P16 DNA Methylation Coupled with Somatic Copy Number Variations in the Development of Gastric Carcinomas Authors: Ziqian Yang, Jing Zhou, Lewen Deng, Juanli Qiao, Liankun Gu, Dajun Deng Published: 2026 ABSTRACT Simple SummaryAs one of the genes mostly changed in cancer genomes, CDKN2A/P16 is often inactivated by copy number loss and promoter DNA methylation. It is unknown whether these two inactivation mechanisms are relevant. In this study, we determined the levels of CDKN2A/P16 copy number and DNA methylation in gastric cancer (GC) and paired normal tissue samples from 200 patients. We found for the first time more P16 deletions in GC tissues without vs. with P16 methylation. Both P16 copy number and DNA methylation levels are significantly decreased during GC development and are associated with GC metastasis. Our findings indicate that both somatic copy number deletion and promoter DNA methylation complementarily inactivate CDKN2A/P16 in GC development and promote GC metastasis. 1. Introduction CDKN2A encodes both P16 and P14 proteins. DNA methylation of CpG islands around the transcription start site (TSS-CGI) of the P16 gene is prevalent in precancerous lesions across organs and drives cancer development and metastasis [1,2,3,4,5]. Our recent findings indicate that both P16 somatic copy number deletion and amplification (SCNdel and SCNamp, respectively) are prevalent in precancerous esophageal squamous cell dysplasia and noncancerous surgical margin (SM) tissues from gastric carcinoma (GC) patients. P16 SCNdel is significantly associated with poor prognosis in patients with ESCdys and GC, whereas P16 SCNamp is associated with good prognosis in patients with these diseases [6,7,8]. Both genetic and epigenetic inactivation of the P16 gene are frequent early driving events in cancer development in humans [9,10]. TSS-CGI hypermethylation not only epigenetically inactivates gene transcription [11] but also causes chromatin condensation. Because P16 inactivation leads to dysfunction of the G1-S checkpoint in the cell cycle via the RB1 pathway [12,13], RB1 loss of function by P16 inactivation may consequently cause replication stress and genome instability [14,15,16]. However, whether somatic copy number variations (SCNVs) are coupled with DNA methylation of the TSS-CGI of genes, including P16, has not been studied previously. In this study, to investigate the relationship between P16 SCNVs and DNA methylation in cancer development, we compared the P16 SCNV frequency in GC and SM samples with and without P16 TSS-CGI methylation from 200 patients. We revealed for the first time that P16 SCNdel is more prevalent in gastric tissues, mainly in GCs, without P16 methylation than in those with P16 methylation, whereas P16 SCNamp is more prevalent in gastric tissues, mainly in SMs, with P16 methylation than in those without P16 methylation. P16 methylation and the SCNV complementarily promote GC development and metastasis. 2. Methods and Materials 2.1. Patients The cohort was composed of 200 GC patients, including 140 males and 60 females; 106 and 94 patients with and without lymph metastasis, respectively, who underwent gastrectomy at the Peking University Cancer Hospital from January 2013 to November 2016. GC, SM (5 cm away from the main cancer mass, from the greater curvature on the distal side), and white blood cell (WBC) samples were freshly collected from these patients and stored in a freezer at −80 °C for approximately 8–10 years at the biobank of the hospital. No cancer cells were observed in these SM samples under an optical microscope. Clinicopathological information and overall survival data were collected. Among these 200 patients, 80 were included in our previously published study, in which only data on P16 SCNV, but not P16 methylation data, were available [8]. Detailed information for all 200 patients is listed in Data File S1 and summarized in Table 1. The Institute Review Board of the Peking University Cancer Hospital and Institute approved the study and the patients provided written informed consent to participate. 2.2. Preparation of Genomic DNA Genomic DNA was extracted from the abovementioned frozen GC, SM, and WBC samples via the phenol/chloroform technique and used for P16 copy number analysis. The GC and SM DNA samples were further modified with sodium bisulfite with an EZ DNA Methylation-Gold Kit (Zymo Research, Tustin, CA, USA) following the manufacturer’s instructions and used for P16 methylation analysis. 2.3. Quantification of P16 Methylation Using the MethyLight Assay An established 115 bp MethyLight assay [17] was used to quantify the proportion of methylated P16 alleles in triplicate. The COL2A1 gene, which contains no TSS-CGI, was used as an internal reference. When the copy number of methylated P16 relative to COL2A1 was greater than 2.74 × 10−4 in one of three PCR tubes for a bisulfite-modified DNA sample, the sample was defined as P16 methylation-positive (P16M), P16 methylation-negative otherwise (P16U). 2.4. Quantification of P16 Copy Number Using the P16-Light Assay CDKN2A/P16 copy number (CN) was quantified using droplet digital PCR based on P16-Light [8,18], in which the copy number of GAPDH was used as an internal reference gene. 2.5. Definitions of CDKN2A/P16 SCNamp and SCNdel As we defined previously, the average CN of P16 in WBCs from each patient was used as the diploid reference. The difference in the average P16 copy number between the tested (GC or SM) sample and the paired WBC sample was calculated for each patient. When the difference in the copy number was statistically significant (p < 0.05) according to Student’s t test and the absolute fold change was greater than 20%, we defined the sample as P16 SCNamp- or SCNdel-positive, as we previously reported [8]. 2.6. Statistical Analysis We used the Wilcoxon test or Mann–Whitney test to compare the proportion of methylated P16 alleles and Student’s t test or chi-square test to compare P16 copy number between different SM and GC samples or subgroups. Log-rank univariate analysis was used to compare patient overall survival between groups in the K–M analysis. All tests were two-sided, and a p value less than 0.05 indicated statistical significance. 3. Results 3.1. Basic Results of P16 SCNV and P16 Methylation Analyses in GC and Paired SM Samples from 200 Patients The results of P16 CN and TSS-CGI methylation analyses were obtained with P16-specific droplet digital PCR and MethyLight assays for GC, SM, and WBC samples from all 200 patients (Figure 1). A considerable frequency of P16 SCNamp was observed in both SMs and GCs (11.5% vs. 10.5%). Moreover, P16 SCNdel mostly occurred in GCs rather than SMs (30.5% vs. 6.5%, p < 0.001) (Table 1). The average P16 CN value was significantly greater in SMs than in GCs (Figure 2A). Similarly, the percentage of P16-methylated samples and the overall prevalence of P16 methylation were significantly greater in SMs than in GCs (81.0% vs. 56.0%, p < 0.001; median, 0.0013 vs. 0.0004, p = 0.002; Figure 2B), although the P16 methylation level for P16M samples was slightly lower in SMs than in GCs (median, 0.0018 vs. 0.0020; Table 2). 3.2. P16 SCNamp Coupled with P16M, Whereas P16 SCNdel Coupled with P16U in Gastric Tissues We further compared the level of P16 CN (relative to that in WBCs) in GC and SM samples with and without P16 methylation. We found that the average P16 CN was significantly greater in P16M GC samples (n = 112) than in P16U GC samples (n = 88) (Figure 3A). A similar but nonsignificant difference was also observed in the SM samples (Figure 3B). Notably, more P16 SCNdel was detected in P16U GC samples (34/88 = 38.6%) than in P16M GC samples (27/112 = 24.1%, p = 0.027). These results suggest that P16 SCNdel is closely coupled with P16U, whereas P16 SCNamp is coupled with P16M in gastric cancer tissues from GC patients. 3.3. P16 SCNVs and P16M in GC or SM Samples Are Complementarily Associated with GC Metastasis Both the average P16 CN value and the P16 SCNamp-positive rate were significantly greater in SMs from patients without vs. with metastasis (2.14 vs. 2.05, p = 0.002 for P16 CN; 19.1% vs. 4.7%, p = 0.004 for P16 SCNamp; Table 1). In addition, the average P16 CN value and P16 SCNamp-positive rate were significantly greater in poorly differentiated GC samples than in well- or moderately differentiated GC samples (1.96 vs. 1.76, p = 0.004 for P16 CN; 22.9% vs. 45.1%, p = 0.006 for P16 SCNdel; Table 1). A significant difference in the prevalence of P16 methylation was observed between pTNM I–II and III–IV stage GCs (0.0013 vs. 0.0036, p = 0.037; Table 1); a marginally significant difference also occurred in GCs with or without lymph node metastasis (0.0025 vs. 0.0013, p = 0.097; Table 1); and a difference in the P16 methylation level in SMs was found between poorly and well/moderately differentiated patients (0.0021 vs. 0.0013, p = 0.018; Table 2). In addition, we further investigated whether the overall survival of patients was associated with P16 SCNVs and methylation. We found that the overall survival of patients with P16 SCNamp-positive SMs or SCNdel-negative GCs was longer than that of patients with SCNamp-negative SMs or SCNdel-positive GCs, but the difference was not statistically significant (Figure 4A). Similar differences in overall survival were also observed between patients with P16 methylation-high and P16 methylation-low/no GCs (Figure 4B). In the combination analysis, no synergistic effect was detected between P16 SCNdel and methylation-high GCs or SMs (Figure 4C). Taken together, our findings demonstrate that both P16 SCNV and methylation in gastric samples are consistently associated with GC metastasis. 4. Discussions Tumor suppressor genes are frequently inactivated both genetically and epigenetically in cancer genomes. Inactivation of one copy of tumor suppressor genes by germline point mutations is often subsequently accompanied by epigenetic inactivation of the wild-type copy of these mutant genes by DNA methylation in adult cells, which causes familial cancer [19]. For example, TSS-CGI hypermethylation serves as a frequent “second hit” for wild-type copies of these genes in inherited tumors and consequently causes hereditary diffuse GC and lobular breast cancer [20,21]. However, it is not known whether the SCNVs of tumor suppressor genes, including P16, are derived from TSS-CGI-methylated or nonmethylated genes in sporadic cancer genomes. Here, we report for the first time, to the best of our knowledge, that more P16 SCNdel was detected in P16U GC samples, whereas more P16 SCNamp was detected in P16M gastric samples. In addition, our findings indicate that GC metastasis is significantly associated with a decrease in P16 CN and an increase in P16 TSS-CGI methylation in gastric samples from GC patients. As with the CDH1 gene, one-allele inactivation of CDKN2A/P16 leads to a high predisposition to familial cancers, indicating that it is a haplotype-insufficient gene [19,20,21]. CDH1 is frequently inactivated in adult cells harboring a germline mutation of one CDH1 allele [20,21]. Similar phenomena also occur in P16. For example, one P16 allele is inactivated by a frame-shift mutation and another P16 allele is inactivated by DNA methylation in human colon HCT116 cancer cells [22]. It is not known whether there is causality between frame-shift mutation and DNA methylation in P16. Although we observed a correlation between P16 SCNamp and TSS-CGI methylation in gastric tissues from 200 patients in this study, we do not know whether both P16 SCNamp and TSS-CGI methylation occur at the same alleles or within the same cells. Whether these are two consequent or independent events is worthy of further study. We previously reported that P16 methylation or SCNdel increases the risk of GC metastasis [5,7,8]. In the present study, we simultaneously analyzed the states of P16 methylation and SCNVs in GC and SM samples and reported that the levels of both P16 CN and methylation correlated with GC lymph metastasis, suggesting a true role for P16 inactivation in cancer metastasis. Our findings are consistent with studies using mouse models [7,11,23]. That P16 SCNamp is coupled with DNA methylation might also contribute to the lack of significant difference in overall survival if amplified P16 alleles are silenced by DNA methylation. A correlation between GC metastasis and P16 CN or DNA methylation was observed at the same time point without interference from clinical management or socioeconomic status whereas a correlation between overall survival and P16 alterations was observed at different time points with interference from these factors. The interference effect may account for the inconsistency between metastasis and overall survival. Copy number deletion of CDKN2A/B is linked to early recurrence of meningioma. Although genome-wide DNA methylation and homozygous CDKN2A/B deletions are used for the classification of central nervous system tumors [24,25], whether a synergetic effect exists between CDKN2A/P16 TSS-CGI methylation and heterozygous deletion and whether P16 DNA methylation alone are prognosis factors are unknown. In our previous study involving 80 GC patients [5], which was also included in the present study, we reported that the P16 SCNamp in SMs correlated not only with a low risk of GC metastasis but also with long overall survival. However, in this study involving 200 GC patients, we observed that the P16 SCNamp in SMs correlated only with a low risk of GC metastasis, but we did not observe a significant difference in overall survival. The small number of patients with P16 SCNamp in SMs (23/200 = 11.5%) may account for the fluctuation. 5. Conclusions Our study involving 200 GC patients revealed that P16 SCNVs are coupled with the methylation status of the P16 TSS-CGI in GC development and that SCNdel and TSS-CGI methylation complementarily inactivate P16 and are associated with GC metastasis. While P16 SCNdel coupled with TSS-CGI nonmethylation in gastric tissue samples, P16 SCNamp coupled with TSS-CGI methylation in the samples implies that the amplified P16 alleles may be silenced by DNA methylation during GC development.