Prognostic and Clinicopathological Significance of Downregulated p16 Expression in Patients with Bladder Cancer: A Systematic Review and Meta-Analysis

p16, encoded by the CDKN2A gene, is a tumor suppressor that has been widely studied in cancer research. However, the relationship of p16 with prognostic and clinicopathological parameters in patients with bladder cancer remains unclear. Data inclusion criteria were articles reporting on the relationship between p16 expression and the prognosis or clinicopathology in patients with bladder cancer. Meta-analyses were performed with Stata software. Hazard ratios (HRs) or odds ratios (ORs) and 95% confidence intervals (95% CI) were calculated to evaluate the relative risks. The source of heterogeneity was analyzed by subgroup analysis. A total of 37 studies with 2246 cases were included and analyzed. The results identified an important link between downregulated p16 expression and poor prognosis in patients with bladder cancer in terms of recurrence-free survival (RFS), overall survival (OS), progression-free survival (PFS), and some clinicopathological parameters including clinical staging, pathological degree, and lymph node metastasis. Subgroup analysis also showed that low p16 expression could function as a warning sign for RFS and PFS in patients with early-stage (Ta–T1) bladder cancer. In conclusion, p16 might play an essential role in the deterioration of bladder cancer and could serve as a biomarker for the prediction for patients' progression and prognosis.


Introduction
Bladder cancer is the most frequent malignancy of the urinary tract and the ninth most common cancer worldwide [1]. About 95% of bladder cancers are histologically transitional cell carcinoma, with rare cases of squamous cell carcinoma and adenocarcinoma. However, the pathogenesis of bladder cancer is still unclear, and its occurrence and development appear to be affected by multiple genes [2]. Serrano et al. first cloned the cDNA of the gene encoding the tumor suppressor protein p16 (CDNK2A) in 1993; since then it has been widely studied in the field of cancer research [3].
Previous studies have reported ubiquitous downregulation of p16 gene expression in bladder cancer, as a result of various alterations including complete deletion, point mutation, or promoter methylation [4][5][6]. Furthermore, p16 could compete with cyclin D1 for binding to Cyclin Dependent Kinase (CDK) 4/6, thus blocking the phosphorylation of retinoblastoma (Rb) protein and inhibiting release of the transcription factor E2F, preventing cell conversion from G 1 phase to S phase, and eventually suppressing cell proliferation. These results suggest that abnormal expression of the p16 gene in cells might be associated with tumorigenesis [6,7].

Inclusion and Exclusion
Criteria. Inclusion criteria were as follows: (1) patients diagnosed with bladder cancer; (2) immunohistochemical (IHC) detection of p16 expression levels in the tissues; (3) relationships between abnormal expression of p16 and prognostic indicators such as recurrence-free survival (RFS), progression-free survival (PFS), and overall survival (OS) or associations between p16 and clinicopathological features that were evaluated; (4) hazard ratio (HR), odds ratio (OR), relative risk (RR), and 95% confidence intervals (CI) that could be obtained directly from the full article or indirectly calculated with relevant software based on the data provided in the graphics and tables; (5) only the newest studies or the ones with higher quality were retained if the data were repeated in different studies; and (6) studies in English or Chinese.
Exclusion criteria were as follows: (1) cell or animal studies, case reports, letters, reviews, and meta-analyses; (2) articles with similar content or using the same data or those with small sample sizes ( ≤ 10) and those with no directly or indirectly extractable HR, OR, and 95% CI data; and (3) articles that could not be understood because of language barriers.

Data Extraction.
Two independent investigators (Xiaoning Gan and Rongquan He) reviewed the articles that met the criteria and extracted data on author, year of publication, nationality, sample size, patient age, detection method of p16, antibody source and dilution, clinical stage, pathological degree, other costudied prognosis-associated genes, cut-off value, outcome, and extraction method of the study subjects. Discrepancies between the two independent investigators in  terms of data extraction were resolved by discussion among all the authors.

Statistical Analysis.
Effects of p16 on the related prognostic indexes were detected by merging the HRs and 95% CI of the included literatures, which were evaluated through the Forest plot and related parameters after the merging. The HRs and 95% CI values mainly came from direct extraction of the original text or survival curve through extraction and calculation by software.
The relationships between p16 and the clinicopathological parameters were derived from the binary variable data extracted from the original articles. ORs and 95% CI values came from the binary variable data calculated by Stata software. The data were then combined, and their statistical significance was evaluated by Forest plot and related parameters, to clarify the relationship between p16 low-expression and clinicopathological parameters.
Publication bias was detected by Begg's funnel plot and Egger's test with Stata software. A two-sided value < 0.05 was considered to indicate statistical significance. Statistical analyses were carried out with StataSE 12.0, Engauge, Photoshop CS5, and Microsoft Office 2007.

Eligible Studies.
A total of 364 articles were identified from the databases, including 190 English and 174 Chinese articles, 222 of which were excluded because of discrepancies between the study theme and their abstracts. The full text of the remaining 142 articles was then reviewed for their fit with the current study, after which a further 105 articles were excluded because they met one or more of the exclusion criteria, such as the cell or animal studies, reviews, and letters and studies with identical data and no extractable HR, OR, and 95% CI data from the full text or language barrier. The remaining 37 articles [4,5, with 2246 cases were included in our study and consisted of 21 English [4, 5, 8-19, 21-25, 41, 42] and 16 Chinese [20,[26][27][28][29][30][31][32][33][34][35][36][37][38][39][40] articles. The screening process was demonstrated in Figure 2.
Cumulative meta-analysis based on year of publication and sample size demonstrated that the results tended to stabilize with increasing sample size, but there was no obvious relationship between the results and year of publication.
Based on sensitivity analysis, the study by Yang et al. [13] was initially excluded because of a large difference in HR compared with the overall average, which was attributed to the selection of a different calculation method in the original article. Binary variable data were extracted and the HR and 95% CI were therefore recalculated with Stata software.     In addition, subgroup analysis of early-stage data from 430 subjects from eight studies also demonstrated that low expression of p16 significantly affected RFS in patients with early-stage (Ta-T1) bladder cancer (HR = 1.96, 95% CI = 1.23∼3.14, and = 0.005; 2 = 47.9%, = 0.088).
Cumulative meta-analysis and sensitivity analysis indicated relatively low overall heterogeneity and no study with high sensitivity.
Subgroup analysis was also performed based on clinicopathological stages. However, limitations of sample size led to the impossibility of determining if the effects of p16 expression on OS were associated with these parameters in patients with bladder cancer ( Ta-T1  Subgroup analysis based on cut-off value indicated that the effects of p16 on OS in patients with bladder cancer were associated with cut-off value (cut-off value ≤ 10%: HR = 1.83, 95% CI = 1.17∼2.86, and = 0.008; 2 = 3.2%, = 0.006; cutoff value > 10%: HR = 1.40, 95% CI = 0.66∼2.96, and = 0.384; 2 = 0%, = 0.951).
Cumulative meta-analysis revealed no obvious characteristics because of the limited range of publication dates and the sample sizes.
Sensitivity analysis identified two studies [41,42] as having the highest heterogeneities. Further investigation revealed that this heterogeneity was caused by different methods of measuring p16 (fluorescence in situ hybridization) and studying the influence of hemizygous or homozygous deletion of p16 on patient prognosis. These two studies were finally excluded because of their incompatible study objectives, leaving a total of 347 subjects from six studies in the final analysis of PFS. The results showed that low expression of p16 was correlated with poor PFS in patients with bladder cancer, and the heterogeneity was eliminated (HR = 1.84, 95% CI = 1.13∼3.01, and = 0.015; 2 = 0%, = 0.487) (Figure 3(c)).

Relationship between Low Expression of p16 and DSS/CSS in Patients with Bladder Cancer.
A total of 187 subjects from three studies were included in the DSS/CSS analysis [9,17,18]; limitation of the sample size caused the impossibility of demonstrating an association between low expression of p16 and DSS/CSS (HR = 1.52, 95% CI = 0.85∼2.71, and = 0.149; 2 = 0%, = 0.825). Analysis of the results for lymph node metastasis showed OR = 2.20, 95% CI = 1.26∼3.83, and = 0.006; 2 = 27.2%, = 0.240. The small sample size caused the impossibility of demonstrating any significant influence of pathological parameters such as muscle invasion, tumor number (multiple/single), and tumor size on the effect of p16 expression ( Table 2).

Retrospective Review.
Three studies [6,24,25] were retrospectively reviewed because of differences between their prognosis results and the data required by the meta-analysis. As shown in Table 3, low expression of p16 was associated with poor prognosis in patients with bladder cancer. However, some of the values were <0.05 because of the small sample sizes.

Publication
Bias. Publication bias was detected by Begg's funnel plot and Egger's test (Figure 4). The points representing studies were symmetrically arranged in a funnel shape in the funnel plot, and the values calculated from Egger's test with higher detection effectiveness were >0.05, indicating no publication bias. The only exception was for RFS; the funnel plot was asymmetrical and with a few points outside the funnel. Publication bias was also detected by Egger's test (G1-G3 group: = 0.031; Asia group: = 0.020), indicating potential publication bias in terms of RFS.

Discussion
p16, also known as tumor suppressor gene I (multiple tumor suppressor, MTS I), is located in 9p21 and is composed of two introns and three exons [45]. It is a key gene in cell cycle regulation, with its expression product being involved in the negative regulation of cell proliferation. Studies have shown that downregulation of p16 gene expression resulted in the loss of its inhibitory effects on CDK4/CDK6, which in turn may lead to malignant, abnormal cell proliferation and accelerated tumor development [7,46,47]. Elucidation of the relationship between low expression of p16 and prognosis and clinicopathology in patients with bladder cancer is therefore important for its early diagnosis, treatment, and prognosis.
Pan et al. performed a meta-analysis of the prognostic significance of abnormal p16 and p21 expression in bladder cancer in 2006 [48]. However, the current study analyzed a larger sample size; Pan et al. 's study included 12 articles with 975 cases, compared with 37 articles and 2246 cases 8 Disease Markers   in our study, leading to more accurate and reliable results. Secondly, Pan et al. 's study involved a number of mixed factors with no clear listing of each prognostic index or subgroup discussion. In contrast, the current study included subgroup analyses for the different indicators including RFS, PFS, OS, and DSS/CSS, allowing more thorough insights into the relationships between p16 expression and the prognostic and clinicopathological parameters in bladder cancer patients. Thirdly, Pan et al. found no association between p16 expression and prognosis in early Ta-T1 stage (stage I) bladder cancer, possibly because of the omission of the study by Krüger et al. [5], which explored the significance of p16 as an independent tumor predictive factor for the development of T1 bladder cancer, and demonstrated the important clinical value of low p16 expression in the early diagnosis and prognosis of patients with early-stage bladder cancer.
The current study systematically analyzed the relationships between p16 expression and prognostic index and clinicopathological parameters in patients with bladder cancer and showed that low expression of p16 was closely correlated with poor prognosis ( Figure 5). However, the included studies varied in terms of study subjects, design, sample size, interventions, outcomes, time of study, and publication date. We used cumulative meta-analysis, sensitivity analysis, and subgroup analysis to explore the effects of the main variables in the included studies. Overall, the results confirmed that the relationship between low expression of p16 and prognosis in patients with bladder cancer was affected by clinicopathological stage, geographic origin of the study subjects, detection method, and cut-off values. Based on these findings, we further analyzed the relationships between p16 expression and clinicopathological parameters and demonstrated associations between low expression of p16 and clinical stage and lymph node metastasis, implying that the p16 gene tended to exert its regulatory effects during the early stage of bladder carcinogenesis. Low expression of p16 was also correlated with poor PFS and RFS in early-stage (Ta-T1) bladder cancer. These results thus confirmed an important role for p16 in the occurrence and development of bladder cancer. Meanwhile, through Phase I and II clinical trials, studies have revealed that CDK4/6 is an attractive target in p16 related pathway for anticancer therapy [49][50][51]. Furthermore, previous study also suggested that p16 functional peptide, as a molecular targeting agent, showed effective reactions for the treatment of renal cell carcinoma [52].   The results of these researches and our current metaanalysis had the effect of mutual authentication. Therefore, a better understanding of the mechanism underlying the development and progression of bladder cancer may play a significant role in prevention, target therapy, and prognosis, particularly if more sensitive and specific correlative biomarkers can be discovered and verified. The current study had some limitations. First, tumors are the result of both environmental and genetic factors, and p16 may thus be only one of several factors involved in the whole process of bladder carcinogenesis. Secondly, heterogeneity may result from differences in intervention measures (surgery, radiotherapy, chemotherapy, or combination), immunohistochemical techniques (different antibodies, evaluation standards, etc.), and the HR extraction methods used in the included studies. Finally, the exclusion of articles because of language barriers and of studies that were not published because of a lack of sufficient data may have led to potential publication bias.
In conclusion, the results of the current study provide evidence for a relationship between p16 expression and prognosis and clinicopathological features in patients with bladder cancer. The results of this meta-analysis will help to inform about the development of clinical guidelines promoting best medical care for patients with bladder cancer. Further studies are required to investigate the combined influence of genetic and environmental factors on the development and progression of bladder cancer.