Mitochondrial echoes in the bloodstream: decoding ccf-mtDNA for the early detection and prognosis of hepatocellular carcinoma.

Introduction
Mitochondrial DNA (mtDNA), a circular genome distinct from nuclear DNA, was first fully sequenced in 1981, revealing 37 genes that encode 12 S and 16 S rRNAs, 22 tRNAs, and 13 protein-coding subunits essential for oxidative phosphorylation (OXPHOS) [1]. These include cytochrome c oxidase subunits I-III (MT-CO1, MT-CO2, MT-CO3), ATP synthase subunits 6 and 8 (MT-ATP6, MT-ATP8), cytochrome b (MT-CYB), and NADH dehydrogenase subunits 1–6 and 4 L (MT-ND1-ND6, MT-ND4L) [1]. Mitochondria play a central role in cellular metabolism, regulating key processes such as the tricarboxylic acid (TCA) cycle, fatty acid β-oxidation, amino acid metabolism, and nucleotide biosynthesis [2]. While adenosine triphosphate (ATP) can be generated via glycolysis or OXPHOS, cancer cells preferentially adopt aerobic glycolysis (the Warburg effect) to sustain rapid proliferation [3, 4]. However, growing evidence suggests that OXPHOS may be upregulated in specific cancer subtypes, underscoring its therapeutic potential [5–8].
Beyond energy metabolism, mitochondrial homeostasis is critical in tumorigenesis, influencing reactive oxygen species (ROS) production, apoptosis, and immune regulation [9]. Mitophagy, a selective form of autophagy, removes dysfunctional mitochondria to maintain cellular integrity and metabolic balance [10]. Moreover, mitochondria participate in immune responses and inflammation through metabolic reprogramming and the release of mitochondrial damage-associated molecular patterns (DAMPs) [11]. Given their multifaceted roles, mitochondrial dysfunction has been implicated in various pathologies, including cancer, neurodegenerative diseases, aging, and metabolic disorders [12]. Despite these insights, the regulatory mechanisms governing mitochondrial function in hepatocellular carcinoma (HCC), the most prevalent primary liver malignancy characterized by high recurrence and mortality rates, remain incompletely understood, thereby hindering the development of effective mitochondria-targeted therapies [13].
Recent studies highlight the presence of circulating cell-free mitochondrial DNA (ccf-mtDNA) in the bloodstream, either as free-floating fragments, within intact mitochondria, or encapsulated in extracellular vesicles (EVs) [14–16]. The clinical potential of ccf-mtDNA as a minimally invasive biomarker for HCC detection and prognosis is increasingly recognized. However, despite emerging evidence linking ccf-mtDNA to cancer progression, a comprehensive review focusing on its application in HCC remains lacking. This review aims to consolidate current knowledge on the origin, biological relevance, and clinical utility of ccf-mtDNA as a biomarker for HCC, while critically evaluating its advantages and limitations in the context of cancer diagnosis and prognosis.

Mitochondrial dynamics and regulation of mitochondrial homeostasis and quality
Mitochondria are essential organelles that regulate key metabolic processes, including the TCA cycle, OXPHOS, fatty acid β-oxidation, amino acid metabolism, and nucleotide biosynthesis [2]. Among these, OXPHOS and fatty acid β-oxidation serve as primary sources of mitochondrial reactive oxygen species (mtROS), which can induce DNA damage [17]. Notably, mtDNA is particularly susceptible to ROS-mediated damage due to its proximity to mtROS production sites, lack of protective histones and introns, inefficient proofreading mechanisms, and limited repair capacity. These vulnerabilities contribute to mitochondrial dysfunction and genomic instability [18]. To mitigate these risks, mammalian cells employ quality control mechanisms to preserve mitochondrial function and maintain cellular homeostasis (Fig. 1).

Mitochondrial homeostasis is maintained through a highly dynamic network of processes collectively termed mitochondrial dynamics. These processes encompass reshaping, rebuilding, and recycling events that regulate mitochondrial stability, abundance, distribution, and integrity, enabling cells to adapt to fluctuating metabolic demands and stress conditions [19].
Reshaping mechanisms, including mitochondrial trafficking, fusion, fission, and cristae remodeling, primarily regulate the morphology and function of individual mitochondria without altering overall mitochondrial mass. These reversible modifications allow cells to rapidly adjust mitochondrial activity in response to physiological and environmental cues [20].
In contrast, rebuilding and recycling processes directly influence mitochondrial mass and composition. These include mitochondrial biogenesis, which generates new mitochondrial components, and the degradation and turnover of dysfunctional mitochondria through mitochondrial-derived vesicles (MDVs) and mitophagy. Mitophagy selectively eliminates damaged mitochondria, preventing the accumulation of dysfunctional organelles and preserving mitochondrial quality [21–23]. Additionally, programmed cell death contributes to mitochondrial homeostasis by facilitating the clearance of defective mitochondria. Upon cellular damage, mitochondria release their components, which can be recycled by surviving cells to support mitochondrial regeneration and maintain cellular function [24].

Mitophagy-mediated recycling in mitochondrial homeostasis and quality control
Mitophagy is a selective autophagic process that eliminates dysfunctional or superfluous mitochondria to maintain mitochondrial quality and cellular homeostasis. This process is activated by various stimuli, including mitochondrial membrane depolarization, oxidative stress, and mitochondrial permeability transition [25]. One of the best-characterized mechanisms is the PINK1-Parkin pathway. Upon mitochondrial damage, PINK1 accumulates on the outer mitochondrial membrane (OMM) due to impaired import across the inner membrane and recruits the E3 ubiquitin ligase Parkin. Parkin subsequently amplifies mitophagic signaling by ubiquitinating numerous OMM proteins, such as VDAC1, MFN1/2, and MIRO1/2 [26–30]. These ubiquitin chains are recognized by autophagy adaptor proteins including p62, NBR1, NDP52, OPTN, and TAX1BP1, which facilitate the recruitment of LC3-positive phagophores [31, 32].
In parallel, Parkin-independent mitophagy is mediated by several receptors located on the OMM or inner mitochondrial membrane (IMM), such as BNIP3, NIX, FUNDC1, PHB2, and MCL-1. These receptors contain LC3-interacting regions (LIRs) that allow direct engagement with ATG8 family proteins [33–42]. Their activity is modulated by cellular stress, for instance, FUNDC1 is dephosphorylated under hypoxia to enhance LC3 binding, while PHB2 becomes accessible upon OMM rupture.
Subsequent phagophore formation involves membrane precursor delivery by the ULK1 complex and ATG9 vesicle trafficking, followed by LC3 lipidation via ubiquitin-like conjugation systems, anchoring LC3-II to expanding autophagosomal membranes [23, 32, 43, 44]. GTPase-activating proteins (e.g., TBC1D15/17) regulate Rab GTPases to coordinate membrane dynamics at mitochondria-phagophore contact sites [45]. Finally, mature mitophagosomes fuse with lysosomes to degrade engulfed mitochondria and recycle their contents, completing the mitophagic process [46, 47].
This multilayered system ensures the turnover of defective mitochondria, prevents excessive ROS accumulation, and maintains mitochondrial integrity. Despite its biological importance, monitoring mitophagy in vivo remains technically challenging, requiring the development of robust molecular markers and imaging tools to support mechanistic investigation and therapeutic targeting [48].

Mitochondrial homeostasis and quality control through fusion- and fission-mediated reshaping
Mitochondrial morphology is dynamically regulated by a balance between fusion and fission events, which in turn influence mitochondrial distribution, function, and quality. These processes are tightly coordinated with mitochondrial motility and cytoskeletal interactions.
Long-distance mitochondrial transport along microtubules is mediated by kinesin and dynein motors, linked to the OMM via TRAK1/2 adaptors and the Rho GTPases MIRO1/2 [49–51]. These interactions enable precise positioning of mitochondria in response to cellular energy demands. At short range, actin filaments and myosin motors guide fine-scale mitochondrial movement and facilitate docking at sites of high ATP consumption, such as synapses [52, 53]. In neurons, syntaphilin immobilizes axonal mitochondria by binding to microtubules [54]. Importantly, mitochondrial trafficking is regulated by cytosolic signals including Ca²⁺, redox status, and metabolic cues [55, 56].
Mitochondrial fusion allows exchange of mtDNA, proteins, and metabolites between individual organelles, thereby mitigating damage and restoring functional capacity. Fusion is initiated by MFN1 and MFN2, which tether and merge OMMs. OPA1, localized to the IMM, then mediates inner membrane fusion and cristae remodeling [57–60]. OPA1 also exists in long and short forms, regulated by proteolytic cleavage, to fine-tune mitochondrial dynamics and bioenergetic efficiency.
Conversely, mitochondrial fission segregates dysfunctional or depolarized fragments, facilitating their clearance via mitophagy. The cytosolic GTPase DRP1 is recruited to the OMM by adaptor proteins such as MFF, MID49, and MID51. DRP1 oligomerizes into ring-like structures that constrict the mitochondrion, aided by actin filaments and ER-mitochondria contact sites [61, 62]. The final scission is mediated by DNM2, which pinches off the constricted membrane [63, 64]. Notably, ER tubules and actin polymerization mark future fission sites before DRP1 recruitment, suggesting a hierarchical model of preconstriction and fission [65].
The dynamic reshaping of mitochondria through fusion and fission is not merely structural but also tightly linked to mitochondrial function. Fusion promotes OXPHOS by maintaining membrane potential and diluting mtDNA mutations [66–69]. Fission, meanwhile, is essential for mitochondrial quality control, inheritance during mitosis, and the isolation of damaged segments for degradation [70–72]. Dysregulation of these processes leads to mitochondrial fragmentation or hyperfusion, which have been implicated in neurodegeneration, cancer, and metabolic diseases.
Together, mitophagy and mitochondrial dynamics form an integrated system for maintaining mitochondrial fitness. Their precise orchestration ensures organelle turnover, metabolic adaptability, and protection against cellular stress.

Origins of ccf-mtDNA
Mitochondrial quality control is vital for maintaining cellular homeostasis. When these mechanisms fail or mitochondria sustain irreversible damage, mtDNA can escape from the mitochondrial matrix and enter the extracellular space, forming ccf-mtDNA. In circulation, ccf-mtDNA exists in various forms and is transported by different carriers, including intact circulating mitochondria and vesicular structures such as EVs and MDVs [73]. It may appear as circular DNA, fragmented sequences, or DNA bound to proteins [74].
Emerging evidence suggests that a substantial portion of ccf-mtDNA is encapsulated within EVs (Fig. 2), which shield it from degradation by circulating nucleases and attenuate recognition by immune receptors [75, 76]. Additionally, ccf-mtDNA can be released through neutrophil extracellular traps (NETs), further contributing to its presence in the bloodstream [77, 78].

Free-form ccf-mtDNA
The existence and abundance of free-form ccf-mtDNA, whether as intact circular molecules or fragmented DNA, remain under debate. However, most studies agree that the predominant sources of ccf-mtDNA are intact mitochondria or mtDNA (full-length or fragmented) encapsulated in EVs or MDVs. Under physiological conditions, processes such as apoptosis, which mediate controlled cell turnover, rarely lead to the extracellular release of mtDNA due to efficient sequestration of cellular contents into membrane-bound apoptotic bodies cleared by phagocytosis. This process prevents direct mtDNA leakage and suppresses immune activation [79, 80].
Nonetheless, growing evidence supports the presence of free-form ccf-mtDNA and its potential role in various diseases and cancers [81, 82]. Despite these insights, the mechanisms governing the generation and extracellular transport of free-form ccf-mtDNA remain poorly understood and merit further investigation.

MDVs and mitochondrial-derived EVs (mitoEVs)
MDVs represent a major pathway for ccf-mtDNA entry into circulation [83]. Within cells, MDVs may interact with organelles such as lysosomes and peroxisomes or undergo further processing for release via EV pathways [22]. These vesicles selectively incorporate diverse cargo, including proteins, lipids, and mtDNA, and participate in mitochondrial quality control, immunomodulation, energy transfer, and intracellular transport [83].
MDVs originate through budding from the outer or inner mitochondrial membrane (or both), producing single- or double-membrane vesicles released into the cytosol in response to various stimuli [84]. The biogenesis of specific MDV subtypes, such as TOMM20-positive vesicles, is regulated by proteins including MIROs and DRP1, as well as mitochondrial membrane lipid composition [85]. Stressors such as nutrient deprivation, toxins, oxidative stress, lysosomal dysfunction, infection, and inflammation activate pathways that reshape mitochondrial dynamics and promote MDV formation [86–88]. In HCC, chronic hypoxia and tumor-associated inflammation, hallmarks of the tumor microenvironment, have been shown to induce mitochondrial stress, depolarization, and vesicle formation, thereby enhancing the release of mtDNA-containing MDVs into circulation and contributing to the distinct fragmentomic profiles observed in HCC patients [89–92]. Additionally, viral proteins such as HBx and the HCV core protein can directly interfere with mitochondrial dynamics and biogenesis, exacerbating mtDNA instability and vesicle-mediated release [93–95]. These HCC-specific fragmentomic signatures may provide mechanistic insight into the observed variability in ccf-mtDNA size and composition across patients. For instance, hypoxia-induced changes in mitochondrial membrane permeability may alter the mode and content of mtDNA release, resulting in shorter or oxidized mtDNA fragments that could serve as potential HCC-specific biomarkers [96, 97]. Moreover, HCC-associated metabolic reprogramming, such as enhanced glycolysis and suppressed OXPHOS activity, can increase mitochondrial stress and impair mitophagy, thereby affecting mtDNA fragmentation and export [3, 5, 66, 67]. Tumor burden may further influence these profiles, as spatial gradients of hypoxia, necrosis, and nutrient deprivation within tumors can drive regional differences in mitochondrial damage and vesicle-mediated mtDNA release [89–92]. Together, these mechanisms underscore how viral, metabolic, and microenvironmental factors in HCC shape the distinctive ccf-mtDNA landscape and reinforce its potential as a disease-specific biomarker.
MDVs are heterogeneous, with subtypes defined by cell type and function, including roles in mitochondrial fission, peroxisome biogenesis, oxidative stress resistance, infection resistance, and innate immune signaling [88, 98–100]. While MDV formation occurs at basal levels under homeostasis, it increases markedly under pathological stress [86]. Formation requires Rab9 and SNX9, although the precise regulatory mechanisms remain unclear. Rab9, localized to the trans-Golgi network and late endosomes, regulates endo-lysosomal trafficking, whereas SNX9 interacts with clathrin to modulate endocytosis [101, 102].
MDVs play a pivotal role in mitochondrial antigen presentation by transferring mitochondrial antigens to MHC I molecules at the ER, enabling surface presentation [103, 104]. This pathway influences T cell and macrophage responses [105, 106]. MDVs also contribute to mitochondrial component clearance via EVs, preventing the release of oxidized DAMPs that trigger inflammation [107].
Single-membrane MDVs containing MAPL and TOMM20 primarily target peroxisomes and EV secretion pathways [85, 108–110], while double-membrane MDVs encapsulate outer and inner membrane contents, including PDH, and regulate iron homeostasis, damaged protein disposal, and OXPHOS complex maintenance [99, 111, 112]. These MDVs can also package mtDNA, contributing to immune activation in pathological conditions such as FH deficiency [113, 114].
Single-membrane MDVs bearing MAPL are directed to peroxisomes, while those with TOMM20 are released as exosomes via multivesicular bodies (MVBs) [115]. Similarly, double-membrane MDVs can also be secreted as exosomes [116], and microvesicles have been implicated in MDV release [117]. Recently, a novel subpopulation of mitoEVs, double-membraned vesicles containing multiple mitochondrial proteins, has been identified using an optimized approach, distinguishing them from known EVs such as microvesicles and exosomes [118, 119].
Collectively, these findings indicate that mtDNA can be enclosed within vesicular structures and released into circulation via EV-mediated mechanisms.

Intact and functional circulating mitochondria
Beyond mtDNA fragments and MDV/mitoEV-associated cargo, recent studies suggest that intact mitochondria can also be released into circulation, often as a byproduct of cell death. During this process, extracellular mitochondria may display structural abnormalities such as swelling, fragmentation, or cristae loss. These mitochondria may be vesicle-encapsulated or membrane-free, especially under conditions such as lysosomal dysfunction or necroptosis [120–122].
Intact mitochondria have been detected in large EVs across multiple contexts, including human and animal studies showing their presence in endothelial cells [120, 123, 124], myeloid-derived regulatory cells interacting with T cells [125], and platelet-derived EVs [126]. In vitro, whole mitochondria have also been observed in the supernatants of cultured cells [127]. These extracellular mitochondria may act as DAMPs, triggering inflammation in recipient macrophages and dendritic cells [128]. Interestingly, EVs from lysosome-impaired cells that retain structurally intact mitochondria do not elicit immune activation, indicating potential for mitochondria-based therapies [121].
Intact and functional mitochondria have also been found in large vesicles called “blebbisomes,” which act as mobile communication hubs up to 20 μm in diameter [129]. Additionally, mitoEVs have been shown to carry intact mitochondria [118], supporting the notion that mitochondria in circulation are not limited to fragmented mtDNA.
Despite these findings, the mechanisms by which mitochondria maintain integrity outside the cell and their relative abundance remain unclear. Notably, it is proposed that upon uptake by recipient cells, extracellular mitochondria may regain functional activity [130]. Some hypotheses suggest that circulating mitochondria may serve physiological roles independent of respiration.
Together, these studies demonstrate that intact and potentially functional mitochondria can enter circulation via EV-mediated pathways or as free organelles.

Neutrophil extracellular traps (NETs)
NETs are fibrous extracellular structures composed of a DNA scaffold interlaced with histones and neutrophil-derived enzymes such as NE and MPO. NET formation (NETosis) is a hallmark of neutrophil activity and occurs via multiple death-associated pathways triggered by stimuli such as tumor cytokines, infections, and drugs [131]. These triggers activate stress responses that lead to the extrusion of nuclear or mtDNA along with proteases, forming NETs.
Three NET formation mechanisms have been described: lytic, viable, and mitochondrial NETs [132]. Most studies focus on lytic NETs, while viable and mitochondrial NETs remain underexplored [133]. Notably, mtDNA-containing vesicles released during NETosis contribute to the ccf-mtDNA pool [133].
Despite these advances, many aspects of ccf-mtDNA biology remain unclear, including the characteristics of mtDNA enclosed in MDVs, mitoEVs, intact mitochondria (free or vesicle-associated), and NET-derived vesicles. The functional roles of these distinct forms and their cellular interactions are poorly defined, partly due to the lack of methods for isolating specific ccf-mtDNA subtypes. Furthermore, how different forms of ccf-mtDNA affect transport to recipient cells and their post-uptake processing remains unresolved.
Clarifying the origin, distribution, and fate of these ccf-mtDNA forms is essential for understanding their roles in physiology and disease.

Impact of mtDNA leakage on the immune response
Mitochondrial quality control mechanisms are vital for preserving cellular homeostasis. When these mechanisms fail or mitochondria undergo irreversible damage, mtDNA can escape from the mitochondrial matrix and serve as a DAMP, initiating an inflammatory response. Owing to their bacterial ancestry, mitochondria retain a compact 16.5-kilobase circular genome enriched with CpG motifs, rendering mtDNA highly immunogenic [113, 134].
Escaped mtDNA, structurally akin to bacterial DNA, has been detected in several forms, including nucleoids, circular double-stranded DNA, linear single-stranded DNA, and fragmented species. These variants are particularly abundant during apoptosis or upon mtDNA release via MDVs and the mitochondrial permeability transition pore, a process potentially involving VDAC [135–139]. Once released, mtDNA can act as a DAMP both intracellularly and extracellularly, activating major innate immune pathways such as TLR9, the NLRP3 inflammasome, the AIM2 inflammasome, and the cGAS-STING axis [90, 140–142]. The specific pathway engaged depends on the localization of mtDNA, within lysosomes, the cytosol, or the extracellular milieu, as well as the responding cell type, whether immune or non-immune. These signaling events lead to the release of inflammatory cytokines, recruitment of immune cells, and persistent inflammation, ultimately modulating disease progression [142] (Fig. 3).

In addition to mtDNA, mitochondria carry other immunogenic DAMPs, including cardiolipin, TFAM, N-formyl peptides, ATP, and cytochrome c, all of which can activate immune responses upon release [143]. Structurally, mitochondria consist of an outer phospholipid membrane populated by nuclear-encoded proteins and an inner membrane enriched with cardiolipin, the site of OXPHOS. The inner membrane also encases mtDNA, shielding it from immune surveillance. Notably, mtDNA is tightly associated with TFAM, which organizes it into nucleoids, further influencing its immunogenicity [144].
As noted earlier, extracellular mtDNA has been detected in various forms, including free mtDNA, intact functional mitochondria, and mtDNA enclosed within EVs (such as MDVs and mitoEVs), all of which can activate innate immune responses [142, 145]. Intriguingly, under conditions of lysosomal dysfunction, intact mitochondria can be released via large EVs and taken up by macrophages without triggering inflammation [121]. This phenomenon has been proposed as an alternative mitochondrial quality control mechanism, challenging the prevailing view that mitochondria- or mtDNA-containing EVs are uniformly pro-inflammatory [121, 146]. Moreover, intact mitochondria encapsulated within EVs have been shown to fuse with the mitochondrial networks of recipient cells, thereby enhancing respiratory function [147].
These seemingly divergent outcomes, where some mtDNA- or mitochondria-containing EVs induce inflammation while others do not, may stem from differences in EV subtypes, biogenetic pathways, surface markers, cargo contents, or uptake mechanisms (e.g., endocytosis versus ligand-receptor interactions) [22]. Further studies are warranted to clarify the molecular underpinnings of these differential responses and their implications for immune modulation. These divergent immunological responses underscore the complexity of ccf-mtDNA biology, highlighting how variations in structural form and packaging can differentially influence immune activation or tolerance.
While the downstream immune consequences of mtDNA leakage have been increasingly characterized, the upstream processes governing how mtDNA is fragmented and exported into circulation remain incompletely understood. To fully harness the potential of ccf-mtDNA as a disease-specific biomarker in HCC, it is essential to clarify the biological mechanisms that shape its fragmentation patterns and structural features. The following section addresses these mechanisms, with a particular focus on how HCC-specific factors influence the generation of distinctive ccf-mtDNA profiles.

Mechanisms governing ccf-mtDNA fragmentation and HCC specific profiles
Recent advances in ccf-mtDNA profiling have revealed distinctive fragmentomic signatures in HCC. However, the biological underpinnings of these patterns remain incompletely defined. Understanding the molecular basis for ccf-mtDNA fragmentation is essential to contextualize its biomarker potential and to distinguish HCC-specific profiles from nonmalignant or inflammatory processes.
Fragmentation of mtDNA can arise from both passive and active processes. Passive release typically occurs during cell death pathways, such as apoptosis, necrosis, and pyroptosis, where mitochondrial integrity is compromised, and mtDNA is exposed to cytoplasmic and extracellular nucleases. In contrast, active release mechanisms include vesicle-mediated export via MDVs, exosomes, and NETs, which selectively incorporate and traffic mitochondrial components, including fragmented or oxidized mtDNA, into the extracellular space [83, 88, 132].
One major contributor to mtDNA fragmentation is oxidative stress, particularly in tumor settings such as HCC. Mitochondrial DNA lacks protective histones and is in close proximity to the electron transport chain, making it highly susceptible to damage by ROS. Oxidatively damaged mtDNA is prone to cleavage by nucleases such as ENDOG, FEN1, and DNAse I, generating shorter, often oxidized fragments that can persist in circulation [67, 91, 134]. Fragmentation is further promoted by disruptions in mitophagy, particularly when PINK1/Parkin or receptor-mediated pathways (e.g., BNIP3, NIX, FUNDC1) are impaired, leading to the accumulation and release of damaged mitochondrial material [24, 29, 31].
In HCC, several tumor-specific factors intensify these processes. First, chronic hypoxia and inflammation, hallmarks of the HCC tumor microenvironment (TME), promote mitochondrial depolarization and increase mitochondrial membrane permeability, facilitating mtDNA extrusion via vesicles or pores such as the mitochondrial permeability transition pore (mPTP) and VDAC channels [89–92, 135]. Second, viral proteins, particularly HBx from HBV and the core protein of HCV, have been shown to impair mitochondrial dynamics and induce mitophagy defects, further contributing to mtDNA instability and vesicle-mediated release [93–95]. Third, metabolic reprogramming, characterized by enhanced glycolysis and reduced OXPHOS, increases mitochondrial stress and reduces mitochondrial turnover capacity, facilitating the accumulation of fragmented mtDNA [3, 5, 66].
These biological perturbations result in HCC-specific fragmentomic profiles, typically enriched in short (< 150 bp) and oxidized mtDNA species, with altered nucleotide composition and specific breakpoints. High-throughput sequencing has revealed that these fragments display non-random fragmentation patterns, possibly reflecting preferential cleavage at G-rich or CpG-rich regions, mtDNA D-loop instability, or transcription-replication conflicts at the origin of replication [91, 148]. Importantly, such patterns differ from those observed in benign liver conditions, suggesting their potential use for disease discrimination and therapeutic monitoring.
Additionally, the nature of mtDNA packaging influences fragmentation profiles. mtDNA encapsulated in MDVs or EVs may be selectively processed or protected, whereas free mtDNA released during necrosis is subject to rapid degradation and oxidation, creating a heterogeneous size distribution. Furthermore, spatial heterogeneity in the tumor, such as localized hypoxia, necrosis, or differential immune infiltration, may produce region-specific mtDNA damage, contributing to patient-to-patient variation in fragmentomic signals [90, 92, 149].
In sum, the generation of ccf-mtDNA in HCC reflects a convergence of mitochondrial stress, vesicle trafficking, metabolic derangement, and viral oncogenesis. These factors not only influence the abundance and stability of ccf-mtDNA but also sculpt disease-specific fragmentation patterns that may offer mechanistic insight and diagnostic or prognostic utility. A deeper understanding of these processes will be instrumental in developing robust fragmentomic algorithms for clinical application.

The potential of ccf-mtDNA as biomarkers for HCC
MtDNA damage and somatic mutations within tumor tissues have emerged as critical hallmarks of carcinogenesis and tumor progression, including in HCC [150]. However, the invasive nature and limited feasibility of tissue sampling hinder its widespread use in routine clinical practice. As a result, ccf-mtDNA has attracted growing interest as a minimally invasive biomarker for liquid biopsy (Fig. 4), offering promise for HCC detection and prognosis across diverse etiologies [91, 148, 151].

ccf-mtDNA as a diagnostic biomarker for HCC
Initial investigations focused on identifying mtDNA mutations in tumor tissues as potential tumor-specific markers. A landmark 2002 study reported that mutations in the D-loop region were clonally expanded and detectable in circulating blood in 13 of 19 HCC patients, suggesting the potential utility of ccf-mtDNA for tumor detection [152]. However, methodological limitations, including small sample size and low-resolution detection techniques, precluded conclusive diagnostic correlations.
Subsequent advances in next-generation sequencing (NGS) have enabled detailed characterization of circulating DNA. A 2015 study revealed the presence of aberrantly short and long DNA fragments, including ccf-mtDNA, in the plasma of HCC patients. Notably, the shorter fragments preferentially harbored tumor-associated copy number alterations, with elevated ccf-mtDNA levels observed in HCC compared to controls, indicating their potential as molecular diagnostics [153]. Contrastingly, later studies reported reduced ccf-mtDNA content in hepatitis C virus (HCV)-associated HCC, with levels correlating inversely with tumor burden and predictive of HCC risk in HCV-cirrhotic patients [154]. Similar trends were observed in hepatitis B virus (HBV)-related HCC, where ccf-mtDNA content was significantly lower in HCC patients than in non-HCC controls, supporting its potential as a risk stratification marker in HBV-infected populations [155].
More recent evidence further supports the diagnostic relevance of ccf-mtDNA. A 2021 study using capture-based NGS identified tumor-specific mtDNA mutations in the plasma of HCC patients, but not in colorectal cancer (CRC) patients, highlighting the cancer-type specificity of ccf-mtDNA signals [156]. In 2022, another study confirmed significantly reduced plasma ccf-mtDNA levels in HCC patients compared to healthy individuals [157].
Moreover, increased ccf-mtDNA levels have been reported in other malignancies, including cholangiocarcinoma, CRC, pancreatic, and prostate cancers [14]. However, limited representation of HCC cases in such studies restricts their applicability. Notably, the emergence of “fragmentomic” profiling, focusing on the size, end motifs, and distribution patterns of ccf-mtDNA fragments, has shown promise for classifying cancer tissue-of-origin, including in non-small cell lung cancer (NSCLC), HCC, CRC, serous ovarian cancer (SOC), breast cancer (BC), and clear cell renal cell carcinoma (ccRCC) [92].
Building upon these fragmentomic insights, it is important to consider how ccf-mtDNA compares with the more extensively studied cell-free nuclear DNA (cf-nDNA). Unlike cf-nDNA, ccf-mtDNA is characterized by a higher copy number per cell, cancer-type-specific mutation profiles, and unique fragmentation patterns that may confer additional diagnostic and prognostic value [14, 91]. Emerging evidence suggests that in certain contexts, including hepatocellular carcinoma, ccf-mtDNA-based fragmentomic analyses may outperform cf-nDNA in tissue-of-origin classification [14]. Therefore, an integrative approach that combines ccf-mtDNA with cf-nDNA profiling holds potential to enhance the sensitivity, specificity, and clinical utility of liquid biopsy-based strategies for early cancer detection and outcome prediction [91].
Despite growing support for its diagnostic utility, several challenges remain. Reports of both elevated and diminished ccf-mtDNA levels in HCC underscore the need for standardization and validation in large, etiologically stratified cohorts. Additionally, while tumor-specific mtDNA mutations reflect clonal evolution, their sensitivity and specificity in distinguishing HCC from benign liver diseases require further investigation. Although NGS has improved analytical precision, its cost and accessibility limit current clinical deployment. Furthermore, while fragmentomic approaches offer improved resolution, the clinical interpretability and predictive accuracy of these features remain to be fully validated. Resolving these issues will be pivotal for the reliable integration of ccf-mtDNA into HCC diagnostic workflows.
In addition to biological insights, the clinical translation of ccf-mtDNA as a biomarker depends heavily on methodological standardization and technical optimization. Multiple detection platforms have been employed, including quantitative PCR (qPCR), digital PCR (dPCR), and NGS, each differing in their sensitivity, dynamic range, and ability to capture quantitative and qualitative features such as mutation burden and fragment length profiles [14, 91, 148, 151]. However, substantial variability in sample handling, plasma processing, DNA extraction protocols, and library preparation methods can significantly affect ccf-mtDNA yield and quality, complicating comparisons across studies [14, 91]. Another technical challenge lies in differentiating authentic mitochondrial sequences from nuclear mitochondrial DNA segments (NUMTs), which can contaminate results and introduce artifacts, particularly in NGS-based workflows [14, 92].
Moreover, while recent advances have enabled fragmentomic profiling of ccf-mtDNA, examining size distributions, end motifs, and sequence context, there remains a lack of consensus on the optimal analytical pipelines and thresholds for clinical interpretation. The influence of preanalytical variables, such as blood collection tubes, centrifugation protocols, and storage conditions, also remains poorly characterized. Importantly, validation of ccf-mtDNA biomarkers across large, etiologically stratified HCC cohorts is still limited, and most studies have focused on early discovery rather than robust clinical performance metrics. Establishing reference ranges, quality control standards, and reproducible analytical frameworks will be essential for regulatory approval and widespread adoption in clinical practice. Addressing these challenges is critical for realizing the full diagnostic and prognostic potential of ccf-mtDNA in HCC.

ccf-mtDNA as prognostic biomarker for HCC
Beyond diagnosis, ccf-mtDNA is gaining recognition as a prognostic biomarker for HCC. Early studies linked dynamic fluctuations in ccf-mtDNA levels to liver graft injury. One such study demonstrated that early postoperative increases in ccf-mtDNA were associated with liver injury following transplantation, while delayed surges in patients with early allograft dysfunction occurred independently of traditional liver enzymes, suggesting a distinct mechanistic role for mtDNA in graft surveillance [158].
Emerging data have expanded this utility to non-transplant HCC contexts. In 2021, a novel approach employing ccf-mtDNA profiling was proposed for prognostication and therapeutic stratification in intermediate-stage HCC patients receiving transarterial chemoembolization (TACE). Elevated ccf-mtDNA levels were correlated with worse overall survival, particularly in HBV-related HCC patients treated with TACE in combination with traditional Chinese medicine. Integrating mtDNA mutation burden and fragmentomic characteristics into a composite prognostic scoring system revealed a robust association with unfavorable overall and progression-free survival [159, 160]. A subsequent validation study confirmed these findings in a larger cohort, underscoring the clinical relevance of aberrant ccf-mtDNA dynamics and fragment patterns in predicting HCC prognosis and treatment response [149].
While these findings are promising, further studies are needed to establish the clinical robustness and reproducibility of ccf-mtDNA as a prognostic tool. Many existing studies are limited by small sample sizes, retrospective design, and heterogeneous treatment backgrounds, which undermine generalizability. Additionally, the definitions of prognostic endpoints, such as overall survival, time to recurrence, and progression-free survival, often vary between studies, complicating direct comparisons.
To address these limitations, prospective longitudinal studies incorporating serial ccf-mtDNA measurements are essential to determine whether changes in mtDNA levels or fragment profiles reliably reflect tumor burden, therapeutic response, or relapse risk in real time. Standardizing the timing of sample collection in relation to key clinical interventions (e.g., before and after TACE, systemic therapy, or surgical resection) will be critical for establishing consistent and clinically meaningful monitoring frameworks [91, 148]. Furthermore, integrating ccf-mtDNA profiles with established clinical parameters, such as tumor stage, alpha-fetoprotein (AFP) levels, or cf-nDNA signatures, may enhance prognostic accuracy by supporting multi-parametric risk stratification strategies.
Equally important is the development of standardized protocols for ccf-mtDNA quantification, fragmentomic characterization, and bioinformatic analysis. Current methodological variability, including differences in blood collection, DNA extraction, sequencing, and data interpretation, poses a significant barrier to reproducibility and cross-study validation. The lack of consensus on detection platforms, quality control metrics, and analytical thresholds hinders clinical translation and regulatory acceptance. Addressing these technical and procedural challenges will be pivotal for the reliable implementation of ccf-mtDNA as a prognostic and predictive biomarker in HCC.
Together, these findings highlight the potential of ccf-mtDNA not only as a diagnostic tool but also as a dynamic, noninvasive biomarker for prognostication and therapeutic monitoring (Table 1). Future research should prioritize prospective validation, harmonization of analytical workflows, and refinement of fragmentomic algorithms to support the clinical application of ccf-mtDNA-based assays in HCC management.

Considerations for ccf-mtDNA detection
The utility of ccf-mtDNA as a diagnostic or prognostic biomarker in HCC is intimately tied to the reliability and reproducibility of its detection. Currently available platforms, such as qPCR, dPCR, and NGS, differ markedly in their sensitivity, cost-effectiveness, and clinical applicability. qPCR remains the most accessible and cost-efficient method but may lack sufficient sensitivity to detect low-abundance or highly fragmented mtDNA. dPCR offers superior precision and absolute quantification capabilities, making it advantageous for detecting subtle changes in mtDNA copy number or mutation load. However, its higher cost and lower throughput may limit widespread clinical adoption. NGS enables detailed fragmentomic and mutational profiling, allowing for insights into tumor-specific mtDNA features, but is technically complex, costly, and sensitive to contamination by NUMTs, which can confound interpretation [14, 91, 148]. To aid interpretation and methodological planning, Table 2 summarizes the key technical and biological considerations for ccf-mtDNA detection in HCC.

Beyond platform selection, pre-analytical factors can significantly affect ccf-mtDNA yield and integrity. These include blood collection tube type, time to plasma separation, centrifugation speed, and storage conditions. For example, prolonged delays before plasma isolation may lead to leukocyte lysis, releasing intracellular DNA and artificially inflating ccf-DNA measurements. Inadequate standardization of DNA extraction protocols may also contribute to batch effects or cross-study variability. Moreover, biological variables such as underlying liver inflammation, fibrosis, and hepatocyte turnover, common in HCC patients with chronic liver disease, may independently elevate ccf-mtDNA levels, thereby obscuring tumor-specific signals. These confounding factors underscore the importance of carefully controlled sample processing and appropriate patient stratification when interpreting ccf-mtDNA results [14, 91, 92].
Future clinical implementation of ccf-mtDNA biomarkers will require harmonized workflows, from sample collection through data analysis. Establishing reference standards, optimizing detection sensitivity, and minimizing pre-analytical variability will be essential steps toward achieving clinical-grade performance. Only through such standardization can ccf-mtDNA achieve its full potential as a reliable and scalable tool in HCC management.

Conclusion and future perspectives
With rising efforts to reduce HCC-related mortality, recent research has increasingly focused on the identification of novel biomarkers. Among these, somatic alterations in mtDNA have emerged as promising candidates for both diagnostic and prognostic applications. Nevertheless, the limitations of conventional diagnostic tools and the intrinsic heterogeneity of HCC pathogenesis underscore the need for minimally or noninvasive biomarkers that can be assessed from easily accessible clinical specimens, enabling real-time, dynamic disease monitoring. In this context, circulating cell-free DNA (ccf-DNA), particularly ccf-mtDNA, has gained attention for its potential as a liquid biopsy. Compared to nuclear DNA, mtDNA possesses distinct advantages, its smaller genome, circular structure, and higher copy number per cell facilitate more efficient detection in circulation, particularly in plasma and serum where total DNA yields are often low.
Over the past decade, ccf-mtDNA has emerged as a noninvasive and clinically actionable biomarker for HCC. Its diagnostic value is supported by evidence demonstrating tumor-specific mtDNA mutations, quantitative alterations, and fragmentomic features that may aid early detection. In parallel, its prognostic utility has been underscored by associations between ccf-mtDNA profiles and patient outcomes, including overall survival and treatment response, particularly in patients undergoing TACE. Notably, the relevance of ccf-mtDNA extends across various HCC etiologies, including HBV- and HCV-associated HCC, enhancing its potential for broad clinical application (Table 3).

Despite these promising attributes, several key challenges must be addressed prior to routine clinical implementation. First, inconsistent findings across studies, some reporting elevated ccf-mtDNA levels in HCC patients, while others indicate a reduction, highlight the need for larger, well-controlled cohorts to clarify these discrepancies. Second, the lack of standardized methodologies for DNA extraction, quantification, and sequencing limits cross-study comparability and reproducibility. Third, the influence of non-tumor-related factors, such as systemic inflammation, liver injury, or comorbid conditions, may confound ccf-mtDNA measurements, reducing diagnostic specificity. Furthermore, the absence of universally defined cutoff values for distinguishing physiological from pathological ccf-mtDNA levels complicates clinical interpretation.
In addition to technical limitations, ambiguities in clinical study design remain problematic. Many existing studies are hindered by small sample sizes, inconsistent definitions of disease stage, and heterogeneous treatment regimens. These factors may introduce bias and obscure the true clinical relevance of ccf-mtDNA. For instance, therapeutic interventions, including surgical resection, systemic therapy, or locoregional treatment, can alter ccf-mtDNA levels, necessitating appropriate stratification and longitudinal assessment. Similarly, inconsistent staging criteria across studies may mask stage-dependent variations in ccf-mtDNA dynamics.
Emerging research has begun to explore EVs containing mtDNA, such as MDVs and mitoEVs, as alternative liquid biopsy targets. However, the clinical significance of mtDNA encapsulated within these EVs remains largely unexplored. Moreover, it is yet unclear whether extracellular mtDNA reflects mitophagy or mitochondrial turnover in tumor tissues, processes that may have divergent roles in tumorigenesis and therapy resistance.
In conclusion, while ccf-mtDNA represents a compelling biomarker candidate for HCC diagnosis and prognosis, its clinical translation remains in its early stages. Future research should prioritize the standardization of detection platforms, validation in large and diverse patient cohorts, and determination of clinically meaningful thresholds. Integrating ccf-mtDNA assessments with clinical variables such as tumor stage, etiology, and treatment modality will be crucial for refining its utility as a reliable, noninvasive biomarker for precision HCC management.