The methylation of TSGs is one of the most important epigenetic changes, which plays a vital role in driving tumorigenesis [22]. Aberrant methylation of ctDNA could reflect very consistent changes in cancer [23, 24]. In recent years, ctDNA has gained increasing attention as a noninvasive alternative to tissue biopsies and a potential surrogate for the entire tumor genome. Most previous studies were based on methylation-specific polymerase chain reaction (MSP) to analyze the methylation status of ctDNA, which is a semi-quantitative technique. NGS, known as high-throughput sequencing, can not only detect the whole genome sequence but also the target regions, which especially provide more information of methylation at single base resolution [25]. Nevertheless, only a few previous studies have applied NGS technology to quantitatively detect the methylation status of gene-specific methylated CpG sites in ctDNA for patients with PC [26]. To date, the usefulness of detection of TSGs methylation in ctDNA using NGS has not been established in PC patients.
In the current study, methylation levels of five candidate TSGs (NPTX2, RASSF1A, EYA2, p16 and ppENK) were detected in ctDNA of PC patients via NGS. NPTX2 has been known as a tumor suppressor gene, with low expression attributed to its promoter hypermethylation in tumor proliferation and metastasis [27]. Sato et al [28] reported that the hypermethylation rate of NPTX2 was 98% in PC tissue. RASSF1A, a putative tumor suppressor gene, plays an important role in the regulation of cell cycle, apoptosis, and microtubule stability through the regulation of Ras signaling[29]. In Dammann’s [30] study, the methylation statues of RASSF1A and p16 were detected in patients with pancreatic adenocarcinomas, endocrine tumors and pancreatitis samples. The hypermethylation rate of RASSF1A was 64% (29/45), 83% (10/12) and 44% (8/18), respectively. And p16 was hypermethylated in 44% (19/44) of adenocarcinomas, 17% (2/12) of endocrine tumors, and 18% (3/17) of pancreatitis cases. EYA2 is an important regulatory protein that is important for cell growth and regulation [31]. Vincent et al [13] found loss of tumoral EYA2 expression in 63% of pancreatic cancers. Further research found that silencing of EYA2 in pancreatic cancer cell lines correlated with promoter methylation and histone deacetylation. p16, a cell cycle regulator, controls the G1 phase of cell cycle to the S phase, which is also associated with the transcriptional silencing in PC [32, 33]. ppENK encodes a tonically active inhibitory factor, which was called met-enkepahlin and inhibited the growth of PC [34]. In a study by Fukushima et al [35], 93.3% (14/15) of pancreatic cancer showed methylation of the ppENK gene and 26.7% (4/15) showed methylation of p16 gene. And they further found that the prevalence of methylation of CpG islands of the ppENK and p16 genes increases with PanIN grade (PanIN1A: 7.7% and 12%, PanIN1B: 7.3% and 2.6%, PanIN2: 22.7% and 4.5%, PanIN3: 46.2% and 21.4%, respectively). Our data demonstrated that the methylation levels of RASSF1A, EYA2, p16 and ppENK in ctDNA were significantly higher in PC patients than those in healthy controls, and the methylation levels of EYA2, p16 and ppENK were markedly higher in PC as compared to pancreatic benign diseases. In addition, we found that the methylation status of these five TSGs in patients with PC was not correlated with metastasis and the stage of the disease. Therefore, our results indicated that the methylation levels of the above candidate TSGs in ctDNA can be used as a novel biomarker in the early detection and differential diagnosis of PC.
Several previous studies have reported that methylation of multiple cancer-related genes could be used as a different diagnostic biomarker for PC. For example, Melnikov et al [36] reported that the sensitivity and specificity of combined detection of five genes (PLAU, VHL, SOCS1, THBS1 and CCND2) promotes methylation in cfDNA for the diagnosis of PC were 76% and 59%, respectively. Henriksen et al [37] developed a diagnostic prediction model for PC, including age > 65, cfDNA methylation of BMP3, RASSF1A, BNC1, MESTv2, TFPI2, APC, SFRP1 and SFRP2 genes, which had an AUC of 0.86 with sensitivity 76% and specificity 83%. Eissa et al [38] demonstrated that combined detection of ADAMTS1 and BNC1 promoter methylation in cfDNA exhibited an AUC of 0.95, sensitivity of 97.4% and specificity of 91.6% in the early diagnosis of PC. In the present study, we analyzed the diagnostic performance of specific methylated CpG sites and genes in ctDNA between PC patients and healthy controls. The sensitivity and specificity of ctDNA RASSF1A, EYA2, p16 and ppENK methylation in the diagnosis of PC were 70.6%, 85.7%, 70.6%, 88.6% and 72.2%, 68.4%, 77.8% and 84.2%, respectively. The combination of these four hypermethylated TSGs with CA19-9, which was the most efficient biomarker for PC diagnosis in our series, was associated with increased sensitivity (100%) but a discount of specificity (77.8%). In addition, we found that there was no significant difference in the methylation levels of five candidate TSGs in ctDNA between stage I–Ⅱ and stage Ⅲ-IV. In light of these results, we believe that these TSGs methylation levels in ctDNA could be potential biomarkers in the diagnosis of early stage PC.
The differential diagnosis of benign and malignant pancreatic diseases is another widely concerned point in pancreatic surgery. ctDNA methylation was considered a meaningful biomarker for differential diagnosis of PC. Liggett et al [39] detected methylation levels of 56 tumor-related genes by using microarray technique and identified 17 gene promoters as informative (8 for distinguishing normal controls from chronic pancreatitis and 14 for distinguishing chronic pancreatitis from PC. The sensitivity and specificity were 91.2% and 90.8% for discriminating PC from chronic pancreatitis. In the current study, EYA2, p16 and ppENK is highly methylated in ctDNA of PC than those of pancreatic benign diseases, with sensitivity of 78.8%, 83.7%, 89.6%, specificity of 71.1%, 76.9% and 47.2% in differential diagnosis of PC. The concordance improved when a combination of three TSGs (EYA2, p16 and ppENK) and CA19-9 was used for the differential diagnosis of PC from pancreatic benign diseases (AUC, 0.92; sensitivity, 87.9%; specificity, 85.7%).
ctDNA is associated with the tumor burden. Therefore, one of the major potential applications of ctDNA is monitoring treatment efficacy and predictive prognosis in various cancers. Previous studies have demonstrated the recurrence risk of ctDNA-positive patients was significantly higher than that of ctDNA-negative patients[40]. Higher levels of plasma ctDNA in PC patients may be indicative of metastasis and recurrences [41]. Henriksen et al [42] detect the methylation of 28 tumor-related genes in cfDNA of 95 pancreatic cancer patients. They established a survival prediction model and found that the number of hypermethylated genes in cfDNA was negatively correlated with patient prognosis. Singh et al [43] reported that higher methylation levels SPARC and NPTX2 genes represent poor prognostic in PC. In a recent study, the dynamics of circulating NPTX2 methylation are closely associated with disease progression and response to therapy in metastatic pancreatic cancer[44]. To investigate the value of TSGs methylation in ctDNA to predict response to surgery, we observed the dynamics change of TSGs methylation levels at postoperative week 1. In our study, methylation levels of NPTX2, EYA2 and ppENK were significantly decreased after surgical removal of the primary tumor. That may imply the methylation status of those TSGs represents changes in tumor burden throughout the clinical disease course. Thus, we propose that NPTX2, EYA2 and ppENK methylation in ctDNA may be a significant prognostic factor associated with recurrence in PC. Nevertheless, its value in prognosis and recurrence prediction needs to be validated by further studies.
Undoubtedly, there are some limitations to this study. First of all, this study is only a single institution research with limited cases, especially patients in the advanced stage. Secondly, the value of hypermethylated ctDNA in predicting the prognosis and recurrence of PC can not be evaluated because of the limited cases and short time of follow-up time.
In conclusion, our preliminary results suggest that quantitative analysis of methylation pattern of NPTX2, RASSF1A, EYA2, p16 and ppENK in ctDNA by NGS could be a valuable non-invasive tool for detection and monitoring treatment efficacy of PC.