FormalPara Key Points

Circulating cell-free DNA is present in different forms in the bloodstream.

Cell-free DNA has immense potential to be considered a diagnostic and prognostic marker.

Non-invasive procedures using cell-free DNA can be used to assess disease in real time.

1 Introduction

Mendel and Metais [1] first discovered cell-free nucleic acids in 1948. Later, they were observed in many disease conditions such as inflammatory and autoimmune diseases, lupus erythematosus, myocardial infarction, diabetes, and cancer [2, 3]. A higher concentration of cell-free DNA (cfDNA) was detected in cancer patients than in healthy individuals [3]. cfDNA levels reflect disease status [3–5], but the concentration alone is not sufficient to be used as a biomarker [6–8]. cfDNA released by the tumor contains the same mutations as those in tumor cells. Hence, cfDNA concentration and genetic alterations detected in cell-free DNA can, together, be used as cancer biomarkers. Biopsy of a tumor with intra- and inter-tumor heterogeneity does not provide a complete genetic picture of the tumor, as only a part of the tumor is tested. Furthermore, the diversity between primary tumors and metastatic tumors makes it difficult to highlight mutations, which can increase the metastatic ability of cancer cells. Circulating tumor DNA (ctDNA) includes DNA released not only by primary but also by metastatic tumors [9]. Therefore, it is possible to analyze ctDNA and obtain information about mutations present in primary and metastatic tumors [10]. Since procedures are non-invasive, testing can be repeated to follow the progress of disease or response to treatment [11, 12]. Mutations contributing to drug resistance can be detected, allowing therapeutic decisions to be changed accordingly [13–16]. Hence, ctDNA has immense potential as a cancer biomarker. Inclusion of non-invasive ctDNA assays would be helpful in cancer diagnostics, assessing disease status, and monitoring treatment response. This review summarizes the origin of cfDNA, recent advancements in the detection of cfDNA, a possible role in metastasis, and its importance as a non-invasive diagnostic assay for cancer.

2 Types of Cell-Free DNA (cfDNA)

cfDNA are present in various forms such as unbound DNA fragments, nucleosomes, vesicle-bound DNA, and virtosomes (Fig. 1) [17].

Fig. 1
figure 1

Circulating cell-free DNA. DNA can be released into the bloodstream either through cell death, i.e. apoptosis (yellow) or necrosis (green) or it can be released by viable cells (purple). Cell-free DNA can be present in the form of unbound DNA, nucleosomes, vesicle-bound DNA, or virtosomes

Full size image

2.1 DNA Fragments

Free DNA fragments are not bound to any other molecule or surface. Upon release, they are digested by DNases in blood. However, in cancer, the DNase concentrations are low and DNase inhibitors have also been detected, which could also contribute to increasing levels of cfDNA [18]. Circulating DNA has been reported to have a half-life of less than 2 h [19, 20].

2.2 Nucleosomes

Nucleosomes are complexes consisting of DNA wrapped around an octamer of histone proteins and are stabilized by histone H1. Nucleosomes are linked to each other by 100 bp of linker DNA. During apoptosis, chromatin is cleaved into oligo- and mono-nucleosomes, packed into vesicles, and released from the cell before being engulfed by neighboring cells. In cancer, during chemotherapy or radiotherapy, the rates of cell death are high, and the phagocytic mechanism becomes saturated, leading to increased levels of nucleosomes in circulation [4]. Increased nucleosome concentrations correlated with cfDNA concentrations; this has been linked to metastasis and disease progression in breast cancer [21]. Circulating nucleosome levels have been shown to predict therapeutic efficacy in lung cancer, acute myeloid leukemia, colorectal cancer (CRC), and pancreatic cancer [22–25].

2.3 Vesicle-Bound DNA

Vesicle-bound DNA include DNA contained in apoptotic bodies, exosomes, and microparticles [17]. Apoptotic bodies are formed by cell blebbing during apoptosis. Nucleic acids are then packaged in these bodies and released from the cell to be ingested by phagocytes. Holmgren et al. [26] showed that apoptotic bodies can be taken up by phagocytic cells, causing horizontal transfer of DNA. Exosomes are small vesicles produced by many cell types. They may contain nucleic acids as well as protein. Exosomes released by tumor cells contain mutated nucleic acid fragments that may contribute to the growth and spread of primary and metastatic cancers [27].

2.4 Virtosomes

Virtosomes are complexes of newly synthesized DNA, RNA, proteins, and lipids and are only released by living cells [17, 28]. The nucleic acids cannot be readily digested with nucleases, unless the complex is first treated with proteinase or lipase [28]. The protein portion consists of DNA: dependent DNA and RNA polymerases [17, 28]. The complex is released in a controlled energy-dependent manner [17, 28]. The DNA is synthesized in nucleus and moves into cytosol where it is joined by newly synthesized protein, lipid, and RNA before being released from the cell [28]. Virtosomes can easily enter and change the biology of recipient cells. Hence, virtosomes released by cancer cells may enter and transform normal cells [28].

3 Origin

Although the origin of ctDNA is not clear, several studies have proposed that ctDNA could be released by apoptosis or necrosis or shed by viable tumor cells (Fig. 1). Jahr et al. [29] investigated the origin of plasma cfDNA using size distribution analysis. Although the size of DNA was frequently in the range of multiples of 180 bp, similar to DNA from apoptotic cells, DNA fragments of high molecular weight, similar to necrotic cell DNA, were also detected in some samples. Chromatin fragments were observed to be released by cells undergoing induced cell death in animal models and were eventually detected in the bloodstream of the models. Therefore, ctDNA may be released by dying tumor cells [29]. A mechanism of cfDNA release was proposed by Diehl et al. [30], whereby invasive tumors that outgrow their blood supply contain larger regions of necrosis than benign tumors. The necrotic cells are then engulfed by macrophages that release digested DNA into the surrounding medium. Active release by lymphocytes has also been viewed as a possible origin of cfDNA [31]. Evaluations of associations between cfDNA and alterations in the balance of angiogenesis activators and inhibitors have suggested that cfDNA may reach plasma at least in part due to matrix metallopeptidase 2 (MMP2) overexpression [32].

4 Genometastasis

In 1999, García-Olmo et al. [31] proposed that ctDNA containing dominant oncogenes could transform susceptible cells in distant organs, thereby contributing to metastasis of tumors. They termed this process ‘genometastasis’ (Fig. 2). Rat colon cancer cell line DHD/k12-PROb cultured in plasma from tumor-bearing rats showed that the cfDNA may have properties for integrating into host cell genome [33]. Furthermore, horizontal transfer of DNA may occur via uptake of apoptotic bodies or virtosomes by phagocytic cells [26, 28]. In three different studies, when immortalized and normal cells were cultured in patients’ plasma, immortalized cells were transformed into cancer cells while normal cells were not transformed [34–36]. No transformation was seen when cell lines were exposed to normal plasma [36]. Therefore, ctDNA may play a role in cancer metastasis through oncogenic transformation of initiated cells.

Fig. 2
figure 2

Genometastasis. Circulating DNA has been suspected has having a role in metastasis. Studies in this direction have shown that circulating tumor DNA has the ability to transform susceptible cells in distant organs

Full size image

5 Methods for Detection of cfDNA

5.1 Quantitative Polymerase Chain Reaction (PCR)

Quantitative real-time polymerase chain reaction (PCR) enables the evaluation of nucleic acid concentration by measuring the emitted signal from a fluorescent-labelled probe during amplification of the target gene [37]. Amplification of various genes such as human telomerase reverse transcriptase (hTERT) and beta globin have been used to measure cfDNA concentration [5, 38]. Furthermore, some studies used real-time PCR to measure DNA integrity [39, 40].

Methylation-specific PCR is an assay for determining the status of CpG islands in which bisulfite modification is performed followed by PCR amplification of the gene of interest using primers specific for either a methylated or an unmethylated DNA sequence [41]. A combination of real-time PCR and methylation-specific PCR was used to determine the methylation status of various genes and to quantitate methylated cfDNA sequences [42–44].

In digital PCR, DNA is serially diluted to isolate single molecules; each molecule is then individually analyzed for mutations [45]. Digital PCR was used to screen for alterations in PIK3CA mutations in breast cancer [12]. Sensitivity of this technique was 93.3 % and specificity was 100 % [12]. Screening of HER2 copy number was carried out in patients with breast cancer using digital PCR [46]. The positive predictive value was 70 % for this assay, and the negative predictive value was 92 % [46]. KRAS and BRAF variants were detected using duplex digital PCR in patients with CRC [47]. Exon 19 deletion and L858R mutation were detected in EGFR gene in lung cancer patients with 92 % sensitivity and 100 % specificity [14].

5.2 Next-Generation Sequencing

Next-generation sequencing (NGS) has been used to analyze ctDNA in many cancer types to identify mutations. Leary et al. [48] devised a technique called ‘personalized analysis of rearranged ends’ (PARE) to detect chromosomal rearrangements in solid tumors that were then used to develop biomarkers for the monitoring of tumor through plasma DNA. However, limitations of this technique include dependence of sensitivity on amount of sequence data, cost, possibility of detecting false positives, and the unavailability of information on source of ctDNA from these analyses [49].

De Mattos-Arruda et al. [10] used massive parallel sequencing to analyze DNA extracted from primary and metastatic tumors and plasma samples from a 66-year-old patient with breast cancer. A total of 15 mutations were detected in both the primary and metastatic tumors, while two mutations were detected in metastatic tumors but not in the primary tumor. All of these variants were detected in ctDNA, thereby capturing the heterogeneity of tumors. Furthermore, on monitoring the mutant allele fractions of ctDNA during treatment, it was found that ctDNA analysis gave an earlier indication of disease progression than radiologic and biochemical assessments [10]. In another study, massive parallel sequencing of circulating cfDNA was used to study metastatic cancers in patients receiving treatment. It was observed that mutant alleles increased with development of therapy resistance [16].

Newman et al. [50] developed a new technique called cancer personalized profiling by deep sequencing (CAPP-SEQ) for quantifying ctDNA. They designed a probe panel to detect recurrently mutated genes using whole exome sequencing data for lung cancer patients from The Cancer Genome Atlas (TCGA). The panel targets included 521 exons and 13 introns from 139 recurrently mutated genes. They identified ctDNA in 100 % of stage II–IV non-small cell lung cancer (NSCLC) patients, with 96 % specificity for mutant allele fractions as low as 0.02 % [50]. Rothé et al. [51] evaluated plasma as an alternative for tissue biopsies with NGS of plasma DNA and tumour DNA. Mutations were detected in p53, PIK3CA, PTEN, AKT1, IDH2, and SMAD4 genes [51]. In two samples, mutations were detected in tumor but not in plasma. This may have been due to the presence of a mutation allele frequency below the detection limit of the method used. Two samples showed mutations in plasma but not in tumor. These mutations may have been present in tumour clones that were not captured by biopsy [51].

Whole genome sequencing of cfDNA has been reported to detect novel mutations not detected by targeted sequencing [52]. However, whole genome sequencing is expensive and requires expertise in analysing the enormous amount of data it generates. Targeted sequencing is limited to mutations in genes. Sequencing of genes/exonic regions of a gene that are frequently mutated in individual cancers may help circumvent the problem.

5.3 BEAMing

Diehl et al. [30] developed BEAMing (beads, emulsion, amplification, and nagnetics) to detect somatic mutations in DNA. In this technique, the DNA segment of interest is first amplified using primers containing known tag sequences. Emulsion PCR amplifies the DNA sequences again using primers targeting the tags. The DNA sequences are then covalently bound to magnetic beads. Labelled nucleotides are added to the sequences by single base extension, and flow cytometry is used to sort beads containing mutant allele from those containing wild-type allele [30]. The technique is highly sensitive, with the ability to detect and measure DNA even when there is one mutated DNA present in 10,000 DNA molecules [53]. BEAMing was used to detect APC variants in CRC [30], EGFR variants in lung cancer [53], and PIK3CA variants in breast cancer [54].

5.4 Peptide Nucleic Acid Clamp-Mediated PCR

Peptide nucleic acid (PNA) is a synthetic nucleic acid analog with a peptide backbone consisting of N-(2-amino-ethyl)-glycine units, instead of phosphodiester backbone [55]. In PNA-mediated PCR reactions, PNA complementary to wild-type sequence acts as a clamp. Mismatch of one base pair can distinguish between a wild-type sequence and a mutated sequence. Hence, PNA binds specifically to wild-type sequences, preventing amplification, while mutated sequences are amplified using primers [55]. EGFR variants were identified using PNA clamp-mediated PCR in NSCLC patients. The mutation detection rate correlated with sex, smoking history, and stage. However, this method was limited in terms of its sensitivity and the multiple steps, lasting several days [56]. Däbritz et al. [57] combined PCR clamping and melting curve analysis to detect KRAS codon 12 variants in pancreatic cancer patients. The sensitivity of the assay was 1–5 × 105; 28 % of plasma samples were found to have alterations in the KRAS gene. The most common alteration was glycine to valine, seen in 83 % samples [57].

Even though techniques such as quantitative PCR and BEAMing are sensitive, they require prior knowledge of the variants present in the patient (Table 1). NGS methods do not require this. Different panels can be used to screen for mutations in multiple genes. This would enable the detection of various cancer-associated genetic alterations in patient blood. Therefore, sequencing circulating tumor DNA using NGS techniques would be useful in the noninvasive detection of genetic alterations.

Table 1 Comparison of different methods available for cell-free DNA analysis
Full size table

6 cfDNA: A Non-Invasive Biomarker

cfDNA has many clinical applications as a diagnostic, predictive, and prognostic biomarker for cancer (Fig. 3). The presence and concentration of somatic sequence variants, changes in methylation status, copy number variants, and microsatellite instability provide data on disease status. As cfDNA changes with disease status during therapy, periodic assessment can be carried out to evaluate disease progression and treatment response. Furthermore, variants in tumor DNA can contribute to drug resistance. Analysis of ctDNA for such variants may aid in deciding the appropriate treatment for the patients.

Fig. 3
figure 3

Clinical applications of circulating tumour DNA. Circulating tumour DNA is easily accessible as it is obtained from the peripheral blood of patients. It is being explored as a potential biomarker for cancer through the measurement of circulating tumor DNA concentration and detection of genetic alterations such as mutations, methylation, and microsatellite alterations. Monitoring these parameters in periodically collected blood samples would give information about disease progression and treatment response. It also allows screening for drug-resistance mutations, thereby aiding in treatment decision making. Minimal residual disease can be detected with the evaluation of cell-free DNA. cfDNA cell-freeDNA, ctDNA circulating tumor DNA

Full size image

6.1 Concentration

cfDNA levels have been measured via many different techniques such as radioimmunoassay, DNA dipstick, and real-time PCR in various types of cancers [3, 5, 58, 59]. The levels reported vary across different studies, but cfDNA concentration is generally higher in cancer patients than in healthy individuals [3–5, 21, 58–61]. A comparison of different cancer types showed that patients with lung cancer had the highest levels of nucleosomes, and prostate cancer patients had the lowest levels [4].

Comparisons of median cfDNA concentrations in plasma and serum showed cfDNA levels were higher in serum than in plasma in both patients and healthy controls [62]. However, this could be due to the release of DNA by cell lysis during clotting. Plasma DNA levels correlated with tumor size [63], degree of tumor invasion [3], disease stage, survival, and disease progression under therapy [3, 5, 13, 62]. Increased cfDNA levels were associated with aggressive disease and worse prognosis in lymphoma [38, 61]. Nucleosome concentrations correlated with circulating DNA concentrations and caspase activities in breast cancer [21]. Increased levels of serum nucleosomes, cfDNA, and protease activities were related to metastasis and hence may be linked to progression of breast cancer [21]. Patients with metastatic cancer had higher ctDNA levels than those with non-metastatic disease [3].

cfDNA levels were observed to decrease after therapy or surgery [3, 11, 58]. During chemotherapy nucleosomes levels increased initially and later decreased; while following tumour resection it decreased soon after surgery [4, 60, 64]. This could be due to the clearance of nucleosomes from circulation [64]. The post-therapeutic concentration of circulating nucleosomes correlated with response to treatment [4, 64]. Decreased ctDNA levels correlated with improved clinical condition, while increased or unchanged levels indicated a lack of response to treatment [3]. Furthermore, levels were lower in disease-free individuals than in patients experiencing relapse [58]. While no cases have been observed showing the presence of circulating tumor cells (CTCs) with an absence of ctDNA, a ctDNA presence has been found in patients with an absence of CTCs [13]. In some cases, ctDNA concentration was more than fiftyfold that of CTCs. [13]. One study has reported an instance in which a patient had large numbers of CTCs but low levels of cfDNA [65].

Many groups have reported cfDNA concentration to be very high in malignant cancer, moderately high in benign disease, and low in healthy individuals [66–68]. However, few studies have reported cases in which DNA concentration alone is not sufficient to discriminate between benign and malignant cases [6, 7, 21]. For instance, in one study, plasma DNA concentration measured using real-time PCR were not significantly different between healthy individuals and those with gastroesophageal reflux disease or esophageal and lung cancer [7]. It has also been reported that some patients with highly metastatic cancer may not have detectable amounts of plasma DNA [9]. The techniques used and concentrations measured also vary across studies. Hence, it is difficult to establish a cut-off value to distinguish between healthy, benign, and malignant conditions. However, as cfDNA levels reflect the progression of disease and effect of therapy [5], they can be used to monitor disease status combined with other parameters such as mutations in ctDNA and DNA integrity.

6.2 DNA Integrity

The DNA integrity index (DII) is determined by amplifying DNA fragments of different sizes by PCR and calculating the ratio between concentrations of long PCR products and concentrations of short PCR products. Many studies have tried to establish an appropriate DII ratio to distinguish between cancer patients, those with benign disease, and healthy individuals.

Different studies have suggested different integrity indices as discriminating values. A 300/60 ratio was suggested as an optimal value in one study [40]. Pinzani et al. [39] compared DII ratios of 180/67, 306/67, and 476/67 in cutaneous melanoma patients. All ratios showed higher values in cancer patients. Although a combination of the three values gave high sensitivity, the 180/67 value was found to be the most suitable by receiver operating characteristic (ROC) curve analysis [39].

The detected ctDNA concentration inversely correlated with amplicon size [30]. The size distribution of neoplastic cfDNA differs from that of non-neoplastic cfDNA [40]. Some studies have reported tumor cfDNA to be more fragmented than normal cfDNA [30, 69]; however, DNA integrity has also been reported to be higher in patients than in healthy controls [39, 40]. Jiang et al. [70] studied the size profile of cfDNA using massive parallel sequencing in patients with hepatocellular carcinoma (HCC). They observed that short as well as long fragments of cfDNA were present in patient plasma. The short fragments were also observed to contain more copy number aberrations [70].

The size distribution of cfDNA has been observed to be biphasic in cancer patients [9]. The biphasic size distribution also correlated with CTCs and levels of mutated DNA fragments [9]. DNA integrity was observed to be more representative of tumor progression than absolute serum DNA values [71]. DNA integrity was higher in patients positive for lymphovascular invasion, micrometastasis, or lymph node metastasis than in negative patients [71]. In patients with post-operative recurrence of disease, serum DNA integrity was similar to that of patients with stage III or IV breast cancer. Evaluation of serum DNA integrity achieved 69 % sensitivity with 80 % specificity for detecting stage II–IV breast cancer and sensitivity of 74 % with 80 % specificity for lymph node metastasis [71].

6.3 Genetic Alteration

In cancers, accumulation of genetic alterations occurs in genes that are involved in cellular processes such as growth and apoptosis. Genetic alterations detectable in tumors can also be detected in cfDNA [72]. Detection of cancer-related mutations in cfDNA indicated that the cfDNA was tumor derived [2]. Since ctDNA may be released by primary as well as by metastatic tumors, analysis of ctDNA may present a whole paradigm of various mutations present in primary as well as in metastatic tumors [10]. Furthermore, monitoring the mutant allele fractions of ctDNA during treatment can give an earlier indication of disease progression than radiologic and biochemical examinations [10]. Tumor cells in cancer patients acquire variants that make them resistant to the effects of chemotherapeutic drugs. Therefore, patients being considered for treatment with such drugs should be tested for drug-resistance mutations, as these would otherwise compromise the efficacy of the treatment.

6.3.1 Mutations

EGFR variants have been detected in ctDNA in various cancers. Detection of EGFR mutation in tumor tissue and plasma was 94.3 % concordant, with 99.8 % assay specificity and 65.7 % sensitivity [73]. The concentration of mutated sequence in plasma correlated with response to treatment. Patients showing complete or partial remission had decreased concentrations, while increased concentrations indicated no response [14]. Furthermore, EGFR mutation re-emerged in a patient whose treatment was stopped after an initial response [14]. Detection of mutated DNA fragments in plasma can predict the response to treatment. These data indicate that monitoring ctDNA during the treatment course can present real-time information on patient response to treatment [74]. The presence of EGFR mutations has correlated with response to gefitinib and progression-free survival time. EGFR variants were detected more frequently in patients who responded to gefitinib and in those with longer survival [75]. Emergent variants were identified in 96 % of patients who initially responded to EGFR blockade treatment and later progressed [13].

The KRAS gene is also involved in the EGFR pathway; it has been studied in CRC and pancreatic cancer. p.Gly12Ala, p.Gly12Cys, p.Gly12Ser, p.Gly12Asp, p.Gly12Val, p.Gly12Arg, and p.Gly13Asp are some of the variants detected in codons 12 and 13 of the KRAS gene [47, 76]. Mohan et al. [52] found that KRAS gene amplifications in ctDNA correlated with the clinical course of disease. Mutations and amplifications in the KRAS gene are mutually exclusive. However, they also observed that KRAS amplification may occur during EGFR-targeted treatment of CRCs [52]. The sensitivity of a liquid biopsy for the presence of variants in codons 12 and 13 of the KRAS gene was 87.2 %, and specificity was 99.2 % [13]. The proportion of the mutated alleles seems to be highly variable, ranging from <1 to 64 %. This would have implications with regard to the technique to be used for the most sensitive detection. PIK3CA variants have also been detected in plasma post-operatively in patients who showed no clinical symptoms [12]. Hence, cfDNA can be used to detect minimal residual disease.

6.3.2 Microsatellite Instability

Microsatellites are DNA sequences in which a short motif is repeated many times. They are subject to expansion and contraction due to defects in mismatch repair machinery and can be used to detect loss of heterozygosity (LOH) in cancer [77]. They were first reported in head and neck cancer and lung cancer in 1996 [78, 79]. Microsatellite alterations were associated with disease status and patient survival in lung cancer [5] and melanoma [80, 81]. In oral squamous cell carcinoma, the presence of microsatellite alterations on chromosomes 2, 3, and 21 in serum DNA correlated with poor prognosis [82, 83].

6.3.3 Loss of Heterozygosity

The detection rate of LOH in chromosomes 1p, 19q, and 10q and methylation status of MGMT and PTEN gene promoters in serum DNA in high-grade astrocytomas or oligodendrogliomas were 51 and 55 %, respectively, with 100 % specificity [2]. LOH on chromosomes 6q, 8p, 9p, 10p, and 18q and DNA methylation correlated with Gleeson score and prostate-specific antigen (PSA) levels in prostate cancer. A combination of DNA assays and a PSA assay gave a sensitivity of 89 % [84]. Microsatellite alterations were detected in DP1 (D5S346) on 5, 486 (D7S486) on 7q, 522 (D7S522) on 7q, D11 (D11S904) on 11p, p53V (intron1 in p53) on 17p, and BRCA1 (D17S855) on 17q in ovarian cancer patients [85]. High clonality was found between most serum and peritoneal fluid and their corresponding tumor samples. A higher frequency of novel alleles were found in early-stage patients than in later-stage patients [85].

6.3.4 Methylation

Methylation is an event in which a methyl group is added to the carbon at the fifth position of a cytosine ring, resulting in the formation of a 5-methylcytosine [86]. DNA methylation helps in the epigenetic regulation of gene expression. In cancer, genes that could inhibit tumorigenesis are often hypermethylated to stop their expression [86]. Several groups have studied the methylation status of such genes in cfDNA [87–90].

P16 was found to be aberrantly methylated in breast cancers [91], HCC [87, 89], NSCLC [92], esophageal squamous cell cancer [93], and CRC [94]. In lymphoma, p16 methylation co-occurred with fragmented DNA in plasma, and both parameters correlated with disease stage [95]. After surgical resection of HCC, the median plasma p16 methylation reduced 12-fold [87]. P16 and p15 was concurrently methylated in HCC [96] and head and neck squamous cell carcinoma (HNSCC) [97]. A panel of genes including p16, DAPK, GSTP1, and MGMT was methylated in lung cancer [98] and head and neck cancer [99, 100]. APC was aberrantly methylated in lung cancer [101].

In breast cancer patients undergoing neo-adjuvant therapy, serum RASSF1A methylation became undetectable during therapy in patients achieving response but persisted in patients who had partial or minimal response [102]. RASSF1A methylation significantly correlated with overall survival and chemotherapy response in melanoma patients [103]. Methylated RASSF1A and/or APC genes in serum DNA were independently associated with poor outcomes in breast cancer [104]. Methylation of APC, RASSF1A, GSTP1, and ESR1 correlates with the presence of CTC in breast cancer [105, 106]. RASSF1A methylation was also detected more frequently than APC and ESR1 [105]. The presence of methylated APC, RASSF1A genes, and CTCs correlated with American Joint Committee on Cancer (AJCC) staging. Methylated GSTP1 was mainly detected in patients with large primary tumors and was highly correlated with positive human epidermal growth factor receptor (Her)-2/neu status [106].

The presence of high amounts of methylated DNA sequences and CTCs correlated with a more aggressive tumor biology and advanced disease [105, 106]. Hypermethylation of GSTP1 and RARB2 was detected only in stage IV patients with prostate cancer, indicating that these two may have roles in invasion and metastasis [84]. Methylation analysis of RARB2 and RASSF1A provided 95 % diagnostic coverage in breast cancer patients and 60 % in patients with benign lesions [107]. RASSF1A gene promoter methylation was more frequently detected in samples of glial tumors than in metastatic central nervous system (CNS) cancers [88]. Hence, methylated genes can be used to distinguish between malignant and benign cases [88, 107]. Methylation of APC, RARB2, and RASSF1A was detected more frequently in patients with advanced disease [44].

Alpha-feto protein (AFP) levels have been used as a biomarker for HCC [108]. However, since it has low diagnostic sensitivity, Chan et al. [43] evaluated the use of serum methylated RASSF1A sequences for HCC diagnosis and prognosis. Serum RASSF1A correlated with the presence of cirrhosis in hepatitis B virus (HBV) carriers but not in HCC patients. The diagnostic sensitivity and specificity for the combined analysis of RASSF1A methylation and AFP levels were 77 and 89 %, respectively, compared with 65 and 87 %, respectively, for AFP measurement alone. Hence, combined RASSF1A and AFP testing may allow early diagnosis of HCC [43].

Hypermethylated HLTF and HPP1 genes were detected frequently in patients with CRC with increased lactate dehydrogenase (LDH) levels [109]. Methylation of HLTF, HPP1, and HMLH1 were associated with tumor size, advanced disease stages, and shorter survival in the overall population [109, 110]. Methylation of HLTF and HPP1 was strongly correlated with cell death [109]. Methylation analysis of KIF1A, DCC, NISCH, and RARB genes was able to differentiate patients with cancerous tumors from those with non-cancerous tumors with 71 % specificity and 73 % sensitivity. However, each gene was less sensitive as independent assays [111]. XAF1 gene was hypermethylated in 69.8 % of gastric cancer serum samples. Methylation of XAF1 gene was a predictor of poor survival and associated with tumor recurrence [112]. Methylation of ATP4B gene was detected in the plasma of patients with gastric cancer [113].

7 Conclusion

Analysis of cfDNA from patient blood samples to obtain information about their disease status represents a convenient method of disease monitoring. Assessment of prognostic significance using cfDNA assay is non-invasive and enables simpler periodic assessment. As it provides information on each patient’s disease status in real-time, treatment decisions can be made according to the patient’s condition. An appropriate panel of genes can be used to screen for gene alterations for cancer diagnosis and prognosis, and for detection of drug-resistance mutations. Development of such cfDNA-based assays for clinical application would transform the current concepts of cancer diagnosis and prognosis and the assessment of therapeutic response. Future research work needs to focus on better and cost-effective isolation techniques for cfDNA, more robust detection of the spectrum of mutations in the cfDNA, and should compare primary tumor heterogeneity with mutations seen in cfDNA to assess the role of cfDNA as a better sample for molecular analysis than core biopsy or fine needle aspiration cytology.