Surgery Prolongs the Survival of Patients with Metastatic, Recurrent, or Unresectable Locally Advanced Gastrointestinal Stromal Tumors Response to Imatinib Mesylate Treatment: A Systematic Review and Meta-analysis

Background The role of surgery in patients with metastatic, recurrent or unresectable locally advanced gastrointestinal stromal tumors (GIST) who respond to imatinib mesylate (IM) treatment is still not formally dened. Therefore, we systemically searched and analyzed the available literature to evaluate the oncologic benets of surgery in this specic population. Methods A systematic literature search of the PubMed, Embase, and Cochrane Library databases was performed to identify relevant articles on July 16, 2020. Pooled data analysis was also performed using Review Manager. Totally 10 studies including 1188 patients (410 patients in the surgery group and 778 patients in the no surgery group) were included in the nal analysis. No signicant differences in baseline clinical characteristics were found except that patients in the surgery group were signicantly younger (WMD, -5.02, 95% CI, -8.38 to -1.67, P = 0.003). In the overall population, pooled data showed a signicant improvement in overall survival (OS) (HR, 0.62; 95% CI, 0.53 to 0.73; P < 0.0001) and progression-free survival (PFS) (HR, 0.57; 95% CI, 0.44 to 0.72; P < 0.0001) with surgery. In the subgroup analysis, the impact of surgery on patient response to IM treatment was further conrmed (OS: HR, 0.67; 95% CI, 0.55 to 0.81; P < 0.0001; PFS: HR, 0.62; 95% CI, 0.46 to 0.82; P = 0.009).


Introduction
Gastrointestinal stromal tumors (GIST) are the most common mesenchymal malignancies of the digestive tract, originating from the interstitial cells of Cajal. 1 GIST compose approximately 20% of soft tissue sarcomas, and the annual incidence of these tumors is 11-20 per million people. 2,3 It is estimated that at the time of diagnosis, 15%-50% of patients develop metastatic disease. For patients with locally advanced GIST after undergoing complete resection, nearly 40% suffer recurrence within 2 years. [4][5][6] Up to 90% of GIST have gain-of-function mutations in either KIT or platelet-derived growth factor receptor alpha (PDGFRA) receptor tyrosine kinases, with exon 11 of KIT being the most common mutation site. 7,8 Since the introduction of imatinib mesylate (IM), a low-molecular tyrosine kinase inhibitor that blocks the kinase activity of both KIT and PDGFRA, the outcomes of patients with GIST have greatly improved. 9,10 A 400-mg daily dosage of IM has become the rst-line treatment for metastatic, recurrent or unresectable locally advanced GIST patients, achieving an objective response rate of approximately 80% and a median progression-free survival (PFS) of 33 months in responding patients. 11,12 Although up to 80% of patients exhibit an initial response to IM treatment, 13,14 some may eventually become resistant to IM, mainly due to the acquisition of second mutations, such as exons 13,14,17 in KIT. 15 The role of surgery in patients with advanced GIST is still not formally de ned. In 2003, Heinrich and colleagues assumed that removing residual disease can extend the time of response by reducing the risk of acquiring secondary mutations in imatinib-sensitive patients. 16 In recent decades, although several single-institution series [17][18][19][20][21][22][23] have shown that surgery can be safely performed on metastatic or recurrent GIST patients in response to IM treatment, resulting in a better oncologic outcome, most of them were not comparative studies. Two randomized clinical trials (NCT00956072 in Europe and ChiCTR-TRC-00000244 in China) failed to draw a rm conclusion regarding whether surgery prolongs the survival of patients in response to IM treatment because of failing to meet target accrual. Therefore, we systemically searched and analyzed the available literature to evaluate the oncologic bene ts of surgery in patients with metastatic, recurrent, or unresectable locally advanced GIST who respond to IM treatment.

Methods
We performed a systematic review and meta-analysis of relevant comparative retrospective, prospective studies and randomized controlled trials according to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines.

Study selection
A systematic literature search of the PubMed, Embase, and Cochrane library databases was performed to identify relevant articles on July 16, 2020, with the following search terms: ("gastrointestinal stromal tumor*" OR "GIST*") AND ("surgery" OR "resection" OR "excision" OR "cytoreduction") AND ("recurrent" OR "metastatic" OR "advanced" OR "unresectable") AND ("imatinib mesylate" OR "imatinib"). Searches were limited to human studies and English-language publications. The reference lists of all retrieved studies and review articles were also searched manually. In the case of repeated studies by the same authors, the most recent or complete report was included.

Inclusion and exclusion criteria
The following inclusion criteria were used: (1) histologically con rmed CD117-positive, metastatic, recurrent, or unresectable locally advanced gastrointestinal stromal tumors; (2) treatment with 400 mg/day IM (600-800 mg/day for patients with KIT exon 9 mutation) administered for at least 3 months and evaluated according to the Response Evaluation Criteria in Solid Tumors (RECIST); and (3) at least one of the oncologic outcomes (overall survival (OS) or PFS) mentioned in the main assessment results. OS was measured from the date of initiation of 400 mg/day IM to the date of all-cause death. PFS was measured from the date of initiation of 400 mg/day IM to the date of disease progression or tumor recurrence with 400 mg/day IM or all-cause death with or without surgical intervention. Conference abstracts, reviews, letters, case reports, and noncomparative studies were excluded.
Data extraction and quality assessment Data from the selected studies were extracted and summarized independently by two investigators (WH Li and XY Li), and any disagreements were discussed and resolved among all authors. The extracted data included basic information ( rst author, year of publication, study design, sample size, follow-up period, extension of disease, whether underwent surgery), clinicopathological characteristics (gender, age, tumor size, primary and metastatic tumor site, genotype of primary tumor, response to rst-line imatinib, R0 resection rate), and oncologic outcomes (OS, PFS). If survival was not reported in the text, it was independently calculated from survival curves by two investigators (WH Li and JH Peng), or we contacted the corresponding authors via email to ask for relevant data.
The quality of all observational studies included was assessed with the Newcastle-Ottawa Scale (NOS) by examining three factors: method of patient selection, comparability of the study groups, and number of outcomes reported. A score of 0-9 was allocated to each study based on these three parameters.
Studies achieving six or more stars were considered to be of high quality. One randomized controlled trial was considered to be of high quality for the purposes of this study.

Statistical analysis
All meta-analyses were performed using Review Manager 5.4 (Cochrane Collaboration, Oxford, UK). For continuous variables, we used the weight mean difference (WMD) to analyze and for dichotomous variables, we used the the odds ratio (OR) to analyze. Both of them were reported with 95% con dence intervals (CIs). If the results were expressed as medians with ranges, the standard deviations were calculated using the method described by Hozo and colleagues. 24 The hazard ratio (HR) with a 95% CI was applied for comparison of the oncologic outcomes.
Heterogeneity between studies was assessed by the chi-square test and I 2 statistic, in which a P value < 0.10 or I 2 > 50% was de ned as substantial heterogeneity. The random effects model was used if there was heterogeneity between studies; otherwise, the xed effects model was used. Subgroup analyses were performed focusing on the patients who respond to IM treatment. Funnel plots were used to evaluate potential publication bias. Sensitivity analysis was performed to evaluate the robustness of our outcomes by removing low-quality studies. A two-tailed P value less than 0.05 was considered signi cant if the 95% CI did not include the value 1.

Study selection and quality
A total of 10 studies 25-34 including 1119 patients (361 patients in the surgery group and 758 patients in the no surgery group) met the eligibility criteria and were included in the nal analysis ( Figure 1). Manual searches of the references lists of the included studies did not yield any further studies. The quality of the included studies ranged from 3 to 7 on the NOS. The detailed quality assessment and scores for each study are provided in supplementary material Table S1.

Patient characteristics
The main features and demographic data of the studies included are shown in Table 1. Among these 10 studies, 7 studies were aimed at patient response to IM treatment, and the other 3 studies contained a small number of patients with progressive disease (PD). A total of 1089 patients (97.32%) with good response (complete response or partial response) or stable disease and 30 patients (2.68%) with PD were included in analysis, with an R0 resection rate ranging from 41.1% to 100%. Notes: a Extension of disease: 1 = Metastasis, 2 = Local recurrence, 3 = Locally advanced initially unresectable.
b Mean or median. c This study focused on patients response to IM treatment without patient with progressive disease. However, detailed information is not available. Table 2, patient clinical characteristics were compared between the surgery group and the no surgery group. Patients in the surgery group were signi cantly younger than patients in the no surgery group (WMD, -5.02, 95% CI, -8.38 to -1.67, P = 0.003). There were no signi cant differences observed regarding gender, tumor size, primary and metastatic tumor site, genotype of primary tumor, or response to rst-line imatinib between the two groups.

Subgroup analysis
Subgroup analyses were performed to further investigate the impact of surgery on patients who respond to IM treatment. Studies that contain patients with PD were excluded. A total of 7 studies 25,[28][29][30][31][32]34 including 863 patients (240 patients in the surgery group and 621 patients in the no surgery group) were included in the subgroup analyses. Subgroup analysis results were consistent with the overall results (OS: HR, 0.67; 95% CI, 0.55 to 0.81; P < 0.0001; Figure 3A; PFS: HR, 0.62; 95% CI, 0.46 to 0.82; P = 0.009; Figure  3B).

Sensitivity analysis and publication bias
One randomized control trial 28 and 4 retrospective studies 27,30,33,34 with scores of six or higher on the NOS were included in the sensitivity analysis. The results related to OS and PFS did not changed signi cantly in the sensitivity analysis (shown in supplementary material Figure S1 and Figure S2).
As shown in Figure 4, a funnel plot was constructed for the studies included in this meta-analysis that reported OS. All studies lie inside the 95% CIs, with an even distribution around the vertical axis, indicating no obvious publication bias.

Discussion
In this meta-analysis, we investigated the oncologic outcomes of surgery for patients with metastatic, recurrent, or unresectable locally advanced GIST response to IM treatment. Our pooled data analyses and subgroup analyses showed that surgery prolongs the OS and PFS of patients in response to IM. In the sensitivity analysis of high-quality studies, there was no change in terms of the signi cance of OS and PFS. We found no signi cant differences in baseline clinical characteristics between the two groups except that patients in the surgery group were signi cantly younger. To our knowledge, this is the rst meta-analysis to systematically compare the oncologic outcomes of surgery plus IM treatment with IM treatment alone in patients with metastatic, recurrent, or unresectable locally advanced GIST.
The potential bene t mechanism of surgery for patients with GIST after receiving IM treatment has been widely discussed. One of the most popular points of view is that surgery prolongs the effects of IM and prolongs survival time by removing residual disease before secondary resistance develops or by halting disease progression by eliminating resistant clones. 16 On the other hand, the ultimate goal of treatment of unresectable locally advanced or metastatic patients through active multidisciplinary therapy (MDT) is to achieve status of no evidence of disease (NED), of which R0 resection surgical intervention is an essential part to achieve it. Without subsequent surgical treatment, tumor progression becomes inevitable because the effect of IM on these residual tumor cells is cytostatic rather than cytotoxic. 7 Due to the natural limitation of selection bias, in previous retrospective studies, 25,30 patients in the surgery group tended to have signi cantly better tolerance for traumatic treatment (patients were younger or had better performance status by the Eastern Cooperative Oncology Group criteria) and a higher rate of complete response or partial response. Additionally, a higher proportion of patients have primary lesions in the stomach, metastatic lesions in the liver and KIT exon 11 mutation genotype, meaning that more patients in the surgery group are at low risk. In the present meta-analysis, no signi cant differences in baseline clinical characteristics between the two groups were found except that patients in the surgery group were signi cantly younger, which means that the selection bias of retrospective studies included has been minimized, making the results more valuable and credible.
The role of surgery in patients with metastatic, recurrent, or unresectable locally advanced GIST has been widely examined, and there is a consensus that those who respond to IM treatment bene t more from surgery than those with PD. 23,35−37 Obviously, due to the reduction of the operation di culty and the scope of resection, patients with responsive disease (RD) will have a higher R0 resection rate than those with PD, which has been revealed to be associated with improved long-term survival. 38,39 A multiinstitutional study by Bauer et al. 39 found that patients with metastatic GIST treated with IM may bene t from surgery if complete resection can be achieved. Otherwise, surgery does not provide a survival bene t. In the present study, 30 patients (2.68%) with PD were included in the overall population, which may affect the accuracy of the results, but it was considered to be slight and acceptable. After removing the studies that contain patients with PD in the subgroup analysis, the result that surgery provides a signi cant improvement in OS and PFS in patients with RD was further con rmed.
Several limitations should be acknowledged in the present meta-analysis. First, with only one RCT exception, all the studies included were observational and were carried out in different clinical centers with varying protocols and different levels of surgical expertise. Second, the limited sample size might weaken the accuracy and representativeness of our ndings. The studies included in the analysis were mostly performed at tertiary cancer centers and major institutions, which might not re ect patient populations in the community. Third, statistical heterogeneity existed in some results, such as age, sites of metastasis, response to rst-line imatinib and PFS. Calculations using the random effects model for overall incidence estimation might minimize but do not abolish this. Future systematic reviews should evaluate more subgroups or focus on speci c groups, such as patients with liver metastasis, when enough literature is available. Although our study con rms the survival bene t of surgery in patients with metastatic, recurrent, or unresectable locally advanced GIST response to IM treatment, the ndings must be validated in a prospective, multicenter clinical trial with a large population in the future.

Conclusion
This meta-analysis indicates that surgery prolongs the OS and PFS of patients with metastatic, recurrent, or unresectable locally advanced GIST response to IM treatment. Future prospective, multicenter RCTs need to be undertaken to fully assess the role of surgery in patients with metastatic, recurrent, or unresectable locally advanced GIST and update the ndings of this analysis.

Declarations
Ethics approval and consent to participate All analyses were based on previously published studies; thus, no ethical approval or patient consent were required.

Consent for publication
Not applicable.