Gastrointestinal Stromal Tumor

Risk stratification is preferred to anatomical staging.

Most GISTs harbor gain-of-function mutations of the KIT or PDGFRA oncogene and progress by the stepwise inactivation of tumor suppressor genes. See Diagnostic molecular pathology, below, for full details, which are of clinical significance.

Localized GIST presents as a well-circumscribed mass of highly variable size (ranging from incidental, submillimeter lesions to > 20 cm). In larger lesions, the cut surface may show foci of hemorrhage, cystic change, and/or necrosis. Gastric GISTs often feature an intraluminal component and may produce umbilicated mucosal ulcers. In the small bowel, GISTs more frequently present as external masses. Some GISTs feature a narrow pedicle linked to the serosal surface, the interruption of which may contribute to the generation of extragastrointestinal GISTs.

Advanced disease most often presents as a main lesion associated with multiple smaller nodules that may extend from the diaphragm to the pelvis. Invasion of surrounding organs such as the spleen and pancreas can be observed in aggressive tumors. SDH-deficient GISTs are often associated with a distinctive multinodular pattern of growth.

Microscopically, GISTs exhibit a broad morphological spectrum. Anatomical location (gastric vs small bowel) seems to influence the histological appearance. Most gastric GISTs are spindle cell tumors, with epithelioid morphology seen in approximately 20–25% of cases. Some cases feature a combination of spindle cell and epithelioid histology. Nuclear pleomorphism is uncommon. Distinctive histological patterns among spindle cell GISTs exist. One example is the sclerosing type, seen especially in small tumors that often contain calcifications. The palisaded-vacuolated subtype is one of the most common, whereas some examples show a diffuse hypercellular pattern. Very rarely, sarcomatoid features with substantial nuclear atypia and high mitotic activity can be observed. Epithelioid GISTs may show sclerosing, discohesive, hypercellular (sometimes with a pseudopapillary pattern), or sarcomatous morphology with substantial atypia but low mitotic activity. Myxoid stroma is rarely observed.

Small intestinal and colonic GISTs are usually spindle cell tumors with diffuse sheets or vague storiform arrangements of tumor cells. Tumors with low biological potential often contain extracellular collagen globules (skeinoid fibers). Intestinal GISTs may feature anuclear areas (somewhat mimicking Verocay bodies or neuropil) composed of cell processes. Nuclear palisading, perivascular hyalinization, and regressive vascular changes (e.g. dilated and thrombosed vessels, hemosiderin deposition, and fibrosis) similar to those in schwannomas can be seen. Rectal GISTs most often feature spindle cell morphology.

SDH-deficient GISTs characteristically show epithelioid morphology and are typically multinodular with plexiform mural involvement. Unlike in conventional GISTs, lymphovascular invasion and lymph node metastases are common.
Extremely rarely, morphological progression to high-grade (KIT-negative) sarcomatous morphology can be observed either de novo or after therapy with imatinib (dedifferentiated GIST). Dedifferentiation can also be associated with heterologous epithelial, myogenic, or angiosarcomatous differentiation.

Immunophenotypically, most GISTs show strong and diffuse expression of KIT (CD117), which appears as cytoplasmic, membrane-associated, or sometimes perinuclear dot-like staining. However, a small minority (< 5%), especially GISTs with PDGFRA mutations, may lack KIT expression or show very limited staining. The chloride-channel protein ANO1/DOG1 is an equally sensitive and specific marker and may rescue diagnostically as many as 50% of KIT-negative GISTs. KIT and DOG1 are also expressed in the interstitial cells of Cajal, whose precursors are believed to be the histogenetic origin of GISTs. Most spindle cell GISTs (especially gastric tumors) are positive for CD34, whereas epithelioid examples are less consistently positive. Some GISTs express h-caldesmon; a minority express SMA; and rare examples show positivity for desmin, keratins (CK18), or S100. SDH-deficient GISTs exhibit loss of SDHB protein expression irrespective of which SDH gene is mutated. SDHA loss is specific for SDHA-mutant tumors. Loss of expression of neurofibromin (NF1; using an antibody specific to the C-terminus) may help in identifying NF1-associated GISTs.

Cytologic features of GISTs largely depend on the predominant histologic pattern. Spindle cell GISTs yield cellular smears composed of elongated cells with scant cytoplasm and indistinct cell borders. The nuclei are typically uniform, elongated, and often contain small, inconspicuous nucleoli. Epithelioid GISTs tend to exfoliate as loose clusters or single, round to oval cells. The cytoplasm is typically abundant and eosinophilic. The nuclei may show more variation in size and shape compared to spindle cell GISTs, and the nucleoli may be more prominent. In both spindle cell and epithelioid GISTs, mitotic figures are rare. However, in high-risk GISTs, mitotic figures may be identified, along with nuclear atypia and necrosis.

The cytologic diagnosis of GIST is often challenging. The differential diagnosis includes other mesenchymal tumors such as leiomyoma, leiomyosarcoma, schwannoma, and solitary fibrous tumor. Immunocytochemistry plays a crucial role in establishing the diagnosis. Most GISTs show strong and diffuse positivity for KIT (CD117), DOG1, and often CD34. SDH-deficient GISTs show loss of SDHB expression.

The best-documented prognostic parameters for GIST are mitotic activity, tumor size, and anatomical site (see #6402Table #6402). Mitotic counting is for an area of 5 mm2, which in most modern microscopes corresponds to 20–25 fields with the 40× objective and standard eyepiece diameter. This prognostic assessment applies best to KIT/PDGFRA-mutant GISTs. In general, intestinal GISTs and SDH-deficient GISTs are more unpredictable. Tumors with low mitotic counts can metastasize, whereas tumors with higher mitotic counts may remain indolent for extended periods. Many patients with SDH-deficient GISTs with liver metastases can survive for years or decades without specific treatment, in contrast to patients with KIT/PDGFRA-mutant GISTs, which are rapidly progressive when metastatic. Tumor rupture is an additional adverse factor in GIST. The grading principles for soft tissue sarcomas do not apply to GIST. In order to refine risk assessment for consideration of adjuvant therapy, it has been suggested to include size and mitotic counts as continuous variables to be incorporated along with anatomical site into prognostic tools such as nomograms or prognostic contour maps.

Mutation status represents a prognostic as well as predictive factor. In general, KIT-mutant tumors tend to behave more aggressively than PDGFRA-mutant or triple-negative (KIT/PDGFRA/BRAF-wildtype) tumors. The best outcome seems to be associated with PDGFRA exon 12, BRAF, and KIT exon 12 mutations. An intermediate risk seems to be associated with KIT/PDGFRA/BRAF-wildtype status and with KIT exon 17, PDGFRA exon 18 (p.Asp842Val), and PDGFRA exon 14 mutations. The worst outcome seems to be associated with KIT exon 9 and 11 and PDGFRA exon 18 (non-p.Asp842Val) mutations.

Mutation status also predicts response to imatinib, with KIT exon 11–mutant tumors exhibiting the highest rate of response and PDGFRA exon 18 (p.Asp842Val) mutants showing primary resistance. Molecular status also influences imatinib dose selection, with KIT exon 9 mutants benefiting from a higher dose (800 mg instead of 400 mg). Subsequent mutations are associated with acquired resistance to imatinib. Secondary KIT gene mutations are most often found in the ATP-binding pocket of the kinase domain (exons 13 and 14) or in the kinase activation loop (exons 17 and 18). Both KIT/PDGFRA/BRAF/SDH-wildtype and NF1-associated GISTs are also characterized by a lack of sensitivity to imatinib.

About 85% of GISTs harbor gain-of-function mutations of the KIT or PDGFRA oncogene located on chromosome 4 (4q12), encoding for type III receptor tyrosine kinases. With exceedingly rare exceptions, they are mutually exclusive and result in the constitutive activation of either KIT or PDGFRA. Normally, KIT and PDGFRA are activated by the binding of their respective ligands (i.e. stem cell factor and PDGFA). Downstream oncogenic signaling involves the RAS/MAPK and PI3K/AKT/mTOR pathways.

About 75% of GISTs harbor activating mutations of KIT, most often in exon 11 (66% overall) or exon 9 (6%); mutations in exons 13 and 17 are rare (~1% each). Even more uncommon are mutations in exon 8. KIT exon 11 mutations include deletions (45%), substitution mutations (30%), and insertion/deletion (indel) mutations (15%) including duplications. Nearly all KIT exon 9 mutations are duplications (p.Ala502_Tyr503); 80% of GISTs with such mutations arise in the small intestine. KIT exon 13 and 17 mutations are most often p.Lys642Glu and p.Asn822Lys, respectively.

About 10% of GISTs harbor PDGFRA activating mutations (most often in the stomach), usually in exon 18 (8% overall); mutations in exons 12 and 14 are rare. The most common PDGFRA mutations are p.Asp842Val (55%) and p.Val561Asp (10%). Patients with PDGFRA-mutant tumors have a lower risk of metastasis than patients with KIT-mutant tumors. Given these differences in metastatic risk, nearly 85% of advanced GISTs harbor KIT mutations and only 2% harbor PDGFRA mutations.

Many GISTs that are wildtype for KIT and PDGFRA harbor alterations in SDH subunit genes (5–10% overall); 60% harbor inactivating mutations (nearly always germline); and 40% harbor SDHC promoter methylation (epimutation), leading to SDH dysfunction (SDH-deficient GIST). Patients with SDH-deficient GISTs are younger than those with tyrosine kinase receptor gene–mutant tumors; nearly all pediatric GISTs are SDH-deficient. Tumors from patients with Carney triad usually show SDHC epimutation. SDHA is the most commonly mutated subunit gene (~35% of SDH-deficient GISTs), followed by SDHB, SDHC, and SDHD. Rare GISTs are associated with mutations of NF1 (which are usually germline alterations in patients with NF1 or rarely somatic mutations), BRAF, or KRAS. Like KIT and PDGFRA mutations, these alterations also result in RAS/RAF/MEK pathway activation.

Most GISTs (with the exception of SDH-deficient tumors) progress through a stepwise acquisition of chromosomal alterations, each of which probably inactivates tumor suppressor genes: loss of 14q (as many as 70%), followed by loss of 22q (~50%), 1p (~50%), and 15q (~40%). MAX is the 14q GIST tumor suppressor gene, inactivated early (in microscopic and low-risk tumors). Inactivating mutations in CDKN2A, TP53, and RB1 are found in GISTs of higher-risk categories. DMD inactivation is a late event in GIST progression, identified in nearly all metastatic GISTs. Very rare GISTs harbor NTRK3 or FGFR1 gene fusions.

Essential: an intramural, submucosal, or subserosal mass; spindle cell, epithelioid, or mixed morphology; KIT and/or DOG1 immunopositivity; SDHB loss in SDH-deficient GISTs.

Desirable: KIT or PDGFRA gene mutations in approximately 85% of tumors.