Do Protocadherins Show Prognostic Value in the Carcinogenesis of Human Malignant Neoplasms? Systematic Review and Meta-Analysis

Background: Protocadherins (PCDHs) have been reported as tumor suppressor genes, implying that these genes may be involved in tumor suppression in a variety of cancers. However, a thorough understanding of the functions and mechanisms of PCDHs remains limited. Our aim was to investigate the methylation profile of PCDHs in human malignant neoplasms. Methods: This systematic review has been recorded in PROSPERO (#42019117844) and conducted according to PRISMA’s checklist; search was conducted in LILACS, PubMed, Science Direct, Scopus, and Web of Science databases, manually, with search queries and without date or language restrictions. Results: We found 91 articles, of which 26 were used for this meta-analysis and categorized according to the origin of the neoplasia. In total, 3,377 cases were compiled, with PCDH10, PCDH17, and PCDH8 being the most studied; males were 2.22 times more affected than females. Studies have shown significant heterogeneity (p <0.001), with the odds ratio varying between cases and controls [2.20 (95% CI = 1.11– 4.35) to 209.05 (95% CI = 12.64– 2,457.18)], and the value of association between methylation and cancers studied was 26.08 (95% CI = 15.42–44.13). Conclusion: In this systematic review, we have demonstrated using meta-analysis that PCDHs could emerge as potential tumor suppressor genes and that a significant increase in methylation may be useful for early detection of different cancers. This work may help in the identification of new prognostic biomarkers in malignant neoplasms.


Introduction
Cadherin is a calcium-dependent adhesion protein that is a member of a large family of cell adhesion molecules. Cadherins can be classified into three groups: classical, desmosomal, and protocadherins (PCDHs). PCDHs are predominantly expressed in the nervous system and are primarily involved in the maintenance of brain functions. However, in recent years, it has been revealed in few studies that members of the PCDH family may have other functions, including functions in some types of cancer (Narayan et al., 2009;Harada et al., 2015;Keeler et al., 3678 be involved in tumor suppression in a variety of cancers (Imoto et al., 2006;Narayan et al., 2009;Yu et al., 2009;Hu et al., 2013;Keller et al., 2015;Shi et al., 2015;Deng et al., 2016;Lee et al., 2016;Ye et al., 2017;Zhong et al., 2017).
A detailed understanding of the functions and mechanisms of PCDHs remains limited, and, so far, there has been no systematic literature review in which the underlying mechanisms of PCDH activity in carcinogenesis have been discussed. Thus, understanding the participation and actual role of PCDHs in important cancer signaling pathways is fundamental for better elucidation of the regulatory and progressing intracellular mechanisms of carcinogenesis. Hence, our aim in this review was to investigate the expression of PCDHs in primary malig-nant neoplasms.

Materials and Methods
This systematic review was conducted to summarize the current knowledge on PCDH expression in cancer, according to the Checklist of Preferred Reporting Items for Systematic Re-views and Meta-Analyses (PRISMA) (Additional file 1). The protocol has been registered in the International Prospective Register of Systematic Reviews (PROSPERO) database with the registration number CRD42019117844.

Information sources and search strategy:
The studies to be considered for inclusion were identified using a search strategy in each of the following electronic databases: LILACS, PubMed, Science Direct, Scopus, and Web of Science (Online Resource 1) (Additional file 2). A partial search for gray literature was per-formed using Google Scholar, OpenGrey, and ProQuest Dissertations and Theses Global. The gray database and bibliographic search included all articles published until October 21, 2019, without time restriction. Duplicate references were removed using the reference manager software (EndNote ® , Thomson Reuters). In addition, the reference lists of selected articles were manually selected for possible relevant studies that might have been missed during electronic database search.

Eligibility Criteria
Articles were selected without restrictions in year or language of publication; and the inclusion criteria adopted in this review were as follows: laboratory studies involving PCDH analysis in cancer and studies involving humans. The exclusion criteria were as follows: case reports, animal studies, literature reviews, and studies in which the clinical-pathological parameters were not presented.

Study selection
The selection of articles was performed in two phases. In phase 1, two reviewers independently (TTBD and EMCL) verified the titles and summaries of all of the identified electronic database citations. A third reviewer was involved as needed to make the final decision (KMAP). Studies that did not meet the inclusion criteria were omitted (Additional file 3). In phase 2, the same selection processes was carried out for full papers to confirm if the articles were suitable, according to inclusion criteria. TTBD and EMCL independently participated in phase 2 as well. The reference list of all of the included articles was reviewed by an examiner; however, both examiners read the selected articles. Any disagreement at any stage was resolved by discussion and mutual agreement.

Data collection process
One author (TTBD) collected important information from each article selected. A second reviewer (EMCL) cross-checked the information collected and confirmed its accuracy. Any disagreement between them was resolved by discussion and mutual agreement. For all studies included, the following information was recorded: author(s), year of publication, country, sex, age, smoking status, alcohol consumption pattern, HPV infection, sample size (study group and control group), tissue type (control and cancer), histological type, tumor differentiation, clinical staging, tumor size, tumor volume, lesion location, serum PSA (where applicable), disease progression, margin status, metastasis, biochemical recurrence (where applicable), follow-up time, death rate, survival, methodological design, and main results.

Risk of bias in individual studies
The risk of bias analysis of the studies was performed independently by two authors (TTBD and EMCL), in accordance to a questionnaire developed by the Joanna Briggs Institute (2017). The authors rated each item of the questionnaire as "yes", "no", "inaccurate", or "not applicable". The risk of bias was rated as high (up to 49% yes), moderate (50% to 69% yes), and low (over 70% yes).

Summary measures and synthesis of results
The primary outcome for this systematic review was the use of PCDH as a biomarker for cancer diagnosis. A secondary outcome would be the role of PCDH as a tumor suppressor. Any type of measurement was considered in this review (categorical and continuous variables). Data were tabulated in an Excel® spreadsheet (Microsoft Corporation, Redmond, WA, USA) for relative and absolute frequencies and then exported to the Statistical Package for Social Sciences (version 17, IBM, Armonk, New York, USA) to perform statistical analysis. For meta-analysis, data were exported to MedCalc software for case-control meta-analysis. Der-Simonian Laird analyses were used to calculate the combined random effect odds ratios. A 95% confidence interval was used for all evaluations. A total of 26 articles were included in the meta-analysis, which were also categorized according to the origin of the neoplasia (upper respiratory tract, gastrointestinal tract , and genitourinary tract).

Study selection
At the end of phase 1, we could obtain 91 bibliographic references from the electronic database. However, 14 articles were present in more than one database (duplicate in the forty-one studies that were selected as sample, and the number of cases in the sample ranged between 2816 and 107,217. The total number of controls were 1,934, and 7 studies reported no control group (Losi et al., 2011;Harada et al., 2015;Hou et al., 2015;Lv et al., 2015;Wu et al., 2017;Cao et al., 2018).

Immunoexpression of PCDHs in malignant neoplasms is negative or low, and methylation is a specific tumor event
The immunohistochemical profile was evaluated in 13 articles in which negative or low immunoexpression was reported (Haruki et al., 2010;Losi et al., 2011;Ma et al., 2013;Harada et al., 2015;Dang et al., 2016;Lin et al., 2016;Chen et al., 2017;Wu et al., 2017;Cao et al., 2018;Li et al., 2018). The methylation profile of PCDHs in malignant neoplasms was evaluated in 33 studies articles), after the removal of which, 77 articles remained for independent analysis by the reviewers. In addition, a manual search was conducted, and we found 25 additional articles from Google Scholar; however, we excluded 2 articles as these were duplicated. A flowchart of the study selection has been presented in Figure 1.
A full-text review was independently conducted by the reviewers, and 41 articles that met the inclusion criteria were selected. The studies that were excluded were as follows: laboratory studies involving PCDH analysis in other pathologies; studies on blood, lymphatic, or brain malignant neoplasms; studies that did not use tumor samples for analysis; animal studies; and studies without clinical-pathological parameters described.

Study characterization
All of the studies included in this review were published in English language between the years 2006 and 2018 and were conducted in seven different countries, with most of the studies published in China (n = 34), followed by Japan (n = 3), Italy (n = 2), the Netherlands (n = 1), and the Czech Republic (n = 1). A summary of the descriptive characteristics of the included studies has been presented in Table 1.

Malignant neoplasms show significant increase in methylation
In total, 26 articles were included in this metaanalysis, which were also categorized according to the origin of neoplasia, and 3,377 cases and 1,638 controls were compiled. In the studies, we observed that there was significant heterogeneity (p <0.001), with a fairly variable odds ratio between cases and controls [2.20 (95% CI = 1.11. 4.35) to 209.05 (95% CI = 12.64. 2.457.18)], and the value of association between methylation and the cancers studied was found to be 26.08 (95% CI 15.42. 44.13) ( Figure 3). Depending on the site of origin of neoplasm, the tumors of the upper respiratory

PCDH17
Methylation PCDH17 was more methylated in case than in controls and methylation in serum was correlated with advanced stage, higher grade, lymph node metastasis and tumor progression. In addition, patients with methylated PCDH17 had shorter progression-free survival and overall survival than patients without, and methylation in serum was an independent prognostic factor for worse progression-free survival and overall survival of patients.
Wu et al. PCDH10 promoter methylation plays a significant role in the progression of non-small cell lung cancer and may be a promising prognostic marker for patients with curatively resected pathological stage I.

PCDH20
Methylation and others Methylation of this PCDH20 promoter was observed in primary tumor and was associated with a shorter overall survival. Moreover, the PCDH20 methylation status was an independent prognosticato tract and gastrointestinal tract (GIT) showed significant heterogeneity, p = 0.0003 and p = 0.00001, respectively, while the tumors of the genitourinary tract showed no heterogeneity (Figure 4).

Risk of bias in individual studies
A summary of the risk of bias of the 26 studies included in this meta-analysis has been presented in Table 2. Three studies were classified as having a moderate risk of bias, while in other studies the risk of bias was low. The criteria for inclusion in the sample were clearly defined (Table  2) in five studies (Yu et al., 2009;Haruki et al., 2010;Lin et al., 2012;Danese et al., 2013;Lin et al., 2013). Confounding factors were uncertain, and strategies for dealing with these factors were not applied in all studies. The other items were entirely scored as "yes"
PCDH10 is a non-fragmented delta2-PCDH (subgroup of the delta-PCDH subfamily), with a unique cytoplasmic domain (Hirano et al., 1999); and unlike other members of the PCDH family, PCDH10 was found to exhibit widespread expression (Wolverton and Lalande, 2001).
PCDH8 and PCDH17, in contrast, may have multiple tumor suppression functions, such as involvement in cell-cell adhesion, signal transduction, and growth control, although the exact function of both the PCDHs is poorly understood. PCDHs are often inactivated by DNA methylation in different neoplasms and function as tumor suppressors. In addition, it was observed that the immunoreactivity of PCDHs was downregulated in specimens Zhang et al., 2016;Cao et al., 2018;Li et al., 2018); and protein expression was re-duced in methylated tumors, in comparison to non-cancerous tissues or tissues in which there was no methylation (Harada et al., 2015).
Lung cancer accounts for 13% of new cancer cases and is the leading cause of cancer mortality, representing the highest age-standardized mortality rate (26.6 deaths per 100,000 popula-tion) (GBD 2015 Mortality and Causes of Death Collaborators, 2016). Hypopharyngeal carci-noma (HFC) is predominantly squamous cell carcinoma and is the reason for about 5% of head and neck cancers. Nasopharyngeal carcinoma (NFC) is a highly malignant disease, which is more prevalent in southern China, with the overall survival period being 5 years in 70% of the patients afftected (Chou et al., 2008). The evaluation of the role of PCDHs in this group of neoplasms was performed and reported in seven studies (16.3%).
PCDH8 functions as a tumor suppressor (Li et al., 2018). PCDH8 methylation was detected in tumor tissues and not in non-tumor tissues Li et al., 2018). PCDH8 expression was dysregulated or completely silenced in HFC tumor tissues (Li et al., 2018) and NFC cell lines , and the low expression was correlated to the advanced pathological stage of HFC.
PCDH10 hypermethylation was detected in lung  cancer tissues; however, no association was found between the clinicopathological features, except for the smoking pattern (Tang et al., 2013). In addition, it was observed that there was no difference in recurrence pattern in patients (with or without PCDH10 methylation) (Harada et al., 2015). PCDH20 expression was detected in normal lung and nasopharyngeal tissues, but was silenced or downregulated in non-small cell lung cancer and nasopharyngeal cell lines (Imoto et al., 2006;; the same was also observed in nasopharyngeal carcinoma tissues . However, PCDH20 methylation was not associated with the clinicopatho-logical features in NFC , but was associated with the clinical outcome in severe lung cancer, indicating that the inactivation of PCDH20 could occur regardless of the stage of cancer and that malignancy progression may be involved in this fact (Imoto et al., 2006). Thus, PCDH20 may have several tumor suppression functions, potentially contributing to tumor growth control, signal transduction, and cell-cell adhesion .
From this meta-analysis, we observed that the combined odds ratio in upper airway tract tumors of 6.3 (95% CI = 0.34-118.44) for PCDH methylation and that these proteins could be biomarkers for early detection of tumors. However, the evidence is not strong because of the heterogeneity in results. Thus, further studies are needed to confirm the effectiveness of using PCDH as a biomarker of the tumors discussed herein.
Digestive tract tumors represent a large fraction of human cancers and can occur anywhere in the gastrointestinal tract, but the most common sites are the colon and the rectum. The role of PCDHs in this group of neoplasms has been evaluated in fifteen studies, and the most re-searched are PCDH7, PCDH8, PCDH9, and PCH10.
Genetic deletion is pivotal for PCDH10 inactivation in colorectal cancer (CRC), and PCDH10 silencing or downregulation occurs more frequently in tumor cell lines and primary tumors compared to normal mucosa (Jao et al., 2014). Allelic loss of PCDH10 is also associated with the progression of clinical staging and distant metastasis. Additionally, this type of loss is an independent prognostic factor for predicting poor survival in patients with CRC. Thus, PCDH acts as a tumor suppressor in CRC and plays a role in the restriction of liver metastasis (Jao et al., 2014); PCDH10 methylation can be detected in patient's blood circulation (Danese et al., 2013).
The immunoexpression of u-PCDH is silenced in CRC and its downregulation occurs in both tumor samples and cell lines (Losi et al., 2010). u-PCDH has also been shown to retain β-catenin in the cell membrane of normal colon enterocytes, which implies that the release of β-catenin from this site and translocation to the nucleus in tumor cells has occurred (Losi et al., 2010).
Reduced PCDH17 expression was observed in esophageal squamous cell carcinoma and was closely linked to hypermethylation in the CpG-rich region (Haruki et al., 2010). However, other mechanisms, including posttranslational modification, may contribute to this reduction (Haruki et al., 2010). Poorly differentiated tumors are negative for PCDH17. In addition, in non-neoplastic esophageal epithelia, PCDH17 was expressed in the spinous cell layers, implying that this fact may be dependent on the status of cell differentiation and that the silenced expression acts on the de-differentiation of esophageal neoplasms (Haruki et al., 2010).
Gastric cancer (GC) is the fourth most common gastrointestinal tumor and presents a high mortality rate (Isomoto et al., 2008). In this cancer, PCDH7 immunoexpression showed a gradual reduction in the normal tissue to intraepithelial neoplasia and GC, and an even lower immunoexpression in CG lymph node metastasis (Chen et al., 2017). This immunoexpression was inversely or negatively correlated with Lauren's classification, lymph node metastases, and TNM stage (Chen et al., 2017). This demonstrates that PCDH7 suppresses GC progression (Chen et al., 2017).
It has also been observed that the loss of expression and high methylation of the PCDH8 promoter have been detected in the tumor cells of GC. PCDH8 methylation was also found in peri-lesional tissues, but not in normal gastric tissue, and was associated with lymph node metastasis (Zhang et al., 2012). Loss of PCDH9 expression is associated with epithelial dif-ferentiation and cell metastasis in gastric cancer .
There is a significant correlation between methylation and loss of PCDH10 expression in GC (Yu et al., 2010), and it is silenced or downregulated in GC cell lines. This is the major regulatory mechanism of PCDH10 inactivation (Yu et al., 2009;Deng et al., 2014). However, there was no association between staging and methylation, which may occur because PCDH10 methylation is supposed to be the cause rather than the result of carcinogenesis. This suggests that PCDH10 methylation can be used as a marker for early stage cancer diagnosis (Yu et al., 2010), and the counting of methylated CpG islands has clinical prognostic assessment applicability (Deng et al., 2014;Hou et al., 2015).
Liver cancer is the sixth most commonly diagnosed cancer, the fourth leading cause of cancer death worldwide in 2018, and is most prevalent cancer in Africa and Asia (Bray et al., 2018). PCDH10, PCDH17 and PCDH20 expression levels were downregulated in this neoplasm.
Low PCDH10 expression correlates with tumor size, TNM stage, smoking status, and alcohol consumption pattern (Bing et al., 2019). The ectopic expression of PCDH10, when silenced by methylation, can suppress tumor cell growth, migration, invasion, and colony formation (Fang et al., 2013). In addition, PCDH20 can inhibit metastasis by preventing cell migration (Lv et al. 2015), because it is a sensitive clinical parameter to predict survival in indolent and early stage cases (Wu et al., 2017).
The PCDH17 knockout can inhibit the proliferation, migration, and invasion of hepatocarci-noma cells through the overactivation of the EGFR/MEK/ERK signaling pathway (Dang et al., 2016). PCDH8 methylation correlates with alpha-fetoprotein (AFP) levels in patients with AFP levels >50 ng/mL . Thus, methylation is the main mechanism of loss of expression in hepatocarcinoma .
Thus, we observed from the studies significant heterogeneity (p <0.001), the value of odds of association between methylation and cancer being 10.06 (95% CI = 5.15-19.66). Thus, PCDHs could be biomarkers for the early detection of GIT cancers. However, the evidence is not strong because to the results, for the reason that the results were not sufficiently homogeneous. This fact can be explained by the heterogeneity of the lesions that involve this group of neoplasms with different etiopathogenesis. Bladder cancer (BC) is a heterogeneous disease, the outcome of which is difficult to predict, and ranks 10th in cancer incidence worldwide (Kaufman et al., 2009;Grossman, 2011;Mossanen and Gore, 2014). BC can be classified into two types, that is, noninvasive muscle BC (NIMBC) and invasive muscle BC (IMBC), based on histopathological and clinical aspects (Sun and Trinh, 2015). In many studies, the role of PCDHs in this type of cancer has been evaluated, and the immunohistochemical and methylation profiles were the adopted methodologies. The immunoreactivity levels of PCDH7 and PCDH10 proteins are lower in NIMBC tissues than in normal bladder epithelial tissues Lin et al., 2016). Low PCDH7 immunoexpression has been associated with high pathological grade, recurrence, and tumor progression; this fact is an independent prognostic factor for clinical outcome  and is associated with BC invasiveness .
Prostate cancer (PC) is the fifth leading cause of death in men and is a molecularly heterogeneous disease. Therefore, there is no accurate predictive marker for outcome assessment in cancer patients (Jao et al., 2014). Methylation could occur in PCDH8, PCDH10, and PCDH17 in both PC tumor tissue and serum, and is associated with poor prognosis and shorter biochemical relapse-free survival (Losi et al., 2011;Viu et al., 2014;. Ovarian cancer (OC) is the second most lethal cancer among women worldwide (Siegel et al., 2017), and PCDHs may represent as a possible direction in the search for tumor biomarkers. PCDH17 methylation is important in OC carcinogenesis (Baranova et al., 2018). In addition, low PCDH8 immunoexpression in OC tissues predicts a poor prognosis (Cao et al., 2018).
Renal carcinoma (RC) is one of the most commonly diagnosed urinary malignancies, and the most common histological subtype is clear cell RC, accounting for 80% to 90% of cases (Low et al., 2016;Capitanio and Montorsi, 2016). Aberrant methylation of tumor suppressor genes is involved in tumor initiation and progression (Shenoy et al., 2015). The frequency of PCDH8 and PCDH17 methylation is high in this cancer, and the PCDHs are associated with malignancy clinicopathological characteristics and unfavorable prognosis .
In this meta-analysis, we have shown strong evidence for the occurrence of methylation of some PCDHs in the group of neoplasms discussed in this article, and this is because the results were sufficiently homogeneous. Thus, methylation is a specific event for the tumors discussed and can occur in PCDH7, PCDH8, PCDH10, and PCDH17.
Most of the studies had the subjects and configuration described in detail, reliably measured the results, and used statistical analysis. On the other hand, we did not identify confounding factors, nor did they describe strategies to minimize these factors.
This review has certain limitations that must be taken into consideration. None of the articles reported the sample size calculation; all census-type samples were collected for convenience. Age was reported heterogeneously in the studies, and it was not possible to categorize individuals to verify the most affected age group. There was a heterogeneous geographic distribu-tion of studies, focusing on Asia. In this review, the research was limited to clinical samples and laboratory techniques, and it was not possible to speculate the definitive clinical utility of PCDHs. In addition, there was no standardization of the methodologies adopted. In most of the studies, no follow-up of patients with regard to recurrence and disease progression was reported. Finally, the heterogeneity of malignancies and PCDHs made it difficult for us to compare the large number of studies included.
In this systematic review, we have demonstrated that PCDHs are often silenced by DNA methylation in different malignant neoplasms, of which genitourinary tract tumors are more prevalent. Thus, PCDHs could emerge as potential tumor suppressor genes, and a significant increase in methylation may be useful for early detection of different types of cancers. Moreover, the existence of methodologies with great heterogeneity reinforces the importance of studies to evaluate the role of this group of proteins in the carcinogenesis of other neoplasms.