GIST has a relatively limited spectrum of histologic appearances. The majority are well-circumscribed, but some show infiltrative margins. Most (70 %) are composed of a relatively uniform population of spindle cells, and in 20 % of cases of a uniform population of epithelioid tumor cells; the remaining cases show mixed spindled and epithelioid cell morphology. The spindled tumor cells of GIST are arranged in short fascicles and appear bland (i.e., lack of significant cytologic atypia or pleomorphism) with indistinct cell borders, which imparts a syncytial appearance to the cytoplasm (Fig. 2a). The nuclei are elongated with tapered ends, vesicular chromatin, and inconspicuous nucleoli. The cells have moderate amounts of pale eosinophilic fibrillary cytoplasm (Fig. 2b). Paranuclear vacuoles are common in gastric GIST. Dense eosinophilic collagen fibrils, known as skenoid fibers, are often present in small bowel tumors. The vasculature can range from minimal to thick hyalinized hemangiopericytoma-like vessels. A lymphocytic infiltrate is often present, and nuclear palisading of tumor cells can be seen. Epithelioid GIST virtually always arises in the stomach, and is usually PDGFRA-mutant or less often SDH-deficient (see discussion below). These tumors have a nested or sheetlike growth pattern, and are composed of cells with round nuclei, vesicular chromatin, variably prominent nucleoli, and abundant cytoplasm which can be eosinophilic or less often clear, and may have distinct cell borders (Fig. 3a). A gastric tumor with a multinodular or plexiform growth pattern through the muscularis propria should raise suspicion for underlying SDH deficiency (discussed below). Some tumors, either epithelioid or spindled cell, may have a prominent myxoid stroma, making recognition of a tumor as GIST difficult.

Approximately 80 % of GIST harbor activating KIT mutations [1, 9, 10, 25] and 10 % mutations in PDGFRA [26], resulting in constitutive kinase activation independent of the presence of the receptor ligands (SCF for KIT and PDGFA for PDGFRA). KIT and PDGFRA are members of the type III receptor tyrosine kinase family and share a common structure that is comprised of an extracellular ligand-binding domain, a transmembrane domain, a juxtamembrane domain, and a cytoplasmic kinase domain. Ligand binding to the extracellular domain triggers receptor dimerization, phosphorylation, and signal transduction via MAPK, PI3K, and p90RSK pathways. It appears that KIT and PDGFRA mutations constitute the earliest molecular events detectable by current techniques in GIST tumorigenesis, and they are present in extremely small tumors [27].

The vast majority of KIT-mutant GIST harbors mutations in exon 11 (70 %), which encodes the juxtamembrane domain, whose normal function is to prevent the kinase activation loop from moving into the active conformation [9, 28]. Exon 11 mutations can result from substitutions, insertions, or in-frame deletions, and cause KIT to switch into an active conformation despite the absence of natural ligand SCF. Among the various mutagenic mechanisms, exon 11 deletions appear to portend a worse prognosis and are associated with shorter progression-free and overall survival compared to insertions or substitutions [29–32]. KIT exon 11-mutant GIST can arise anywhere in the GI tract, and are generally highly sensitive to imatinib, at least initially. The second largest group of KIT-mutant GIST harbors mutations in exon 9, which encodes the KIT extracellular domain, causing a conformational change that simulates ligand binding [33, 34]. Exon 9 mutations are seen in small and large intestinal GIST, but infrequently in gastric GIST. KIT exon 9-mutant GIST are less sensitive to tyrosine kinase inhibitors than exon 11-mutant tumors, largely because the kinase domain remains unaltered just as in wild-type KIT, and higher doses of imatinib are needed to achieve similar responses to those seen in exon 11-mutant tumors. Much rarer are mutations in KIT exon 17, which encodes the activation loop of the kinase domain, stabilizing KIT in its active conformation, and mutations in KIT exon 13, which encodes the adenosine triphosphate-binding region of the tyrosine kinase domain [35]. Both KIT exon 17-mutant and exon 13-mutant GIST are usually spindled in cytomorphology, and arise slightly more frequently in the small intestine than in the stomach [35]. KIT mutations occur extremely rarely in exon 8 [36, 37], and these tumors appear to have a predilection for the small intestine and show a mixed spindled and epithelioid morphology.

The most common PDGFRA mutations occur in exon 18, followed by exons 12 and 14. Exon 18 encodes the activation loop, while exons 12 and 14 encode the juxtamembrane domain and the adenosine triphosphate-binding domain, respectively [26, 38, 39]. PDGFRA-mutant GIST most often arises in the stomach and shows an epithelioid cytomorphology [40, 41]. Additionally, these tumors commonly have a myxoid stroma and may be negative for KIT expression by immunohistochemistry, making recognition difficult. The clinical behavior of PDGFRA-mutant GIST is generally more indolent compared to KIT-mutant GIST.

Familial GIST has been reported in multiple families [1, 25, 42]. These patients carry KIT or PDGFRA germline mutations and are affected with nearly 100 % penetrance. They present with tumors at multiple sites within the GI tract, often in a background of hyperplasia of the interstitial cells of Cajal.

In addition to oncogenic KIT and PDGFRA mutations, which are thought to represent early events in the molecular pathogenesis of GIST, comparative genomic hybridization analysis and cytogenetics studies have identified a signature of secondary chromosomal aberrations that are associated with disease progression, such as losses at 1p, 9p/9q, 11p, 15q, and gains at 5p, 8q, 17q, and 20q [43]. Chromosome 14 abnormalities occur in up to two-thirds of cases, predominantly monosomy or partial loss of 14q. Loss of the long arm of chromosome 22 is seen in approximately 50 % of tumors [35–38]. Losses at 14q and 22q do not appear to contribute to malignant behavior. However, gains on chromosome 8q (MYC locus), 3q (region of SMARCA3), and 17q have been associated with aggressive behavior [38, 39, 44, 45]. These findings are present in both KIT and PDGFRA-mutant GIST, as well as GIST that arises in patients with type 1 neurofibromatosis (NF1), but are not seen in SDH-deficient GIST. Gene expression profiling studies have identified genetic changes associated with an aggressive clinical course, including inactivation of the tumor suppressor gene CDKN2A [42, 46–48], TP53 mutations [43, 49–51], abnormalities in genes involved in the PI3 kinase pathway [52], and rarely amplification of MDM2 and CCND1 [45, 53].

Tumors measuring less than 1 cm in greatest dimension are considered “micro-GIST.” These small lesions are usually incidentally detected, if detected at all, and are in fact very common, as shown in systematic studies of stomachs at autopsy and surgical resection which estimate an overall frequency of 30 % among the general population [54, 55]. Micro-GIST typically shows a spindled cell morphology and has a hyalinized or calcified stroma. KIT mutations are detected in the vast majority of micro-GIST. The clinical course is benign and these tumors virtually never metastasize. Importantly, micro-GIST should not be confused with synchronous metastatic lesions measuring <1 cm and occurring in association with a larger dominant mass.

The term “wild-type” GIST is commonly used for tumors without identifiable KIT or PDGFRA mutations. This group accounts for approximately 10–15 % of adult GIST and 90 % of pediatric GIST. Recent advances in our understanding of the pathobiology of GIST have shown that this group of “wild-type” tumors actually represents a heterogeneous group of clinicopathologically and molecularly distinct GIST. This group includes not only sporadic tumors with distinct mutation signatures, but also lesions arising in patients with the nonhereditary Carney triad syndrome (gastric GIST, paraganglioma, and pulmonary chondroma) and the hereditary Carney-Stratakis syndrome (gastric GIST and paraganglioma), the latter two being part of the group of SDH-deficient GIST discussed in more detail below, NF1-associated GIST, BRAF-mutant GIST, and a small group whose molecular pathogenesis has yet to be elucidated. “Wild-type” GIST are largely resistant to imatinib, and therefore correct classification is paramount in order to select the appropriate therapy. Furthermore, since some of these tumors arise in association with inherited syndromes, correct classification is critical for clinical follow-up (i.e., detection of other tumor types), germline testing, and genetic counseling.

This recently described clinicopathologically and molecularly distinct group of tumors includes the majority of pediatric GIST, GIST arising in patients with Carney triad, Carney-Stratakis syndrome, and a subset of apparently sporadic adult “wild-type” GIST (some previously referred to as “pediatric-type” GIST) [56, 57]. SDH-deficient GIST represents 7.5 % of all gastric GIST [58] and 42 % of all “wild-type” GIST. They have a female predilection, and arise exclusively in the stomach, usually in the antrum, where they may present as multiple discontiguous lesions. Histologically, this group of GIST shows a multinodular or plexiform growth pattern (Fig. 4a) and predominantly epithelioid morphology (Fig. 4b) [56, 57, 59]. In contrast to KIT– and PDGFRA-mutant GIST, vascular invasion may be seen and lymph node metastases are relatively more common. Although tumors lack KIT and PDGFRA mutations, immunoreactivity for KIT and DOG1 is usually strong. Additionally, tumor cells show loss of protein expression of SDHB (Fig. 4c), which is normally ubiquitously expressed in all cell types (see section “Immunohistochemistry in the Evaluation of GIST”); the mechanism underlying this “deficiency” of SDHB expression is discussed below. The above histologic and immunohistochemical features are all helpful clues to identify this distinct subgroup, which has important clinical and syndromic implications. SDH-deficient GIST tends to be resistant to imatinib, but may respond to second and third-generation tyrosine kinase inhibitors, and the clinical course of this group of tumors is relatively indolent, even in the setting of metastatic disease [56, 57].

The SDH enzyme complex is a member of the tricarboxylic acid cycle and electron transport chain that catalyzes the oxidation of succinate to fumarate, and is made of four normally and ubiquitously expressed subunit proteins SDHA, SDHB, SDHC, and SDHD [60]. Loss of SDHB expression in tumor cells reflects dysfunction of the entire SDH complex, which can be caused either by mutations in any of the genes coding for the four subunit proteins or by other functional deficiencies due to other mechanisms, such as hypermethylation or epigenetic events. In contrast, loss of SDHA expression is only seen in SDHA-mutated tumors. SDH complex dysfunction driven tumorigenesis is incompletely understood. Several studies have suggested that increased levels of the metabolite succinate alter the global gene methylation profile [61]. SDH-deficient GIST appears to have increased levels of methylated DNA when compared with KIT-mutant tumors. Succinate accumulation inhibits the TET family of DNA hydroxylases, which catalyze the production of the gene expression altering molecule 5-hydroxymethylcytosine (5-hmC). Reduced 5-hmC levels have been demonstrated in SDH-deficient GIST compared to KIT– and PDGFRA-mutant tumors [61, 62]. In addition, succinate accumulation stabilizes the hypoxia-inducible factor 1α, which enhances the transcription of target genes including vascular endothelial growth factor [63]. SDH-deficient GIST has also been strongly associated with overexpression of the type 1 insulin-like growth factor receptor (IGF1R) [64]. The mechanism underlying IGF1R overexpression is currently unknown.

The Carney-Stratakis syndrome is inherited in an autosomal dominant fashion with variable penetrance and young adulthood onset of gastric GIST and paraganglioma [18, 65]. Affected patients have loss-of-function germline mutations in SDHB, SDHC, or SDHD [65, 66]. The Carney triad syndrome is nonhereditary, usually affects young women, and manifests with gastric GIST, paraganglioma, and pulmonary chondroma [67]. GIST arising in the setting of Carney triad also shows SDH complex dysfunction reflected by loss of SDHB protein expression. However, these patients generally do not have SDH mutations [17, 19, 68, 69]. Recent evidence has demonstrated that the Carney triad is associated with hypermethylation of the SDHC promoter, resulting in loss of SDHC expression [70]. While only 20–25 % of patients with SDH-deficient GIST harbor mutations in SDHB, SDHC, or SDHD, mutations in SDHA have been found in one-third of these tumors, making SDHA the most commonly mutated subunit. SDH-deficient GIST with SDHA mutations have an older age distribution (third to fifth decades) and less female predominance compared to other SDH-deficient GIST [71–74]. Despite the presence of germline SDHA mutations, SDHA-mutant GIST is almost never familial and therefore has a low penetrance.

The diagnosis of SDH-deficient GIST has important clinical implications, both prognostically and predictively. These tumors pursue a relatively indolent clinical course, even in the presence of nodal and distant metastases. They respond poorly to imatinib, but many second- and third-generation tyrosine kinase inhibitors such as sunitinib, sorafenib, and dasatinib have greater efficacy [56, 75, 76]. Furthermore, the commonly used standard risk-stratification system (based on tumor site, tumor size, and mitotic index) for predicting malignant potential in GIST fails to predict clinical behavior for SDH-deficient tumors; thus it should not be applied [17, 56, 58]. Identifying SDH-deficient GIST also identifies a subset of patients who benefit from germline testing for SDH mutations and long-term clinical follow-up for the detection of the other aforementioned syndromic tumors [75, 77]. From a practical standpoint, we would advise that the possibility of SDH-deficient GIST should be considered when a gastric GIST with epithelioid morphology and multinodular plexiform architecture is encountered, and immunohistochemical loss of SDHB protein expression is an extremely useful screening tool in this regard.

BRAF mutations in GIST may arise de novo or after treatment with tyrosine kinase inhibitors. Up to 13 % of “wild-type” GIST harbor the BRAF exon 15 V600E substitution mutation [78–80]. BRAF-mutant GIST have a slight female predilection, and most often arise in the small bowel. Their biological behavior and clinical course has not been well defined to date, but limited evidence suggests a high risk of malignancy based on risk-stratification criteria [78], with the caveat that BRAF mutations are also present in some micro-GIST with no mitotic activity [79]. BRAF-mutant GIST are typically composed of spindled cells and are morphologically indistinguishable from conventional KIT-mutant GIST [80]. BRAF belongs to the RAF family of serine/threonine protein kinases in the RAS–RAF–ERK signaling pathway, which activates the MAPK pathway and controls cell cycle regulation and cellular response to growth signals. The V600E substitution activates the BRAF kinase domain. Thus, mutated BRAF may act as a primary oncogenic driver event. Moreover, since BRAF is located downstream of KIT, its activation leads to KIT-independent growth. Not surprisingly, BRAF-mutant “wild-type” GIST are resistant to imatinib, and BRAF mutations may contribute to the development of secondary resistance to imatinib in KIT- and PDGFRA-mutant GIST [78]. As a result, detection of this mutation also carries significant treatment implications. There is some evidence showing tumor regression in BRAF-mutant GIST treated with BRAF inhibitors [81].

Patients with NF1 have a higher risk of developing GIST than the general population, and tumors usually occur at a younger age compared to cases of sporadic GIST [20, 23, 24]. NF1-associated GIST are “wild-type” and arise most frequently in the small intestine. They are usually small, almost always display a spindled cytomorphology with a low mitotic rate, and have a good prognosis [20]. In the GI tract of patients with NF1, GIST are more common than neurofibromas. Patients with NF1 typically present with multiple primary GIST, often arising in a background of hyperplasia of the interstitial cells of Cajal. The tumors display strong KIT immunoreactivity despite a lack of KIT mutations. The pathogenesis of GIST associated with NF1 remains unknown.