- DNA methylation
In this issue of Clinical Cancer Research, Wallner et al. suggest that methylated DNA released in the circulation could possibly be used as a prognostic indicator and for tumor detection (1).
Tumor markers are biological substances that indicate the presence of cancer. They are usually produced by malignant tumors and are detected in the blood, urine, or other biological fluids. A huge variety of substances have been investigated as potential tumor markers since the initial description of Bence-Jones proteins in the urine of patients with multiple myeloma well over a century ago. At present, most tumor markers used in clinical practice are tumor antigens, enzymes, hormones, receptors, growth factors, and biological response modifiers that are detected by biochemical assays. The rapid growth in the understanding of the molecular biology of cancers, however, means that in the near future, we will be able to add a host of novel substances to this panel. Among them is circulating, epigenetically altered, free DNA.
Ideally, if a tumor marker is to be used in screening, diagnosis, monitoring response to therapy, and detecting recurrence during follow-up, it should be tumor-specific, quantitatively reflect tumor burden, and should be produced in sufficient amounts to allow the detection of minimal disease. Clinical usefulness is heavily dependent on two variables—high sensitivity (i.e., the proportion of patients who have cancer, will relapse or die who have a positive/abnormal test) and high specificity (i.e., the proportion of patients who do not have cancer, will not relapse or die who have a negative/normal test). These need to be well established for the intended use of the marker before it can be adopted into clinical practice.
DNA Methylation
Among the molecular alterations and epigenetic events described in human neoplasia, changes in DNA methylation are one of the most common (2). It involves the methylation of cytosines that have a guanosine located at their 5′ end. These CpG dinucleotides are uncommon in the vertebrate genome except in small stretches of DNA termed CpG islands, usually 500 to 2,000 bp in length, that are frequently located in and around the transcription start sites of human genes (2). It is increasingly recognized that the CpG islands of a growing number of genes (which are mainly unmethylated in nonneoplastic, normally differentiated cells) are methylated to varying degrees in many types of human cancer. In fact, such promoter hypermethylation is as common as the disruption of classic tumor suppressor genes by mutations and perhaps even more so (2).
Other DNA alterations, like point mutations, often occur at different sites within a given gene in individual tumors. In contrast, promoter hypermethylation usually occurs over the same region of a given gene. This greatly simplifies the design of a test system. More importantly, there is no need to test the methylation status of a given gene in the primary tumor before devising a means for detecting the hypermethylation marker. Finally, in comparison with frequent chromosome changes in cancer, e.g., allelic loss, CpG island hypermethylation constitutes a positively detectable signal, as opposed to a loss of signal seen in loss of heterozygosity.
Aberrant CpG island hypermethylation generally does not (or only rarely) occur in nonneoplastic, normally differentiated cells. Therefore, the tumor-derived DNA can be detected with a very high degree of sensitivity, even in the presence of a vast excess of DNA derived from normal cells. MethyLight technology can detect a single hypermethylated allele against a background of 10,000 unmethylated alleles (3).
Circulating Methylated DNA as a Marker
Over the last decade, it has become clear that hypermethylation can be detected in tumor-derived DNA found in the serum and plasma of patients with cancer (4, 5). More recently, analysis of serum/plasma DNA methylated at specific CpG islands has been shown to predict prognosis and/or response in patients with cancer undergoing specific treatment regimens (refs. 1, 6–21; Fig. 1 ). Several groups have used a common strategy to identify prognostic markers in serum using DNA methylation. A panel of genes (known to be aberrantly methylated in cancer specimens) have been analyzed in serum samples of patients with distant metastases, primary localized tumor, and healthy age-matched controls. Genes which were frequently and rarely methylated in serum from metastasized versus healthy control patients, respectively, were then analyzed in pretreatment serum samples from patients with localized cancer, for whom a sufficient follow-up time was available (1, 8). The methylation pattern adds to the characterization of the underlying tumor, and may in time, serve as a predictor for sensitivity to certain drugs.
Source, potential biological role and possible clinical use of CpG DNA in cancer (see text). IFN, interferon; MMP, matrix metalloproteinase; pDC, plasmacytoid dendritic cells; B, B cells.
The precise mechanism by which DNA is released into the bloodstream still remains uncertain. Pathologic cell death of malignant cells results mainly from necrosis, autophagy, and mitotic catastrophe. This is associated with a high integrity of DNA compared with apoptosis, which involves well programmed enzymatic splicing of DNA, resulting in small fragments ranging from 185 to 200 bp. Highly proliferative, aggressive tumors release tumor DNA into the intercellular space that is then drained by lymphatics into the bloodstream (see Fig. 1). The findings that (a) following chemotherapy, an initial peak in free-circulating tumor DNA is subsequently followed by a massive decline in levels (22) and (b) the integrity of free-circulating DNA (indicating that the DNA is derived from cells which have died by nonapoptotic mechanisms) is of prognostic significance (23), supports this explanation. This supports the finding that circulating tumor DNA is a surrogate marker for active, fast-growing tumors with intense neovascularization, and dense lymphovascular invasion.
The article by Wallner et al. in this issue of Clinical Cancer Research suggests that methylated DNA released into the circulation has potential in both tumor detection and as a prognostic indicator (1). In a validation set of 104 patients, methylation of all the three genes studied in-depth (HPPI, HLTF, and hMLH) significantly correlated with tumor size. Methylation of HPP1/HLTF was significantly more frequent in patients who had metastases and were independent poor prognostic markers. Some healthy individuals showed methylation of certain genes. This may be attributable to increased turnover of cells both in the colon and elsewhere or may be an indication of an early carcinogenic process. As serial serum samples become available from healthy individuals on long-term follow-up, it will be possible to address the question as to whether tumor-free individuals with tumor-specific methylated DNA in their serum samples are at an increased risk of developing the cancer.
Possible Role of Free-Circulating CpG DNA
Despite an increased appreciation of the potential role of tumor-specific DNA as a prognostic marker, its precise role in metastasis still remains elusive. Although, there may be a correlation between metastasis-predisposing conditions, such as circulating DNA and the occurrence of circulating tumor cells, it is clear that free-circulating DNA does not result from direct lysis of circulating tumor cells. It has been calculated that 1,000 cancer cells/mL would be necessary to provide the amount of DNA found in the plasma of patients with pancreatic cancer—a figure considerably more than what has been actually found (24). Lack of ras mutations in micrometastatic cells derived from peripheral blood mononucleocyte culture of patients with colorectal cancer supports this view (25).
What then is the role of circulating DNA? Is there any evidence that extracellular tumor DNA has any intrinsic activity, either paracrine (to tumor and surrounding cells) or endocrine (to cells which may have a secondary effect on tumor; Fig. 1)? The innate immune system senses the invasion of pathogenic microorganisms through the Toll-like receptors (TLR), which recognize specific molecular patterns present in microbial components. Microorganisms (prokaryotes) and tumor cells share a common feature: both contain unmethylated CpG motifs at a higher frequency than eukaryotes and normal cells, respectively. The innate immune system detects unmethylated CpG motifs using TLR-9 (26). The release of unmethylated CpG DNA during an infection provides a danger signal to the innate immune system, triggering a protective immune response that improves the ability of the host to eliminate the pathogen (27). Activation of the TLR-9 subsequently leads to the activation of a cascade including nuclear factor-κB that culminates in the activation and proliferation of immune cells.
Although tumor cells show hypermethylation at CpG islands associated with various genes, the overall 5-methylcytosine content in tumor cells is much lower, mainly because of an increased hypomethylation at repetitive elements (28). Although there is no direct proof, it is quite likely that the massive increase in free-circulating hypomethylated DNA, via activation of TLR-9, boosts the immune system. TLR-9 has also been identified, however, in human breast cancer cells and clinical breast cancer samples (29). Stimulation of TLR-9-expressing breast cancer cells with TLR-9 agonistic CpG oligonucleotides dramatically increases their in vitro invasion in both Matrigel assays and three-dimensional collagen cultures. CpG oligonucleotide treatment results in a decrease in expression of tissue inhibitor of metalloproteinase-3 and increase in levels of active matrix metalloproteinase in TLR-9-expressing, but not TLR-9-negative, breast cancer cells. In addition, it has been shown that TLR-9-CpG-DNA interaction occurs at an acidic pH in tumor cells (30). We therefore hypothesize that fast-growing tumor cells (containing much more unmethylated CpG DNA compared with their normal counterparts) undergo necrosis caused by insufficient neoangiogenesis, and thereby release free tumor DNA. Supported by the milieu (low pH in the perinecrotic area in the center of the tumor), hypomethylated DNA binds to TLR-9 on the surrounding tumor cells and may thereby act as a strong stimulator of these cancer cells, enabling and facilitating (like a hormone produced from the tumor) the process of invasion and metastasis of circulating tumor cells. This scenario would clearly explain the somewhat paradoxical finding that variables indicating a stimulation of the immune system are an independent poor prognostic marker in breast cancer (31).
Some evidence also supports the hypothesis that circulating altered DNA per se may cause de novo development of tumor cells in organs that are known to be susceptible to cancer metastases. This so-called “hypothesis of genometastasis” suggests that malignant transformation might develop as a result of transfection of susceptible cells in distant target organs with dominant oncogenes that circulate in the plasma and are derived from the primary tumor (32, 33). The genometastasis theory is supported by some previously reported evidence. In an in vivo study, healthy rats receiving plasma from a tumor-bearing rat subsequently exhibited the presence of the tumor marker gene in their lung DNA (34). In addition, plasma (from cancer patients) used as culture medium has the ability to transform cells. When such cells are s.c. injected into syngeneic rats, tumors develop (33). Experiments done by Anker et al. (35) have shown that the supernatant of colorectal cancer cell cultures can transfect fibroblast cultures. The hypothesis that epigenetically altered DNA, once integrated into cells, could sufficiently alter their behavior is supported by the fact that methylated oligonucleotides applied in vitro as well as in vivo are able to regulate gene expression in hepatocellular cancer (36).
Conclusion
Despite the demonstration that free-circulating, aberrantly methylated DNA is an independent prognostic and predictive tumor marker in several cancers, several issues need to be addressed before these markers can be used in the clinic. These include standardization of sample collection, DNA isolation, DNA modification (bisulfite or restriction enzyme), and assay (e.g., real-time PCR, sequencing, etc.); ensuring test reproducibility in terms of intraassay and interassay variability, determining biological variability within the same patient; sensitivity and specificity, and finally, efficacy when compared with other markers already in use (e.g., CA125, CEA, etc.). Finally, it needs to be determined whether free-circulating methylated DNA may be a direct or indirect target for novel therapeutic interventions.
Footnotes
- Received October 18, 2006.
- Accepted October 24, 2006.
References
Volume 12, Issue 24
