The prevalence of bone metastasis in patients with gastric cancer : A systematic review and meta-analysis

Objective: To assess the prevalence Methods: Literatures that reported prevalence rates of bone metastasis in patients with gastric cancer were identified via PubMed, China National Knowledge Infrastructure (CNKI), Wanfang database and Weipu database from inception to October 2018. The methodological quality of the included studies was evaluated by Agency for Healthcare Research and Quality (ARHQ). Subgroup analyses were performed stratified by areas, years of publication, Tumor Node Metastasis (TNM) stages, pathological types and clinical characteristics. Publication bias was assessed with Egger’s test and Begg’s funnel plot. Results: A total of 59 studies were eligible for inclusion and the methodological qualities of included studies were moderate to high. The pooled prevalence of bone metastasis in patients with gastric cancer was 8.1% (95%CI: 7.0%, 9.1%). Of all the included studies, thirty-three studies were conducted in China, and the pooled prevalence of China was 10.6%. From 1996 to 2000, the prevalence was 10.8% (95%CI: 5.6%, 16.1%), and it increased to 17.4% (95%CI: 8.5%, 26.3%) from 2001 to 2005. After that till now, the prevalence drastically decreased and maintained at a low level around 5.9%. The prevalence showed an increasing trend from 2.7% (95%CI: 0.0%, 5.4%) to 38.5% (95%CI: 20.6%, 56.3%) from TNM stage Ⅰ to stage Ⅳ. The prevalence was 4.5% (95%CI: 1.7%, 7.4%) in the well and moderately differentiated group, 19.0% (95%CI: 13.8%, 24.1%) in the poorly differentiated and undifferentiated group, 31.6% (95%CI: 2.9%, 60.3%) in the signet ring cell group, and 19.3% (95%CI: 11.8%, 26.9%) in the mucinous group, respectively. Conclusion: The pooled prevalence of bone metastasis a random-effect model was selected. To determine possible causes of heterogeneity, subgroup analyses were conducted by areas, years of publication, tumor node metastasis (TNM) stage, pathological type and clinical features of bone metastases. Publication bias was assessed by visually inspecting a Begg’s funnel plot and applying Egger’s test to evaluate sources of variability. For all tests, p value < 0.05 was considered to be statistically significant.


Introduction
The prevalence of gastric cancer is still in the forefront of malignant tumors. People aged over 50 years old tend to suffer from it, with the main contributing factors being precancerous lesions, Helicobacter pylori infection, diet, environment and genetic factors, etc. [1,2]. There are no obvious clinical symptoms in the early stages, while some symptoms of upper gastrointestinal tract, like upper abdominal discomfort, pain, emerge gradually. If not treated timely, it would be worse, which also would raise the risk of getting bone metastasis. Meanwhile, most patients feel painful brought by periostitis and periosteal thickening. Bone metastasis seems more common for patients diagonised with cancers related to breast, lung or prostate, compared with gastric cancer [1,2]. The terminal stage of gastric cancer is the peak period of incidence and multiple bone metastasis seldom is found at early stage. Seldom, mucosal gastric cancer at early stage could give rise to multiple bone metastasis. Once dragonized the disease, patients with gastric cancer have a higher risk of death, with a reduced survival time, no more than 1 year [2].
The earlier gastric cancer is diagnosed and treated, the better prognosis is. As there is no obvious uncomfortable symptom in early cases, the patients with advanced gastric cancer occupy a considerable proportion in all newly diagnosed patients. The liver or lung metastasis in patients with gastric cancer often attracts much attention, while the bone metastasis is more likely to be ignored and not diagnosed at an early stage [3]. It is clear that biological behavior of advanced gastric cancer leads to the sinister prognosis. Bone metastasis caused by vascular tumor thrombus in patients with gastric cancer is not rare [4]. The advanced gastric cancer mostly has invaded the serosa, abdominal and peritoneum. And local recurrence, liver metastasis, bone metastasis and other related events would also occur. Bone metastasis in patients with gastric cancer develops at a rapid speed and patients often die because of disseminated or diffuse intravascular coagulation (DIC) in short time [4].
Different surveys have different results on the prevalence of bone metastasis in patients with gastric cancer. It is reported that the clinical prevalence is 0.46%~6.93% in China [5]and 1.2%~1.4% in foreign countries [6,7]. After the autopsy, the prevalence is as high as 15.9%~17.6% [8]. To date, no meta-analysis on the prevalence of bone metastasis in patients with gastric cancer has been conducted. Accordingly, it seems that an international and pooled estimate based on the various populations is necessary.
The main objective of this systematic review and meta-analysis is to summarize all available data to give a description of a picture on the prevalence of bone metastasis in patients with gastric cancer. A better understanding of metastatic behaviors of patients with skeletal metastases from gastric cancer is helpful for developing diagnostic, therapeutic or follow-up strategies, so as to further improve the quality of life and prognosis.

Search Strategy
We conducted a systematic search of scientific databases, including PubMed, China National Knowledge Infrastructure (CNKI), Wanfang database and Weipu database to find relevant papers published from inception to October 2018. The search medical subject heading keywords and all fields were "gastric cancer" AND "bone metastasis". In addition, a manual search was supplemented by verifying a secondary review of the reference lists of key publications to confirm additional relevant citations.
The work has been reported in line with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) and AMSTAR (Assessing the methodological quality of systematic reviews) Guidelines.

Inclusion and Exclusion Criteria
The criteria of included studies were as follows: (1) with sufficient information to estimate the pooled prevalence of bone metastasis in patients with gastric cancer; (2) published in either English or Chinese language.

Study Selection and Data Extraction
Initially, two investigators independently screened all the titles and abstracts according to the keywords. Then full texts of the selected studies were further reviewed. Finally, the studies which met the inclusion criteria were included. The whole potentially relevant information from the included studies was independently reviewed by two investigators (Fang Zheng, Yuhui Zhang) by using a standardized form which was designed in advance. The following information was extracted from each suitable study: first author's name, year of publication, survey year, survey age, location, total sample size, number or prevalence rate of bone metastasis in patients with gastric cancer. When there was any disagreement during the whole procedure, a third investigator (Xingyu Chen) made the final decision.

Assessment of Methodological Quality
Two investigators independently evaluated the methodological quality of the included studies, using Agency for Healthcare Research and Quality (ARHQ) [9]. Of all the 11 items of ARHQ, item 11 "Clarify what follow-up, if any, was expected and the percentage of patients for which incomplete data or follow-up was obtained" was not applicable for the prevalence study. So ARHQ items 1-10 were used for the assessment of methodological quality in our meta-analysis. (Table 1) Table 1. Criteria used to assess the methodological quality of the studies.
Item yes no unclear 1) Define the source of information (survey, record review). 2) List inclusion and exclusion criteria for exposed and unexposed subjects (cases and controls) or refer to previous publications.
3) Indicate time period used for identifying patients. 4) Indicate whether or not subjects were consecutive if not population-based. 5) Indicate if evaluators of subjective components of study were masked to other aspects of the status of the participants. 6) Describe any assessments undertaken for quality assurance purposes (e.g., test/retest of primary outcome measurements). 7) Explain any patient exclusions from analysis. 8) Describe how confounding was assessed and/or controlled. 9) If applicable, explain how missing data were handled in the analysis. 10) Summarize patient response rates and completeness of data collection.

Data Analyses
All statistical analyses were made using Stata software (version 11.0; Stata Corporation, College Station, Texas, USA) and the meta package was used to produce the pooled estimates, forest plot and publication bias assessment. The pooled prevalence estimate of bone metastasis in patients with gastric cancer and 95% confidence intervals (CIs) were calculated. A fixed-effect model was used when no heterogeneity was present (p>0.1, I 2 <50%). Or else (p<0.1, I 2 >50%), a random-effect model was selected. To determine possible causes of heterogeneity, subgroup analyses were conducted by areas, years of publication, tumor node metastasis (TNM) stage, pathological type and clinical features of bone metastases. Publication bias was assessed by visually inspecting a Begg's funnel plot and applying Egger's test to evaluate sources of variability. For all tests, p value < 0.05 was considered to be statistically significant.

Literature Search
The authors had retrieved 1215 relevant studies by the title or abstract, among which 1088 were abandoned for the low relevance, and then reviewed the remaining ones further, among which 69 were excluded for the duplication and insufficiency. Finally, the essay analyzed 59 studies up to the standard . The flow chart of study selection process was shown in Fig 1.

Study Characteristics and Quality Assessment of Included Studies
Of the 59 studies, the years of publication ranged from 1996 to 2018. The countries were China, Sweden, Japan, Turkey, Korea, Italy, America and the areas covered Asia, Europe, America. The sample size of included studies ranged from 30 to 19022. The prevalence of bone metastasis in patients with gastric cancer ranged from 0.00% to 55.56%. The characteristics of included studies were summarized in Table 2. The clinical features of bone metastasis in patients with gastric cancer were summarized in Table 3. The methodological qualities of included studies were moderate to high.
Twenty-four studies were assessed with 5-7 items as "yes" and five studies were assessed with more than 7 items as "yes". The assessment results were shown in Table 4.

The Result of Meta-analysis
Since there was heterogeneity among the studies (p=0.000, I 2 =97.5%), the random-effect model was used to conduct the analysis. The pooled prevalence of bone metastasis in patients with gastric cancer was 8.1% (95%CI: 7.0%, 9.1%) and the forest plot was shown in Fig.2. The subgroup analyses were conducted by different factors and were presented in Table 5.

Publication Bias
Based on the result of Egger's test, there was no significant publication bias in studies (t=-0.14, p=0.886). The Begg's funnel plot of publication bias was shown in Fig.3.

Discussion
The prevalence of bone metastasis of gastric cancer varies greatly from 0 to 55.56%, related to the following factors. Firstly, there is no unified diagnostic criteria for bone metastasis, for which the detection rate also varies and the positive rate of autopsy is higher than that of non-autopsy.
Secondly, sample selection also affects the results. Whether being done surgery has a significant effect on the incidence of bone metastasis. Most of the patients have been advanced gastric cancer when they are found. Meanwhile, the recurrence rate and the incidence of distant metastasis are also affected by the operation methods and habits. Overall, the incidence of bone metastasis after operation is significantly lower than that without surgery. Finally, it is related to the difference of the follow-up time and the frequency of rechecking. At the same time, the quantity of the sample will also affect the overall data making it bias to the larger sample. These factors directly or indirectly affect the incidence of bone metastasis in gastric cancer.
To our knowledge, this is the first systematic review and meta-analysis examining the prevalence of bone metastasis in patients with gastric cancer. According to previous records, the risk factors of bone metastasis in patients with gastric cancer included young age, poorly differentiated and undifferentiated adenocarcinoma, especially mucinous adenocarcinoma, multiple lymph node metastasis, multiple metastatic carcinoma, tumor located in the gastric body, etc [16]. The present meta-analysis showed that the prevalence was as high as more than 30% in TNM stage Ⅳ. Moreover, the prevalence of patients in poorly differentiated and undifferentiated group and in the mucinous group were nearly 20%. The signet ring cell group had the highest prevalence of 31.6%. The pathological types have close relationship with vascular invasion.
Low differentiated adenocarcinoma had its unique tumor habits, such as diffuse growth of tumor cells in a free loose state, tumor cells falling off easily leading to vascular embolization and into the circulatory system leading to bone metastasis [20]. Peritoneal implant and distant metastasis mostly occurred in parallel [26]. Low differentiated had a higher incidence of metastasis than high differentiation, and the biological behavior of it determines the poor prognosis, more susceptible to bone metastasis [29]. The rate of lymph node metastasis was 78.5% in postoperative gastric cancer patients with positive vascular invasion and 54.7% in patients with positive venous thrombosis [30].
Bone metastases in patients with invasion of lymph accounted for 76.9%, and patients with venous invasion accounted for 30% [39]. So vascular invasion should arouse sufficient attention. It was reported that there were 65.4% of the low differentiated gastric cancer occurring bone metastasis and the tissue types of bone metastasis were certainly low differentiated gastric cancer [40]. The inherent biological behavior of the tumor determines its clinical development and final outcome. Most scholars believe that the prevalence of bone metastasis in patients with gastric cancer is inversely related to the degree of tissue differentiation and is proportional to clinical TNM stage [54]. So the later pathological stage of gastric cancer is, the higher the prevalence of bone metastasis is. It is consistent with the results of current meta-analysis.
Despite we have conducted a comprehensive searching of the epidemiology of bone metastasis in patients with gastric cancer, several limitations should be considered in this meta-analysis. The available publications/studies were only from several countries. The data of unavailable countries are required to reflect the wide variation. Some characteristics of the patients, such as gender or age of onset, risk factors, etc., were not included in the subgroup analyses. These might exert an important influence on the prevalence. Some included studies had noted methodological flaws, especially related to selection and recruitment of samples. Control group with other diseases such as thyroid disease, breast disease, kidney disease was also selected in some studies. As a result, the estimates of prevalence may have been influenced in unpredictable ways and need continuous perfectibility for verifying our results.
In conclusion, the pooled prevalence of bone metastasis in patients with gastric cancer was 8.1%. The prevalence increased with the exacerbation of gastric cancer from TNM stage Ⅰ to stage Ⅳ and was inversely related to the degree of tissue differentiation. At the same time, it was also affacted by area and years. The current study provides basic information which is useful fordeveloping clinical strategies. Flow diagram of the study selection process. Figure 1 Flow diagram of the study selection process.

Figure 2
Forest plot of the pooed prevalence of bone metastasis in patients with gastric cancer.

Figure 2
Flow diagram of the study selection process.

Figure 3
Funnel plot assessing publication bias in the prevalence of bone metastasis in patients with gastric cancer.

Figure 3
Forest plot of the pooed prevalence of bone metastasis in patients with gastric cancer.

Figure 4
Forest plot of the pooed prevalence of bone metastasis in patients with gastric cancer.

Figure 5
Funnel plot assessing publication bias in the prevalence of bone metastasis in patients with gastric cancer.

Figure 6
Funnel plot assessing publication bias in the prevalence of bone metastasis in patients with gastric cancer.