Traditionally, Response Evaluation Criteria in Solid Tumors (RECIST), which are based on changes in tumor size, have been universally used to measure tumor response following systemic treatment [26, 27] (Table 1). GIST responding to a targeted agent demonstrate homogeneous hypoattenuation, resolution of enhancing tumor nodules, and decreased tumor vascularity on contrast-enhanced CT compared with heterogeneous, hyperattenuating, enhancing lesions visualized on pre-treatment CT (Fig. 2).

CT imaging features correlate with pathologic change, characterized by necrosis with decreased cellularity inside the tumor, myxohyaline degeneration, and pseudocyst formation [28]. These changes can be observed 1 or 2 months after starting treatment. GIST do decrease in size, but the median tumor shrinkage required to satisfy the partial response (PR) criteria by RECIST can take 3–4 months or longer [28]. Moreover, intratumoral hemorrhage or myxoid degenerative change in responding tumors can result in an increase in tumor size (Fig. 3). This paradoxical response can be mistaken for progression of disease (pseudoprogression) if only tumor size is considered when evaluating the treatment response. Similarly, GIST can progress with the development of new intratumoral nodules within the responding tumor without changing the overall tumor size [29]. This nodule-within-a-mass pattern can result in underestimation of disease progression while the tumor size remains stable (Fig. 4).

New CT response evaluation criteria were proposed by Choi et al. to address the issues arising from the use of traditional size-based criteria in GIST [30] (Table 1). The Choi criteria incorporate changes in tumor size as well as in CT density, reflecting the morphologic changes in CT characteristics, to evaluate treatment response. These changes include a 10 % decrease in the sum of tumor size of the unidimensional tumor size or a 15 % decrease in tumor density, as determined by the CT attenuation coefficient in Hounsfield units (HUs), at the first follow-up (2 months) (Table 1). The Choi criteria have been shown to correlate well with the responses noted by positron emission tomography (PET) and can best categorize patients into good responders and poor responders, and the category of response is an excellent predictor of progression-free survival [31].

Caution is needed if intratumoral hemorrhage or calcification occurs in responding tumors following treatment, as these developments could increase tumor density. Unenhanced images may be helpful in overcoming misinterpretation of the findings in this setting (Fig. 5).

Following imatinib treatment, fluid overload can occur as a side effect of the treatment. On imaging, fluid overload can present as ascites, pleural effusion, pericardial effusion, or edema. Such fluid overload should not be mistaken for progression of peritoneal disease [21].

Once the tumors respond to treatment, the role of imaging is to identify disease recurrence and progression in a timely manner. The tumor’s development of resistance to the treatment is believed to be responsible for the recurrence [32].

Disease progression/recurrence in GIST can be detected by an increase in tumor size, increase in tumor density (enhancement on imaging), appearance of new lesions, presence of distant metastasis, or emergence of a new, intratumoral enhancing nodule.

GIST can recur following margin-free resection, particularly within the first 5 years after treatment [33]. For GIST patients with a high risk of recurrence (e.g., tumor size larger than 5 cm, high mitotic count of more than 5/50 high-power fields), close follow-up with CT imaging is recommended at a 3- to 4-month interval for the first 3 years, twice yearly for up to 5 years, and then annually thereafter.

For GIST patients at low risk for recurrence, CT follow-up twice a year for 5 years is recommended [34], although no standard surveillance protocol has been established in this group.

In cases of inconclusive CT findings, or if imaging findings cannot support the clinical findings, PET imaging can be useful.

The use of volumetric measurement on CT images has been explored (Table 1). Schiavon et al. reported that the three-dimensional (3D) ellipsoidal model was more sensitive for identifying PR than RECIST was, and 3D criteria were also useful for predicting overall survival [35].

The effect of a new antivascular agent can be monitored using CT perfusion, a fast imaging technique after administration of the iodinated contrast agent that provides quantitative tissue perfusion information such as blood flow, blood volume, and permeability [36]. In GIST, imatinib was postulated to have antivascular activity and to induce tumor apoptosis, which was correlated with a decrease in perfusion CT parameters such as blood volume and blood flow in PET responders [37]. Schlemmer et al. studied the value of perfusion CT on metastatic GIST patients treated with sunitinib or imatinib. Good responders based on Choi criteria were compared with poor responders and showed decreased perfusion parameters such as volume of distribution, blood flow, blood volume, and permeability. This distinct tendency between good responders and poor responders was also observed in hepatic lesions [38]. Although further validation of these findings is needed in a large group of patients, perfusion CT appears to have potential as a functional imaging modality to assist in more accurate assessment of treatment response.

Recently, the role of dual-energy CT (DECT) has been explored in GIST [39, 40]. The use of iodine-related attenuation has shown good correlation with the Choi criteria [39], and new response criteria have been proposed as a potential novel predictor of clinical outcome [40]. Moreover, DECT has a benefit in the possible elimination of unenhanced CT in protocol, reducing radiation exposure to the patient. However, one should keep in mind that this new technique does have technical limitations, especially in obese patients or patients with a large abdomen, and DECT is still in its developmental stage [41].

PET is a metabolic imaging tool, measuring glucose metabolism using fludeoxyglucose (FDG). Once transported into the intracellular space, the FDG is trapped within the cell without being used as a cellular energy source like glucose is.

Malignant cells usually have increased glycolysis, which presents as increased FDG uptake on imaging. FDG uptake can be evaluated subjectively by visual analysis, semiquantitatively by measuring the standardized uptake value (SUV), and quantitatively by calculating the absolute rate of cellular metabolism using a kinetic model on dynamic sequence [42]. Currently, maximum SUV (SUVmax) is the most used due to its universal availability and semi-quantitative nature.

Combining PET with contrast-enhanced CT (PET-CT) and taking advantage of the features of contrast-enhanced CT can improve the accuracy of tumor detection, tumor characterization, and tumor localization [43]. Combining contrast-enhanced CT with PET can be technically challenging and is not yet universally available, but its use has been increasing.