Surgery plus TKIs therapy for gastrointestinal stromal tumors

The following keyword combinations and their variants were searched in PubMed, Web of Knowledge, and Ovid databases by July 31, 2022: “gastrointestinal stromal tumors,” “recurrence,” “metastasis,” “surgery,” and “TKIs.” The reference lists of relevant studies were manually checked for missing studies.

The eligibility of the identified studies for inclusion in the review was assessed by carefully examining the title, abstract, and keywords of each record retrieved. Studies were limited to studies published in English and Chinese. Clinical studies on any aspect of the comparison between surgery plus TKIs and TKIs alone for the treatment of recurrent or metastatic GISTs were also included. In accordance with the PICOS model [13], the following inclusion criteria were applied: (a) Participants: patients with GISTs; (b) Interventions: patients received surgery plus TKIs therapy; (c) Comparison: patients received TKIs; (d) Outcome: progression-free survival, overall survival; and (e) Study design: retrospective study.

Three co-authors (ZK, GB, and YT) separately evaluated the methodological quality of every study on the basis of the Methodological Index Criteria for Non-Randomized Studies (MINORS) [14]. Variables such as the following were recorded: authors, publication journal and year, number of patients, sex, age, tumor size, use of TKIs, surgical management, overall survival (OS), and progression-free survival (PFS). In case of necessity, we contacted the corresponding authors of the study for more information.

The quality of retrospective studies was estimated in accordance with the Newcastle–Ottawa Scale (NOS) [15]. Study quality assessment evaluated each case-control study based on three elements, including selection, comparability, and exposure outcome conditions. The maximum score for the scale is a total of 9. “Good” is defined as a total score of 7–9, “fair” a total score of 4–6, and “poor” a total score of <4.

A formal meta-analysis of all included studies was performed to compare the survival rates between the surgery plus TKIs group and the TKIs only group. Heterogeneity was analyzed using Q statistics and I² statistics, which measures the extent to which differences between studies can be attributed to inter-study variation rather than chance. P<0.05 for the Q statistic and I²>50% were considered to indicate strong heterogeneity. A fixed effects model was used to calculate pooled estimates for OS and PFS. However, with P < 0.05 and I² > 50% for heterogeneity, a random effects model was used. Survival outcomes such as 3- and 5-year OS rate data were analyzed by dichotomous variables with an estimation of odds ratio (OR) together with a 95% confidence interval (CI). In all the tests, a P value smaller than 0.05 was considered statistically significant. Begg’s and Egger’s tests [16, 17] were used to estimate the publication bias. The statistical analyses in this review were conducted using the Cochrane Collaboration’s Review Manager software (version 5.0) and STATA software (version 14.0).

After a thorough examination, 8 studies [8, 9, 18,19,20,21,22,23] were included in our meta-analysis (Figure 1). The initial search yielded 83 publications. Excluding the duplicates, irrelevant topics, non-original studies, studies without a control group or TKIs alone group or surgery alone group, 54 full-text articles were assessed for qualification. Finally, 8 studies [8, 9, 18,19,20,21,22,23] with therapies—TKIs plus surgery and TKIs alone in patients with recurrent or metastatic GISTs—were eligible and included in our meta-analysis (Fig. 1).

Figure 1.

A total of 842 patients were recruited for these eight studies, 326 of whom underwent surgery after treatment with TKIs and 516 of whom received TKIs alone. 304 GIST patients included in seven studies were reported surgical margins, and the surgical management was illustrated in Table 1. R0 resection rates were demonstrated to be as low as 60.5%. The patients in 5 studies received TKIs preoperatively and postoperatively, the other 3 studies received TKIs postoperatively. The year of publication ranged from 2005 to 2022. The recruitment period ranged from 2001 to 2018. The characteristics of the included studies are presented in Table 1 and Table 2.

The four studies [8, 18, 21, 23], including 453 patients, reported 3-year PFS. The pooled analysis indicated the TKIs plus surgery group did have better PFS than the TKIs only group (OR = 4.02; 95% CI: 1.45 to 11.16; P=0.008, Fig. 2A). The publication bias confirmed by the Begg’s and Egger’s tests did not appear significant (Pr > |z|= 0.734, P > t = 0.306, Fig. 2B and Fig. 2C).

Figure 2A.

Figure 2B.

Figure 2C.

The four enrolled studies [8, 18, 20, 22] comprised 441 patients, of whom 302 received TKIs and 139 proceeded with TKIs plus surgery. The 3-year OS was significantly higher in the TKIs plus surgery group than in the TKIs group (OR 2.37, 95% CI 1.45–3.88; p = 0.001, Fig. 3). No significant publication bias appeared (Pr > |z| = 0.734, P > t = 0.207).

Figure 3.

5-year OS was reported in six included articles [9, 18,19,20,21, 23]. The 5-year OS was 69.0% in the TKIs plus surgery group compared with 49.1% in the TKIs group, and the 5-year OS was significantly higher in TKIs plus surgery group than in the TKIs group (OR = 2.69, 95% CI 1.49–4.86, P=0.001, Fig. 4). Publication bias was not significant (Pr >|z| = 0.133, P > t = 0.156).

Figure 4.

The three included studies [20, 21, 23] focused on gastrointestinal stromal tumor liver metastasis (GLM). The role of surgery combined with TKIs had a statistically better 5-year OS compared to treatment with TKIs alone (OR = 2.34, 95% CI 1.30–4.22, P=0.005, Fig. 5). No significant publication bias appeared (Pr > |z| = 0.296, P > t = 0.103).

Figure 5.