The journey of cf-mtDNA from serendipitous observation in 1948 to a clinically actionable biomarker in 2025 encapsulates the iterative cycle of discovery, mechanistic interrogation, and translational validation that defines modern precision medicine. The study of cf-mtDNA can be traced back to 1948, when Mandel and Metais first detected cell-free DNA (cf-DNA) in human plasma. cf-DNA comprises nuclear DNA (nDNA), mitochondrial DNA (mtDNA), viral DNA, and messenger RNA. Since this initial observation, cf-DNA has rapidly evolved into a non-invasive molecular marker with broad applications. In 2000, Zhong et al. reported the presence of specific cf-mtDNA mutations in the serum of diabetic patients (Zhong et al., 2000), marking the beginning of cf-mtDNA as a disease biomarker. Subsequent studies have confirmed the presence of cf-mtDNA in both plasma and serum. Wang et al. developed a fragmentomic assay based on aberrant cf-mtDNA patterns for non-invasive early detection of colon cancer (Wang et al., 2025). Additional investigations have demonstrated the diagnostic value of cf-mtDNA in systemic lupus erythematosus (Halfon et al., 2025), acute kidney injury (Malik, 2023) and embryo quality assessment (Steffann et al., 2015). Thus, cf-mtDNA has exhibited substantial translational potential across multiple disciplines.

Mitochondria contain their own non-chromosomal DNA—mtDNA. This 16.5 kb, non-methylated, circular double-stranded molecule encodes 22 tRNAs, 2 rRNAs, and 13 protein subunits essential for oxidative phosphorylation. These proteins constitute complexes I–IV of the electron transport chain and ATP synthase (complex V), driving mitochondrial respiration and ATP production. Studies have shown that oxidative phosphorylation capacity differs across rodent tissues, depending on mitochondrial mass and complex activity (Benard et al., 2006). A major regulatory region—the D-loop—harbors transcription promoters. Analogous to chromatin, mtDNA is packaged into nucleoids, mtDNA–protein complexes that associate with the inner mitochondrial membrane (Falkenberg et al., 2007). Over 50 nucleoid-associated proteins bind mtDNA, among which mitochondrial transcription factor A (TFAM), a high-mobility group protein, is the principal component (Lee and Han, 2017). TFAM maintains mtDNA transcription and protects against reactive oxygen species (ROS) damage (Kang and Hamasaki, 2005; Kanki et al., 2004). RNAi-mediated TFAM knockdown reduces mtDNA content in proportion to TFAM levels (Kansaku et al., 2018), and both TFAM concentration and mtDNA-binding density are modulated to achieve precise transcriptional control (Bestwick and Shadel, 2023). Despite these protective mechanisms, mtDNA repair pathways are limited compared with nDNA; for example, mitochondria lack nucleotide excision repair. Consequently, the mitochondrial genome is vulnerable to ROS and toxic insults.

The fragment size of cf-mtDNA remains controversial in research findings. Due to the lack of histone protection, mtDNA is more susceptible to degradation (Jiang et al., 2015), potentially resulting in shorter cf-mtDNA fragments compared to cf-nDNA. Studies report that cf-nDNA fragments peak at approximately 167 bp, suggesting their association with histones and circulation as intact nucleosomes in blood (Lo et al., 2010) Notably, research has demonstrated that (Wang et al., 2011) testicular germ cell cancer patients exhibit elevated levels of fragmented cf-mtDNA (79 bp and 220 bp) in plasma compared to healthy controls. Importantly, the 79 bp mtDNA fragments showed greater diagnostic value than 220 bp fragments, indicating that shorter cf-mtDNA fragments may play a critical role in human diseases.

Second-generation sequencing enables single-nucleotide resolution measurement of plasma DNA fragments. Lo et al. (Lo et al., 2010) identified a predominant plasma cf-mtDNA peak at 140 bp, shorter than the 167 bp cf-nDNA fragments. However, their fragment size analysis was technically limited by conventional DNA library preparation methods (Wang et al., 2011), which involve multiple purification steps leading to poor recovery of short DNA fragments (<100 bp). Optimized plasma DNA isolation and library preparation protocols (Zhang R. et al., 2016) have revealed significantly shorter average cf-mtDNA lengths than previously reported (Jiang et al., 2015). Collectively, these findings highlight the biological and diagnostic significance of shorter cf-mtDNA fragments, while underscoring the critical need for optimized detection methodologies to fully capture their clinical potential.

Extracellular mtDNA exhibits diverse forms of existence. When mtDNA escapes from damaged tissues or is actively secreted by viable cells into circulation, it may exist either as free circulating mtDNA or in vesicle-encapsulated forms (including platelet-bound forms) (Torralba et al., 2016; Lindqvist et al., 2018). Following extracellular release, mtDNA responds to various cellular signals including stress and injury. If not promptly cleared, cf-mtDNA can function as an autoantigen to induce inflammatory responses. Furthermore, elevated cf-mtDNA concentrations correlate with numerous chronic diseases and may serve as prognostic biomarkers for disease progression and survival outcomes (Lindqvist et al., 2018; Shockett et al., 2016; Pyle et al., 2015).

The high mutation rate and low recombination frequency of mtDNA have driven the evolution of unique natural selection mechanisms in female germ cells to prevent accumulation of deleterious mutations. Toby Lieber et al. (Li et al., 2019) demonstrated in female Drosophila that fragmented mtDNA facilitates selective clearance of mutant mtDNA through a quality control mechanism. Extracellular vesicles (EVs) have been established as crucial mediators in inflammatory processes (Hezel et al., 2017), serving as effective carriers for mitochondrial components. While larger EVs (e.g., microvesicles) can transport entire mitochondria, smaller EVs (e.g., exosomes) primarily deliver mtDNA and other nucleic acids to recipient cells (Berridge and Neuzil, 2017; Sansone et al., 2017). However, the precise mechanisms governing EV-mediated mtDNA uptake by target cells remain elusive. Microvesicles, ubiquitous in various bodily fluids, play pivotal roles in intercellular communication. Upon detachment from parent cells, they facilitate long-distance molecular transport (Turola et al., 2012). Notably, mitochondria-containing microvesicles can function as danger signal transducers. Studies have confirmed that mesenchymal stem cells and astrocytes secrete large microvesicles containing functional mitochondria, which can be recognized by epithelial cells, immune cells, and neurons (Hayakawa et al., 2016; Sinha et al., 2016). P. Sansone et al. further demonstrated that EVs can mediate complete mtDNA transfer between cells, potentially altering the endogenous mitochondrial repertoire of recipient cells (Sansone et al., 2017). Exosome biogenesis and secretion are cell type- and purpose-dependent. For instance, maximal exosome release coincides with apoptotic processes. Microvesicles express surface receptors enabling target cell recognition and subsequent membrane retention, endocytosis, or lysosomal degradation (van Niel et al., 2018). Platelets represent the most abundant and complex extracellular mitochondrial carriers in circulation. These anucleate, discoid structures-derived from bone marrow megakaryocytes-typically harbor approximately 4 mitochondria each. With nearly one trillion platelets present in human blood, their production, activation, and clearance are intimately linked to mitochondrial dynamics and interactions with other platelet organelles. These findings collectively establish that cf-mtDNA release occurs through multiple regulated pathways, with vesicle-mediated transport representing a sophisticated intercellular communication system that influences both physiological homeostasis and pathological processes, while simultaneously presenting novel opportunities for biomarker development and therapeutic.

During cell death, mtDNA can be actively released into the cytoplasm through selective mitochondrial membrane permeabilization. Concurrently, studies have demonstrated (Land, 2012) that mtDNA may also undergo passive release into the extracellular environment during cell death. As reported (Patrushev et al., 2004), oxidative stress-induced mitochondrial damage leads to mtDNA fragment escape into the cytoplasm through opening of the mitochondrial permeability transition pore (mPTP)-a process that can be pharmacologically inhibited by cyclosporine A. When released into extracellular spaces, these mtDNA fragments function as damage-associated molecular patterns (DAMPs). Notably, human lymphocytes have been shown to actively secrete mtDNA into extracellular compartments (Ingelsson et al., 2018), subsequently triggering inflammatory responses. This observation confirms the existence of specific regulatory mechanisms governing mitochondrial genome release. White and Kile (2015) demonstrated that viral infections can induce mtDNA stress, leading to cytoplasmic mtDNA release through pathways distinct from apoptotic mtDNA extrusion. However, the precise regulatory mechanisms controlling mtDNA translocation from cytoplasm to extracellular space remain unclear (Nakahira et al., 2015). Current evidence identifies cellular stress as the primary driver of mtDNA release. Caielli et al. (2016) proposed an alternative release mechanism requiring fusion between mitochondrial and plasma membranes, though this hypothesis awaits experimental validation. Below we systematically review potential cellular origins of mtDNA as documented in current literature.

The evolutionary origins and structural properties of mtDNA underpin its critical role in sterile inflammation and disease pathogenesis. cf-mtDNA has garnered significant attention as a DAMPs in cancer, trauma, and other pathological conditions. Evolutionarily derived from bacterial DNA, mtDNA shares structural similarities including double-membrane association, circular genome architecture, autonomous replication, and unmethylated CpG motifs. These conserved features enable extracellular mtDNA to mimic pathogen-associated molecular patterns (PAMPs) under cellular stress or injury conditions, functioning as potent DAMPs that are recognized by pattern recognition receptors (PRRs) of the immune system to activate innate immunity and inflammatory responses.

Notably, PAMPs and DAMPs are frequently detected by overlapping receptor systems, explaining the mechanistic parallels between sterile inflammation and pathogen-induced inflammation. Collins et al. (Collins et al., 2004) first demonstrated the immunogenicity of mtDNA in 2004, showing that murine splenocyte exposure to mtDNA triggered tumor necrosis factor (TNF) secretion, while joint injection induced arthritis. Subsequent studies (Zhang et al., 2010b; Gan et al., 2015; Tsuji et al., 2016) have consistently confirmed that exogenous mtDNA administration elicits both localized and systemic inflammatory responses. Specifically, mtDNA DAMPs stimulate neutrophils to activate immune cells, promoting proinflammatory cytokine release and propagating injury to distant organs (Simmons et al., 2013). The inflammatory cascades are primarily mediated through three key pathways: i, TLR9 recognizes hypomethylated CpG motifs in mtDNA, engaging MYD88 to activate MAPK and NF-κB pathways (enhanced by IRF7) (Itagaki et al., 2015); ii, NLRP3 inflammasome assembly through mtDNA binding; and iii, cGAS-STING-IRF3 signaling axis activation that amplifies interferon-stimulated gene (ISG) expression. Our preliminary data corroborate that cf-mtDNA engages cell surface TLR9 to activate NF-κB/MAPK pathways, driving both inflammation and apoptosis (Liu et al., 2020). Furthermore, mtDNA directly interacts with activated leukocytes to modulate antimicrobial responses (Zhang et al., 2010a). While mtDNA DAMPs clearly function in intercellular signaling, their generation and release mechanisms remain enigmatic. Emerging evidence (Kuck et al., 2015) reveals an association between mtDNA and mitochondrial TFAM, with TFAM acting as a TLR9 co-factor to amplify TNF-α release (Julian et al., 2013). In conclusion, cf-mtDNA represents a unique evolutionary relic that bridges mitochondrial dysfunction to systemic inflammation through multiple parallel immune activation pathways, offering novel therapeutic targets for DAMPs-mediated pathologies.

cf-mtDNA has gained increasing attention as a clinically informative biomarker due to its unique biological properties, broad distribution in body fluids, and close association with mitochondrial dysfunction, inflammation, and cellular stress. Beyond plasma, cf-mtDNA has been detected in pleural effusions (Sriram et al., 2012), cerebrospinal fluid (CSF) (Liimatainen et al., 2013), synovial fluid (Leon et al., 1981), and other extracellular compartments, underscoring its versatility for minimally invasive disease monitoring. Compared with nDNA, mtDNA displays enhanced resistance to nuclease-mediated degradation (Manuelidis, 2011), enabling its stable persistence in extracellular environments and facilitating reliable detection.

Accumulating clinical evidence demonstrates disease-specific alterations in cf-mtDNA abundance and structure. In neurological disorders, CSF cf-mtDNA levels are markedly reduced in Alzheimer’s disease (Podlesni et al., 2013), Parkinson’s disease (Pyle et al., 2015), and advanced multiple sclerosis (Lowes et al., 2019), with the degree of reduction correlating with neuronal dysfunction and disease progression (Gambardella et al., 2019). In cardiovascular disease, elevated plasma cf-mtDNA levels have been consistently observed in coronary artery disease (Liu et al., 2016) and acute myocardial infarction (Bliksoen et al., 2012), reflecting tissue injury and inflammatory activation. In oncology, cf-mtDNA alterations show diagnostic and prognostic relevance across multiple malignancies, including ovarian cancer (Meng et al., 2019), head and neck cancer (Jiang et al., 2005), and lung cancer (Hosgood et al., 2010). Additional associations have been reported in trauma (Lam et al., 2004), infection and inflammation (Arshad et al., 2018), toxic or oncogenic exposures (Budnik et al., 2013), and aging (Pinti et al., 2014), positioning cf-mtDNA as a dynamic, multi-system biomarker.

Notably, cf-mtDNA signatures may vary by disease context. In breast cancer, conflicting reports describe both increased (Mahmoud et al., 2015) and decreased cf-mtDNA levels, likely reflecting differences in disease stage, analytical methods, and preanalytical handling. Recent evidence indicates that cf-mtDNA is enriched within EVs, particularly exosomes, compared with freely circulating DNA (85), highlighting vesicle-mediated mtDNA transport as a biologically relevant component of cf-mtDNA signaling. However, the inability to reliably distinguish free from vesicle-encapsulated mtDNA during isolation remains a major source of variability, emphasizing the need for standardized workflows.

A central technical limitation in cf-mtDNA analysis is interference from nuclear mitochondrial pseudogenes (NUMTs). Extensive integration of mtDNA-derived sequences into the nuclear genome—up to 612 loci with high sequence homology—has been documented (Chiu et al., 2001), with some NUMTs approaching full-length mtDNA (50). Without NUMTs-aware assay design, PCR-and sequencing-based approaches may generate false-positive signals. Bioinformatic identification of mtDNA regions with minimal nuclear homology has enabled the development of mtDNA-specific primers (e.g., hmito3, hmito5) alongside nuclear controls (hB2M1, hB2M2) (Malik et al., 2011). Complementary strategies, including Φ29 polymerase–mediated rolling circle amplification, targeted mtDNA sequencing, and pre-PCR template dilution, further mitigate NUMTs contamination (Calvignac et al., 2011; Li et al., 2012; Wolff, 2014). Importantly, the reported mtDNA-specific primers (e.g., hmito3, hmito5, ND1/ND2-based assays) were designed with short amplicon lengths (typically <100 bp), specifically accommodating the highly fragmented nature of cf-mtDNA. This design strategy minimizes NUMTs interference while preserving efficient amplification of short cf-mtDNA fragments, thereby avoiding systematic underestimation of mitochondrial targets in cell-free samples.

Preanalytical variability represents an additional critical determinant of cf-mtDNA measurement accuracy. Blood processing protocols substantially influence cf-mtDNA quantification (Chiu et al., 2003; Ng et al., 2002). Single-step centrifugation yields artificially elevated mtDNA levels due to residual cellular debris, whereas double centrifugation more effectively produces cell-free plasma (Baysa et al., 2019). Moreover, cf-mtDNA exists in both free and vesicle-associated forms, with filtration or ultracentrifugation markedly reducing detectable mtDNA levels (Wolff, 2014). These findings underscore the necessity of standardized centrifugation and vesicle-handling procedures. Structural differences between mitochondrial and nuclear genomes further introduce “dilution bias” in mtDNA/nDNA ratio-based quantification (Malik et al., 2009). Template fragmentation by ultrasonication effectively eliminates this bias, enabling consistent and accurate mtDNA copy number assessment across dilution conditions (Chiu et al., 2001).

Collectively, cf-mtDNA is supported by strong biological plausibility and growing clinical evidence, but its translation into routine diagnostics depends on rigorous standardization of preanalytical, analytical, and computational methodologies.

In summary, circulating double-stranded mtDNA molecules present in bodily fluids have demonstrated remarkable diagnostic and prognostic value since their discovery. Current evidence indicates these molecules originate primarily through cellular oxidative stress, mitochondrial dysfunction, and autophagic secretion. However, widespread clinical adoption necessitates rigorous standardization of preanalytical variables-particularly sample handling and cf-DNA extraction methods to ensure reproducibility across laboratories and enable meaningful interstudy comparisons.