Clinical significance of p16INK4A and p14ARF promoter methylation in renal cell carcinoma: a meta-analysis

The inactivation of p16INK4A and p14ARF via promoter methylation has been investigated in various cancers. However, the clinical effects of p16INK4A and p14ARF promoter methylation on renal cell carcinoma (RCC) remain to be clarified. The pooled data were calculated and summarized. Finally, an investigation of 14 eligible studies with 1231 RCC patients and 689 control patients was performed. Methylated p16INK4A and p14ARF were observed to be significantly higher in RCC than in control subjects without malignancies (OR = 2.77, P = 0.005; OR = 11.73, P < 0.001, respectively). Methylated p16INK4A was significantly associated with the risk of RCC in the tissue subgroup, but not in the serum and urine subgroups. Methylated p16INK4A was significantly associated with tumor size. We did not find that p16INK4A promoter methylation was associated with sex, tumor grade, lymph node status, and tumor histology. Methylated p14ARF was significantly correlated with sex and tumor histology. Three studies reported that p16INK4A methylation was not significantly correlated with the prognosis of RCC. The results suggested that p16INK4A and p14ARF promoter methylation may be correlated with the carcinogenesis of RCC, and that methylated p14ARF, especially, can be a major susceptibility gene. We also found the different clinicopathological significance of 16INK4A and p14ARF in RCC. Additional studies with sufficient data are essential to further evaluate the clinical features and prognostic effect of p16INK4A and p14ARF promoter methylation in RCC.


INTRODUCTION
Renal cell carcinoma (RCC) is one of the most common cancers of the human urinary system. Based on cancer statistics, approximately 62,700 new cases will be reported in clinics, with approximately 14,240 deaths in the USA in 2016 [1]. Clear cell renal cell carcinoma (ccRCC) is the most common histological type of RCC, accounting for 70% to 75% of all RCCs [2]. Most patients with RCC are symptom-free in the early stage, and more than 50% of RCCs are found coincidentally by physical examination and imaging [3]. Approximately 30% of the patients with RCC have developed metastases, and the average 5-year survival rate is just 12.3% [4].
Epigenetic and genetic changes are identified to be significantly associated with cancer [5,6]. DNA methylation is an important mechanism of epigenetic alterations involved in gene expression, which is closely associated with the carcinogenesis and progression of various carcinomas [7][8][9]. The transcription repression of the gene via CpG island methylation of the promoter can lead to the downregulation of gene expression [10,11]. Located at chromosome 9p21, cyclin-dependent kinase inhibitor 2A (CDKN2A) has two alternative splicings, www.impactjournals.com/oncotarget/ Oncotarget, 2017, Vol. 8, (No. 38), pp: 64385-64394 Meta-Analysis encoding the cell cycle regulatory proteins p16 INK4A and p14 ARF , which have a key function in regulating the activities of the retinoblastoma (RB) and p53 genes, respectively [12,13]. p16 INK4A and p14 ARF as tumor suppressor genes are involved in the regulation of cell division and apoptosis, and the maintenance of cellular homeostasis [14]. The inactivation of p16 INK4A and p14 ARF through promoter methylation has been reported in many cancers [15][16][17]. Promoter methylation of p16 INK4A and p14 ARF has been shown in different sample types of RCC, including blood, urine, and tissue samples [18][19][20][21].
Although some studies involving p16 INK4A and p14 ARF promoter methylation included patients with RCC, the studies published in this field have had small sample sizes. In addition, whether p16 INK4A and p14 ARF promoter methylation is associated with clinical characteristics of RCC remains to be determined. Therefore, in this study, we performed a systematic meta-analysis to further evaluate the clinical significance of p16 INK4A and p14 ARF promoter methylation in RCC.

Study characteristics
One hundred sixty-six potentially relevant studies were identified by the initial literature search. According to the inclusion criteria, a total of 14 studies involving 1231 RCC patients and 689 control patients [18][19][20][21][22][23][24][25][26][27][28][29][30][31] were included in the current analysis ( Figure  1). Of these studies, which involved p16 INK4A and p14 ARF gene promoter methylation, nine studies evaluated the association between p16 INK4A promoter methylation and RCC risk, five studies assessed the correlation between p14 ARF promoter methylation and RCC risk, ten studies evaluated the relation between p16 INK4A promoter methylation and clinicopathological features, and four studies evaluated the relation between p14 ARF promoter methylation and clinicopathological features. The general characteristics of included studies are presented in Table 1.

Association between p16 INK4A and p14 ARF promoter methylation and RCC risk
When cancer patients were compared to control subjects, the result of p16 INK4A promoter methylation with strong heterogeneity was conducted using a random-effects model (I 2 = 51.7% and p = 0.029); under a fixed-effects model, no obvious heterogeneity was found for p14 ARF promoter methylation (I 2 = 0.0%; and p = 0.667) ( Figures  2 and 3).
A significant association was found between p16 INK4A promoter methylation and RCC (OR = 2.77, 95% CI = 1.36 -5.66, P = 0.005), including in 582 of the cancer patients and 422 of the controls ( Figure  2). The pooled OR of p14 ARF promoter methylation in RCCs was significantly higher than in controls (OR = 11.73, 95% CI = 4.11 -33.47, P < 0.001), including in 317 of the cancer patients and 267 of the controls ( Figure 3).

Meta regression and sensitivity analyses of p16 INK4A promoter methylation in cancer patients versus control patients
Meta regression based on sample type (tissue, serum, or urine), ethnicity (Caucasian or Asian), and testing method (MSP or non-MSP) was performed to find the potential sources of heterogeneity ( Table 3).
A sensitivity analysis was also conducted to evaluate the stability of the overall OR and the change of heterogeneity by deleting a single study. When a study from Hoque 2004 et al. ( [21], urine) was removed, the pooled OR was not significantly changed (OR = 2.06, 95% CI = 1.41-3.02), with no obvious heterogeneity (I 2 = 23.2%, and P = 0.237).

Relation of p16 INK4A and p14 ARF promoter methylation and clinicopathological features
We further determined whether p16 INK4A and p14 ARF promoter methylation status was associated with clinicopathological characteristics, such as sex, tumor grade, tumor stage, tumor size, lymph node status, and tumor histology. The fixed-effects model was used in relation to clinicopathological characteristics in cancer (all p > 0.1) ( Table 4).

Association of p16 INK4A and p14 ARF methylation and gender in cancer
The pooled OR from four studies suggested that p16 INK4A promoter methylation was not significantly correlated with gender in RCC (OR = 0.66, 95% CI = 0.31-1.38, P = 0.266), including in 201 males and 91 females ( Table 4). The pooled OR from three studies involving 168 males and 76 females suggested that p14 ARF promoter methylation was significantly correlated with gender in RCC (OR = 0.48, 95% CI = 0.25-0.94, P = 0.032) ( Table 4), indicating that it was lower in males than in females.

Association of p16 INK4A and p14 ARF methylation and the pathological T category of primary tumor (pT) in cancer
The pooled OR from six studies including 132 pT2-4 patients and 203 pT1 patients suggested that p16 INK4A promoter methylation was significantly correlated with tumor size in RCC (OR = 2.43, 95% CI = 1.10-5.35, P = 0.028) ( Table 4), indicating that it was higher in pT2-4 than in pT1. The pooled OR from four studies suggested that p14 ARF promoter methylation was not significantly correlated with tumor size in RCC (OR = 0.92, 95% CI = 0.44-1.91, P = 0.815), including 94 pT2-4 patients and 180 pT1 patients ( Table 4).

Association of p16 INK4A and p14 ARF methylation and lymph node status in cancer
The pooled OR from five studies and two studies showed that p16 INK4A and p14 ARF promoter methylation was not significantly associated with lymph node status in RCC  Table 4).

Association of p16 INK4A and p14 ARF methylation and tumor histology in cancer
The pooled OR from seven studies comprising 289 ccRCC and 144 non-ccRCC patients suggested that p16 INK4A promoter methylation was not significantly associated with tumor histology in RCC (OR = 0.54, 95% CI = 0.29-1.00, P = 0.051) ( Table 4). The pooled OR from four studies involving 185 ccRCC and 100 non-ccRCC patients demonstrated that p14 ARF promoter methylation was significantly correlated with tumor histology in RCC (OR = 0.38, 95% CI = 0.18-0.81, P = 0.012) ( Table 4), suggesting that it was lower in ccRCC than in non-ccRCC.

Prognostic value of p16 INK4A and p14 ARF gene promoter methylation in RCC
The detailed overall survival (OS), and diseasefree survival (DFS) data on p16 INK4A or p14 ARF gene promoter methylation as a prognostic factor for RCC were insufficient. The mean follow-up time for the participants ranged from 28 months [22] to 76 months [23] in this meta-analysis. Dulaimi et al. 2004 [25], Vidaurreta et al. 2008 [23], and Martino et al. 2012 [22] reported that p16 INK4A methylation was not significantly associated with the prognosis in DFS or OS (Table 1). Dulaimi et al. 2004 [25] reported that p14 ARF methylation was not significantly associated with the prognosis in OS (Table 1). More studies with sufficient data are necessary to further evaluate the prognostic value of p16 INK4A and p14 ARF promoter methylation in RCC.

Publication bias
The Egger test was used to evaluate potential publication bias. The Egger test showed low publication bias for p16 INK4A promoter methylation in cancer patients versus control patients, and in cancer in relation to lymph node status and tumor histology (P = 0.011, P = 0.02, P = 0.015, respectively) (Tables 2 and 4).

DISCUSSION
The p16 INK4A is formed from an alternative transcript of exons 1α, 2, and 3, whereas p14 ARF is translated from alternative reading frames (ARF) consisting of exons 1β, 2, and 3 [32]. The silencing of p16 INK4A and p14 ARF can result in uncontrollable cell proliferation and tumor growth [32,33]. Methylated p16 INK4A and p14 ARF have been investigated in various cancers, including RCC [18], esophageal squamous cell carcinoma [34], melanoma [35,36], and gliomas [37]. Although numerous studies have been conducted to evaluate the role of p16 INK4A and p14 ARF promoter methylation in RCC, the results are still inconsistent and controversial. Kasahara et al. [38] found that the frequency of the methylated p14 ARF was 0% in RCC. Hori et al. [20] found that the frequency of the methylated p14 ARF was 70.5% in RCC. Morris et al. [27] reported that the frequency of p16 INK4A promoter methylation was 0% in RCC. Arai et al. [24] reported that the frequency of p16 INK4A promoter methylation was 73.3% in RCC. Therefore, we conducted this study of all available articles to further evaluate the effects of p16 INK4A and p14 ARF promoter methylation in RCC.
Analysis of the pooled OR showed that p16 INK4A and p14 ARF promoter methylation were significantly higher in patients with RCC than in control subjects, suggesting that p16 INK4A and p14 ARF inactivation via promoter methylation may play an important role in the tumorigenesis of RCC. Interestingly, p14 ARF promoter methylation had a higher OR value (OR = 11.73) than that of p16 INK4A promoter methylation (OR = 2.77) in cancer patients versus control patients, suggesting that RCC can be more susceptible to p14 ARF promoter methylation.
When RCCs were compared to nonmalignant samples, the heterogeneity of p16 INK4A promoter methylation was high (I 2 = 51.7%, P = 0.029). According to sample type (tissue, serum, or urine), ethnicity (Caucasian or Asian), and testing method (MSP and non-MSP), subgroup analyses and meta-regression were used to explore the possible sources of heterogeneity. Analysis showed that subgroup analyses and meta-regression failed to find heterogeneity. Moreover, based on subgroup analyses of sample types, a significant association was observed between p16 INK4A promoter methylation and tissue subgroup, but not in the serum and urine subgroups. The results should be carefully considered as only one study or two studies with a small number of samples involved in subgroup analyses. A sensitivity analysis was also performed in our study; when we deleted a study (Hoque 2004 et al., urine) [21], the overall OR was not significantly changed, with no significant heterogeneity, suggesting that our result was stable and reliable. We further determined whether p16 INK4A and p14 ARF promoter methylation were correlated with clinicopathological features. Methylated p14 ARF was significantly associated with gender, in which it was lower in males than in females, suggesting that female RCC patients can be more susceptible to p14 ARF promoter methylation, whereas methylated p16 INK4A had a similar frequency in males and females. Methylated p16 INK4A was significantly associated with tumor size, in which it was higher in pT2-4 patients than in pT1 patients, suggesting that p16 INK4A promoter methylation may play a key role in the pathogenesis of T2-4, whereas methylated p14 ARF was not significantly correlated with tumor size. Methylated p14 ARF was significantly associated with tumor histology, and it was lower in ccRCC than in non-ccRCC, suggesting that p14 ARF promoter methylation had a decreased risk of ccRCC; whereas methylated p16 INK4A had a similar frequency in ccRCC and Non-ccRCC. In addition, our findings showed that p16 INK4A and p14 ARF promoter methylation were not significantly associated with tumor grade, tumor stage, and lymph node status.
The prognostic data involving the pooled hazard ratio (HR) were insufficient and not available, as only three studies reporting showed that p16 INK4A and p14 ARF gene promoter methylation were not significantly correlated with the prognosis of RCC in OS or DFS [22,23,25]. More studies with sufficient data need to be done in the future.
The current study had several potential limitations. First, analysis of p16 INK4A promoter methylation showed a slight publication bias in cancer versus control, and in cancer in relation to lymph node status and tumor histology. The articles with positive results are more often published than articles with negative results. The study was restricted to literatures published in English, which can lead to bias. In addition, because fluid samples from serum, plasma, and urine were limited, additional studies will be essential to evaluate the value of fluid detection in the future. Finally, the primary ethnic groups were Asian and Caucasian; thus, further studies using a larger variety of ethnic groups are warranted.
In conclusion, our study showed that RCC had a higher p16 INK4A and p14 ARF gene promoter methylation than did nonmalignant control patients. RCC had a higher p16 INK4A promoter methylation in pT2-4 than in pT1. However, RCC had a lower p14 ARF promoter methylation in males than in females, and was also lower in ccRCC than in non-ccRCC. Further large-scale studies with well-designed research are necessary to validate the role of p16 INK4A and p14 ARF promoter methylation in the prognosis and clinical effects of RCC patients in the future.

Literature search
This meta-analysis was conducted based on the preferred reporting items for systematic reviews and meta-analyses (PRISMA) statement criteria [39] (Supplementary Table 1

Inclusion criteria
Eligible studies were selected in this meta-analysis if they met the following criteria: 1) patients were diagnosed with primary RCC; 2) although tissue specimens used must include surgically resected primary tumor samples, other samples, such as serum, plasm and urine, were used; 3) CDKN2A methylation included p16 INK4A and p14 ARF promoter methylation; 4) studies with sufficient data on p16 INK4A and p14 ARF promoter methylation frequency were selected to assess the association between p16 INK4A and p14 ARF promoter methylation and RCC; 5) to avoid duplicated publications, only the most recent paper or the most complete paper was included in the current study.

Data extraction
We collected information from each eligible report regarding first author's name, country, ethnicity, testing method, sample type, methylation frequency, the number of samples, gender, tumor grade, clinical staging, pT, lymph node status, tumor histology, OS, and DFS. The whole data extraction was conducted independently by two authors, and minor disparities were solved by discussion.

Data analyses
All statistical analyses were performed using STATA software (version 12.0, Stata Corporation, College Station, TX, USA). The pooled OR and 95 % confidence interval were calculated to assess the strength of the association between p16 INK4A and p14 ARF genes promoter methylation and RCC. Heterogeneity among studies was examined by Cochran test and the I 2 test [40]. If I 2 greater than 50% or p value less than 0.1 was considered as a measure of significant heterogeneity, then the random-effects model was applied in this study; otherwise, the fixed-effects model was used [41,42]. The meta-regression and subgroup analyses were conducted to explore the source of heterogeneity. A sensitivity analysis was also performed to assess the contributions of an individual study on the overall OR by omitting one study [43]. Any possible publication bias was detected using the Egger linear regression test [44].

Author contributions
YR and GW contributed to the conception and design. YR, LX and BS contributed to the completion of articles, the extraction of data, the calculation of data and the design of figures and tables. All the authors approved the final manuscript.