Circulating cell-free DNA for non-invasive cancer management

Abstract

Cell-free DNA (cfDNA) was first identified in human plasma in 1948 and is thought to be released from cells throughout the body into the circulatory system. In cancer, a portion of the cfDNA originates from tumour cells, referred to as circulating-tumour DNA (ctDNA), and can contain mutations corresponding to the patient’s tumour, for instance specific TP53 alleles. Profiling of cfDNA has recently become an area of increasing clinical relevance in oncology, in particular due to advances in the sensitivity of molecular biology techniques and development of next generation sequencing technologies, as this allows tumour mutations to be identified and tracked non-invasively. This has opened up new possibilities for monitoring tumour evolution and acquisition of resistance, as well as for guiding treatment decisions when tumour biopsy tissue is insufficient or unavailable. In this review, we will discuss the biology of cell-free nucleic acids, methods of analysis, and the potential clinical uses of these techniques, as well as the on-going clinical development of ctDNA assays.

Biology of cell-free DNA

cfDNA is believed to be released during normal cell functions, such as secretion and export in exosomes, as well as during cell death programs, such as apoptosis and necrosis [1,2]. It can be found in plasma [3] as well as other body fluids such as urine [4,5] , cerebral spinal fluid (CSF) [6] , pleural fluid [7] , and saliva [8,9] , among others. Once outside of the cell, cfDNA is steadily degraded by nucleases [10–12] , possibly with help from macrophages [13] , and excreted in urine via the renal system [14–17]. The half-life of cfDNA is very short according to studies done in prenatal medicine, between 16 min–2.5 h in plasma [18–20] , therefore it is often highly damaged and degraded. Despite this, the fast turnover of cfDNA can be useful for real-time tracking of genetic and genomic changes.

Multiple studies have demonstrated that the majority of cfDNAs are short molecules around 167 bp, although longer fragments also exist [21].167 bp corresponds approximately to the length of DNA wrapped around a histone, indicating cfDNA is likely associated with nucleosomes [22–24]. These fragments are believed to be released from cells in exosomes and extracellular vesicles, as well as during apoptosis [1]. Dying cells and tumours undergoing cell necrosis release additional longer fragments, up to 10 kb [21]. The majority of cfDNA in the blood originates from hematopoietic cells [25]; however tissue-specific methylation signatures of the cfDNA fragments has also identified fragments from many other organs [24–26].

Circulating tumour DNA

As cfDNA originates from many organs, ctDNA released from tumours within these organs can also be found within cfDNA samples. Data from xenograft experiments has shown that ctDNA is slightly shorter than cfDNA [27,28], approximately 134–144 bp [29,30]. Although the reason behind this shortening is not known, it is possible that this is the size of the DNA wrapped around a nucleosome without the linker DNA that connects histones or that increased time in circulation resulted in degradation. Interestingly, foetal DNA and cfDNA originating from transplanted tissue is also shorter than maternal and recipient cfDNA, indicating this property is not specific to ctDNA and can occur in different types of cfDNA [31,32].

A large amount of work in the cfDNA field has been focused in maternal-foetal medicine, and here it was observed that the amount of foetal DNA found in the mother’s blood correlates with the stage of development of the fetus – the further in development, the greater the foetal cfDNA content [33,34]. This has also been shown to be true of tumours where the size [35,36] and stage [37,38] correlates with the amount of ctDNA. Stage IV or advanced tumours have 100–1000 copies of ctDNA per 5 mL of plasma, compared with only 10 copies in early stage cancers [38]. Additional increases are also seen in patients with metastatic disease. Parkinson et al. attempted to quantify this in patients with relapsed high-grade serous ovarian carcinoma (HGSOC) and found an increase of 6 copies of ctDNA per mL plasma for additional each cm 3 of disease [39].

Not much is currently known about the function of ctDNA, however several studies have indicated possible roles in immunomodulation [40–42] or transformation of nearby cells through internalization of ctDNA [43]. The latter studies have yet to be confirmed in an in vivo setting, however NIH-3T3 cell lines merely coming into contact with KRAS-mutant plasma have been transformed in vitro [44]. There remains much more to learn about the function of ctDNA and cfDNA, which hopefully will be revealed through additional research.

Methodologies for cfDNA analysis

PCR-based techniques

The first ctDNA matching a patient’s tumour mutation was identified using allele-specific PCR in 1994 [45]. Although still used in the field, digital PCR, which involves single-molecule reactions, is favoured as it provides absolute quantification of the number of transcripts in a sample [46,47]. This development has allowed the detection of rare mutations even at low abundance and has significantly increased the sensitivity of ctDNA analysis [48–50]. Droplet digital PCR (ddPCR) involves the distribution of DNA templates into thousands or millions of droplets each containing only one DNA fragment. There are many different commercially available ddPCR platforms that use various numbers of droplets and slightly different chemistries [51]. These assays mostly rely on fluorescent probes and can be multiplexed to detect the presence of a few loci or hotspots at a time. ddPCR has the potential to be applied in the clinic to monitor tumour dynamics and in the detection of minimal residual disease [52]. Despite the limitations in the number of mutations that can be tested for at a time, ddPCR is still a regularly used cfDNA technology as it provides quick turnaround time and high levels of sensitivity compared to sequencing strategies.

Next-generation sequencing-based techniques

Next-generation sequencing (NGS) techniques provide the opportunity to discover mutations without prior knowledge of the tumour mutation landscape. One type of sequencing strategy is amplicon based, where primers are designed to amplify hundreds of regions and sequenced to very high coverage to achieve a 1% sensitivity for de novo mutation calling or 0.1% sensitivity for genotyping [53,54]. Many of these panels are designed to be patient or cancer specific based on prior knowledge of the mutations within the tumour or using literature findings from tumour profiling. One key advantage of an amplicon-based strategy is the high efficiency of converting cfDNA molecules into sequencing reads, as a result of the high efficiency of amplification by PCR. Although due to challenges in retaining the relative ratio between genomic regions throughout the PCR amplification process, this method has less sensitivity for detecting copy number alterations.

Another NGS strategy is hybridisation capture-based approaches. Larger regions of the genome can be interrogated, however there is a trade-off between the breadth and depth of sequencing for the same cost. As a result, the sequencing coverage is generally reduced in larger panels leading to lower sensitivity. More sensitive patient specific assays can be designed containing only a few genes or loci of interest, however more commonly panels are developed to include multiple genes with important prognostic and diagnostic information for a specific cancer type. Cancer personalized profiling by deep sequencing (CAPP-Seq) is an example of this, and was first used to build a panel for non-small cell lung cancer (NSCLC) [55]. Larger panels encompassing hundreds of genes across many cancer types also exist and panels derived for tumour profiling can be adapted for cfDNA [56–58]. More recently, whole exome sequencing has been done on cfDNA samples, allowing identification of additional mutations and resistance mechanisms [59–63]. One of the major improvements in hybrid capture sequencing technology over the last few years has been the introduction of uniquemolecular identifiers (UMIs). UMIs are added to sequencing adaptors early in the library preparation to tag each template DNA fragment with a unique molecular barcode [64]. Once barcoded, any PCR products and reads deriving from that template will retain the barcode allowing them to be identified and collapsed into a consensus sequence, minimizing artefacts arising from PCR and sequencing errors. Many cfDNA sequencing strategies have successfully been adapted to incorporate UMIs [65–67].

Apart from point mutations and indels, epigenetic changes and chromosomal aberrations cannot be easily detected using capture-based methods. As these could also be drivers of cancer, methods designed specifically to detect these changes have been developed or adapted for cfDNA. For instance, bisulphite or other methylation sequencing methods can be used on cfDNA [24,26,68].ctDNA studies involving methylation analysis have included methylation specific PCRs [69] , arrays [70] , and bisulphite sequencing [71–73]. For copy number, the two main methods that have been employed are ddPCR [74–78] and whole genome sequencing at very low coverage (~0.1X), known as shallow whole genome sequencing (sWGS) [79] or Plasma-Seq [80,81]. By comparing the coverage across the genome to normal samples, chromosomal alterations originating from ctDNA can be detected and quantified [82], including focal amplifications and gene rearrangements. Computational modelling approaches have been developed using sWGS data to estimate the proportion of ctDNA within the cfDNA sample [81,83], a concept similar to tumour purity in biopsy samples. However, in this case knowing the fraction of molecules derived from the tumour can be used to identify patients with high disease burden [81] and to screen samples for techniques that do not require very high sensitivity and can provide additional information, such as exome sequencing [60]. A summary of the features and applications of PCR-based and NGS approaches for cfDNA can be found in Table 1.

Table 1.

Comparison of PCR-based approaches and next-generation sequencing for their use in cfDNA analysis and applications.

Methods	PCR-based approaches	Next-generation sequencing
Features	Highly sensitive (MAF 0.001%)	Less sensitive (MAF 0.1–10%)
	Fast to run	Long to run
	Limited multiplexing	High levels of multiplexing
	Easy to interpret results	Requires bioinformatics and computational resources
	Low cost	High cost
	Absolute quantification	Approximate quantification
Applications	Identification of single or few mutations for molecular diagnostics and guiding therapy	Identification of multiple mutations for molecular diagnosis and guiding therapy
	Tracking mutations during therapy	Tracking mutations during therapy
	Limited to mutations in assay	Discovery of new mutations
	Measuring mutation load to predict prognosis	Measuring mutation load to predict prognosis
	Monitoring for minimal residual disease	Monitoring for minimal residual disease requires UMIs and sequencing to redundancy
	Copy number alteration detection in single or few genes	Broad copy number alteration detection across whole genome
	Studies of emerging resistance with well-defined mechanisms	Studies of all emerging resistance
	Tracking recurrence of known mutations	Tracking recurrence of known or unknown mutations

Abbreviations: PCR polymerase chain reaction, MAF mutant allele fraction, UMI unique molecular indexes, cfDNA cell-free DNA.

Method sensitivity

By comparing the number of mutant alleles to wildtype alleles for multiple loci the average mutant allele fraction (MAF) within the sample can be derived, a metric which correlates with tumour burden [84]. The MAF will vary between alleles are they are represented at different proportions due to tumour heterogeneity. Those at higher allele fractions are likely stem mutations, while lower allele fractions are likely private mutations [84,85], although factors such as copy number can also influence the allele fraction. Despite these differences in allelic frequency, the average MAF of a sample still correlates with disease volume. Parkinson et al. managed to quantify this for HGSOC and found the MAF increases by 0.08% for each additional cm 3 of disease [39].

The sensitivity of the technologies varies greatly, and a useful way of expressing that sensitivity is using the range of MAF they are able to detect. sWGS requires between 5– 10% MAF to be detected [81], very high amounts for cfDNA, and cannot generally be used for detecting early or residual disease. However, this is dependent on cancer type: early HGSOC, which often presents with a high number of copy number variants (CNVs), has been assessed using sWGS with CNVs detected in 6/16 patients (37.5%) [86]. Other cancer types, which do not have large CNVs, may not be detected using this strategy.

Larger panels and exome sequencing have a similar limit of detection of 5% MAF [62], while smaller multiplexed panels can detect mutations down to 1% MAF [56,87,88]. This can be further improved by using cancer or patient specific panels, which can detect 0.01–0.5% MAFs. The CAPP-Seq assay for NSCLC, for instance, has a MAF limit of 0.02% and was able to find mutations in 100% of stage II or greater cancers, and in 50% of stage I [55].

PCR based assays have very high sensitivity; qPCR and ddPCR are able to detect 0.001–0.01% MAF [89], and in some samples only one copy in 5 mLs of plasma [90]. When compared with known mutation status based on tumour assays, KRAS ctDNA assays developed for clinically relevant mutations within colorectal cancer achieved 87.2% sensitivity and 99.2% specificity [91] and another similar assay developed against known mutations in early-stage breast cancer reached 93.3% sensitivity [92].

Using known mutations identified within a patient’s tumour sample can also improve the sensitivity of NGS assays [18,93] and reduce the background error rate [53]. As evidence for the mutation already exists, it is easier to be confident calling mutations in the corresponding cfDNA. As discussed above, using UMIs when making sequencing libraries increases sensitivity and can allow detection of MAFs down to 0.1% [54,65,66]. Another method to improve ctDNA detection is through size selection. As previously discussed, ctDNA is shorter than normal cfDNA. Physical size selection of the shorter molecules has been shown to enrich the sample for tumour derived fragments and improve detection rates after sequencing [29,94,95]. These strategies are just beginning to be explored, but they have to potential to significantly increase the sensitivity of these types of assays and making them a possibility for use in the clinic.

Factors affecting mutant allele fraction

Although high levels of ctDNA is correlated with high disease burden, pre-analytical, analytical and physiological factors can affect the amount ctDNA fraction within a cfDNA sample. cfDNA undergoes quick turnover in the blood [13,19,20] , reducing overall yields. Additionally, release of DNA from blood cells increases the normal cfDNA content diluting ctDNA within the sample. Specialized blood collection tubes, such as Streck and PAXgene cfDNA tubes, have been developed to protect cfDNA from degradation extending the time samples can be stored and limiting DNA release [96–98]. These are an easy and inexpensive way to ensure cfDNA remains intact and ctDNA within the sample is undiluted. Location and vascularization of the tumour can also affect ctDNA release into the bloodstream. This is most commonly observed in patients with primary brain tumours, where low amounts of ctDNA within the blood is thought to be due to the blood brain barrier [38,99]. To overcome this limitation, cfDNA samples can be obtained from CSF, which yields better results than plasma [100]. This strategy can also be used in patients with brain metastasis and central nervous system restricted disease [101]. This strategy also applies to other tumour types, cfDNA can be extracted from other liquid sample sources in close proximity to the tumour, for instance through collecting urine [15,102] or saliva [103] to increase the tumour-derived fraction.

Clinical need

cfDNA analysis will not replace tumour biopsies, however they can provide additional insight into and important clinically relevant information about the mutational landscape of tumours and metastases. In cases where it is not possible to biopsy or repeated biopsies are required cfDNA analyses are of enormous benefit, as they can provide information not otherwise obtainable. Furthermore, there are additional benefits to using cfDNA when biopsies have been taken, for instance for tumour heterogeneity, metastasis, and longitudinal studies.

Tumour heterogeneity and metastatic sites

Tumours are often heterogeneous and can have regional differences in mutation profiles [104–107] leading to actionable or prognostic mutations being missed during biopsy. cfDNA can capture these mutations to provide clinically relevant information missed due to heterogeneity [56,84,85,100,108,109], and identify mutations missed during sequencing of tumour biopsies [110,111]. This is also true of multiple tumours [112] or metastatic sites [113] ; variants missed due to spatial separation from the biopsy site can be detected using cfDNA [59,87,114].

Identifying tissue of cancer origin

Studies of DNA methylation can be used to identify the tissue of origin, a process that is especially important in cancers of unknown primary [115], and cfDNA methylation and nucleosome positioning studies can be used the same way [24–26,116]. For instance, a pregnant woman undergoing non-invasive prenatal testing was found to have copy number changes not associated with the fetus , methylation deconvolution indicated an increased contribution from B-lymphocytes and she was subsequently given a diagnosis of follicular lymphoma [117]. Using cfDNA can provide insight into the tumour tissue type, without requiring a biopsy.

Measuring disease burden

The presence of ctDNA itself is indicative of disease and the amount of ctDNA can also be an indicator of the amount of disease. As previously discussed, the amount of ctDNA is correlated with tumour stage, and many groups have observed that higher levels of ctDNA has been associated with worse survival outcomes in patients [69,118–122]. Thus, the amount of ctDNA can be used as a measure disease burden, along with imaging studies. In a colorectal cancer study, patients with detectable ctDNA after treatment had reduced (48%) 2-year recurrence free survival compared to those without (100%) [69]. Similarly, in prostate cancer patients who had AR copy number gains detected by cfDNA analysis prior to abiraterone treatment had worse outcomes and exhibited primary resistance [77]. Lower ctDNA levels correlate with treatment response [18,53,55,114,120] and this response can be detected earlier than with clinical detection methods [114,123,124]. This provides additional information about how a patient will respond to treatment. Overall, ctDNA has been shown to be a better predictor of prognosis than other tumour markers [39,114] and that ctDNA concentration increase correlates with poorer clinical and radiological outcomes [39,114,125,126].

Patient stratification

Total levels of ctDNA provide insight into disease burden and prognosis, but more specific molecular studies can provide the underlying genetic profile. Using molecular profiling tests prior to treatment provides the possibility of stratifying patients based on prognosis for the administration of adjuvant therapy [77,124] or for selection of specific targeted therapies.

Tracking specific mutations

Specific mutations can also be tracked during treatment to signal recurrence or relapse of the tumour before clinical evidence of disease, as is possible with total ctDNA burden [120]. On average, ctDNA can be detected between 7.9 and 11.0 months prior to clinical relapse in primary cancers [127–129], with similar results being found for metastases [114]. These types of assays require very sensitive approaches such as ddPCR, rely on previous knowledge of the tumour mutation profile, and are developed in a patient specific manner. This poses a challenge for clinical development as individual assays need to be created and optimized for each patient, however they have been shown in research to be very powerful methods for detecting even minimal residual disease. In a large study of early-stage colorectal cancer, ctDNA levels measured at follow up after tumour resection was predictive of recurrence-free survival after 3 years. In the ctDNA positive group 0% were recurrence-free, while in the ctDNA negative group 90% were recurrence-free, this number was improved to 100% after serial samples were taken into account [130].ctDNA positivity is indicative of minimal residual disease, the level of which can be tracked and quantified through these types of assays. A similar study was completed on patients with breast cancer, where personalized assays were able to predict recurrence and measure minimal residual disease [127]. This can also be done using patient specific rearrangements such as gene fusions [128,131,132], as rearrangements have low background noise compared to somatic mutations.

Detecting tumour evolution and acquisition of resistance

Tracking recurrence of individual mutations or rearrangements present in the original tumour provides useful information for recurrence of tumour cells containing that mutation, but does not provide additional information about recurrence with new mutations or emergence of other clones. By measuring several mutations, the changes in their ratios can provide some insight into the tumour’s evolution and continued heterogeneity during treatment [84,133–135]. This can also be extended to identifying the appearance of resistance mechanisms. Serial studies of colorectal cancer found positive selection of KRAS mutations during anti-EGFR therapy and decline in their representation after withdrawal [108,136]. Similar results have been recorded in NSCLC patients treated with EGFR inhibitors, where resistance mutations were identified in ctDNA prior to clinical progression [50,137,138]. Another study completed this year identified resistance mutations not previously observed in the tumour as driving resistance to treatment [139], indicating convergent evolution was taking place and could be identified using ctDNA. Additional studies are currently on-going involving exome sequencing to discover novel resistance mechanisms in ctDNA developed during treatment [59,84], however as these mutations need to be present at > 5% MAF to be detected, this is not a technique that can be used for low frequency variants or for minimal residual disease.

Clinical development

Non-invasive prenatal testing

There’s a huge potential for cfDNA in the clinic and already many cfDNA assays have made an impact. The initial assays were developed for identifying foetal cfDNA sequences in maternal blood plasma. This included finding foetal Y chromosome cfDNA in maternal blood, indicating a male fetus, and other tests for sex determination [33]. Non-invasive sex determination is particularly important for carriers of X-linked conditions, and cfDNA testing has resulted in a decrease in the number of invasive tests, such as amniocentesis, to confirm foetal sex [140]. Detection of chromosomes can also be done to discover aneuploidy, and has successfully been used to detect Down Syndrome (trisomy 21) [74], a test that was made available in the clinic in 2011 [141–143] and is now in widespread clinical use [144,145]. These techniques were refined to look at specific gene sequences and cfDNA tests have been developed monogenic disorders as well [146].cfDNA studies in maternal-foetal medicine have paved the way for the development of more highly specific clinical cancer assays.

PCR for guiding treatment

The most famous example of a clinically validated cfDNA test is the allele-specific PCR for the detection of EGFR mutations in NSCLC [147]. Plasma has been shown to be effective and comparable to tissue biopsies in NSCLC [148–150] and as the tumours can be difficult to biopsy [151], clinical approval has been driven faster than for other types of cancer. Two forms of the test exist and are approved for selection of gefitinib [152], erlotinib [153] or osimertinib [153,154]: Therascreen EGFR was approved by the European Medicines Agency in 2015 [155] and Cobas EGFR was approved by the US Food and Drug Administration (FDA) in 2016 [153].PCR-based tests are highly sensitive and specific, are easier to gain approval for than NGS assays, and additional ctDNA assays for other genes and cancer types are likely to follow soon.

Sequencing panels for guiding treatment

Although the majority of clinical development has been in PCR-based assays, some clinical sequencing panels have been developed. A 34 gene panel developed by Inviata for NSCLC found mutations in 79% of 174 patients, following this 28 patients (17%) went on to receive personalized treatment [156]. Another broader 54-gene panel detected ctDNA in 58% of patients with multiple types of cancer, with 68% of those having an actionable mutation by an FDA-approved drug [57]. Similar proof of concept studies have been done showing the feasibility of matching metastatic patients to targeted therapies for clinical trials, however they have not been approved as of yet [157]. Numerous companies have begun to capitalize on the prospect of ctDNA screening, including Guardant Health, who have developed a 73-gene ctDNA assay called Guardant 360 [67,158]. Assays such as these have begun to be used to guide enrolment in clinical trials [159], in particular for cases where a positive finding results in enrolment.

Screening for early detection of disease

cfDNA analysis has the potential to be used as a screening tool prior to clinical onset of disease [103,160,161]. Screening of viruses associated with cancers, such as Epstein–Barr Virus (EBV) – a virus associated with nasopharyngeal carcinoma, has been shown to be an effective way to use cfDNA to screen for early stage and asymptomatic cancers in the clinic, as viral cfDNA levels are much higher than ctDNA. One study involved screening of > 20,000 asymptomatic people for EBV DNA in their plasma and lead to the diagnosis of nasopharyngeal carcinoma in 34 individuals [162]. Identification of these patients led to early diagnosis and improved 3-year progression-free survival [162]. Screening for ctDNA is also being trialled, recently a screening panel of 58 genes has been developed for early detection of cancer [161], although it has only been used in patients with known stage I or II disease to date, this study demonstrates the potential of screening ctDNA. Techniques such as this have the potential to improve patient outcomes and continued development will soon improve non-invasive cancer screening.

Future developments

Sequencing developments

Promising improvements are being made in sequencing technology and bioinformatics methods that will benefit and improve cfDNA analysis. Longer cfDNA fragments are not easily detected during current sequencing methods [163], however long read sequencers, such as Nanopore, have improved ability to identify structural variants [164]. Additionally, using alternative library preparation methods, such as single-stranded library preparation, has been shown to increase the diversity of sequenced fragments to include the shorter molecules often missed during standard double stranded preparations [24,165,166]. These methods could also allow further study of cfDNA fragmentation, which could provide complementary information to MAF for determining the amount of ctDNA in a sample [88,167] or to understand the fragmentation and clearance from blood through urine [20,168]. Additional studies into the longer and shorter fragments could provide additional insight about the biology of cfDNA and ctDNA.

Advances in biological knowledge

Although it is known that ctDNA is shorter than cfDNA, the exact mechanism of how this occurs has yet to be elucidated. Studies into these size differences, as well as how cfDNA is protected in body fluids would provide important information about these mechanisms as well as tumour specific processes resulting in the release of ctDNA. There are numerous factors influencing the amount of cfDNA in circulation and the proportion of ctDNA other than the size of the tumour. The amount of exercise done prior to blood draw can increase the amount of lymphocyte derived cfDNA greatly [169,170], which dilutes the ctDNA within a sample making detection of mutations more difficult. However, why and how this happens remains unknown. Additional factors, such as blood supply to the tumour presumably could also have an effect on the amount of ctDNA in the blood, and again these factors have not been studied in detail.

Research by Snyder et al. indicates an association between cfDNA and nucleosomes, however functional studies have not yet been done to confirm how long this association lasts and if recovered cfDNA is still bound [24]. Several groups have also suggested studies should be done into the other proteins associated with cfDNA [171,172], however little research has been done into this area as of yet outside of studies of exosomes.

Exosomes and cell-free RNA

Exosomes are membrane bound vesicles containing proteins, DNA, and RNA that is packaged and excreted from cells for transport or signalling between cells [14,173]. After being excreted, exosomes can enter bodily fluids, such as blood, and can subsequently be isolated and purified. Exosomes can be a rich source of information often lost from other cell-free samples as the membrane protects the contents within from degradation. They may contribute to cancer progression and treatment failure, and can contain ctDNA [173–175]. Numerous studies of the content of exosomes have been published [176,177], including combined cfDNA and cell-free RNA (cfRNA or exRNA) studies [178]. Exosomes are a promising way to study cfRNA [178,179]. Sequencing of the exosomal RNA from plasma in normal healthy individuals [180,181] and cancer patients [182,183], has identified that the majority of RNA within exosomes is microRNA (miRNA), with some fragments of coding sequence (mRNA) present as well.

RNA can also be detected at a lower level in a cell-free state in plasma without the time consuming and difficult procedures of isolating exosomes. cfRNA extracted this way also mainly contains miRNAs [184], however, important diagnostic and prognostic mRNA fragments can also be identified [185–187].Sequencing of cfRNA has been done by several groups [188–190], although the most promising avenue currently being explored for cancer is in the identification of tumour specific RNA transcripts and fusion genes, which are especially difficult to detect through DNA methods. That said, by further sequencing and characterizing of the exosomal and cfRNA profiles, it is possible that they could provide more information about cancer both for diagnostics and for fundamental studies of cancer biology and progression.

Clinical progress

Progress in the development in clinical tests and validation has been rapid, and is likely to continue in that fashion as they are approved in additional cancer types and as more drugs are developed requiring molecular stratification. The only approved tests currently use plasma as a starting source for cfDNA, however there is potential for this to expand to other sources of cfDNA and to cfRNA.

Most studies are currently focused on identification of mutations and cancer evolution; although many other potential areas of research and clinical use exist that have yet to be explored in depth, such as in understanding response to immunotherapy [120,125,191]. Further studies into these more complex areas will be able to provide additional information, not only about cancer, but also about cfDNA itself, and will help us to fully understand the potential and limitations of cfDNA analysis.

Conclusion

cfDNA analysis is a non-invasive method for obtaining important and clinically relevant molecular information about cancer that is rapidly gaining momentum in the research and clinical world. Technological developments in molecular biology and NGS approaches continue to advance the sensitivity of cfDNA analysis techniques, increasing the potential for its use not just in prognostic and diagnostic settings where disease burden is high, but also for minimal residual disease, emergence of resistance, and early screening. Clinically validated cfDNA assays are beginning to make an appearance in oncology, and will likely become widespread for monitoring response to treatment and molecular stratification. As research into additional sources of cfDNA and cell-free analytes continues, a wider variety of options for detection will become available and an increased understanding of the biology of cfDNA and ctDNA can be gained.