Presence of cancer cell DNA in clinical specimens containing exfoliated tumor cells has offered opportunity to use non-invasively obtained clinical samples for methylation analysis [23
]. Thus, methylation-based biomarkers based on analysis of exfoliated cancer cells have shown great promise in early detection and prognosis of cancer. The examples of clinical samples readily available for detection for DNA methylation in cancer cells are sputum for lung cancer [29
], urine sediments for bladder cancer and prostate cancer [40
], stool for colorectal cancer [42
] and plasma/serum DNA for all types of cancers [43
]. Dramatic progress has been made over last few years in identifying genes that are found methylated in cancer tissue but not in corresponding non-cancer tissues [18
]. A number of the genome-wide approaches to the DNA methylation biomarkers to various types of cancer have been developed, producing a decent list of candidate targets. However, only limited progress has been made in identifying methylation markers that could be used in reliable detection and prognosis of cancers based on analysis of exfoliated cancer cells. Thus, there is an urgent need to step up efforts for identify methylation markers that may be used in clinical samples obtained non-invasively from cancer patients or high risk populations. Although such investigations offer enormous promises, they are not without significant challenges. It is anticipated that more research emphasis and resources will be directed to this avenue of biomarker research.
The feasibility of using methylation of a particular gene as a biomarker is initially explored by analysis of the DNA from cancer tissues biopsies and corresponding non-cancerous tissue. Genes that are found to be methylated at higher frequency/higher prevalence in cancer tissue and at very lower frequency/lower prevalence in normal tissue may have potential as biomarkers. However, studies conducted by our laboratory and other laboratories suggest that the promise of using such markers in blood, sputum etc will depend or biochemical composition of these body fluids. For examples, genes that are specifically methylated in cancer cells but not in the normal epithelial cells and fibroblasts might be excellent biomarkers in biopsies but may not be as promising in blood or sputum analysis, if they are found methylated in normal lymphocytes [30
]. Additionally, there could be enzymatic degradation of DNA in body fluids such as sputum that can affect the integrity of the DNA sequences differently in different genes [30
]. These might be some of the possibilities, which can complicate application of DNA methylation based biomarkers in body fluids. Thus, it is important to evaluate candidate methylation based biomarkers using biological samples with similar biochemical composition. This has been a major challenge and the primary reason why out of large number of promising DNA methylation based biomarkers, only a relatively small number of biomarkers have shown significant promise in clinical application.
summarizes a list of studies that have used aberrant methylation of a gene or a panel of genes based on the analysis of body fluids from cancer patients for non-invasive detection of multiple types of cancers. Lung cancer is the number one cancer killer in the United States [44
]. Early detection of lung cancer could reduce lung cancer mortality. There are multiple reports that have tested methylation of P16INK4A
alone or with a panel of other genes in plasma or serum of lung cancer patients as biomarkers [45
]. It is important to note that most of these studies showed moderate concordance in terms of presence of methylation of genes between tumor tissue and plasma/serum of the cancer patients. In general, none these studies reported methylation frequencies of 25% or higher for any one of the multiple markers tested in plasma/serum. Notably, none of these studies reported false positive methylation in plasma/serum from non-cancer patients, with any of the genes tested. In certain studies [55
were reported to be methylated in plasma/serum at a frequencies of higher than 30%. However, substantially lower frequencies were reported in other studies. This was attributed to differences in primer design and/or to differences in the methodologies used. Besides, plasma or serum, sputum from cancer patients has also been used as potential medium for non-invasive detection of lung cancer. Studies by Belinsky et al.
] have tested methylation of multiple genes in sputum samples from cancer and non-cancer patients. Based on their studies, methylation of P16INK4A
in sputum DNA may have significant potential as a biomarker of lung cancer risk. Recent studies from our laboratory identified a methylated panel of genes that appear to be more promising than those previously reported as potential lung cancer biomarkers based on analyses of biopsies as well as sputum samples from cancer patients [30
]. These novel DNA methylation based biomarkers should be tested in plasma/serum to determine their potential as lung cancer biomarkers.
Presence of Methylated DNA Sequences in Body Fluids from Cancer Patients
further summarizes a list of studies that have used aberrant methylation of a gene or a panel of genes in detection of breast cancer based on analysis of body fluids from breast cancer patients [58
]. There was significant variation in results from different studies, even with the same markers. These variations in results were attributed to differences in stage and grade of the cases analyzed. However, as a general observation, RASSF1A
appeared to show significant promise as a biomarker of breast cancer when serum/plasma samples were used for analysis. Recent studies from our laboratory identified a methylated panel of genes showing characteristics useful for detection of breast cancer cells in body fluids from cancer patients [31
further summarizes a list of studies that have used aberrant methylation of a gene or a panel of genes in detection of multiple cancer types such as prostate ovarian cancer, hepatocellular cancer, gastric cancer, colorectal cancer, bladder cancer, pancreatic cancer, esophageal cancer, cervical cancer and hematologic cancers. Prostate cancer is one of the most common male malignancies in the industrialized countries [65
]. There has been great deal of interest in early detection of prostate cancer. Prostatic intraepithelial neoplasia (PIN) is considered to be possible precursors to prostate cancers. Multiple genes have been found to be ethylated in prostatic tumors and PINs and thus they may be good candidates for early detection of prostatic cancer in high-risk populations [66
was found to be commonly methylated in prostatic cancers and plasma of cancer patients [41
methylation was also found in ejaculates (41
). There is need for investigations of additional specific and sensitive biomarkers that can be used in non-invasive detection of prostate cancer. The highly lethal nature of ovarian cancer is related to the absence of symptoms in the majority of women with early stages of the disease [70
]. Early detection of ovarian cancer is critical to increase the possibility of favorable prognosis. Thus, better biomarkers for ovarian cancer may be helpful in early detection and treatment of ovarian cancer. Ibanez et al.
] have reported application of a methylation panel of six genes (BRCA1
) in peritoneal fluid washings from cancer patients. Most importantly, none of the genes were observed to be methylated in peritoneal fluid washings from non-cancer patients. Similarly, as reported in the same study, detection of tumor cell-specific BRCA1
hypermethylation in serum, plasma and peritoneal fluid from early stage ovarian cancer patient should enhance early detection of ovarian cancer. Muller et al.
] have reported application of a methylation panel of 15 genes in peritoneal fluid washings from ovarian cancer patients. Using a high through put assay Wei et al.
] have identified number of highly prognostic DNA methylation biomarkers. Such biomarkers need to be tested in serum, plasma and peritoneal fluid from cancer patients to identify those with great clinical application.
Hepatocellular cancer is one of the most common malignancies worldwide (74
). In hepatocellular cancer the usual outcome is poor, because only 10 – 20% of hepatocellular carcinomas can be removed completely using surgery. If the cancer cannot be completely removed, the disease is usually deadly within 3 to 6 months. further summarizes a list of studies that describe genes that might be useful as biomarkers in hepatocellular cancer. Besides promoter methylation of p16INK4A
not many other markers have been reported to be useful in detection of hepatocellular cancer based on plasma analysis [75
]. Specifically, p16INK4A
methylation appears to be at very high frequency>50% in plasma of cancer patients with no false positive in plasma samples from non-hepatocellular cancer patients. RASSF1A
promoter hypermethylation was detected in 93% of HCC tissues. Of the paired plasma from the HCC patients, aberrant methylation was detected in 43% of the patients. No RASSF1A
methylation was detected in the plasma in the absence of methylation in the corresponding tumor. The presence of RASSF1A
promoter hypermethylation in plasma DNA was found to associate with HCC size. Thus RASSF1A
promoter methylation in plasma should be evaluated as a screening tool and/or prognosticator of HCC patients.
Gastric cancer is the fourth most common cancer worldwide [79
]. It is a disease with a high death rate (700,000 per year) making it the second most common cause of cancer death worldwide after lung cancer. It represents roughly 2% (25,500 cases) of all new cancer cases yearly in the United States but the incidence is slowly rising. P16INK4A
along with CDKN2B
were found to be useful makers based on serum analysis in gastric cancer patients [80
]. Colorectal cancer includes cancerous growths in the colon, rectum and appendix. With 655,000 deaths worldwide per year, it is the third most common form of cancer and the second leading cause of cancer-related death in the Western world [84
]. Methylation of P16INK4A
along with that of CDH1
has been reported in plasma samples from colorectal cancer patients [85
]. Although, frequency of methylation of these genes in plasma samples of primary cancer patients did not appear to be high, it is substantially higher in recurrent cancers. The feasibility of amplification of ethylated DNA from stool samples of patients with CRC has been reported. In a previous study [91
, and PGR
genes were found to be methylated differentially in the stool of patients with CRC. Of these markers, SFRP2
was found to be the most sensitive fecal methylation marker, detecting 77%–90% of CRCs. However, specificity of SFRP2
methylation was quite poor, at 77%. Lenhard et al
] studied the potential of HIC1
gene promoter methylation as a stool-based DNA marker. They showed that HIC1
promoter methylation can be detected frequently and with high specificity in stool samples from patients with CRCs. The combination of HIC1
methylation analysis with fecal occult blood test allowed for the detection of two thirds of CRCs. Assay of methylated DNA markers in stool is a promising approach for colorectal cancer (CRC) screening. A method to capture and enrich hypermethylated CpG islands from stool using Methyl-binding domain (MBD) protein was recently reported [93
]. With MBD enrichment, methylated vimentin was detected in stools enriched with ≥10 ng of cancer cell DNA in stool from CRC patients. In stools from healthy individuals methylated vimentin was not detected, even with MBD enrichment.
Bladder cancer is currently the second leading cause of genitourinary cancer mortality after prostate cancer [84
]. Prognosis for non-invasive cancer is very good but that for invasive cancer very poor. Thus development of reliable prognostic and diagnostic markers to improve strategies for disease management for patients with bladder cancer is crucial. There is a clear need of noninvasive procedures for detection of bladder cancer, whether at initial diagnosis or during follow-up. Multiple genes have been found to be ethylated in bladder cancer with high specificity and sensitivity. Methylation of p14ARF
has been reported in bladder cancers as well as plasma from the cancer patients [94
]. Application of using DNA methylation markers in urine may have significant potential in initial diagnosis or during follow-up to surgery or treatment. However, even though some promising methylation markers showed excellent tumor specificity in bladder cancer biopsies, unexpectedly they showed very poor tumor specificity in urine analysis [96
]. Thus, development of novel methylation based markers with high sensitivity as well as specificity for bladder cancer, based on urine analysis are needed that urine can also be used in urine samples.
Head and neck cancer refers to a group of biologically similar cancers originating from the upper aerodigestive tract and are very common in United States [84
]. Recent studies by Wong et al.
] have shown that methylation of p16INK4A
in plasma might be potential useful biomarkers in screening high-risk populations for early HNSCC and monitoring their to response to treatment. Although, significant amounts of work has been done in identifying genes that are methylated in head and neck cancer specimens, only few reports exist that describe application of these methylation based biomarkers to body fluids from cancer patients.
Pancreatic cancer is the fifth leading cause of cancer-related death, more than 31,000 deaths are anticipated in 2004 [98
]. Since pancreatic cancer is typically detected late in its development, early detection markers are needed for disease diagnosis, risk assessment, and treatment follow up. Hypermethylation of 16INK4A
genes was detected in 60% and 80% of the plasma samples taken from patients whose tumors harbored the same methylation, respectively. No methylation was detected in the plasma sample of the patients whose corresponding tumor DNA had no methylation in the p16INK4A
promoter, with 100% specificity of methylation detection using plasma DNA [100
]. Pancreatic juice has been found to be suitable for early detection of pancreatic cancer. A recent study [101
] has reported, methylation of p14ARF
and p16INK4a was reported in the pancreatic fluid of ~50% of cancer patients but undetected in pancreatic fluid of chronic pancreatitis patients. In another study Methylation of six genes (Cyclin D2, FOXE1, NPTX2, ppENK, p16INK4A and TFPI2
) were reported in >80% cancer patients but <10% in pancreatic fluid of chronic pancreatitis patients. In this study based on qMSP it was clearly possible to separate cancer patients from non-cancer patients.
Esophageal cancer ranks sixth in incidence among cancers worldwide, with 400,000 new cases being diagnosed per year [102
]. This malignancy exists in two principal forms, each possessing distinct pathologic characteristics: Esophageal cell carcinoma, which occurs at high frequencies in many developing countries, especially in Asia; and esophageal adenocarcinoma, which is more prevalent in Western countries, with a rapid rate of increase in recent years. Although significant advances have been made in the treatment of esophageal cancers, these aggressive malignancies commonly present as locally advanced disease, with a very poor prognosis. Using qMSP Jin et al.
] reported TAC-1
promoter methylation in plasma samples from >50% of the esophageal cancer patients but at significantly less frequency and levels in plasma samples from non-cancer patients. In another studies promoter methylation of APC
and p16 INK4A
in plasma was reported to have potential as biomarker of esophageal cancer [104
Cancer of the uterine cervix is an important cause of death in women worldwide [106
]. New molecular and biochemical approaches for the recognition and treatment of high-risk patients are needed to improve survival and avoid over treatment of low-risk patients. Aberrant methylation of CpG islands within the promoter regions of several genes such as p16INK4A
-β has been identified in cervical cancer (106
). Recently, Widshwendel 2004 identified five additional genes, namely CALCA
, as being methylated significantly more frequently in cervical cancer than in normal cervical tissue. On the basis of these observations, we examined the methylation status of CALCA
, and TIMP3
genes in serum samples of cervical cancer patients and compared it with clinicopathological characteristics and outcome of the disease. Promotor CpG island methylation of DAPK1
, and MGMT
was detectable, respectively, in 60%, 28%, and 18% of cases of cervical tumor DNA; and in 40%, 10%, and 8% of cases of patients' plasma DNA [107
]. At least one of the tree genes was found to be methylated in 75% tumors and 55% plasma samples. The methylation pattern in primary tumor and plasma was found to be concordant in ~50% of patients with matched tissue and plasma samples.
In the area of methylation biomarkers much less work has been done with hematologic cancers compared to that with solid tumors. In one study, 39 of 43 blood samples (91%) sequentially collected from 12 patients with AML, ALL, or ABL showed CDKN2B
methylation status in excellent concordance with morphologic disease stage. Early detection of CDKN2B
methylation at apparent remission or its acquisition during follow-up may prove valuable for predicting relapse. Overall survival of patients with CDKN2B
methylation was notably shortened among 38 adults with AML and 12 adults with ALL. Aberrant p15
methylation may have important prognostic implications for clinical monitoring and risk assessment [109
]. Non-Hodgkin's lymphoma (NHL) is a group of malignancies with heterogeneous genetic and epigenetic alterations. Discovery of molecular markers that better define NHL should improve diagnosis, prognosis and understanding of the biology. Shi et al.
] developed a CpG island DNA microarray for discovery of aberrant methylation targets in cancer, and applied this method to examine NHL cell lines and primary tumors. This methylation profiling revealed differential patterns in six cell lines originating from different subtypes of NHL. The investigators further identified 30 hypermethylated genes in these cell lines and independently confirmed 10 of them. Methylation of six of these genes was then further examined in 75 primary NHL specimens composed of four subtypes representing different stages of maturation. Each gene (DLC-1
-β) was frequently hypermethylated in these NHLs (87, 78, 61, 53, 40 and 38%, respectively), but not in benign follicular hyperplasia. Although some genes such as DLC-1
were methylated in the vast majority of NHLs, others were differentially methylated in specific subtypes. The methylation of the candidate tumor suppressor gene DLC-1
was detected in a high proportion of primary tumor and plasma DNA samples by using quantitative methylation-specific PCR analysis. This promoter hypermethylation inversely correlated with DLC-1
gene expression in primary NHL samples (110
). Deligezer et al.
] investigated methylation of p16INK4A
gene in plasma DNA of lymphoma patients by the methylation-sensitive restriction enzyme-related PCR. p16INK4A
methylation was found to occur in 73% of patients but in none of the healthy controls. Nucleosomal DNA fragmentation was detectable in 81% of patients. In 67% of patients, copresence of both parameters was observed. Presence of both parameters was associated with the stage of disease which was more pronounced for nucleosomal DNA fragmentation. The results suggest that presence of methylated and apoptotic DNA in plasma of patients with lymphoproliferative diseases is a frequent event and may be used as a marker for early diagnosis and during the follow-up of the disease.
There is enormous promise in identifying biomarkers that can be reliably used using DNA from body-fluids of cancer patients. Usually, most of DNA methylation based biomarkers are identified for their suitability in body fluids, through actually testing them in similar clinical samples. Future studies should include attempts to integrate the methylation data for a gene in cancer cells and also in types of normal cells which are normal constituents of a certain type of body fluid. Also studies should include attempts to extrapolate bioinformatics features to predicting potential as biomarker in certain type of body-fluid. Thus future challenge should be to use all such information and develop algorithm to come up with non-invasive biomarkers. Future challenge should be to develop criteria or algorithms predicting whether a particular biomarker may be used with a particular type of clinical sample.