Risk factors of synchronous peritoneal metastases in colorectal cancer: a meta-analysis

Background: Early detection of synchronous colorectal peritoneal metastasis (CPM) is dicult due to the absence of typical symptoms and the low accuracy of imaging examinations. Better knowledge of risk factors for synchronous CPM may be essential for early diagnosis and strengthening management. This study aimed to clarify the risk factors. Methods: This meta-analysis was based on PRISMA guidelines. A systematic search of PubMed, Embase and Cochrane Library databases was performed. The pooled data was assessed by a random-effects model. Results: 25 studies containing 171932 patients were included. Synchronous CPM was associated positively with female (OR 1.299; 1.118 to 1.509; P = 0.001), T4 (OR 12.331; 7.734 to 19.660; P < 0.001), N1-2 (OR 5.665; 3.628 to 8.848; P < 0.001), poorly differentiated grade (OR 2.560; 1.537 to 4.265; P < 0.001), right-sided colon cancer (OR 2.468; 2.050 to 2.970; P < 0.001), mucinous adenocarcinoma (OR 3.565; 2.095 to 6.064; P < 0.001), signet-ring cell carcinoma (OR 4.480; 1.836 to 10.933; P = 0.001), elevated serum CA19-9 (OR 12.868; 5.196 to 31.867; P < 0.001), PROK1/PROKR2-positive (OR 2.244; 1.031 to 4.884; P = 0.042) and BRAF mutations (OR 2.586; 1.674 to 3.994; P < 0.001). However, it’s associated negatively with rectal cancer and non-mucinous adenocarcinoma, and not associated with KRAS, NRAS, PIK3CA mutations and MSI-H/dMMR. Conclusions: These risk factors are the alerts that could predict the presence of synchronous CPM and node metastasis, 5 studies on differentiation, 6 studies on primary tumor site, 7 studies on histology, 2 studies on serum CA19-9, 2 studies on PROK1/PROKR2, 9 studies on BRAF, 6 studies on KRAS, 2 studies on NRAS, 2 studies on PIK3CA and 4 studies on MSI-H/dMMR status.

factors for synchronous CPM would increase the level of suspicion in patients with no suggestive signs or symptoms, and thus allow physicians to treat these patients more adequately, such as more aggressive preoperative examination, proactive laparoscopic exploration or referring them to specialized centers.
Several studies have tried to determine the risk factors of synchronous CPM, but with heterogeneous outcomes, such as the location of primary tumor [8,15], . Furthermore, many studies only focused on the aspect of clinicopathological characteristics of synchronous CPM, and thus they were lack of systematic and comprehensive analysis of molecular characteristics.
Further comprehensive understanding of its clinicopathological and molecular features may be necessary for early diagnosis and may help to enhance the management of patients at high risk of synchronous CPM. Therefore, a systematic review and meta-analysis of all studies comparing gender, tumor invasion depth, lymph node metastasis, differentiation, location of primary tumor, histology, serum CA19-9, PROK1/PROKR2, BRAF, KRAS, NRAS, PIK3CA and MSI-H/dMMR status between synchronous pmCRC and non-pmCRC was undertaken.

Methods
This systematic review and meta-analysis adhered to the recommendations of the Preferred Reporting Items of Systematic Reviews and Meta-analysis (PRISMA) statement [19]. PRISMA Checklist is available in supplementary Appendix 1.

Study registration
This study was registered at PROSPERO (Prospective Register of Systematic Reviews, www.crd.york.ac.uk/prospero). Number CRD42020198548.

Eligibility criteria
Colorectal peritoneal metastases can be divided into synchronous CPM and metachronous CPM.
Synchronous CPM has different de nitions [5,6,20]. Referring to the international consensus on colorectal liver metastases [21], synchronous CPM is de ned as peritoneal metastases detected at or before diagnosis or surgery of the primary CRC; metachronous CPM is de ned as those detected after curative surgery.
Comparative studies of primary colorectal tumor with or without synchronous PM involving data on clinicopathological and molecular characteristics were eligible for inclusion. The included studies need to use the recognized diagnostic criteria, as follows: the primary tumor's pathological diagnosis was con rmed; the tumor cells were primary in colorectal tumor; and the patient's synchronous PM was con rmed by imaging diagnosis before surgery, intraoperative exploration or histopathological examination.
The exclusion criteria were: (1) case reports, review articles and animal studies; (2) non-English publications; (3) studies that are not related to CRC or PM; (4) metachronous PM; (5) no analysis of the risk factors; (6) no comparator group; (7) no relevant data, including articles published only in abstract form as well as studies without complete data and inability to construct a 2×2 contingency table; (8) mixed primary tumor; (9) non-standardized histological type (10) synchronous CPM was not clearly or correctly de ned.

Data sources and search strategy
We selected relevant studies by searching PubMed, Embase and the Cochrane CENTRAL Register of Controlled Trials. The following combined terms were used in the search: (peritoneal metastasis OR peritoneal metastases OR peritoneal carcinomatosis) AND (colorectal OR colon OR rectal). The latest search was implemented on 14 July 2020 and the earliest search was not limited in the relevant database.

Selection process
Two independent authors (Y.Z and X.Q) checked the title and abstract of each study, and studies that satis ed the potential eligibility were obtained for further full-text assessment. Disagreements were resolved by discussion with senior authors (Y.D or H.W) until consensus was achieved.

Data extraction
By using standardised forms, two independent authors (Y.Z and X.Q) extracted the data from each eligible study. The authors resolved disagreements by discussion with senior authors (Y.D or H.W). The following data were extracted from each eligible study: author, year of publication, country where the study was conducted, setting of centre, type of study, enrollment interval, number of primary CRC patients with or without synchronous PM, clinicopathological and molecular characteristics. In addition, the score of Newcastle-Ottawa Scale (N-O score) for eligible studies was also calculated and extracted.

Statistical analysis
We used Comprehensive Meta-Analysis (version 2.0) and Stata (version 12.0) for all statistical analyses.
All pooled outcomes were determined using a random-effects model (DerSimonian-Laird method). In pooled analyses of associations between various clinicopathological-molecular factors and synchronous CPM, effect sizes were calculated as odds ratios (OR) with a 95 percent con dence interval (CI). The χ 2based Cochran Q test was used to assess heterogeneity between studies, in which P < 0.1 indicates the presence of heterogeneity [22]. We also did I² inconsistency testing to assess the extent of the heterogeneity between studies, with values greater than 50% regarded as moderate-to-high heterogeneity [23]. For signi cant heterogeneity, we tried to do sensitivity analysis or subgroup analysis to nd its potential sources. Sensitivity analysis was performed by omitting each study sequentially to test the in uence of each individual study on the pooled result. Publication bias was evaluated by visual inspection of the funnel plot for symmetry (an asymmetric plot suggested possible publication bias) and quanti ed by means of Begg's test, with P value < 0.05 regarded as signi cant publication bias.
The quality of included studies was assessed using the Newcastle-Ottawa Scale [24], in which a score ≥ 6 indicates the high-quality of studies. The quality of studies was evaluated by examining 3 categories: patient selection, comparability of the 2 study groups, and assessment of exposure (maximum score 9), as was shown in the Newcastle-Ottawa Scale.

Study characteristics
Among the 25 included studies, 7 had a multicentre setting and 18 had a single centre design. Five of the included studies were prospectively performed; the remaining twenty were retrospective. 22 studies were considered of high quality (N-O score ≥ 6), and 3 studies were considered of low quality. Complete characteristics of the included studies are available in Table 1.

Factors not included in the quantitative synthesis
Six of clinicopathological and molecular factors could not be included in quantitative synthesis because they had only a single study of their subgroup, or their methodology did not permit pooling data. The six factors were serum CEA [25], serum CA125 [32], CTGF (connective tissue growth factor) [40], DDR2 (discoidindomain receptor 2) [30], VIM (vimentin) [42], and TP53[28] respectively. We included these factors in table 1 for completeness, but not in the nal quantitative synthesis through meta-analysis.
Finally, 21 studies about 13 factors were included in the quantitative synthesis through meta-analysis, 7 studies on gender, 4 studies on tumor invasion depth, 3 studies on lymph node metastasis, 5 studies on differentiation, 6 studies on primary tumor site, 7 studies on histology, 2 studies on serum CA19-9, 2 studies on PROK1/PROKR2, 9 studies on BRAF, 6 studies on KRAS, 2 studies on NRAS, 2 studies on PIK3CA and 4 studies on MSI-H/dMMR status.

Gender
Seven studies[8, 15,17,18,25,33,41], including data on 160679 patients (30366 synchronous pmCRC, 130313 non-pmCRC) regarding gender, were included for eligibility in the meta-analysis. The pooled analysis indicated that female was associated positively with synchronous CPM compared with male (OR 1.299; 95% CI, 1.118 to 1.509; P = 0.001) (Fig. 2a). There was signi cant heterogeneity (Cochran Q, P < 0.001; I² = 76.9 percent). In order to explore possible sources of heterogeneity, sensibility analysis was performed by omitting each study sequentially to test the in uence of each individual study on the pooled result. When one study [17] was omitted, there was no signi cant heterogeneity (Cochran Q, P = 0.099; I² = 46.0 percent), with no noticeable in uence on the pooled OR and con dence interval. It's noteworthy that the rate of female in the synchronous CPM group was > 50 percent in that one study, but the others were < 50 percent.

Publication bias
No signi cant publication bias was found, according to visual inspection of funnel plot and to Begg's test (supplementary Fig. S1-S5).

Discussion
We found that synchronous CPM was associated positively with female, PROK1/PROKR2-positive, rightsided colon cancer location, poorly differentiated grade, BRAF mutations, mucinous adenocarcinoma, signet-ring cell carcinoma, N1-2, T4 and elevated serum CA19-9 (ascendingly sequenced by the value of odds ratios). Our study has provided an extensive analysis of the association between synchronous CPM and clinicopathological-molecular features, especially the molecular characteristics compared with the previously published studies [44][45][46][47]. In addition, several studies have different conclusions about the association between MSI-H and synchronous CPM, although MSI-H is a poor prognostic factor in metastatic colorectal cancer [48,49]. Nonetheless, we found that MSI-H was irrelevant with synchronous CPM in this study. The meta-analytical techniques could increase the volume, which may cause more su cient statistical power.
Based on the hypothesis that phenotype and the subsequent clinical behavior of CPM is driven by underlying biological mechanisms, readouts of disease biology will contribute to more precise identi cation of suitable patients and guidance of therapy. This is one of the critical future research targets in CPM. The potential mechanisms of risk factors that associated positively with synchronous CPM are discussed as follows. Due to a longer asymptomatic period, right-sided colon tumors are usually larger in diameter when diagnosed than left-sided colon tumors. Larger neoplasms in ltrate the surface of serosa over a larger area, so it may lead to increased abscission of cancer cells into the peritoneal cavity. In addition, typical genetic differences between right-sided and left-sided colon tumors have been found, such as BRAF status, and these genotypes may bring about a fenotype with a different probability to be associated with synchronous CPM [50]. Several studies have shown that mucinous histologic type has a poor prognostic impact, including a higher tendency to peritoneal carcinomatosis and a lower e cacy of oxaliplatin and irinotecan-based chemotherapy [51][52][53]. A more advanced T stage is associated positively with the presentation of peritoneal carcinomatosis, with the potential mechanism that peritoneal carcinomatosis is caused by serosal in ltration of the malignant tumor and subsequent abscission of cancer cells into the peritoneal cavity [54]. Regarding peritoneal tumor spread, CA19-9 was shown to interact with E-and P-selectins expressed on human mesothelial and endothelial cells in the peritoneum [25,55]. Prokineticin1 (PROK1) is a known ligand of Prokineticin-receptor2 (PROKR2) and transduces important molecular signals to induce physiological changes. The PROK1 protein was identi ed as a vascular endothelial growth factor. Increased PROK1 expression is associated with angiogenesis involving hematogenous metastasis [34,37]. Besides direct invasion and hematogenous spread, peritoneal carcinomatosis could be occurred by lymphatic dissemination, which supports the risk factor of N1-2 [54,56,57].
Some studies once de ned the degree of risk of developing colorectal peritoneal carcinomatosis [58,59]. A high risk of synchronous CPM should modify the management strategy of this special type of metastatic disease, with the following suggestions [5,58]. First, in CRC patients at high risk of developing synchronous PM, a more aggressive preoperative examination, such as PET-CT, MRI diffusion-weighted and so on, is suggested to be performed to con rm whether there is synchronous PM. Then, if PM is suspected on preoperative imaging, we could propose laparoscopic exploration of the abdominal cavity to assess the extent of the disease and to obtain histological con rmation. Eventually, if the synchronous PM is diagnosed, surgeons are expected to describe the extent of the disease and to determine whether aggressive treatment including complete CRS plus HIPEC should be carried out in patients.
There are some limitations in this study. Firstly, the non-English studies were excluded, with the language bias. Secondly, the risk associated with T4a vs. T4b stage was not analyzed because the speci c data was absent in the included studies. Thirdly, factors such as T4 stage and N1-2 stage are of little help, because they can be poorly assessed preoperatively. Finally, the number of included studies about CA19-9, PROK1/PROKR2, NRAS, PIK3CA status is small, which may cause limited statistical power.

Conclusions
To our knowledge, this is the rst meta-analysis to reveal the clinicopathological and molecular features of synchronous pmCRC compared to non-pmCRC. These evidence-based predictive factors of synchronous CPM are conducive to enhance the management and select optimal therapeutic strategy.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.