GIST is the most common mesenchymal tumor of the gastrointestinal tract (6). GISTs arise from the interstitial cells of Cajal or their common stem cell (7). GISTs range in size from tiny tumors discovered incidentally during tests for other diseases, measuring less than 1 cm to very large lesions measuring upwards of 35 cm (median 5 cm) (8). Irrespective of tumor size, GISTs share morphologic features and immunoreactivity for KIT and contain an oncogenic mutation in the KIT (80-85%) or platelet-derived growth factor receptor (PDGFRA, 5-7%) genes (9). GISTs can arise in any portion of the gastrointestinal tract, but usually occur in the stomach (60%) or the small intestine (30%) (10, 11).

On gross examination, GIST is a well circumscribed, fleshy, pink, or tan-white mass. Large tumors frequently show hemorrhage, necrosis, and cystic degeneration. Microscopically, GISTs can be divided into three different histologic subgroups. Spindle cell GISTs (70%) are composed of cells with palely eosinophilic, fibrillary cytoplasm, ovoid uniform nuclei, and ill-defined cell borders, often with a somewhat syncytial appearance, arranged in short fascicles or whorls (Fig. 1). Epithelioid GISTs (20%) are composed of rounded cells with eosinophilic to clear cytoplasm arranged in sheets and nests (Fig. 2). The final group shows mixed spindle and epithelioid cells (10%). The frequency of these histological types varies according to location. GISTs of the stomach mostly fall into one of 4 spindle cell subtypes of sclerosing, palisading-vacuolated, hypercellular, and sarcomatous or one of 4 epitheloid subtypes of sclerosing epitheloid variant, dyscohesive epithelioid, hypercellular, and sarcomatous (12). GISTs of the small intestine have great amounts of skeinoid fiber, and are most likely to become malignant if epitheloid type or mixed type is present. Many of GISTs in the large intestine are spindle cell type. GISTs developed in the omentum are similar to histological types of the stomach whereas GISTs of the mesentery are similar to histological findings of the small intestine. The diagnosis of GISTs is mainly based on clinical and histological findings, but immunohistochemical staining is needed to confirm diagnosis (13).

The most important immunohistochemical staining in the diagnosis of GISTs is c-kit (CD 117), and other several antibodies may be helpful in the diagnosis and differential diagnosis. Approximately ~95% of cases are positive for c-kit protein (Figs. 1, 2). c-kit negative GISTs account for ~5% of cases and can cause diagnostic difficulties, but given the rather limited choice in mesenchymal diagnostic considerations at these sites, they can often be diagnosed by excluding other potential mimics by immunohistochemical characterization (14). c-kit staining with polyclonal anti-c-kit antibody is mandatory for diagnosis. Extreme caution should always be taken to avoid false-positive or false-negative c-kit staining results by carefully observing positive control (mast cells or interstitial cells of Cajal) and negative control (smooth muscle cells or endothelial cells). Because c-kit may also be positive for other soft tissue tumors, interpretation of c-kit based on H&E findings is necessary (Table 1).

CD34 is positive in 60-80% of GISTs, and the frequency of CD34 positivity depends on location of GISTs. The frequency of positive CD34 is high in GISTs of the esophagus and colon (95%), but relatively low in the small bowel and extra-gastrointestinal sites. In the small intestine, CD34 is positive in 50% of cases while c-kit is positive in almost 100% of cases. However, GISTs of the colon can readily be misdiagnosed as other soft tissue tumors such as inflammatory fibroid polyp or inflammatory myofibroblastic tumor, which is attributable to the rare occurrence of GISTs in the colon and greater incidence of negative or focal staining of c-kit in GISTs of the colon relative to other organs. Thus, negativity of c-kit staining does not exclude a possibility of GIST, and every effort should be made to obtain the diagnosis of GISTs through proper differential diagnoses. Protein kinase C (PKC)-theta staining is positive in approximately 90% of GISTs. The quality of PKC-theta staining must be managed by observing ganglion cells of the intermyenteric plexus as an internal positive control and smooth muscle or blood vessel as a negative control. When the staining is properly performed, it can serve as an important adjunct tool in the diagnosis of c-kit negative GISTs, particularly developed in the stomach and extragastrointestinal locations (15). H-caldesmon is positive in 60-80% of GISTs, which may be helpful in the diagnosis of c-kit negative GISTs. Smooth muscle actin is positive in 30-40% of GISTs, and the frequency of the positive staining is high especially in the small bowel. S-100 and desmin is positive in 5% and 1-2% of GISTs, respectively. A recently developed antibody against DOG1 (discovered on GIST) was reported to be superior in sensitivity and specificity to c-kit and CD34. However, c-kit negative GISTs express DOG1 in only 36% of cases, limiting its use in this setting (16). Fig. 3 shows the algorithm of diagnosis in GISTs based on the immunohistochemical staining results.

In addition to the gastrointestinal tract, GISTs are also found in extragastrointestinal sites, although rare. Caution should be taken because histological and immunohistochemical findings of EGISTs are different from those of GISTs and consequently, it may be very difficult to make a diagnosis.

In GISTs, c-kit is negative in ~5% of cases. These c-kit negative GISTs are common in the stomach and omentum/mesentery. In such cases, examining other immunohistochemical markers (PKC-theta, CD34, SMA or DOG1) and mutation analyses may be useful in diagnosis. Among c-kit negative GISTs, 75% are positive for PKC-theta, 44% for CD34, 40% for SMA, and 36% for DOG1 (14, 15, 17-19).

Morphologic risk assessment in GISTs provides the basis for clinical management and optimal patient care. The vast majority of studies of GISTs suggest that the two most important prognostic features to assess the risk of aggressive behavior in a primary localized GIST are mitotic activity and tumor size. These two features were the foundation of the consensus approach for risk assessment in GISTs published by Fletcher and colleagues in 2002 (13). Subsequent data collected by Miettinen and colleagues (20), analyzing large series of GISTs, confirmed that tumor size and mitotic activity are essential prognostic parameters; they proposed additional new parameter, location of tumor, in the evaluation of the clinical behavior of localized GISTs, and KGSG adopted this risk stratification with slight modification (Table 2) (12).

It has been suggested that one block per centimeter should be examined histologically. The pathology report should include the size of the tumor, mitosis (per 50 high powered fields [HPF]), location, resection margin status, and the presence or absence of metastases. Presence of metastasis or perforation during operation leads to a diagnosis of malignant GIST. Pathologic report may include histological type including the degree of cellularity and atypia, presence of necrosis or cystic change, invasion into mucosa or adjacent structures.

At present, mutation analysis is not required for the diagnosis of GIST when tumors have a typical histology and immunohistochemical staining pattern. However, because the presence and location of mutations in KIT or PDGFRA can have implications for prognosis and management in patients with advanced disease, mutation analysis should be considered at the time of diagnosis. Mutational analysis for KIT exons 9, 11, 13, and 17 or PDGFRA exons 12, 14, and 18 can be performed with unstained slides from formalin-fixed paraffin-embedded tissue or fresh frozen tissue.

The main treatment of resectable localized GIST is surgery. The goal is complete resection without residual tumor cells (R0).

The initial diagnosis is generally made by endoscopy, endoscopic ultrasound, gastrography, or computed tomography (CT) of the abdomen due to difficulty with obtaining adequate tissues. It should be confirmed by pathologic histological findings after resection. Preoperative histological diagnosis is feasible, but it may be difficult to interpret definitively (21-24). Imaging tests to detect metastasis include chest radiography (or chest CT), triphasic CT of the abdomen and pelvis, and/or magnetic resonance imaging (MRI) if necessary (25). Positron emission tomography (PET) may be performed when evidence of metastasis may be equivocal or for clinical trials (21).

There is no consensus regarding the need of endoscopic ultrasound biopsy or percutaneous biopsy for preoperative diagnosis. The important part of histological diagnosis is not to cause tumor seeding during biopsy. Therefore, unless multiple metastases are present, excisional biopsy with laparotomy is suggested (21, 26). If diagnosis is unknown at the time of resection, post-operative frozen tissue examination must be performed in order to elucidate the treatment strategy for GIST as treatment varies for an adenocarcinoma or lymphoma. Biopsy is necessary when planning neoadjuvant therapy.

Due to the high potential for malignancy of GIST, resection should be the first-line treatment (21, 27). We strongly recommend resection for tumors larger than 2 cm or growing tumors (21). Smaller tumors (<2 cm) confer a lower potential for malignancy and may be observed. However, small tumor size does not exclude the potential for malignant transformation. Therefore, patients should be informed about the possibility of malignancy.

The main objectives of surgical treatment are to acquire negative margins and to resect without causing tumor rupture. In case of inadvertent tumor infiltration into the surrounding organs, a complete en bloc resection with negative margins should be performed (21, 26, 27) regardless of size. Therefore, even tumors are small, endoscopic shell-out procedure or enucleation should be avoided if GIST is suspected. In many cases, wedge resection of gastric GIST and segmental resection of small bowel GIST are appropriate treatments. Subtotal or total gastrectomy may be performed based on size and location. We recommend en bloc resection for omental or mesenteric GIST. Adjacent organs adherent to tumor should also be completely resected en bloc to avoid tumor rupture or intraabdominal seeding (26).

Laparoscopic resection is feasible if intraabdominal tumor rupture or seeding is unlikely. Laparoscopic resection should follow principles of oncologic surgery. Generally, it is reserved for small, favorably located gastric GISTs (28-31). Intra-operative endoscopy or laparoscopic ultrasound may be used to assist in laparoscopic resection if needed.

Unlike adenocarcinoma, GIST rarely metastasizes to local regional lymph nodes. Therefore, lymphadenectomy is warranted only if metastasis is suspected, i.e. enlarged lymph nodes.

For patients in the high- or intermediate-risk group, we recommend follow-up with CT of the abdomen and pelvis every 3 to 4 months for the first 3 yr after surgery and then every 6 months until 5 yr; then annually thereafter (21, 26). For patients in the low or very low risk, we recommend follow-up with CT every 6 months for 5 yr. Ultrasonography may replace CT once a year (21, 26).

The role of PET for this purpose is not established, and clinical studies on the role of PET are ongoing. Most recurrences occur within 2 yr after surgery and the liver and peritoneum are the most common sites of recurrence (32). Due to the high incidence of gastric cancer in Korea, the National Cancer Screening program recommends biennial stomach-cancer screening for men and women older than 40 yr with endoscopy or upper-gastrointestinal series.

Medical treatment with imatinib alone usually does not result in complete response in metastatic GIST. And, responses are not usually maintained indefinitely. It is now well recognized that clones of tumor cells resistant to imatinib develop continuously over time after start of imatinib. Therefore, in cases where partial or stable responses are shown after adequate duration of imatinib therapy (usually 4 to 12 months of treatment), complete surgical resection of residual tumor may be considered to reduce the risk of development of resistant clones by eradicating the residual viable tumor cells. Several retrospective studies suggested that resection of residual lesions could prolong progression-free survival if it is done during the tumors are under control with imatinib (either in PR or SD) (33, 34). It is also emphasized that imatinib should be continued after resection (21). However, the role of resection of residual tumors after imatinib therapy has not been established, and several phase III clinical trials to investigate the role of surgical resection in this setting are ongoing or planned worldwide. Hepatic or peritoneal metastases may be locally treated with radiofrequency ablation (RFA) or chemoembolization. Importantly, management of GIST should be coordinated by a multidisciplinary team of experienced medical and surgical specialists.

Adjuvant treatment with imatinib is given to enhance the possibility of cure by eradicating microscopic lesions that might still be present after complete resection of visual tumors. Given the great efficacy of imatinib on metastatic or recurrent GIST, imatinib appears to have a sufficient potential as adjuvant treatment.

In recently published ACOSOG Z9001 study of patients with tumor diameter ≥3 cm who received imatinib for 1 yr following complete resection, imatinib demonstrated a significant increase in recurrence-free survival (RFS) compared with placebo although improvement in overall survival (OS) not observed (35). The benefit in RFS appears to be related to tumor size, with the most marked improvement in patients who had large tumor (≥10 cm) in the study.

A Korean phase II study, recently presented in an abstract, evaluated the efficacy of 2-yr imatinib adjuvant therapy in patients who were at high risk of recurrence (tumor size ≥5 cm and mitotic index ≥5/50 HPF, tumor size ≥10 cm, or mitotic index ≥10/50 HPF) based on the NIH risk criteria and had KIT exon 11 mutation which had been recognized as an independent poor risk factor of recurrence in a Korean retrospective study (36, 37). The recurrence-free survival of these patients looked much improved compared with historic data.

These results strongly suggest that adjuvant imatinib therapy should be recommended to reduce the risk of recurrence after curative resection of localized GISTs. However, there are several issues to solve with regard to adjuvant treatment with imatinib. In the ACOSOG Z9001 trial, rapid disease recurrence was observed after stopping 1 yr of imatinib adjuvant treatment. The study did not demonstrate a benefit of adjuvant imatinib in terms of overall survival. Adjuvant treatment with imatinib may delay recurrence. But, it may not increase the cure rate or eliminate the disease in patients with metastatic GISTs; therefore the majority of them eventually develop resistance to imatinib, leading to treatment failure despite its great efficacy. Although currently it remains yet to be determined whether adjuvant imatinib improves cure rate or just delays recurrence, adjuvant imatinib can be proposed as an option for patients at a substantial risk of relapse. Only tumor size and mitotic count were included in the 2002 NIH Consensus risk classification. More recent risk classifications incorporate primary tumor site and/or tumor rupture in addition to the tumor size and mitotic count. Current consensus of the experts is to recommend adjuvant imatinib for patients at high risk of relapse and not for those at low risk of relapse. But, there is no consensus for the patients at intermediate risk (20, 38). Patients with tumor rupture should be considered as having high likelihood of micrometastasis and treated with imatinib. In addition to the risk assessment, mutational analysis may be helpful for the selection of patients who are not likely to get benefit from the treatment. For example, GIST with D842V mutation in PDGFRα exon 18 is known to be not responsive to imatinib. With the currently available data, it is recommended to use imatinib as an adjuvant treatment for at least one year. And, most of the experts agree that 1 yr of adjuvant imatinib is not long enough for especially patients at high risk. However, optimal duration of adjuvant imatinib remains yet to be determined. Phase III trials comparing one year versus three years and no treatment versus two years of treatment duration are ongoing.

Outside of a clinical trial, neoadjuvant treatment with imatinib is not recommended unless there are clinically significant grounds to improve results of surgery by downsizing tumors with neoadjuvant therapy at the initial diagnosis (21). However, neoadjuvant treatment with imatinib may be considered if R0 resection is not feasible, for the purpose of preserving organ functions in GISTs of the rectum, esophagus, and duodenum, or for complete resection of gastric GISTs accompanied by severe, local infiltration into the pancreas or duodenum (39-41). When such a neoadjuvant treatment is considered, progression and response of tumors before and during the treatment should be assessed very carefully by CT and/or PET scan, and it should be done by an experienced multidisciplinary team. Early assessment of tumor response by CT and/or PET is recommended so that surgery is not delayed in the case of non-responding tumors. Duration of neoadjuvant therapy with imatinib may vary according to response to the treatment, but surgery should be performed after sufficient shrinkage of tumors is observed (typically after 4 to 6 months and within 12 months of imatinib treatment) (21). Mutational analysis might be helpful to exclude GISTs not sensitive to imatinib.