Stratifying patients based on severity and presentation of disease facilitates identification of appropriate therapy for GIST patients with hepatic metastases. Patients can be categorized into the following groups: (1) synchronous isolated liver metastases, (2) recurrence following surgical resection, (3) disseminated disease, and (4) progression on TKI therapy. Each of these categories has varying treatment options and related survival estimates.

Prior to TKI therapy, surgical resection was the mainstay of treatment as cytotoxic chemotherapy has minimal effectiveness. With the advent of imatinib, medical therapy has become first-line standard of care. Initiation of TKI therapy should begin immediately following diagnosis for best response. Rates of radiological response with TKI therapy are approximately 30 % stable disease, 50 % partial response, 5 % complete response, and 15 % progressive disease [16, 17]. RECIST criteria are thought to underestimate treatment response, as there may be reductions in tumor density that are not appreciated using these criteria. Pathological analysis yields similarly poor rates of complete response, as 85–95 % of surgical specimens having residual disease after TKI therapy. Approximately 50–60 % of specimens demonstrate partial response to TKI therapy, which supports initial TKI therapy as a viable treatment strategy [18]. Progression-free survival with imatinib alone is approximately 2 years [19].

Resistance to TKI treatment can be categorized into primary resistance, which is progression of disease with initiation of imatinib therapy, and secondary resistance, a progression of disease following initial radiographic response to imatinib. Primary resistance is observed in an average of 10 % of patients, but KIT 9 and wild-type mutation can have higher rates of resistance (16 and 23 %, respectively) [20–22]. Secondary resistance is more common. Tumors often develop a secondary KIT mutation or have PDGF-R mutations that confer TKI resistance and result in imatinib failure [23]. In these circumstances, increasing the dose of imatinib is typically the first step to improve response. Second-line therapy is currently sunitinib [24], which not only is a multikinase inhibitor that preferentially targets PDGF-R and VEGF-R but also has activity against CD117. Modest response rates are attained with sunitinib salvage therapy, with an improvement in progression-free survival (PFS) with sunitinib versus placebo following imatinib failure (27.3 weeks vs. 6.4 weeks PFS, p ≤0.0001). Escalation of imatinib therapy with increased dosing, as mentioned above, is also a therapeutic option as it was found to be equivalent to sunitinib. This was, however, found to be less effective in patients with exon 9 mutations (14.3 vs. 6.2 months PFS, p = 0.037), which emphasizes the importance of genetic mutation characterization [25]. Third-line therapy consists of regorafenib, which had a marginal increase in progression-free survival of <5 months compared to placebo [15]. Targeted medical therapy should form the backbone of treatment for patients with GIST liver metastases and surgical resection reserved for appropriately selected patients [18].

As there are various medical and less-invasive treatments for metastatic GIST, optimization of the outcome after resection requires appropriate patient selection. The principles of selection are based on tumor biology and patient performance status. Designation of resectable, unresectable, and borderline resectable disease guides surgical approach. In the pre-imatinib era, surgical resection alone was associated with 5-year survival of 30–60 % and a median overall survival of only 16 months following complete resection. Recurrence rates upward of 60 % were noted, with the liver as the most common site [26, 27]. Discovery of imatinib and its effect on GIST affected both surgical decision-making and patient survival, especially in this patient population with advanced disease.

Additional information that is considered includes the disease-free interval, response to medical therapy, presence of extrahepatic disease, the number and size of the metastases, tumor location(s), and the patient level of fitness for surgery. Appropriate contrasted imaging (CT or MRI) of the chest, abdomen, and pelvis is necessary to adequately stage the patient. PET scanning may be useful when extrahepatic metastases are of concern.

Assuming the patient is fit and the resection is technically sound, two factors mainly determine the success of surgical resection of GIST liver metastases: surgical margin status and TKI response. Margin-negative (R0) resection is associated with an improvement in progression-free survival as compared with margin-positive resection (29 months vs. 7 months, p = 0.002) [28]. Overall survival at 1 year is also significantly improved for patients who undergo R0 resection (100 % vs. 37.5 %, p = 0.001) [29]. Preoperative response to TKI therapy portends improved benefit from surgical therapy, and reduced tumor volume may improve surgical margins. NCCN guidelines recommend surgical resection to obtain microscopically negative margins, but do not specify the need for extensive margins [30]. Unlike the typical patient who undergoes hepatic resection for primary and secondary liver malignancies, GIST patients are typically noncirrhotic, cytotoxic-chemotherapy naïve, and more likely have normal background liver parenchyma. As such, they will likely tolerate more extensive hepatic resection if necessary than patients with conditions such as cirrhosis or chemotherapy-associated steatohepatitis.

Preoperative therapy for surgically resectable and borderline resectable metastatic GIST is now a standard approach. TKI therapy should be continued up until the time of surgery with resumption in the postoperative period if clinically indicated. The optimal timing of surgery continues to be an evolving target. Waiting 3–9 months after the initiation of medical therapy is recommended, since this time frame usually represents the period of greatest radiological response [31]. There are multiple studies suggesting that operating during periods of responsive disease or stable disease correlates to improved outcomes compared to progressive disease [29, 32, 33]. Additionally, delay of surgery may result in secondary mutations resulting in resistance to imatinib [16]. To successfully intervene at the time of greatest response, the surgeon must closely follow radiological response. A detailed algorithm for selecting management approach for patients with metastatic GIST is provided in Fig. 4.

The surgical approach to GIST hepatic metastases varies based on the clinical scenario and anatomical considerations of each individual patient. As GIST liver metastases tend to be well circumscribed, there are options for both nonanatomical and anatomical resections. Additionally, when hepatic lesions are present with the primary tumor in place, a combined resection of both primary tumor and hepatic lesions may be possible in appropriately selected patients.

Either an open or minimally invasive approach can be utilized, assuming surgical skill and patient selection are appropriate. Peripheral small lesions may be more conducive to a minimally invasive resection. Nonanatomical liver resections are possible with small peripheral lesions located a distance away from major vascular structures (Fig. 1). Preoperative TKI therapy may reduce tumor volume making more limited resection possible, but complete radiological response does not correspond with cure, and thus liver-directed therapy should still be considered in this circumstance. The operating surgeon must review and compare the original scans to those acquired post-TKI therapy and assess for cystic change, as tumors that are initially solid may become cystic appearing after initiation of TKI therapy (Fig. 5).

For patients with extensive hepatic metastases, preoperative portal vein embolization (PVE) may allow for complete resection by preoperatively increasing the future liver remnant volume. This may apply to a select group of patients with borderline resectable disease. This approach is the accepted clinical practice for colorectal cancer metastases and other primary hepatic malignancies and can be extrapolated to the management of metastatic GIST. The decision for PVE must be made early, as timing maximal hepatic growth after PVE with the best response to TKI therapy is necessary (Fig. 6). Subsequent, repeat resections for recurrences are possible and beneficial in appropriately selected patients with adequate liver remnant size and function [34].

Surgical debulking is also described with some success in management of GIST metastases, primarily for symptom (bleeding, pain, or obstruction) control [35, 36]. Survival estimates for selected patients undergoing hepatic resection in the presence of low-volume peritoneal disease may be similar to those for patients who undergo hepatic resection for liver-only disease (28.7 months vs. 40.5 months; p = 0.620) [37]. Liver transplantation for metastatic GIST has been minimally employed and has limited utility [38].

Adjuvant therapy is recommended for patients with high risk of recurrence, which includes all patients who undergo resection of any kind of metastatic GIST [39]. Evidence supports the notion that suppressive therapy with TKI may be critical as there are descriptions of blossoming lesions when TKIs are stopped [40, 41]. Table 1 provides a summary of results from studies analyzing hepatic resection for GIST.