Does Gastrointestinal Bleeding Affect the Prognosis of Patients With Gastrointestinal Stromal Tumor: A Meta-analysis.

Background: The effect of gastrointestinal bleeding on the prognosis of gastrointestinal stromal tumors has been widely studied in recent years, but it is still controversial. Therefore, we performed the rst comprehensive meta-analysis to evaluate the effect of gastrointestinal bleeding on the prognosis of gastrointestinal stromal tumors. Methods: We searched PubMed, MEDLINE, Web of Science, EMBASE, and Cochrane Library databases to recruit studies on the effect of gastrointestinal bleeding on the prognosis of patients with gastrointestinal stromal tumor. Eight studies containing 2915 patients were involved in this meta-analysis until May 31, 2021. Pooled hazard ratios (HRs) with 95% condence interval (95% CI) were calculated to estimate the effect using random-effects model. Results: The pooled results revealed that gastrointestinal bleeding was not associated with relapse-free survival (HR = 1.33, 95% CI 0.66-2.68, P < 0.001; random-effects model I2=87.7, P < 0.001) and overall survival (HR = 1.29, 95% CI 0.43-3.87, P < 0.001; random-effects model I2=88.9, P < 0.001) in patients with gastrointestinal stromal tumors. Conclusions: Our present a meta-analysis indicates that gastrointestinal bleeding has no effect on relapse-free survival and overall survival of patients with gastrointestinal stromal tumors, although gastrointestinal bleeding is one of the major clinical symptoms of gastrointestinal stromal tumors.


Background
Gastrointestinal stromal tumors (GISTs) are common mesenchymal tumors, which can occur anywhere in the digestive tract (1). About 70% of GISTs are symptomatic, and GISTs have different clinical manifestations in different locations, such as abdominal pain, abdominal distension, gastrointestinal bleeding, intestinal obstruction (2,3),and many patients often undergo emergency operations because of uncontrollable gastrointestinal bleeding. The main determinants of prognostic risk strati cation recommended in the current versions of NCCN guidelines(4) and ESMO guidelines (5) are tumor size, tumor site, mitotic index and tumor rupture. At present, in the modi ed-NIH classi cation, tumor rupture is considered to be a high risk factor for recurrence(6). Patients with tumor rupture should be included in the high-risk group and need to be treated with Tyrosine Kinase Inhibitors (TKIs) represented by imatinib for at least 3 years.
The study of Nishida T et al de ned six forms of rupture of gastrointestinal stromal tumors (7), but gastrointestinal bleeding was not included. Whether gastrointestinal bleeding is one of the forms of tumor rupture is still controversial. In recent retrospective studies, it is considered that gastrointestinal stromal tumors with gastrointestinal bleeding may be accompanied by excessive growth, high mitotic index, incomplete local tumor capsule and other factors leading to poor prognosis. it can be used as an index to predict the risk of postoperative recurrence of GISTs(8, 9), but some studies have shown that gastrointestinal bleeding can detect tumors as soon as possible and carry out medical intervention in time, which may be a protective factor and will lead to a better clinical outcome (10).
Nowadays, the research on the relationship between gastrointestinal bleeding and the prognosis of GISTs is increasing, which has attracted the attention of clinical treatment and postoperative management, but due to the limitations of sample size and single-center study, the evidence is insu cient. Therefore, we conducted a systematic review and meta-analysis of GISTs with gastrointestinal bleeding in order to evaluate whether gastrointestinal bleeding has an impact on the prognosis of patients with GISTs.

Methods
This study was undertaken according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (11). OR (Gastrointestinal Stromal Neoplasm)) OR (Gastrointestinal Stromal Sarcoma)) AND ((Hemorrhage) OR (Bleeding). The language of publication was restricted to English. The meta-analysis collected data from previously published studies; therefore, approval was not required from the ethical committee or medical institutional board.

Criteria for inclusion and exclusion
The inclusion criteria for eligible studies were as follows: (1) all patients were pathologically diagnosed as gastrointestinal stromal tumor; (2) the relationship between gastrointestinal bleeding and prognosis was reported; (3) the hazard ratios (HRs) with 95% con dence intervals (CIs) for survival outcomes were reported or su cient data were given for calculating the HRs with 95% CIs. The following studies were excluded: (1) letters, reviews, and case reports; (2) duplicate studies; (3) studies with insu cient data; (4) a study of less than 10 patients.
Data extraction and quality assessment Two independent investigators (Yue Zhang. and Qi Liu) extracted the data from eligible studies by using a standardized form. Any disagreements were resolved via discussion with a third investigator (Hao Xu). The extracted information included the name of the author, year of publication, country of study origin, sample size, tumor site, follow up, outcomes and HR with 95% CIs. The clinical outcomes included the recurrence-free survival (RFS) and overall survival (OS). The methodological quality of all the included studies was evaluated by using the Newcastle-Ottawa quality assessment scale (NOS) (12). The NOS assesses the quality of the included studies by using a score of 0 to 9 points. Studies with a NOS score of ≥ 6 points were regarded as high-quality studies.

Statistical analysis
The pooled HR with corresponding 95% CI was utilized to estimate the relationship between gastrointestinal bleeding and patients' prognosis. While the effect of gastrointestinal bleeding was described as the combined. Higgins I-squared tests were applied for checking the heterogeneity of the results. whereas I2 values > 50% and/or xed-effect framework was adopted, otherwise, the random-effect model was employed. Besides, probable publication bias was quanti ed with conducting Begg's funnel plot, respectively. Sensitivity analysis was also done by omission of each single study to investigate the stability of the accumulated results. All analyses were carried out using STATA software version 12.0 (Stata Corporation, College Station, TX, USA) P value < 0.05 was regarded as being statistically signi cant.

Data selection and characteristics
The study selection process ow chart is shown in Fig. 1 and the initial search yielded 3508 studies from the four databases. After reviewing the titles and abstracts, 2608 studies were excluded because they were not relevant to what we were studying. After deleting some duplicates and incomplete records, night records were eventually included in our study. A total of 2915 patients were enrolled in the study, and they were all single-center studies.
Of the eight studies, two were from South Korea and the remaining seven were from China. RFS and OS was selected as the major survival outcome for all the available studies in our meta-analysis. The Characteristics of the included studies are shown in Table 1. Quality assessment The quality of eligible publications was calculated based on the Newcastle-Ottawa Scale (NOS) that evaluated the selection of cohorts, comparability as well as exposure or outcome and had a score ranging from 0 to 9.
Studies with higher or equal to 6 points could be considered as high quality Meta-analysis results Figure 2 shows the main results of this meta-analysis. Due to the signi cant statistical heterogeneity of the studies evaluating RFS (I2=87.7, P < 0.001) and OS (I2=88.9, P < 0.001), a random-effect model was used to incorporate HR. Gastrointestinal bleeding had no effect on the prognosis of patients with GISTs, either in RFS (HR = 1.33, 95% CI 0.66-2.68, P < 0.001; random-effects model I2=87.7, P < 0.001) or OS (HR = 1.29, 95% CI 0.43-3.87, P < 0.001; random-effects model I2=88.9, P < 0.001). Interestingly, notable heterogeneity (>50%) was found for migration proportion outcomes among studies. Due to the small number of records included, only 9 records, further subgroup analysis could not be performed.

Sensitivity analysis
Sensitivity analysis was done through the sequential omission of single studies using a model with randomeffects, and the result pattern was not obviously impacted by any single study (Figure 3).

Publication bias
The assessment of the publication bias for RFS and OS was done through the shape of the funnel plot revealed no evidence of asymmetry ( Figure 4). conducted a retrospective study on a small sample and revealed that rectal bleeding is an independent risk factor for prognosis of rectal stromal tumor (9). Over the next few years, several studies have also demonstrated that gastrointestinal bleeding causes poor DFS or OS in patients with GIST (20)(21)(22). Interestingly, two studies have shown that the prognosis in the group with gastrointestinal bleeding is better than that in the group without bleeding. Bleeding may be a protective factor for recurrence of GISTs (10,23). However, Gyu Young Pih et al. from South Korea concluded in a large retrospective study that gastric bleeding leads to a poor prognosis for GISTs, while duodenal bleeding has no effect on GISTs (24,25).

Conclusions
This meta-analysis combined with all previous studies sought to elucidate the relationship between gastrointestinal bleeding and prognosis in patients with GISTs. The results of the analysis showed that gastrointestinal bleeding had no effect on the prognosis of patients with GISTs either RFS or OS. However, due to the insu cient included samples and other limitations, it is necessary to conduct more studies on the relationship between gastrointestinal bleeding and the prognosis of GISTs Declarations Ethics approval and consent to participate Not applicable.

Consent for publication
Consent for publication was obtained from the participants.

Availability of data and materials
All data generated or analysed during this study are included in this published article.
WR collected and analyzed the data, wrote the paper; Yue zhang and Qi liu analyzed the data; Hao xu conceived and designed this study, analyzed the data; and all authors reviewed the paper. All authors read and approved the nal manuscript. Figure 1 Flow diagram of the study selection process.