Recent preclinical study has suggested a potential possibility that imatinib might promote tumor growth in the presence of secondary resistance mutations [10]. This result imply that intermittent dosing schedule of imatinib rechallenge might be better than continuous dosing schedule in terms of controlling tumors harboring secondary resistance mutations. In addition, in these heavily pretreated patients, even mild grade of toxicity may significantly impair quality of life, and intermittent dosing schedule may have an advantage in this context. Therefore, investigators hypothesize that intermittent dosing schedule of imatinib rechallenge might be feasible and effective in patients with TKI-refractory GISTs. This study will assess the feasibility of intermittent imatinib dosing schedule in patients with GISTs who had failures from both imatinib and sunitinib.

Patients will be randomly assigned to an imatinib arm with either intermittent or continuous dosing schedule with a ratio of 1:1 by using a computer-based system. Imatinib will be administered at a dose of 400 mg/day, once a day with food, in the form of 100-mg tablets. Patients assigned to the continuous dosing arm will receive imatinib without off-schedule, and those assigned to the intermittent dosing arm will received imatinib with one-week on/one-week off dosing schedule. Four weeks of study treatment is considered as one cycle for both continuous and intermittent dosing schedules.

In both arms, the imatinib treatment beyond multiple progressions defined by RECIST version 1.1 is permitted, unless treating physician decided that there is no clinical benefit with imatinib. Imatinib will be discontinued when unacceptable toxicity or patient's withdrawal of consent.

Inclusion Criteria:

• Patients aged 19 years and older
• Patients with metastatic or unresectable GIST which has been histologically confirmed by the detection of CD117 on immunohistochemical staining or genetically confirmed by the detection of mutation in KIT or PDGFRα genes on direct sequencing of tumor DNA.
• Prior clinical benefit from 1st line imatinib defined as CR, PR, or SD at 6 months after the start of 1st line imatinib
• Patients whose disease has progressed with at least both prior imatinib (400mg/day) and sunitinib therapy.
• Eastern Cooperative Oncology Group (ECOG) performance status 0 ~ 3
• Adequate bone marrow function as defined by platelets ≥ 75 x 109/L and neutrophils ≥ 1.5 x 109/L
• Adequate renal function, with serum creatinine < 1.5 x upper limit of normal (ULN)
• Adequate hepatic function with serum total bilirubin < 1.5 x ULN, alanine aminotransferase (ALT) or aspartate aminotransferase (AST) < 2.5 x ULN in the absence of liver metastases, or < 5 x UNL in the presence of liver metastases.
• Expected life expectancy of greater than 12 weeks in the absence of any intervention
• No other malignant disease apart from non-melanotic skin cancer or carcinoma in situ of the uterine cervix or any other cancer except where treated with curative intent > 5 years previously without evidence of relapse
• Written informed consent to the study

Exclusion Criteria:

• Medical or psychiatric conditions that compromise the patient's ability to give informed consent or to complete the protocol or a history of non-compliance
• Last dose of radiotherapy received within 4 weeks before the start of study treatment, excluding palliative radiotherapy
• Obstruction of gastrointestinal tract
• Active gastrointestinal bleeding
• Myocardial infarction within 6 months prior to the study medication, and other clinically significant heart disease (e.g., unstable angina, congestive heart failure or uncontrolled hypertension)
• Evidence of severe or uncontrolled systemic disease or any concurrent condition which in the investigator's opinion makes it undesirable for the patient to participate in the study or which would jeopardise compliance with the protocol
• Female patients who are pregnant or breast-feeding. Female patients must have had a negative pregnancy test within one week before starting imatinib.