imaging study for a suspected or confirmed GIST is a contrast-enhanced computed tomography (CT) scan of the abdomen and pelvis.⁷³ Primary GISTs are typically well-circumscribed masses located within the wall of hollow viscera and may appear heterogenous due to the presence of necrosis or intratumoral hemorrhage, particularly in large tumors (Fig. 33-4). Magnetic resonance imaging (MRI) may help to characterize metastases to the liver or primary perirectal disease (Fig. 33-5). [¹⁸F]fluoro-2-deoxy-D-glucose positron emission tomography combined with CT (FDG PET-CT) may be helpful to characterize masses that are ambiguous on CT, to monitor response to tyrosine kinase inhibitor therapy, and to detect emergence of drug-resistant clones. However, its diagnostic test characteristics are sensitive but not specific, a profile that is not recommended for initial identification or staging of suspected primary disease.^74–76

Endoscopy, Fine-Needle Aspiration, and Biopsy

Endoscopically, a primary GIST may appear as a submucosal lesion, with or without ulceration, present in the upper or lower gastrointestinal tract. They are often indistinguishable from other gastrointestinal tumors of smooth muscle origin, such as leiomyomas (Fig. 33-6). Endoscopic ultrasound (EUS) is not necessary to evaluate a confirmed GIST. However, EUS-guided fine-needle aspiration (FNA) may be attempted to establish diagnosis. Nevertheless, EUS-FNA is not consistently diagnostic, with sensitivity approaching 80%.⁷⁷ Additional cytologic morphology, immunohistochemistry, and reverse-transcriptase polymerase chain reaction analysis for KIT mutations may be required to confirm a diagnosis.⁷⁸

A preoperative biopsy is not routinely necessary for a primary, resectable neoplasm suspicious for GIST, particularly in the setting of ongoing symptoms. In fact, preoperative biopsy, either endoscopic or percutaneous, may rupture a suspected GIST and increase risk of dissemination and bleeding. However, if the differential diagnosis includes entities such as nonoperatively treated lymphoma, if preoperative diagnosis is required for targeted neoadjuvant therapy, or if there is suspected metastatic disease, biopsy is warranted.

The American Joint Committee on Cancer (AJCC) implemented formal staging for GISTs in 2010.⁷⁹ This schema is divided into two main groups based on location: gastric GISTs (also used for omentum), and small intestine GISTs (also used for esophagus, colorectal, mesentery, and peritoneum). This dichotomy is based on evidence of prognostic differences based on anatomic location.^80–82 The tumor, node, metastasis (TNM) classification is combined with mitotic rate to determine individual tumor stage. Germline mutational status is not included in the current AJCC classification. This staging system has been validated and correlates with disease-free survival and overall survival, with tumor size and mitotic index being the most important prognostic factors.⁸³

The three established pathologic prognostic factors related to the risk for GIST metastases are tumor size, mitotic index (per 50 high-power field [HPF]), and tumor site of origin, with mitotic count being the most contributory (Table 33-1).^16,80,84 While tumors under 1 cm appear to be at a low risk of recurrence and progression, no tumors can be definitively called benign and most have malignant potential. Individuals with small bowel or colorectal GISTs have a higher risk of progression than those with gastric GISTs of comparable size and mitotic count. Several risk stratification schema have been suggested to categorize patient prognosis and recurrence risk based on pathologic features (Table 33-2).^82,85–87 However, predicting tumor behavior based on pathologic elements alone is suboptimal, and contemporary risk stratification aims to combine these features with molecular analysis in order to glean a more complete picture of biologic potential.

Additional adverse prognostic factors suggested include high cellular proliferation index (Ki-67),⁸⁸ aneuploidy,^88,89 telomerase expression,^90,91 KIT exon 9 mutations,⁸⁰ and KIT exon 11 deletions involving amino acid W557 and/or K558.⁹² Point mutations and duplications of KIT exon 11 appear to have a favorable prognosis.⁸⁰ Exon 9 and 11 mutational status have also been shown to predict response to medical therapy and thereby influence clinical rates of progression and survival. In advanced GIST with KIT exon 11 mutations, approximately 90% of patients will respond to imatinib, whereas only 50% of exon 9 mutations will have a similar response, although dose-related response rates may be observed in the latter. Most GISTs with PDGFRA mutations have a lower risk of recurrence and respond to imatinib, with the exception of the D842V substitution in exon 18.³⁵ Despite these mutational correlations with recurrence risk, evidence suggests that standard pathologic prognostic factors, particularly mitotic count, predict clinical outcomes more accurately than mutational analysis status, suggesting that no individual factor should be used in isolation.²⁵

Completeness of surgical resection has also been suggested as a prognostic indicator. However, the ideal margin of resection is unknown. While a macroscopically complete resection with negative or positive microscopic margins (R0 or R1 resection, respectively) is associated with a better prognosis than a macroscopically incomplete resection (R2 resection), there are no data to confirm that a positive microscopic margin (R1 resection) impacts survival.²

The modern management of GISTs is a multidisciplinary task. Patients should be treated in a center offering expertise in the surgical and medical management of disease, with available multidisciplinary cancer conferences to discuss nuances in clinicopathologic and immunohistochemical tumor profile and its influence on multimodality treatment sequencing.

Active Surveillance

The management of gastric GISTs less that 2 cm in diameter is controversial, as their natural history is not known with certainty but is thought to be favorable. Two studies have shown that sub-centimeter gastric GISTs (microGISTs) are relatively common, detected in 22.5% of autopsies in adults over the age of 50 in Germany, and in 35% of patients undergoing gastrectomy for gastric cancer in Japan.^93,94 Despite their relative frequency, few of these neoplasms appear to become clinically relevant. Until further data are available, the most appropriate management of such small tumors remains uncertain.

In an Italian study comparing 170 GISTs ≤2 cm in size to 101 >2 cm, the frequency of mitoses observed in individual tumors increased from 20% in <6 mm GISTs to 75% in 1.7- to 2.0-cm GISTs to 100% in >2-cm GISTs.⁹⁵ Of the 170 small GISTs, 135 underwent mutational analysis. Key findings compared to larger GISTs were that 74% of small GISTs had KIT or PDGFRA mutations with an excess of wild-type cases, KIT exon 11 mutations were less common, and novel mutations were observed that have not been reported in clinically relevant GISTs >2 cm.⁹⁵

Gastric tumors stratified as low risk based on clinical, pathologic, and molecular features may be considered for active surveillance alone. Decisions for surveillance alone should involve a multidisciplinary discussion with risks and benefits provided to the patient and documented in the medical record. Clinical and radiologic parameters invoking a transition to more aggressive management should also be reviewed.

If active surveillance is undertaken, current National Comprehensive Cancer Network (NCCN) guidelines suggest endoscopy with esophagogastroduodenoscopy (EGD) +/− EUS every 6 to 12 months. Presence of lesional growth to >2 cm, irregular extraluminal borders, ulceration, heterogenous echogenic foci, presence of cystic spaces, and/or development of symptoms should prompt reconsideration of the management strategy.⁹⁶ European Society for Medical Oncology (ESMO) guidelines recommend annual surveillance with EUS for presumed <2 cm GISTs, and EUS-FNA may be offered.⁹⁷ If a diagnosis of GIST is confirmed, then surgery should be offered. If the patient prefers no surgery, then observation should continue, though an evidence-based optimal surveillance policy is lacking.⁹⁷

Japanese guidelines have been developed based on extensive population-based endoscopic surveillance for gastric cancer and distinguish between EGD and EUS findings. Gastric submucosal tumors <2 cm without the malignant findings on EGD of ulceration, irregular margins, or rapid growth may be followed with EGD surveillance once or twice per year.^98,99 Tumor growth or presence of malignant findings on EGD warrants either surgery or further workup, including CT, EUS, or EUS-FNA. Malignant findings on EUS (heterogeneous parenchyma or irregular margins) or confirmation of GIST by EUS-FNA are indications for surgery. Patients with submucosal lesions without confirmation of a diagnosis of GIST and without malignant findings on EUS may be offered surgery or ongoing surveillance.

Surgery

Surgery remains the standard of care and the only potentially curative option for patients with primary, resectable, localized GISTs. All patients with GISTs ≥2 cm, symptomatic tumors, and all GISTs of non-gastric origin should be considered for surgical resection and adjuvant medical treatment. Oncologic principles of safe surgical practice prevail, and the primary goal of the operation is to resect all tumor with macroscopically and microscopically negative margins (R0). Tumors should be handled with a “no-touch” technique, as rupture or violation of the tumor pseudocapsule during surgery is associated with increased risk of recurrence and bleeding. Formal lymphadenectomy is rarely required, except in pediatric populations where lymph node metastases may be more prevalent.^100,101

The extent of surgical resection is governed by the size of the tumor in relation to anatomic location, and the ability to safely restore gastrointestinal continuity. Sphincter and organ preservation is preferred, although locally advanced GISTs should be approached with an en bloc resection of adjacent organs in order to minimize potential violation of the tumor, if necessary. In such cases, surgical resection is often preceded by medical therapy to promote tumor shrinkage and minimize the extent of resection, as described below. The extent of surgery is usually a wedge or segmental resection of the involved stomach or bowel, without the wide margins necessary for corresponding adenocarcinoma. In a series of 140 patients with gastric GISTs, wedge resections were performed in 68%, partial gastrectomies in 28%, and total gastrectomies in only 4%.¹⁰² Occasionally, more extensive resection (total gastrectomy for a large proximal gastric GIST, pancreaticoduodenectomy for a periampullary GIST, or abdominoperineal resection for a low rectal GIST) may be necessary.

Regardless of whether an open or laparoscopic resection is planned, GISTs are approached in a similar fashion using preoperative imaging as a guide. The abdomen is first thoroughly explored to identify involved organs and remove any previously undetected peritoneal metastatic deposits. Although primary GISTs may demonstrate inflammatory adhesions to surrounding structures, they do not generally invade other organs beyond the site of origin, despite ominous CT appearance. Nasogastric suction, preoperative tattooing, and intraoperative EGD can be considered as adjuncts to assist in the accurate localization of the GIST and in the achievement of R0 resection.

Although endoscopic resection of small gastric GISTs has been reported, this is not recommended.¹⁰³ Unlike early gastric adenocarcinomas limited to the mucosa and amenable to endoscopic mucosal resection, GISTs involve the muscularis propria. Therefore, endoscopic resection risks leaving a positive deep margin, and due to the depth of the lesion may result in gastrointestinal perforation. Such endoscopic resections have been reported for the most part for small GISTs, a group of tumors for which surgery may not even be indicated.¹⁰⁴ Therefore, such techniques should be approached with a great deal of caution and are not recommended by any major international guidelines.

Laparoscopic or laparoscopy-assisted resection of primary GISTs continues to evolve and may be considered if standard oncologic principles can be maintained (Fig. 33-7). Two early studies confirmed both the safety and feasibility of a laparoscopic approach.^105,106 A recent meta-analysis by Koh et al. supports laparoscopic resection as a safe and feasible approach with no differences in margin positivity, recurrence-free survival (RFS) or overall survival (OS) when compared to open resections. Additionally, laparoscopic resections frequently had shortened stays in hospital and lower intraoperative blood loss.¹⁰⁷ However, decisions for technical operative approach must balance perioperative factors with oncologic safety. Evaluation of patient comorbidity, anatomic tumor location, surgeon expertise, and ability to convert to an open approach should be included in preoperative decision-making. When laparoscopic and laparoscopy-assisted approaches are undertaken, modified lithotomy or split-leg positioning should be considered for heightened visualization and organ manipulation. In addition, an angled camera and roticulating laparoscopic stapler may assist in the safe resection and extraction of the tumor.

The goal of surgery for GIST is to perform a margin-negative resection. However, unlike for visceral adenocarcinomas or sarcomas at other sites, wide margins are unnecessary. Wide margins beyond a R0 resection have not been definitively linked with improvements in survival or recurrence, particularly in the era of effective medical therapy. There are also no data indicating that patients with an R1 (microscopically involved) resection require re-excision.⁴¹ Margins may retract after resection and chemical fixation, and the pathologist may resect a staple line ex vivo, thereby converting a microscopically negative margin into a positive one. As such, all cases of positive microscopic margins should be reviewed by a multidisciplinary team to assess the need for re-excision with careful consideration of anatomic feasibility, biologic risk profile, and patient comorbidity.