The liver is the site of metastasis in approximately 65 % of patients with relapsed GIST [83]. A few studies have examined outcomes following resection of liver metastases prior to the imatinib era. DeMatteo et al. [84] looked at 34 patients with either GIST or gastrointestinal leiomyosarcoma who underwent complete resection of hepatic metastases and found a median survival of 38 months, with 30 % of patients alive at 5 years. Nunobe et al. [85] found similar results in their series of 18 patients who underwent hepatectomy for metastatic GIST with a median survival of 36 months and 5-year survival of 34 %. Finally, Shima et al. [86] showed in a series of ten patients a median survival of 39 months following hepatectomy. Of note, de la Fuente et al. [87] compared 43 patients with isolated liver metastases (34 underwent surgery) to 16 patients with liver and peritoneal disease (13 underwent surgery). Patients with isolated liver metastases undergoing surgery trended toward longer OS (40.5 months vs 28.7, p = 0.620) and trended toward having a lower recurrence rate after surgery (16/34 vs 8/13 p = .08), suggesting having liver-only disease may be a predictor of slightly better prognosis. This was further confirmed in long-term follow-up of EORTC 62005 study [72]. In patients with only liver metastases, median OS was not reached compared to 7 years in patients with only peritoneal metastases and 3.7 in patients with both. Interestingly, in the group of patients who were operated on while in remission and achieved a complete macroscopic resection, median OS was not reached again in liver-only group, 8.7 years in peritoneal-only group, and increased to 8.1 years in patients with both.

After the advent of imatinib, it has been shown that the combination of imatinib and hepatic resection can offer long-term disease control in patients with isolated hepatic metastases in multiple studies (Table 2).

Turley et al. [90] studied 39 patients that underwent hepatic resection for metastatic GIST between 1995 and 2010. Thirty-one patients received TKI therapy. Three-year OS in patients receiving TKI was 71.9 % and 0 % in the seven patients who did not receive TKI therapy. They also noted that patients who were exposed to TKI therapy for prolonged periods (median 18 months) preoperatively trended toward worsened OS, suggesting against delaying surgery indefinitely. Their median OS exceeded previous reports for treatment of metastatic GIST with hepatic resection alone of 36–47 months [84–86, 88].

Xia et al. [89]reported 39 patients with metastatic GIST to the liver treated with either 6 months of neoadjuvant imatinib followed by surgery and 2–4 weeks of adjuvant therapy or imatinib alone. The 3-year OS was 89.5 % in the surgery group and 60 % in the imatinib-alone group, which was significant. Also significant was that among patients who responded poorly to 6 months of preoperative imatinib, surgery significantly improved OS compared to those that did not undergo surgery (p = 0.04)

Zhu et al. [93] studied 42 patients with recurrent GIST to either the liver or abdomen treated with long-term imatinib alone without surgery. They found a combined median OS of 48 months. Notably, the median time to progression was longer in the liver group (48 months versus 39 months in patients with only abdominal metastases and 33 months in patients with both), but this was not significant. Median OS was not reached in the liver-alone group with three of the ten patients dying during over 3 years of follow-up. This showed that imatinib alone is a reasonable option in patients with hepatic metastases as well as other metastatic disease.

The NCCN [19] and ESMO [94] both recommend indefinite adjuvant imatinib for patients with resected liver metastases, even if resection was complete.

There presently is no standardized follow-up regimen for patients with GIST. The NCCN guidelines [19] recommend an abdominal/pelvic CT be obtained every 3–6 months after surgical resection of GIST and within 3 months after initiating TKI therapy for patients with advanced GIST. ESMO guidelines [94] also recommend follow-up CT or MRI every 3–6 months in patients receiving adjuvant therapy for the first 3 years. Upon cessation of adjuvant therapy, ESMO further recommends follow-up imaging every 3 months for the first 2 years followed by less frequent imaging if stable.

CT scans are the typical initial imaging modality of choice. GISTs typically appear as a solid contoured mass that enhances brightly with IV contrast on CT scans. Larger tumors can be less homogenous due to hemorrhage and necrosis within the tumor. MRI can be useful in patients that cannot receive IV contrast for CT scans. MRIs are also better at evaluating GISTs in the liver and rectum [95], which is especially important ahead of a planned surgery.

PET scans are highly sensitive for detecting GISTs, but lack specificity. While not a stand-alone imaging modality for surveillance, PET can be used for detecting an unknown primary site or resolving ambiguities from CT [16]. Another potential scenario where PET can be considered is when response to therapy must be determined quickly. Specifically, TKI responsiveness can be seen as early as 1 day after treatment is started on PET scan compared to 1–2 months on standard CT scans [96–98].

GISTs that respond to TKI therapy become more homogenous and hypodense, and sometimes will later shrink in size as well. This is important when considering the two major criteria systems for GIST follow-up: RECIST and Choi. RECIST (Response Evaluation Criteria In Solid Tumors) was developed first. which determines treatment response based upon tumor measurements [99].However, it was known that this was not necessarily well-suited for GIST surveillance, as early response to TKI therapy often does not correlate with decrease in tumor size, and further signs of tumor progression on TKI therapy are often seen as new areas of hyperdensity before an increase in size. This led Choi et al. [100] to develop a different response evaluation system that used both tumor density and size (Table 3).