Abstract
Background
Despite advances in early detection and therapies, cancer is still one of the most common causes of death worldwide. Since each tumor is unique, there is a need to implement personalized care and develop robust tools for monitoring treatment response to assess drug efficacy and prevent disease relapse.
Main body
Recent developments in liquid biopsies have enabled real-time noninvasive monitoring of tumor burden through the detection of molecules shed by tumors in the blood. These molecules include circulating tumor nucleic acids (ctNAs), comprising cell-free DNA or RNA molecules passively and/or actively released from tumor cells. Often highlighted for their diagnostic, predictive, and prognostic potential, these biomarkers possess valuable information about tumor characteristics and evolution. While circulating tumor DNA (ctDNA) has been in the spotlight for the last decade, less is known about circulating tumor RNA (ctRNA). There are unanswered questions about why some tumors shed high amounts of ctNAs while others have undetectable levels. Also, there are gaps in our understanding of associations between tumor evolution and ctNA characteristics and shedding kinetics. In this review, we summarize current knowledge about ctNA biology and release mechanisms and put this information into the context of tumor evolution and clinical utility.
Conclusions
A deeper understanding of the biology of ctDNA and ctRNA may inform the use of liquid biopsies in personalized medicine to improve cancer patient outcomes.
Similar content being viewed by others
Background
Despite advances in early detection and treatment, the number of new cancer cases and deaths is still increasing globally [1]. Moreover, each tumor possesses a unique genetic profile and has the potential to develop drug resistance and spread to distant sites [2]. Hence, new strategies for personalized treatment guided by diagnostic, predictive, and prognostic biomarkers are urgently needed to reverse increasing incidence and mortality rates. Adopting blood-based liquid biopsy into clinical practice could help guide therapeutic strategies in personalized medicine. Robust and accessible biomarkers for immediate assessment of tumor response and monitoring of minimal residual disease (MRD) are crucial to improving patient outcomes. Thus, recently published research articles and reviews have highlighted the potential of liquid biopsy-based biomarkers as a real-time reflection of the tumor burden with diagnostic, prognostic, and predictive information to guide cancer management [2,3,4,5,6,7,8].
Cells and DNA shed by tumors into circulation, also known as circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA), respectively, are considered two major components of liquid biopsy [9]. However, the lack of standardization of CTC detection methods [10], as well as the high false-negative rate of ctDNA assays, points to the need for further technological advancements to support liquid biopsy standardization and improve test performance [9]. Circulating tumor RNA (ctRNA) is an emerging biomarker that could provide unique information not found in CTCs and ctDNA.
Cell-free DNA (cfDNA) are fragments of DNA released into the bloodstream which originate mainly from the apoptosis of hematopoietic cells [6, 8]. DNA released by tumor cells may possess alterations that can provide highly specific markers for detection [6, 8, 11]. Notably, compared to healthy individuals, cancer patients’ blood has been observed to contain increased levels of cfDNA [12] as well as messenger RNA (mRNA) and non-coding RNA (ncRNA) [13, 14]. Cell-free nucleic acids (cfNAs) can be released passively into circulation mainly via apoptosis and necrosis as well as through active secretion via extracellular vesicles (EVs) from viable cells. In this review, we use the term ctNAs to represent the fraction of total cfNAs (DNA and RNA) released exclusively by tumor cells. Understanding the nature and origin of ctNAs provides pivotal clues for exploiting these biomarkers in specific clinical settings. The unique characteristics of ctNA molecules go hand in hand with the process of their release from cells and the kinetics of their clearance [15, 16]. While ctDNA can harbor critical genetic traits of tumorigenesis, ctRNA can reflect intra-tumoral dynamic processes on the cellular and intercellular levels [4, 17].
Aside from blood, other non-invasive approaches using urine, saliva, and semen plasma, along with invasive methods using cerebrospinal fluid (CSF), and pleural and peritoneal effusions, have been utilized to assess ctNAs [18,19,20]. A recent review article discussed the properties of ctDNAs originating from different body fluids providing a comprehensive summary of the peculiarities of ctDNA from various sources [20]. For example, ctDNAs detected in urine are composed of shorter fragments (< 100 bp) that are passed from plasma through the glomeruli (tiny networks of blood vessels in the kidney involved in waste filtration) as well as longer ctDNA fragments shed directly by tumor cells in the urinary tract [20, 21]. Additionally, saliva has been suggested as a potential source of ctDNA from local tumors but is very short (40–60 bp) and less enriched [20]. The concentration of ctDNA can also vary from one compartment to another. For example, ctDNA concentration in the CSF is higher than in the plasma [22], perhaps due to the presence of fewer immune cells (compared to blood) that could contribute to the background cfDNA. Similarly, pleural and peritoneal effusions comprise a richer source of ctDNA than plasma due to the proximity of these fluids to tumors that shed these molecules [20, 23]. In addition, cfDNA levels are relatively higher in seminal plasma than other body fluids depending on sexual activity/abstinence and individual composition of the seminal fluid [20, 24].
Despite the numerous publications on ctNAs, there are still many unanswered questions. What governs the fluctuations in the ctNA levels in the blood? Is it possible to distinguish between ctNAs shed by cells dying in response to treatment and ctNAs actively secreted by treatment-resistant cells? Is the absence of ctNAs in cancer patients due to the assay’s low sensitivity (false negativity), or can these biomarkers be truly absent, and why? And, if detected, why do they not correlate, in some cases, with tumor pathophysiologies such as size and stage? And how big an issue is false positivity? To adopt ctNA assays in clinical practice, we need to understand not only their nature and mechanisms of release from cells but also their fates in circulation.
Many research articles and reviews have focused on cfDNA or ctDNA, but less is known about ctRNA. This review summarizes the biology of ctDNA and ctRNA, their release mechanisms from cells, and the kinetics of degradation. Finally, we put these findings in the context of cancer evolution and clinical utility.
Circulating tumor nucleic acid release mechanisms
Mechanisms involved in ctNA release need to be better understood. There are considerable gaps in our knowledge regarding the presence, fluctuations, and characteristics of ctNAs and their potential roles in tumor resistance and evolution. Recent improvements in the sensitivity and specificity of detection methods [10, 25] have facilitated the progress in understanding the biology of ctDNA [6] and ctRNA [14]. While these data could be divergent, burdened with preanalytical variabilities, and lacking standardization methods [26], unified themes can be gleaned from the information available (Fig. 1). To date, systematic investigations of active and passive ctNA release mechanisms have yet to be, to our knowledge, fully described. In the following section, we discuss the current findings about the mechanisms involved in the release of ctNAs into circulation.
Passive release mechanisms of circulating tumor nucleic acids and their properties
Hematopoietic cell turnover has been suggested as the most significant source of cfDNA in blood and is related mainly to cell death by apoptosis [6, 8]. Cancer cells can undergo cell death, either by apoptosis or necrosis, resulting in the release of ctDNA [15]. Apoptosis and necrosis are considered major contributors to ctDNA release, but their exact contribution is unknown [6, 11, 27]. Uncontrolled proliferation is a well-known characteristic of cancer. Rapid cell proliferation causes local nutrient depletion, hypoxia, inflammation, oxidative stress, acidosis, and the production of corresponding tissue-specific transcription factors and signaling death-inducing molecules [28, 29]. Apoptosis and necrosis are major results of hypoxic and metabolic stress and can cause the passive release of cellular content into the extracellular space [11, 30]. In this section, we discuss processes involved in the passive release of ctDNA and summarize current knowledge about passive ctRNA release, a much less understood phenomenon than ctDNA release.
Apoptotic cell-derived cell-free DNA possesses a ladder-like pattern profile
Apoptosis is a form of programmed cell death for maintaining homeostatic balance and is executed by caspases, leading to cell shrinkage, chromatin condensation, and fragmentation. Cellular contents, including nucleic acids (NAs), are then packaged into apoptotic bodies protecting them from digestion by circulating nucleases. The apoptotic bodies are then cleared by phagocytosis, enzymatically digested, and released as soluble debris [4, 31].
Although the exact proportions of NAs released via different types of cell death are unknown, some characteristics can indicate the source of cfNAs. There is strong evidence that cfDNA fragment size distribution is not random and possesses a ladder-like pattern of sizes as visualized by gel electrophoresis [32]. CfDNA fragment size depends on the number of nucleosomes the DNA is wrapped around [5, 11]. The peak cfDNA fragment size is 167 bp, corresponding to the length of DNA around one nucleosome (147 bp) and a linker DNA (20 bp) protecting DNA from cleavage [33,34,35]. This characteristic of cfDNA is predominantly a result of apoptotic internucleosomal DNA fragmentation. Briefly, caspase-activated DNase (CAD) [36], DNaseI L-3 [37], NM23-H1 [38], and EndoG [39] nucleases are activated after apoptotic stimuli and execute continual DNA fragmentation with specificity for the internucleosomal region of DNA not protected by histones. A subset of cfDNA can still be wrapped around histones and thus circulate as nucleosomes [33]. CfDNA wrapped around nucleosomes are protected from cleavage by DNases [33,34,35].
Necrotic tumor cells release DNA mainly through phagocytosis
Besides apoptosis, necrosis is also considered a significant source of cfDNA in cancer patients [5, 15]. Unlike apoptosis, the purpose of which is to eliminate abnormal or unneeded cells, necrosis is a faster and more direct reflection of the adverse tumor environment in cancer. Necrotic cells exhibit organelle dysfunction and plasma membrane aberration, which may lead to the random release of cellular components exposing the tumor DNA to intracellular and extracellular degradative agents such as nucleases and free radicals [4, 5, 27]. As a result of the non-systematic release and digestion of DNA during necrosis, larger fragments of up to many kilo-base pair (kbp) sizes are thought to be shed in circulation [4, 11]. The resulting sizes are useful characteristics for identifying necrosis-derived cfDNA [4].
Necrosis-induced release of DNA is a complex process given that necrotic tumor cells produce various immune cell attractants and are, together with the leaked content, efficiently eliminated mainly by macrophages. This results in the digestion of the cellular DNA and the release of digested ctDNA into the extracellular space [15, 16]. For example, necrotic Jurkat T-lymphocyte cells (derived from a patient with T cell leukemia) alone released minimal levels of ctDNA, but coculture with macrophage cell lines caused a significant increase in ctDNA levels in the culture medium [40]. Detection of long ctDNA fragments might result from exceeding the phagocytic capacity in digesting the released cell contents [5, 40].
Other passive release mechanisms
Circulating tumor cells are a minor source of circulating tumor nucleic acids
CTCs are possible sources of ctDNA [15, 41, 42]. When CTCs are released into circulation, they face various biophysical pressures such as hemodynamic forces, bloodstream swiftness, collisions with blood elements, and complex formation with non-malignant cells such as leucocytes and thrombocytes [43, 44]. These can result in CTC breakage and release of NAs. But given the rarity of CTCs, the lack of evidence, and the infeasibility of quantifying the rate of these events, this proposed mechanism of ctDNA release may not be of clinical relevance [45]. Indeed, the ctDNA genome equivalents are 100 to 1000 times higher than those of CTCs, so the quantity of ctDNA does not correspond with the number of CTCs [4, 46, 47]. Also, ctDNA has often been present in samples where CTCs were undetectable but not vice versa [16].
Chromosomal instability can result in tumor DNA release from cells
Chromosomal instability represents a common trait of cancer and can result in the passive release of ctDNA [48,49,50]. CtDNA can be released via micronuclei, nuclear sub-entities containing chromosomal DNA that segregated aberrantly during mitosis and assembled their own nuclear envelope. Several micronuclei may be formed, with their levels increased in cancer cells. These sub-organelles have been hypothesized as possible translocators of DNA to the extracellular space [11]. Direct evidence studying isolated micronuclei is needed to confirm this potentially promising source of ctDNA [11].
Chromosome fragments that are not reintegrated into reassembled chromosomes can join together, creating double minutes (DMs). These tiny fragments of extrachromosomal DNA are frequently seen in many cancer types [51]. DMs have been observed as extrachromosomal circular DNA in mice and humans, often containing amplified oncogenes [11, 52]. DMs often lack regulatory sequences and are prone to continuous expression and autonomous replication leading to gene amplification. DMs can exit the nucleus by budding and subsequently be extruded from the cells as micronuclei [50]. Alternatively, micronuclei can be eliminated by autophagy [53] and DNA digested in lysosomes, eventually releasing ctDNA into the extracellular space.
The possible contribution of other cell death types to circulating tumor DNA release is unclear
Cell death is a complex process influenced by many factors and may be accomplished via different mechanisms [54]. Thus, the contribution of different cell death types to the ctDNA pool is difficult to estimate. An outstanding question has arisen, whether cell death types like parthanatos, pyroptosis, ferroptosis, necroptosis, and oncosis contribute to ctDNA release [5, 55]. Briefly, necroptosis is a caspase-independent type of programmed cell death possessing similarity to apoptosis but resulting in membrane rupture and cell content release. Similarly, ferroptosis is a membrane rupture-associated programmed cell death induced by the accumulation of lipid peroxides resulting from the failure of antioxidant glutathione-based systems. Pyroptosis is a caspase-dependent, rapid cell rupture-related form of cell death. Ischemic cell death, or oncosis, is a term for a lethal injury early response induced by ischemia [56] degradation by endonucleases and depends on the expression of specific proteins [57]. The role of these cell death types in ctDNA release is unknown and has yet to be demonstrated.
The rates at which different cell death types occur and contribute to the shedding of ctDNA are difficult to estimate. While specific cell death mechanisms are associated with distinct morphological, biochemical, and immune-related changes, these processes are molecularly interconnected [58, 59]. Crosstalk between cell death pathways occurs, allowing backup mechanisms to exist [60]. For example, the rate of necroptosis may be elevated in some cancers as an alternative cell death mechanism to apoptosis [61, 62], but it can also be attenuated in cancer cells resistant to cell death [61]. Cell death can be activated under specific conditions, e.g., stress [62]. For example, nutrient depletion often triggers increased rates of ferroptosis [59]. Parthanatos can be induced by DNA damage [63], and its rate is substantially elevated due to oxidative stress in the tumor microenvironment or by cancer treatment using alkylating agents [64]. Oncosis can frequently occur in cancers with the altered expression of ion channels and compromised ion gradient [65]. Pyroptosis, a type of cell death associated with inflammation, can be induced by damage-associated molecular patterns (DAMPs, e.g., cfNAs and other products of cell death) and can be observed more frequently in highly inflammatory cancers [99,100,101]
Microvesicles are a less studied but appreciable source of circulating tumor DNA
About 100 to 1000 nm in size, MVs are formed by the outward budding of the cell membrane [83]. MVs are also thought to contribute to cell homeostasis and cell-to-cell communication [94]. Less is known about their role in cancer, but MVs derived from cancer cells have also been shown to contain DNA [81, 83, 94, 103]. MVs contribute to tumor pathophysiology as mediators among cancer and stromal cell in the tumor microenvironment [141].
RNA is subjected to intensive degradation early after apoptotic stimuli
There is strong evidence that cellular RNA is subjected to rapid degradation during apoptosis [142]. It has been shown that global mRNA decay occurs early after apoptotic stimuli before DNA degradation begins following the permeabilization of the outer mitochondrial membrane [142]. This permeabilization leads to the release of DIS3L2 and PNPT1 ribonucleases that degrade mRNA [142, 143], and thus, mRNA has been observed to comprise about 2.1% of the total extracellular RNAs [144]. Similarly, DIS3L2-associated degradation was also observed on pre-miRNAs and Poly(A) ncRNAs [145].
Unlike mRNA, short ncRNAs have been shown to be remarkably stable in the plasma and serum of cancer patients [178]. Several ctDNA-based diagnostic tests have been approved for clinical practice (Table 1), representing an important milestone in liquid biopsy implementation [6, 11].
Tumor heterogeneity, evolution, and clonal hematopoiesis may partly cause ambiguity in the ctDNA measurement data [11, 189]. Still, the need for more standardization of both preanalytical phase and ctDNA detection methods remains the most significant challenge [189]. Interestingly, as the clotting process during serum preparation induces an increase in cell lysis, ctDNA analysis might be hampered by increased levels of high-molecular cfDNA when using serum instead of plasma [190]. Thus, plasma has been suggested as the better specimen type for the ctDNA analysis [5, 15, 26]. In addition, while EVs comprise a more consistent source of some miRNAs biomarkers, other miRNAs have been more efficiently isolated from plasma [191]. As described above, cfDNA occurs mainly as dsDNA [11, 89], and ctDNA is generally more fragmented than cfDNA. Since dsDNA library preparation might not detect highly degraded ssDNA, ssDNA-library-based sequencing might improve ctDNA recovery [12].
CtDNA analysis has the potential to provide valuable information regarding tumor dormancy. Actively released ctDNA may have clinical relevance in cancer patients at risk of having dormant disseminated tumor cells (DTCs) [120]. Given that senescent cells also produce EVs, mainly exosomes [192], these particles might be pivotal for dormant tumor cell detection [193]. Detecting dormant DTCs is technically challenging; thus, develo** blood-based dormancy-related biomarkers (as surrogates of DTC presence) may improve sensitivities for the detection of MRD, which could be difficult to evaluate by conventional means, such as imaging.
Necrosis is associated with the release of large cfDNA fragments (up to kbps in size) from the cells [40, 86]. Necrosis-derived cfDNA is phagocyted and digested to residual fragments by macrophages [15, 16]. The presence of longer fragments of cfDNA in the circulation can indicate increased rates of tumor necrosis [4, 11]. Necrosis is related to the aggressiveness of the disease, and its increased rates have been associated with poor prognosis in several tumors, such as breast, renal, mesothelial, and lung cancers [158].
A ladder-like pattern of cfDNA sizes is a characteristic indication of apoptosis in cancer patients and healthy individuals [5]. As discussed previously, the cfDNA size profile has a size peak of 167 bp, corresponding to the length of DNA around one nucleosome with a linker DNA [33,34,35]. However, ctDNA fragments are shorter than cfDNA derived from non-cancer cells [32]. This is probably caused by cancer-related hypomethylation of DNA, which is more accessible to cleavage by nucleases [12]. Short cfDNA is more common in metastatic breast cancer patients when compared to primary breast cancer [32]. Moreover, specific size populations of cfDNA differ in genetic alteration frequency, and short cfDNA fragments have been identified as the major source of mutant-specific alleles [32]. The association of shorter DNA molecules to transcriptional factors [140] and tissue-specific nucleosome wrap** [141] can also explain ctDNA shortening (see section "Circulating tumor DNA is generally shorter than cell-free DNA"). Thus, the analysis of shorter ctDNA fragments (< 145 bp) may improve ctDNA detection among abundant cfDNA derived from non-cancer cells [4, 33, 35]. The different size profiles of cfDNA may serve as prognostic biomarkers as they vary in different stages and correlate with clinical outcomes [32]. Indeed, a correlation between shorter fragments of cfDNA and shorter progression-free survival and overall survival has been shown in pancreatic cancer patients [194]. Similarly, shorter cfDNA length was associated with poor survival and severity of renal cancer [195]. Altogether, the assessment of cfDNA levels and cfDNA size (e.g., shortening) correlates with advanced stages and cancer progression and thus could aid in predicting patient outcomes [32].
Additionally, tissue-specific DNA fragmentation and nucleosomal occupancy patterns have been proposed as promising tools for the identification of ctDNA tissue of origin [5, 16, 33]. Differentiation of ctDNA size populations might be of diagnostic value, which could improve ctDNA detection and cancer monitoring in different pathophysiological stages [4, 5, 141].
False positivity and negativity are critical challenges in the clinical implementation of ctDNA as a biomarker for guiding treatment and predicting recurrence [11, 15, 196]. False negativity may result from very low ctDNA content of cfDNA shed into circulation, especially in the early stage of the disease. Analysis of cfDNA fragment and epigenetic (e.g., methylation) patterns have been suggested as possible improvements to ctDNA analysis, providing reliable negative results [178]. Conversely, false positivity may arise from tumor heterogeneity but is more likely from clonal hematopoiesis and detection of somatic alterations in DNA released by normal blood cells [197, 198]. The predominance of cfDNA over ctDNA and its release mainly by hematopoietic cells, as well as the partial overlap of genes mutated in clonal hematopoiesis with tumor drivers, can significantly increase the risk of false-positive ctDNA detection and limit copy number alteration detection [178].
Clinical utility of circulating tumor RNA
CtRNAs have been suggested as promising minimally invasive diagnostic and prognostic biomarkers [45, 47, 83]. Thus, the standardization of preanalytical and analytical conditions for ctNA biomarker analysis is a crucial prerequisite for their clinical implementation [8, 15].
Despite persisting technical challenges, cancer heterogeneity, and the slow rate of new blood biomarkers approval, ctDNA analysis has been in the spotlight during the last decade [4, 6]. New protocols, independent parallel experiments, and regulatory guidelines have been suggested that consider validated pre-analytic and post-analytic principles of ctDNA analysis [8, 15, 26, 198]. Current proof-of-concept studies can lay the foundations for prospective studies with larger cohorts [6, 196, 198].
Moreover, international and interdisciplinary partnerships and consortia across academic institutions and industry have been established focusing on liquid biopsy implementation [8]— namely, SPIDIA4P consortium - Standardization and improvement of generic Preanalytical tools and procedures for In-vitro DIAgnostics [10], Cancer-ID [10], ISLB - International Society for Liquid Biopsy [209], ILSA - International Liquid Biopsy Standardization Alliance [7], ELBS – European Liquid Biopsy Society [8], BLOODPAC - US Blood Profiling Atlas of Cancer [8, 10]. The creation of these organizations is a crucial milestone for facilitating the standardization of ctDNA analysis for clinical applications.