The Role of Mitochondrial Dysfunction in Hepatocellular Carcinoma: From Pathogenesis and Drug Resistance to Targeted Therapeutic Strategies Authors: Siyu Zhao, Qian Liu, Jing Li, Shilei Zhang, Shuo Wan, Yuan Yang, Feng Jin, Biguang Tuo, Hai Jin Published: 2026 ABSTRACT AbstractHepatocellular carcinoma (HCC) is characterized by high malignancy and widespread drug resistance at advanced stages, posing formidable challenges in clinical diagnosis and treatment. Mitochondrial dysfunction is a core driver of HCC initiation and drug resistance. However, how its plasticity dynamically regulates tumor growth under various conditions remains to be systematically elucidated. This review systematically delineates the critical roles of mitochondrial dysfunction in HCC pathogenesis, drug resistance, and targeted therapy. Regarding pathogenesis, we integrate four major pathways, namely aberrant oxidative phosphorylation (OXPHOS) coupled with mitochondrial DNA (mtDNA) mutations, glutamine metabolic reprogramming, oxidative stress imbalance, and dysregulated ferroptosis, to elaborate the molecular basis driving HCC malignant progression. In terms of drug resistance mechanisms, mitochondrial plasticity is highlighted as a central adaptive hub that synergistically promotes resistance through multifaceted mechanisms, including remodeling of the tumor immune microenvironment, dynamic regulation of OXPHOS levels, and modulation of ferroptosis sensitivity. With respect to therapeutic strategies, we summarize the current major intervention directions, encompassing ROS/OXPHOS modulators, inhibitors targeting mitochondrial metabolic enzymes and transporters, ferroptosis inducers, as well as nanodelivery systems and combination therapies. Of note, the efficacy of these interventions often depends on the mitochondrial phenotypic state of the tumor; single-target approaches tend to yield limited long-term benefits, whereas combination regimens or dynamic adjustments based on mitochondrial plasticity status hold greater promise for clinical application. Collectively, this review consolidates the evidence positioning mitochondrial dysfunction as a convergence point linking HCC progression, drug resistance, and targeted therapy, and proposes that “targeting mitochondrial plasticity” may represent a novel direction for overcoming drug resistance and achieving precision therapy, thereby providing a theoretical reference for future basic research and clinical translation. Introduction Hepatocellular carcinoma (HCC) ranks as the sixth most common malignancy worldwide and the third leading cause of cancer-related mortality.1 Despite continuous advances in diagnostic and therapeutic modalities, the overall 5-year survival rate of HCC patients remains only approximately 18%, which is largely attributable to its complex pathogenesis, difficulties in early diagnosis, and the near-universal occurrence of acquired resistance in advanced-stage treatment.2,3 Mitochondria are central organelles that govern oxidative phosphorylation (OXPHOS), redox homeostasis, apoptosis, and biosynthesis.4 The liver is exceptionally rich in mitochondria, which occupy approximately 13–20% of normal hepatic volume. Mitochondrial dysfunction has been firmly established to be closely associated with the development and progression of not only HCC but also various chronic liver diseases, including hepatitis B, alcoholic liver disease, and non-alcoholic fatty liver disease.5,6 During the malignant progression of HCC, multiple mitochondrial-related mechanisms have been progressively elucidated. In the context of oxidative stress, aberrant expression of PRDX6 promotes HCC proliferation and metastasis through the PI3K/Akt/mTOR and Notch1 pathways, and its expression level correlates significantly with poor prognosis.7 With regard to novel cell death pathways, cuproptosis—a copper-dependent form of cell death associated with mitochondrial dysfunction—features downregulation of its core regulator FDX1 in HCC, which is also linked to unfavorable prognosis.8 At the level of drug resistance, tyrosine kinase inhibitors (TKIs), represented by sorafenib, achieve a clinical response rate of only approximately 30%, and even initially responsive patients frequently develop acquired resistance within six months.2,9 Multiple forms of mitochondrial dysfunction, including metabolic reprogramming, mitochondrial DNA (mtDNA) damage, mitochondrial dynamics disorders, reactive oxygen species (ROS) overload, and dysregulated ferroptosis, have been demonstrated to collectively participate in the development of acquired resistance.10–12 In response to the aforementioned mechanisms, various intervention strategies have entered the research and development pipeline, including small-molecule compounds that modulate ROS/OXPHOS, inhibitors targeting mitochondrial metabolic enzymes and transporters, and inducers of ferroptosis and cuproptosis. Meanwhile, novel candidate targets such as ESR1 and PRDX6 have also been proposed as potential nodes for HCC diagnosis and therapy.7,13 Furthermore, immunotherapy has shown superior efficacy over sorafenib in a subset of HCC patients, suggesting that targeting mitochondrial pathways may provide a synergistic basis for combinatorial immunotherapeutic strategies.14 Although several reviews have addressed the relationship between mitochondrial dysfunction and HCC, most have focused on single mechanisms and lack a systematic integration across the three major domains of pathogenesis, resistance evolution, and targeted therapy. This fragmentation, to some extent, limits the efficiency of clinical translation and the precise identification of responsive patient subsets. In this review, we systematically consolidate the core roles and therapeutic advances of mitochondrial dysfunction from the above three perspectives, aiming to provide a theoretical basis for novel strategies to overcome drug resistance and achieve precision therapy in HCC. Structure and Function of Mitochondria Mitochondria are double-membrane organelles.15 The outer mitochondrial membrane (OMM) and inner mitochondrial membrane (IMM) divide the mitochondrion into two functional compartments: the matrix and the intermembrane space (IMS).15 The OMM contains porins, such as voltage-dependent anion channels (VDACs), rendering it highly permeable and allowing the free diffusion of ions and small molecules.16 In contrast, the IMM is a structurally tight membrane that relies on highly specific membrane transporters for selective transport.16 This selectivity facilitates the formation of an electrochemical gradient, which is fundamental for ATP synthesis.15 The cristae, formed by invaginations of the IMM, serve as the primary sites of mitochondrial energy conversion, with ATP synthase and the electron transport chain (ETC) densely distributed on their inner surface16 (Figure 1). Figure 1Schematic diagram of mitochondrial structure. Mitochondria are double-membrane organelles, consisting of the OMM containing VDACs, the IMM, the IMS, and the matrix. The IMM harbors ATP synthase and the ETC, which are involved in energy metabolism, while the matrix contains mtDNA and ribosomes.Diagram of mitochondrial structure with labeled components.Abbreviations: VDAC, Voltage-Dependent Anion Channel; OMM, Outer Mitochondrial Membrane; IMM, Inner Mitochondrial Membrane; IMS, Intermembrane Space; ATP, Adenosine Triphosphate; ETC, Electron Transport Chain; mtDNA, Mitochondrial DNA. Schematic diagram of mitochondrial structure. Mitochondria are double-membrane organelles, consisting of the OMM containing VDACs, the IMM, the IMS, and the matrix. The IMM harbors ATP synthase and the ETC, which are involved in energy metabolism, while the matrix contains mtDNA and ribosomes. The primary function of mitochondria is to provide energy for cells through the OXPHOS system located in the IMM.17 The ETC, composed of Complexes I–V, utilizes the energy generated by electron transfer to pump protons (H⁺) out of the mitochondrial matrix, thereby establishing a transmembrane proton gradient that drives ATP synthesis18 (Figure 2). Figure 2Schematic diagram of the structure and function of the TCA cycle and ETC. Acetyl-CoA, generated from glucose via glycolysis, enters the TCA cycle for oxidative decomposition. The NADH and FADH2 produced by dehydrogenation reactions in the TCA cycle transfer electrons to Complex I and Complex II, respectively. Complex I couples electron transfer to CoQ with the pumping of protons (H⁺) from the mitochondrial matrix across the IMM into the IMS, whereas Complex II mediates electron transfer only and does not possess proton-translocating function. Subsequently, reduced CoQ is oxidized at Complex III, with protons being released into the IMS. Electrons are then transferred via Cyt c to Complex IV, where oxygen (O2) is reduced to water (H2O), accompanied by transmembrane proton transport. The transmembrane proton electrochemical gradient (Δψm) established by the above electron transfer process ultimately drives Complex V (ATP synthase) to synthesize ATP; the energy released by protons flowing back into the matrix down their concentration gradient is used for the phosphorylation of ADP to ATP.The diagram illustrates the TCA cycle and electron transport chain within a mitochondrion. Glucose is converted to Acetyl-CoA, entering the TCA cycle in the matrix. The cycle includes citrate, isocitrate, alpha-ketoglutarate, succinyl CoA, succinate, fumarase, malate and oxaloacetate. NADH and FADH subscript 2 are produced, transferring electrons to Complex I and II. Complex I pumps protons from the matrix to the intermembrane space. Electrons move via CoQ to Complex III, then via Cyt c to Complex IV, where oxygen is reduced to water. Protons are pumped across the membrane, creating a gradient. Complex V (ATP synthase) uses this gradient to synthesize ATP from ADP and inorganic phosphate. The diagram shows the inner mitochondrial membrane (IMM), intermembrane space (IMS) and outer mitochondrial membrane (OMM).Diagram of TCA cycle and ETC showing electron transfer, proton pumping and ATP synthesis in mitochondria.Abbreviations: TCA Cycle, Tricarboxylic Acid Cycle; ETC, Electron Transport Chain; CoQ, Coenzyme Q; NADH, Nicotinamide Adenine Dinucleotide; FADH2, Flavin Adenine Dinucleotide; IMM, Inner Mitochondrial Membrane; IMS, Intermembrane Space; Cyt c, Cytochrome c; ATP, Adenosine Triphosphate; ADP, Adenosine Diphosphate. Schematic diagram of the structure and function of the TCA cycle and ETC. Acetyl-CoA, generated from glucose via glycolysis, enters the TCA cycle for oxidative decomposition. The NADH and FADH2 produced by dehydrogenation reactions in the TCA cycle transfer electrons to Complex I and Complex II, respectively. Complex I couples electron transfer to CoQ with the pumping of protons (H⁺) from the mitochondrial matrix across the IMM into the IMS, whereas Complex II mediates electron transfer only and does not possess proton-translocating function. Subsequently, reduced CoQ is oxidized at Complex III, with protons being released into the IMS. Electrons are then transferred via Cyt c to Complex IV, where oxygen (O2) is reduced to water (H2O), accompanied by transmembrane proton transport. The transmembrane proton electrochemical gradient (Δψm) established by the above electron transfer process ultimately drives Complex V (ATP synthase) to synthesize ATP; the energy released by protons flowing back into the matrix down their concentration gradient is used for the phosphorylation of ADP to ATP. Mitochondria possess their own independent genetic material, mtDNA, which encodes components related to the inner membrane respiratory chain complexes.19 The mitochondrially encoded cytochrome b is a core component of Complex III, participating in electron transfer and assisting in the establishment of the proton gradient.20 Therefore, the expression level of mtDNA directly influences cellular OXPHOS efficiency. Moreover, mtDNA mutations can affect processes such as redox balance and apoptosis, and are closely associated with tumorigenesis.4 In hepatocytes, over 90% of mitochondrial proteins are encoded by nuclear DNA, with mtDNA encoding only the remaining mitochondrial proteins.20 In HCC, rapidly proliferating tumor cells exhibit increased oxygen consumption, leading to the formation of a local hypoxic microenvironment.21 Under hypoxic conditions, ETC function is impaired, and aberrant electron flow promotes the massive accumulation of ROS, which directly damages mtDNA.10 Furthermore, the hypoxic microenvironment can enhance the migratory capacity of cancer cells, further aggravating the malignant progression of HCC.22 In summary, understanding the structure and function of mitochondria under normal conditions is fundamental to elucidating how mitochondrial dysfunction drives the malignant progression of HCC. The following sections will systematically discuss the specific molecular mechanisms by which mitochondrial dysfunction drives HCC, focusing on OXPHOS dysregulation, metabolic reprogramming, oxidative stress, and ferroptosis. Molecular Mechanisms of Mitochondrial Dysfunction Driving HCC OXPHOS Dysregulation and mtDNA Mutations Metabolic reprogramming is a hallmark of malignancy that not only drives tumor cell proliferation but also mediates drug resistance.23 Mitochondrial metabolism plays a critical regulatory role in tumor pathophysiology and is essential for cancer cell survival, metastasis, and drug resistance.24 Even under normoxic conditions, tumor cells preferentially utilize the less efficient aerobic glycolysis (the Warburg effect) and upregulate glucose transporters (such as GLUT1) to meet their energy demands.25 Importantly, OXPHOS function is not completely suppressed but undergoes adaptive reprogramming. In the hypoxic microenvironment of HCC cells, impaired ETC function leads to electron leakage and massive accumulation of ROS. Excessive ROS directly damage mtDNA and proteins, and can also trigger the mitochondrial unfolded protein response (UPRmt), thereby accelerating HCC malignant progression.10 Thus, dysregulation of the OXPHOS system and its associated mtDNA damage may represent one of the pathological bases of HCC.26 OXPHOS homeostasis is finely regulated by multiple key molecules. Under hypoxic conditions, OXPHOS can be adaptively upregulated to maintain basal mitochondrial function, enabling HCC cells to withstand hypoxic stress.27 SNX17, a regulator of OXPHOS and mitochondrial homeostasis, drives OXPHOS and mitochondrial gene expression through activation of the STAT3/c-Myc pathway, thereby promoting metabolic reprogramming and malignant progression in HCC.28,29 CRIF1, a mitochondrial protein involved in the assembly of OXPHOS complexes, exerts oncogenic effects by regulating the synthesis and assembly of OXPHOS system polypeptides and activating the ROS/NF-κB pathway.30 In HBV-associated HCC, upregulation of SUV39H1 promotes tumor proliferation and migration by enhancing the expression of OXPHOS-related genes.31 The long non-coding RNA MALAT1 induces mitochondrial dysfunction and promotes HCC progression by suppressing OXPHOS function and ATP synthesis.32 Furthermore, resveratrol promotes mitophagy through the MALAT1/miR-143-3p/RRM2 pathway and inhibits HCC progression.33 Similarly, c-Src kinase promotes HCC malignant progression by impairing the function of mitochondrial OXPHOS Complexes I/IV, leading to defects in energy metabolism.34 mtDNA encodes 13 key protein subunits of the ETC. Therefore, the integrity and normal function of mtDNA are decisive factors for maintaining mitochondrial OXPHOS efficiency.35,36 Lacking histone protection, mtDNA is more susceptible to ROS-induced damage, resulting in the accumulation of mutations and aberrant copy numbers, which in turn promote tumorigenesis and progression.37–39 Extensive mtDNA damage exacerbates oxidative stress levels and drives liver disease progression by disrupting respiratory chain function and energy metabolic homeostasis.39 For example, instability in the D-loop region, which regulates mtDNA transcription and replication, leads to reduced mtDNA copy number and decreased OXPHOS levels, thereby causing aberrant bioenergetics in tumor cells.40,41 Multiple oncogenic signals drive HCC malignant progression by interfering with mtDNA stability. XRCC2, a driver gene in HCC, is responsible for clearing damaged mtDNA and maintaining the transcription of mitochondrially encoded Complex I-related genes.42 Conversely, endoplasmic reticulum stress suppresses XRCC2 expression via XBP1, thereby inhibiting tumor proliferation.42 This study thus established the therapeutic potential of endoplasmic reticulum stress-inducing agents (such as tunicamycin) targeting XRCC2 in HCC.42 It should be noted that the functions of OXPHOS-related molecules in HCC are not always consistent. For instance, Rab3A is upregulated in HCC, but unlike most pro-tumorigenic molecules, Rab3A inhibits HCC metastasis by enhancing mitochondrial OXPHOS; however, O-GlcNAcylation of Rab3A attenuates its GTP-binding activity and metastasis-suppressive function.43 This seemingly paradoxical observation suggests that the function of a given molecule in HCC may depend on its post-translational modification status and the relative activity of its downstream effectors. Such context-dependency is not unique to Rab3A in OXPHOS regulation. For example, the activity of mitochondrial Complex I can be suppressed via HIF-1α-dependent pathways under hypoxia, but can also be enhanced by overexpression of certain oncogenes such as MYC, which may explain the discrepant conclusions regarding whether OXPHOS plays pro- or anti-tumorigenic roles in HCC across different studies.44,45 Therefore, future investigations should re-evaluate the net effect of OXPHOS in specific genetic and microenvironmental contexts, rather than simply categorizing it as either “pro-tumorigenic” or “anti-tumorigenic.” Glutamine Metabolic Reprogramming In addition to energy metabolic reprogramming, aberrant glutamine metabolism also plays a significant role in mitochondrial dysfunction in HCC. The liver serves as the central hub for amino acid metabolism, and disturbances in amino acid homeostasis are closely associated with hepatocarcinogenesis.46 Glutamine is essential for cancer cells, not only by fueling the tricarboxylic acid (TCA) cycle to generate ATP but also by supplying reducing equivalents (NADH and FADH2) to the electron transport chain.47 Enhanced glutamine metabolism is especially prevalent under conditions of impaired mitochondrial function, with more than 60% of glutamine being metabolized within mitochondria.46,48 Moreover, increased glutamine metabolism correlates with poor prognosis in HCC patients, highlighting the significant reliance of tumor cells on mitochondrial glutamine metabolism.49 This metabolic dependency is largely governed by the oncogene Myc/c-Myc. Myc regulates the expression of key genes involved in glutamine catabolism, leading to mitochondrial metabolic reprogramming that sustains TCA cycle activity.50 Specifically, SET and MYND domain-containing protein 2 (SMYD2) stabilizes c-Myc protein and upregulates glutaminase 1 (GLS1) expression, thereby promoting HCC progression via the c-Myc/GLS1 signaling axis. Knockdown of SMYD2 overcomes sorafenib resistance in HCC cells.51 Conversely, canagliflozin (CANA) promotes c-Myc degradation by targeting pyruvate kinase M2 (PKM2), thereby suppressing the expression of its downstream target gene GLS1 and ultimately blocking glutamine metabolism.52 Among the various transporters regulating cellular glutamine flux, SLC1A5 (ASCT2) is the sole member of the SLC1 gene family that accepts glutamine as a substrate, and its expression is markedly upregulated in HCC.53,54 A variant of SLC1A5 facilitates glutamine entry into mitochondria, providing energy for cancer cells and maintaining redox homeostasis, thus playing a central role in tumor metabolic reprogramming.55 Clinically, high SLC1A5 expression in HCC tissues is closely associated with enhanced resistance to transarterial chemoembolization (TACE) and shortened overall survival.54,56 Within mitochondria, glutamate dehydrogenase 1 (GDH1/GLUD1) catalyzes the conversion of glutamate to α-ketoglutarate (α-KG), replenishing a key substrate for the TCA cycle.57 Human glutamate dehydrogenase 1 (hGDH1), encoded by the GLUD1 gene, is highly expressed in the liver. Silencing of GLUD1 induces mitochondrial dysfunction and promotes apoptosis in HCC cells.58 Furthermore, oxoglutarate dehydrogenase-like (OGDHL) encodes a core subunit of the oxoglutarate dehydrogenase complex (OGDHC), which catalyzes the conversion of α-KG to succinyl-CoA.59,60 In HCC, OGDHL deficiency elevates the α-KG/citrate ratio through suppression of OGDHC activity, which subsequently drives reductive carboxylation of glutamine-derived α-KG via retrograde TCA cycling in hepatoma cells, thereby promoting HCC progression61 (Figure 3). Figure 3Amino acid metabolism and therapeutic targets in hepatocellular carcinoma. GLUT1 is highly expressed in HCC cells, facilitating glucose uptake, which is then metabolized via glycolysis to generate acetyl-CoA that enters the TCA cycle, providing energy and biosynthetic precursors for the cells. Meanwhile, extracellular Gln is primarily taken up into the cytoplasm via transporters such as SLC38A1, SLC1A5 (ASCT2), and SLC38A2. In the cytoplasm, a portion of Gln is directly converted to Glu by ASNS. Additionally, Gln can enter the mitochondrial matrix via the mitochondrial glutamine transporter (a variant of SLC1A5), where it is catalyzed by mitochondrial GLS1 to produce Glu. This Glu is subsequently converted to α-KG through a series of aminotransferases including GLUD1/GDH1, GPT2, and GOT2, and then enters the TCA cycle to participate in the generation of ATP and NADPH. Furthermore, mitochondrial-derived Glu can be exported to the cytoplasm via the transporters SLC25A18 and SLC25A22, where it is used together with cysteine derived from cystine taken up via SLC7A11 for the synthesis of GSH. In the cytoplasm, Glu can also be condensed with NH⁺ under the catalysis of GS to regenerate Gln, forming an intracellular glutamine metabolic recycling circuit.The diagram illustrates amino acid metabolism and therapeutic targets in hepatocellular carcinoma cells. GLUT1 facilitates glucose uptake, which is metabolized to pyruvate and then acetyl CoA, entering the TCA cycle. The cycle includes oxaloacetate, citrate, isocitrate, alpha-ketoglutarate, succinyl CoA, succinate, fumarase and malate. Extracellular glutamine is taken up via SLC38A1, SLC1A5 and SLC38A2 transporters. In the cytoplasm, glutamine is converted to glutamate by ASNS. Glutamine enters mitochondria via SLC1A5 variant and is converted to glutamate by GLS1. Glutamate is transformed to alpha-ketoglutarate by GDH1/GLUD1, GPT2 and GOT2, entering the TCA cycle. Mitochondrial glutamate is exported via SLC25A18 and SLC25A22, used with cysteine from cystine via SLC7A11 for GSH synthesis. Glutamate condenses with NH4 plus by GS to regenerate glutamine, forming a recycling circuit.HCC cell metabolism: glucose uptake, glycolysis, TCA cycle, glutamine pathways, enzyme labels.Abbreviations: GLUT1, Glucose Transporter 1; TCA Cycle, Tricarboxylic Acid Cycle; Gln, Glutamine; ASNS, Asparagine Synthetase; Glu, Glutamate; GLS1, Glutaminase 1; GLUD1/GDH1, Glutamate Dehydrogenase 1; GPT2, Glutamate Pyruvate Transaminase 2; GOT2, Glutamate–Oxaloacetate Transaminase 2; α-KG, α-Ketoglutarate; GSH, Glutathione; GS, Glutamine Synthetase; NH4⁺, Ammonium. Amino acid metabolism and therapeutic targets in hepatocellular carcinoma. GLUT1 is highly expressed in HCC cells, facilitating glucose uptake, which is then metabolized via glycolysis to generate acetyl-CoA that enters the TCA cycle, providing energy and biosynthetic precursors for the cells. Meanwhile, extracellular Gln is primarily taken up into the cytoplasm via transporters such as SLC38A1, SLC1A5 (ASCT2), and SLC38A2. In the cytoplasm, a portion of Gln is directly converted to Glu by ASNS. Additionally, Gln can enter the mitochondrial matrix via the mitochondrial glutamine transporter (a variant of SLC1A5), where it is catalyzed by mitochondrial GLS1 to produce Glu. This Glu is subsequently converted to α-KG through a series of aminotransferases including GLUD1/GDH1, GPT2, and GOT2, and then enters the TCA cycle to participate in the generation of ATP and NADPH. Furthermore, mitochondrial-derived Glu can be exported to the cytoplasm via the transporters SLC25A18 and SLC25A22, where it is used together with cysteine derived from cystine taken up via SLC7A11 for the synthesis of GSH. In the cytoplasm, Glu can also be condensed with NH⁺ under the catalysis of GS to regenerate Gln, forming an intracellular glutamine metabolic recycling circuit. Notably, glutamine synthetase (GS) exerts dual roles in HCC.62 This functional divergence of GS in HCC may arise from differential metabolic flux of its substrates. On one hand, GS promotes HCC development and progression through the Wnt/β-catenin pathway and is associated with poor prognosis.62 On the other hand, Dai et al demonstrated that specific knockout of hepatocyte GS in a β-catenin-activated HCC mouse model unexpectedly accelerated tumorigenesis, with the underlying mechanism involving ammonia accumulation and increased generation of glutamate-derived non-essential amino acids, which in turn activated the mTORC1 signaling pathway to drive tumor growth.63 Furthermore, clinical studies have shown that GS-positive HCC patients exhibit better survival rates after radiofrequency ablation than GS-negative patients.64 These seemingly contradictory findings suggest that the role of GS in HCC may depend on the molecular subtype and therapeutic context of the tumor, and its utility as a prognostic biomarker or therapeutic target warrants further validation. Oxidative Stress and Activation of Oncogenic Signaling Mitochondrial ETC dysfunction is a major source of ROS generation, with Complexes I and III serving as the primary sites of production.65 Concurrently, oncogenic mutations in key enzymes of the TCA cycle and hypoxic conditions within the tumor microenvironment can also lead to aberrantly elevated mitochondrial ROS levels.66 ROS interact with various intramitochondrial molecules, thereby triggering mitochondrial dysfunction.67 ROS possess dual functions in intracellular signaling and cytotoxicity, with their biological effects being concentration-dependent.68 At physiological concentrations, ROS are maintained in dynamic equilibrium with the intracellular antioxidant system and act as critical signaling molecules involved in regulating various physiological functions.36,69 However, under pathological conditions such as radiation and hypoxia, the balance between ROS production and elimination is disrupted. Abnormally elevated ROS, serving as important retrograde signaling molecules, can drive hepatocarcinogenesis by disrupting cellular homeostasis and normal structure and function.36,66,70 Of note, moderately elevated ROS promote cancer cell proliferation, migration, and acquisition of drug resistance through the activation of multiple oncogenic signaling pathways.71 Upregulation of MIC19, a core subunit of the MICOS complex, elevates ROS levels in HCC and promotes HCC growth and metastasis via activation of the ROS/NF-κB pathway.72 Ferredoxin 1 (FDX1), a small mitochondrial protein, is downregulated in HCC, and its decreased expression promotes HCC progression by elevating ROS levels and activating the PI3K/AKT signaling pathway.73 In addition, protein kinase Cδ (PKCδ) drives HCC progression by inducing mitochondrial ROS generation and oxidizing HSP60, thereby relieving the inhibition of the mitogen-activated protein kinase (MAPK) signaling pathway by Raf kinase inhibitory protein (RKIP).74 Furthermore, upregulation of NOX4 following incomplete radiofrequency ablation promotes mitochondrial ROS production and induces mitophagy through activation of the Nrf2/PINK1 pathway, thereby enhancing cell survival.75 Ferroptosis Ferroptosis is characterized by inactivation of glutathione peroxidase 4 (GPX4) and accumulation of lipid peroxidation, accompanied by alterations in mitochondrial morphology and volume.76,77 This process originates from electron leakage at mitochondrial respiratory Complexes I/III, and the resulting superoxide radicals trigger lipid peroxidation and ferroptosis through a series of reactions.78 Mitochondria play a central role in ferroptosis by regulating bioenergetics, biosynthesis, and ROS levels.78 ETC inhibitors or mitochondrial uncouplers can suppress cystine deprivation- or erastin-induced ROS accumulation and ferroptosis, further confirming the critical role of mitochondrial OXPHOS dysfunction in inducing ferroptosis.79 Mitochondria also serve as a central hub for iron metabolism, as iron is involved not only in the synthesis of mitochondrial enzyme cofactors but also in driving HCC proliferation.80,81 The liver, as the primary organ for iron storage, is particularly sensitive to ferroptosis.82 Iron transport to the IMM is mediated by Mfrn1 and its homolog Mfrn2; dysfunction of these transporters leads to mitochondrial iron accumulation and oxidative damage.83,84 Intracellular iron overload generates highly toxic ROS via the Fenton reaction, thereby inducing ferroptosis.77 Notably, ferroptosis-related molecules exhibit opposing regulatory effects in HCC. Certain molecules promote HCC progression by suppressing ferroptosis. The mitochondrial translocator protein TSPO inhibits ferroptosis and promotes HCC progression through the P62/KEAP1/Nrf2 pathway.85 S100P promotes lysosomal degradation of ACC1 via the S100P-RAB5C-ACC1 axis, thereby reducing de novo synthesis of polyunsaturated fatty acids and decreasing lipid peroxidation levels, thus suppressing ferroptosis and promoting HCC progression and therapy resistance.86 TMEM147 suppresses ferroptosis through the STAT2-DHCR7-27HC-GPX4 axis, promoting HCC proliferation and metastasis, while 27HC also remodels tumor-associated macrophage (TAM) metabolism to create a pro-tumorigenic microenvironment. Targeting this axis can restore ferroptosis sensitivity, offering a novel direction for HCC therapy.87 Conversely, other molecules exert anti-tumor effects by inducing ferroptosis. Ponicidin promotes ferroptosis through regulation of the KEAP1/Nrf2 pathway and enhances the therapeutic sensitivity of lenvatinib-resistant HCC cells.88 Ailanthone induces ferroptosis by downregulating GPX4, SLC7A11, and FTH1 expression, accompanied by elevated ROS levels, increased lipid peroxidation, and loss of mitochondrial membrane potential, indicating that it inhibits HCC progression through impairing mitochondrial function.89 Mitochondrial iron homeostasis also participates in sorafenib-induced ferroptosis.81 The inhibitor NL-1 of CISD1, an iron–sulfur cluster protein on the outer mitochondrial membrane, enhances the sensitivity of HCC cells to sorafenib when used in combination, by promoting mitochondrial iron accumulation and lipid peroxidation.81 Furthermore, deficiency of GCN5L1, a mitochondrial acetylation regulator, downregulates CISD1 expression, leading to iron accumulation and accelerated lipid ROS generation, thereby promoting ferroptosis and similarly enhancing the sensitivity of HCC cells to sorafenib81 (Figure 4). Figure 4Ferroptosis in hepatocellular carcinoma and its regulation by iron metabolism. SLC7A11 is a cystine/glutamate antiporter that takes up cystine and exports glutamate. Intracellular cystine is reduced to cysteine for GSH synthesis. GPX4, relying on GSH as a cofactor, reduces PLOOH to inhibit lipid peroxidation and ferroptosis. The accumulation of glutamate resulting from SLC7A11 inactivation can be converted to α-KG, which enhances the TCA cycle, promotes mitochondrial membrane hyperpolarization, and triggers ferroptosis. Transferrin-bound Fe3⁺ is internalized via TFR1, reduced to Fe2⁺ by STEAP3, and released into the cytoplasm through DMT1. Cytoplasmic Fe2⁺ generates ROS via the Fenton reaction, driving lipid peroxidation and ferroptosis. In addition, mitochondria take up Fe2⁺ through Mfrn 1/2 and VDAC2/3 located on the outer mitochondrial membrane; excessive Fe2⁺ forms the LIP and induces mitochondrial ROS production. The iron–sulfur cluster protein CISD1 on the outer mitochondrial membrane regulates mitochondrial iron export and antagonizes ferroptosis. Collectively, these mechanisms constitute a regulatory network of ferroptosis that may play an important role in HCC progression.The diagram outlines the ferroptosis pathway in liver cancer, emphasizing iron metabolism. Iron bound to transferrin enters cells via TFR1, is reduced by STEAP3 and released into the cytoplasm through DMT1. This iron generates reactive oxygen species via the Fenton reaction, causing lipid peroxidation and ferroptosis. Mitochondria absorb iron through Mfrn1/2 and VDAC2/3, forming a labile iron pool that induces mitochondrial reactive oxygen species. CISD1, an iron to sulfur cluster protein, manages mitochondrial iron export. SLC7A11, a cystine/glutamate antiporter, imports cystine, which is converted to cysteine for glutathione synthesis. GPX4 uses glutathione to reduce lipid hydroperoxides, preventing lipid peroxidation and ferroptosis. Excess glutamate can be transformed into alpha-ketoglutarate, boosting the TCA cycle and promoting ferroptosis. These mechanisms drive liver cancer progression.Ferroptosis in liver cancer: iron metabolism, lipid peroxidation and protein regulation.Abbreviations: SLC7A11, Solute Carrier Family 7 Member 11; Glu, Glutamine; GSH, Glutathione; GPX4, Glutathione Peroxidase 4; PLOOH, Phospholipid Hydroperoxide; α-KG, α-Ketoglutarate; TCA Cycle, Tricarboxylic Acid Cycle; TFR1, Transferrin Receptor 1; STEAP3, Six-Transmembrane Epithelial Antigen of Prostate 3; DMT1, Divalent Metal Transporter 1; ROS, Reactive Oxygen Species; VDAC 2/3, Voltage-Dependent Anion Channel 2/3; LIP, Labile Iron Pool; CISD1, CDGSH Iron–Sulfur Domain-Containing Protein 1; HCC, Hepatocellular Carcinoma. Ferroptosis in hepatocellular carcinoma and its regulation by iron metabolism. SLC7A11 is a cystine/glutamate antiporter that takes up cystine and exports glutamate. Intracellular cystine is reduced to cysteine for GSH synthesis. GPX4, relying on GSH as a cofactor, reduces PLOOH to inhibit lipid peroxidation and ferroptosis. The accumulation of glutamate resulting from SLC7A11 inactivation can be converted to α-KG, which enhances the TCA cycle, promotes mitochondrial membrane hyperpolarization, and triggers ferroptosis. Transferrin-bound Fe3⁺ is internalized via TFR1, reduced to Fe2⁺ by STEAP3, and released into the cytoplasm through DMT1. Cytoplasmic Fe2⁺ generates ROS via the Fenton reaction, driving lipid peroxidation and ferroptosis. In addition, mitochondria take up Fe2⁺ through Mfrn 1/2 and VDAC2/3 located on the outer mitochondrial membrane; excessive Fe2⁺ forms the LIP and induces mitochondrial ROS production. The iron–sulfur cluster protein CISD1 on the outer mitochondrial membrane regulates mitochondrial iron export and antagonizes ferroptosis. Collectively, these mechanisms constitute a regulatory network of ferroptosis that may play an important role in HCC progression. In summary, ferroptosis exerts dual regulatory roles in liver diseases.90 ACSL4-mediated ferroptosis does not function as an endogenous tumor-suppressive mechanism during HCC formation in chronically injured livers; instead, it promotes carcinogenesis by exacerbating liver injury and fibrosis.91 However, during the therapeutic phase of HCC, induction of ferroptosis exhibits anti-tumor effects, suggesting that ferroptosis plays diametrically opposing roles in HCC initiation and treatment.91 Specifically, ferroptosis accelerates disease progression in acute and early-stage chronic liver disease by exacerbating hepatocyte injury and inflammation, whereas in advanced fibrosis and HCC, it exerts tumor-suppressive functions by eliminating malignant cells.92 This double-edged sword effect suggests that ferroptosis-targeting strategies should be differentially designed according to the disease stage to achieve precise intervention. Interplay Network Among Pathways of Mitochondrial Dysfunction The four pathways described above do not operate in isolation but are interconnected through positive feedback loops that synergistically drive HCC malignant progression. First, ROS generated by mitochondrial ETC dysfunction can directly damage mtDNA and respiratory chain complexes, further exacerbating OXPHOS defects and ROS production, thereby establishing a self-reinforcing vicious cycle.10 Second, bidirectional regulatory interactions exist between glutamine metabolism and both OXPHOS and ferroptosis. Under conditions of OXPHOS impairment, HCC cells rely on glutamine anaplerosis to replenish the TCA cycle, sustaining energy supply and the generation of reducing equivalents (NADH/FADH2).47,61 Meanwhile, glutamate, an intermediate in this process, serves as a critical precursor for glutathione (GSH) synthesis. Alterations in glutamate metabolic flux directly modulate GPX4 activity, thereby influencing cellular sensitivity to ferroptosis.93 Furthermore, intracellular iron overload catalyzes ROS production via the Fenton reaction, which acts synergistically with ROS derived from OXPHOS to collectively trigger lipid peroxidation and ferroptosis.77,81 These positive feedback loops are intricately intertwined, forming a cascade-amplifying network. Perturbation at any node may be progressively amplified through the aforementioned feedback mechanisms and ultimately propagated to the entire network, cooperatively driving HCC malignant progression and therapeutic resistance. Role and Mechanisms of Mitochondrial Dysfunction in Resistance to Targeted Therapy in HCC Mitochondrial dysfunction is closely associated with tumor drug resistance.94 The underlying mechanisms encompass multiple dimensions, including metabolic reprogramming, remodeling of the immune microenvironment, and suppression of ferroptosis.23,95 Therefore, mitochondrial dysfunction not only constitutes the biological basis for the development of drug resistance in HCC but also represents a critical entry point for overcoming resistance. Sorafenib, a first-line therapeutic agent for advanced HCC, induces tumor cell death by inhibiting the activity of mitochondrial Complexes II/III and Complex V.96,97 However, its clinical response rate is only approximately 30%, and the majority of initially sensitive patients develop acquired resistance within six months.2,9 Mitochondrial dysfunction mediates sorafenib resistance in HCC through multiple mechanisms, including remodeling of the immune microenvironment, regulation of biosynthesis, and suppression of ferroptosis. UBQLN1-mediated degradation of PGC1β inhibits mitochondrial biogenesis and reduces ROS production, thereby contributing to sorafenib resistance.98 Sorafenib-induced mitochondrial dysfunction promotes the release of mtDNA into the tumor microenvironment (TME), which in turn induces M2 polarization of tumor-associated macrophages (TAMs) via TLR9 signaling, leading to an immunosuppressive microenvironment that fosters resistance.12 Aloin has been shown to inhibit M2 macrophage polarization through activation of the NR3C2/MT1M pathway in lung squamous cell carcinoma.99 Given the critical role of M2-polarized TAMs in driving sorafenib resistance in HCC, whether aloin or its analogs can reverse HCC resistance by blocking the mtDNA-TLR9-M2 polarization axis warrants further investigation.99 YAP/TAZ enhance GSH synthesis and suppress ferroptosis by upregulating SLC7A11.100 In addition, inhibition of iron uptake and dysregulation of mitochondrial iron release are also implicated in the resistance process, suggesting that targeting mitochondrial iron homeostasis may represent a potential strategy to overcome resistance.81 In contrast to the multi-pathway network involved in sorafenib resistance, the resistance mechanisms of lenvatinib, cabozantinib, and regorafenib exhibit distinct features, yet all converge on the dynamic regulation of mitochondria. Lenvatinib resistance involves an imbalance in mitochondrial dynamics; tumor matrix stiffness upregulates the expression of mitochondrial fission protein 1 (FIS1) via H3K27me3, driving mitochondrial fission and protective autophagy, thereby enabling tumor cells to evade drug-induced apoptosis.101 On the other hand, ABHD6, following lactylation modification, competitively binds to FIS1 and maintains mitochondrial fusion, thereby blocking drug-induced ROS production and apoptosis.102 Cabozantinib resistance is characterized by c-Myc-mediated protective mitophagy. c-Myc overexpression enhances mitochondrial fragmentation and p62 aggregation, promoting protective mitophagy that selectively removes damaged mitochondria and reduces cytochrome c release, consequently suppressing apoptosis.103 Inhibition of mitochondrial fission, p62 aggregation, or autophagy effectively reverses resistance, providing a theoretical basis for combining cabozantinib with autophagy inhibitors such as chloroquine.103 Distinct from the mitochondrial dynamics abnormalities mediated by the aforementioned TKIs, the core mechanism underlying regorafenib resistance lies in ferroptosis evasion. The resistance-driving gene ZNF207 promotes lactate accumulation, which facilitates lactylation of PRDX1 at K67, enhances NRF2 nuclear translocation, and activates downstream anti-ferroptotic pathways.104 In summary, although the resistance mechanisms of the four TKIs manifest in diverse forms, they all converge on mitochondrial dysfunction as a common core mechanism. Among these, mitophagy repeatedly emerges as a protective mechanism in various TKIs resistance contexts, either by eliminating damaged mitochondria to evade apoptosis or by sustaining tumor cell survival through metabolic reprogramming. Therefore, therapeutic strategies targeting mitochondrial function should be precisely tailored based on the distinct resistance mechanisms. For resistance driven by metabolic reprogramming, interventions should target mitochondrial metabolic enzymes and energy pathways; for resistance rooted in mitochondrial dynamics imbalance, regulation of fission/fusion and autophagy processes is warranted; for resistance characterized by ferroptosis evasion, restoring ferroptosis sensitivity should be prioritized. The following sections will elaborate on the corresponding therapeutic strategies in accordance with these relationships. Mitochondria-Targeted Therapeutic Strategies for HCC Given the heterogeneity in HCC pathogenesis and resistance phenotypes, this section follows a “mechanism–target–strategy” correspondence logic to categorize and elaborate therapeutic approaches and resistance reversal strategies targeting distinct types of mitochondrial dysfunction, while also summarizing emerging mitochondria-targeted interventions. Modulation of Mitochondrial Reactive Oxygen Species Levels and Oxidative Phosphorylation Of note, moderately elevated ROS act as signaling molecules that drive tumorigenesis, whereas markedly elevated ROS exert cytotoxic effects and effectively induce tumor cell apoptosis.105 Based on this property, modulating ROS levels has become an important therapeutic strategy for HCC.106 Specifically, this can be achieved either by scavenging excessive ROS to suppress tumorigenesis or by elevating ROS to toxic levels to selectively kill cancer cells.107 In terms of scavenging excessive ROS, the selective NOX1 inhibitor ML171 significantly reduces mitochondrial ROS production and effectively inhibits HCC cell migration and invasion in vitro.108 With respect to elevating ROS to toxic levels, Mito-lonidamine (Mito-LND) and (E)-Akt inhibitor-IV ((E)-AKTIV) interfere with glycolysis and the TCA cycle in hepatocytes, thereby suppressing ATP synthesis while inducing ROS accumulation, leading to effective inhibition of HCC growth both in vitro and in vivo.109 Ionizing radiation (IR) generated during radiotherapy represents another source of ROS; excessive ROS can damage mitochondrial electron transport chain complexes and react with biomolecules including lipids, proteins, and DNA, inducing oxidative stress that results in tumor cell death.69,110 Therefore, ROS act as a “double-edged sword” in HCC, with their ultimate effects depending on the interplay between concentration levels and cellular context. Mitochondrial OXPHOS inhibitors also effectively kill HCC cells. The clinical agent levo-tetrahydropalmatine (I-THP) promotes autophagy by activating the AMPK-mTOR-ULK1 and ROS-JNK-ATG pathways while concurrently inhibiting ERK/AKT signaling, leading to reduced mitochondrial OXPHOS and mitochondrial damage, thereby effectively eliminating HCC cells.111 In addition, the mitochondrial OXPHOS inhibitor FH535 induces HCC cell apoptosis by disrupting mitochondrial transmembrane potential and the ETC, resulting in decreased ATP production.112 Inhibitors Targeting Mitochondrial Metabolic Enzymes and Transporters Targeting glutamine metabolism has emerged as a promising direction for HCC drug development. GS plays a critical role in glutamine metabolism and therapeutic resistance in HCC. Inhibition of GS activity enhances oxidative stress and induces mitophagy, thereby increasing the radiosensitivity of HCC.113 Clinical analyses have shown that HCC patients with negative GS expression exhibit better responses to sorafenib after hepatectomy, suggesting that GS may serve as a biomarker for predicting therapeutic efficacy and guiding combination therapy.114 GLS1 expression is elevated in sorafenib-resistant HepG2 (HepG2 (R)) cells, and exogenous glutamine deprivation or GLS1 enzymatic inhibitors reduce cell viability and increase apoptosis in HepG2 (R) cells.115 Ginsenoside Rk1 inhibits glutamine metabolism by downregulating GLS1 expression, reducing GSH synthesis and promoting ROS accumulation, while also acting through the ERK/c-Myc signaling pathway to induce HCC cell apoptosis.116 Both BPTES and CB-839 inhibit GLS1 expression and exhibit tumor-specific antiproliferative effects.117 BPTES significantly prolongs survival in Myc-driven HCC mouse models.118 However, while CB-839 inhibits HCC cell growth, it concurrently triggers the amino acid response pathway and upregulates asparagine synthetase (ASNS) expression, which in turn enhances HCC cell proliferation, migration, and invasion. This finding suggests that ASNS may serve as a potential combination target to overcome CB-839 resistance.119 To overcome monotherapy resistance and enhance efficacy, combination therapy has become a prevailing trend. The combination of CB-839 with the transporter inhibitor V-9302 synergistically disrupts the GSH/ROS balance and potentiates anti-HCC effects.49 Furthermore, the development of inhibitors targeting mitochondrial SLC1A5 (ASCT2) variants represents a novel strategy to suppress cancer cell metabolism and reduce tumor growth.120 The compound iMQT-020 selectively inhibits cancer cell growth by disrupting the trimeric assembly of SLC1A5 variants, thereby interfering with glutamine uptake and mitochondrial OXPHOS.55 Therapeutic Strategies for Inducing Ferroptosis Given its dual role in liver diseases, ferroptosis is regarded as a promising precision cancer therapeutic strategy.90 As an iron-dependent form of cell death, ferroptosis effectively inhibits tumor growth and metastasis, making it a therapeutically valuable anticancer target.121 Given the central role of mitochondria in iron metabolism, lipid peroxidation, and ROS generation, targeting mitochondrial function has emerged as a potential approach for inducing ferroptosis. Some agents act directly on mitochondria to disrupt their integrity and trigger ferroptotic signaling. For example, Polyphyllin I (PPI) induces ferroptosis and inhibits HCC progression by damaging mitochondrial structure and function while activating the Nrf2/HO-1/GPX4 axis.122 Depleting GSH or inhibiting GPX4 activity represents another important strategy for inducing ferroptosis. Bisdemethylcurcumin (bm-Cur), a novel GSH scavenger, reduces GSH levels through dual mechanisms involving inhibition of GSH synthesis and direct consumption, thereby targeting mitochondrial lipid peroxidation and damage to induce ferroptosis and suppress tumor progression in HCC.123 Meanwhile, elevated expression of suppressor of cytokine signaling 2 (SOCS2) synergistically promotes ferroptosis by facilitating SLC7A11 ubiquitination and degradation while suppressing GPX4 expression, thereby enhancing radiosensitivity.124 In addition, the steroidal saponin SSPH I induces iron overload, promotes lipid peroxidation, and triggers mitochondrial morphological abnormalities in HCC cells, consequently driving ferroptosis and exerting anticancer effects.125 Emerging Mitochondria-Targeted Therapeutic Strategies As mitochondrial dysfunction represents a core driver of HCC, therapeutic strategies targeting this organelle are continuously evolving, primarily in two aspects: first, uncovering novel mitochondria-targeting mechanisms of existing metabolic drugs; and second, leveraging advanced nanotechnologies to construct intelligent delivery systems for precise modulation of mitochondrial function. Certain drugs originally developed for metabolic diseases have been found to possess mitochondria-targeting potential and inhibit HCC growth due to their ability to influence core metabolic pathways. Beyond its glucose-lowering effects, metformin exerts anticancer activity by inhibiting Complex I activity and reducing ATP production.126,127 At high doses, metformin binds to VDAC1, disrupting the IP3R-GRP75-VDAC1 complex and mitochondria-associated endoplasmic reticulum membranes (MAMs), thereby elevating cytoplasmic calcium levels and activating autophagy-associated cell death, which collectively enhances its anti-HCC efficacy.128 Sodium-glucose cotransporter 2 inhibitors (SGLT2i) not only improve liver function but also delay c-Myc-driven HCC progression through mTOR inhibition.129,130 Among these, canagliflozin reduces tumor burden in liver tissues, decreases oxidative stress, and modulates GSH content, thereby exerting anticancer effects.131 In recent years, the development of mitochondria-targeted nanodelivery systems has provided novel approaches for precise regulation of mitochondrial function.36 The ROS-responsive nanosystem co-delivers YC-1 and regorafenib, reversing drug resistance while increasing cellular sensitivity to the drug.132 The ROS/pH cascade-responsive supramolecular nanoplatform precisely targets tumors and releases payloads, eliciting a self-amplifying oxidative storm through a unique ROS-regenerative effect, thereby achieving efficient killing of HCC cells.133 In addition, an injectable POM-loaded coacervate delivery system effectively suppresses HCC metastasis after incomplete radiofrequency ablation by scavenging ROS.134 The carbon-nitrogen-based DNA and photosensitizer co-delivery system (BPSCNs) enables targeted delivery of the P53 gene and the mitochondria-targeted photosensitizer MTTPY, inducing mitochondrial-derived ferroptosis through the HIF-1α/SLC7A11 pathway and enhancing photodynamic therapy (PDT) efficacy.135 The biomimetic nanoplatform PZ@M-T induces ferroptosis and releases mtDNA, activating the cGAS-STING pathway and remodeling the HCC immune microenvironment, thereby potentiating chemo-immunotherapeutic responses in HCC.136 The continuous exploration of metabolic drug targets and iterative innovation in nanodelivery technologies jointly propel the future development of mitochondria-targeted therapy for HCC.137 To facilitate a systematic understanding of the developmental progress of the aforementioned agents, we have summarized the targets, research stages, HCC-related effects, and toxicity profiles of each drug and its combinations (Table 1). Table 1Summary of Preclinical/Clinical Stages and Toxicity Risks of Mitochondria-Targeted AgentsDrug NameTarget/MechanismResearch StageToxicity/Safety ProfileML171NOX1 inhibitorPreclinicalNo apparent toxicity reported108Mito-LNDGlycolysis/TCA cycle interferencePreclinicalNo apparent toxicity reported109(E)-AKTIVAkt inhibitor/glycolysis interferencePreclinicalNo apparent toxicity reported109I-THPAMPK/mTOR/ROS pathway modulatorPreclinicalNo apparent toxicity reported111FH535OXPHOS inhibitorPreclinicalNo apparent toxicity reported112BPTESGLS1 inhibitorPreclinicalPoor solubility, low oral bioavailability138CB-839GLS1 inhibitorPreclinicalTriggers AAR-mediated upregulation of ASNS, potentially promoting tumor progression119IACS-010759Mitochondrial Complex I inhibitorPhase I clinical (terminated)Dose-limiting neurotoxicity137SSPH ISteroidal saponinPreclinicalNo apparent toxicity reported125MetforminComplex I inhibitorPhase II/III trials ongoingHigh-dose gastrointestinal toxicity, lactic acidosis139CanagliflozinSGLT2 inhibitorPreclinicalNo apparent toxicity reported131MitoQTPP⁺-conjugated ubiquinone derivativePreclinicalNo significant neurotoxicity137Mito-ATOTPP⁺-conjugated arsenic trioxidePreclinicalNo significant neurotoxicity137Mito-HNKTPP⁺-conjugated honokiol derivativePreclinicalNo significant neurotoxicity137Abbreviations: TCA Cycle, Tricarboxylic Acid Cycle; OXPHOS, Oxidative Phosphorylation; GLS1, Glutaminase 1; AAR, Amino Acid Response; ASNS, Asparagine Synthetase; SGLT2i, Sodium–Glucose Cotransporter 2 Inhibitor; TPP⁺, Triphenylphosphonium. Table 1Summary of Preclinical/Clinical Stages and Toxicity Risks of Mitochondria-Targeted AgentsDrug NameTarget/MechanismResearch StageToxicity/Safety ProfileML171NOX1 inhibitorPreclinicalNo apparent toxicity reported108Mito-LNDGlycolysis/TCA cycle interferencePreclinicalNo apparent toxicity reported109(E)-AKTIVAkt inhibitor/glycolysis interferencePreclinicalNo apparent toxicity reported109I-THPAMPK/mTOR/ROS pathway modulatorPreclinicalNo apparent toxicity reported111FH535OXPHOS inhibitorPreclinicalNo apparent toxicity reported112BPTESGLS1 inhibitorPreclinicalPoor solubility, low oral bioavailability138CB-839GLS1 inhibitorPreclinicalTriggers AAR-mediated upregulation of ASNS, potentially promoting tumor progression119IACS-010759Mitochondrial Complex I inhibitorPhase I clinical (terminated)Dose-limiting neurotoxicity137SSPH ISteroidal saponinPreclinicalNo apparent toxicity reported125MetforminComplex I inhibitorPhase II/III trials ongoingHigh-dose gastrointestinal toxicity, lactic acidosis139CanagliflozinSGLT2 inhibitorPreclinicalNo apparent toxicity reported131MitoQTPP⁺-conjugated ubiquinone derivativePreclinicalNo significant neurotoxicity137Mito-ATOTPP⁺-conjugated arsenic trioxidePreclinicalNo significant neurotoxicity137Mito-HNKTPP⁺-conjugated honokiol derivativePreclinicalNo significant neurotoxicity137Abbreviations: TCA Cycle, Tricarboxylic Acid Cycle; OXPHOS, Oxidative Phosphorylation; GLS1, Glutaminase 1; AAR, Amino Acid Response; ASNS, Asparagine Synthetase; SGLT2i, Sodium–Glucose Cotransporter 2 Inhibitor; TPP⁺, Triphenylphosphonium. Summary of Preclinical/Clinical Stages and Toxicity Risks of Mitochondria-Targeted Agents Abbreviations: TCA Cycle, Tricarboxylic Acid Cycle; OXPHOS, Oxidative Phosphorylation; GLS1, Glutaminase 1; AAR, Amino Acid Response; ASNS, Asparagine Synthetase; SGLT2i, Sodium–Glucose Cotransporter 2 Inhibitor; TPP⁺, Triphenylphosphonium. Mitochondrial Dysfunction-Based Diagnostic Biomarkers for HCC The detection of molecular indicators associated with mitochondrial dysfunction is progressively transitioning toward clinical application, providing novel tools for precision diagnosis, prognostic assessment, and therapeutic monitoring in HCC. In terms of treatment sensitivity prediction and prognostic evaluation, researchers have developed multiple predictive models based on mitochondrial-related genes. One study established a rapid prediction platform integrating nanodiamond-mediated OXPHOS biosensors with artificial intelligence (AI), enabling efficient assessment of tumor sensitivity to the OXPHOS inhibitor IACS-010759.140 Liu et al utilized transcriptomic and clinical data from HCC patients in the TCGA and GEO databases, and applied univariate Cox regression analysis and LASSO regression analysis to screen for OXPHOS-related genes, constructing a prognostic model for HCC that provides important references for understanding the role of OXPHOS in HCC and predicting patient survival outcomes.141 Furthermore, Xu et al based on the TCGA-LIHC dataset, employed univariate Cox regression analysis and LASSO regression analysis to identify four glutamine metabolism-associated genes (GMAGs)—RRM1, RRM2, G6PD, and GPX7—and constructed a prognostic prediction model. Kaplan-Meier survival analysis and ROC curves demonstrated that this model possesses reliable predictive capability for HCC patients (p < 0.05), providing evidence-based support for individualized treatment and prognostic assessment in HCC.142 Collectively, these models and platforms suggest that mitochondrial metabolism-related gene signatures hold promise as important tools for identifying patient populations most likely to benefit from mitochondria-targeted therapies. With respect to diagnostic and therapeutic monitoring biomarkers, mitochondrial dysfunction not only leads to mtDNA damage but also promotes the release of mtDNA into the cytoplasm and extracellular environment.143 Among these, circulating cell-free mtDNA (ccf-mtDNA) exhibits significantly superior diagnostic performance for early-stage HCC compared with AFP, and may serve as a risk biomarker for monitoring the efficacy of therapies such as TACE.144,145 In addition, elevated SUV39H1 expression regulates the expression of OXPHOS-related genes, and its combined detection with AFP demonstrates higher diagnostic efficacy than AFP alone in patients with hepatitis B virus-associated HCC, although this finding requires further experimental validation.31 Shi et al screened and validated six mitochondrial-related genes—MTHFD1L, NT5DC2, POLQ, RECQL4, TOMM40L, and TXNRD1—that are upregulated in HCC, suggesting their potential as diagnostic biomarkers for HCC.146 Furthermore, several emerging physical parameters have shown potential application prospects. For example, an in vitro study demonstrated that in metformin-treated HepG2 cells, cells with lower metabolic activity exhibited higher mitochondrial temperature, suggesting that mitochondrial temperature may serve as an adjunctive diagnostic indicator for HCC, although its clinical value requires further validation147 (Table 2). Table 2Summary of Mitochondrial Dysfunction-Related Biomarkers and Platforms in Hepatocellular CarcinomaCategoryBiomarker/PlatformApplicationKey FindingsTreatment sensitivity predictionOXPHOS biosensor combined with AI-based prediction platformPrediction of tumor sensitivity to the OXPHOS inhibitor IACS-010759Enables rapid and efficient drug sensitivity assessment, providing technical support for precision medicine140Prognostic assessmentOXPHOS-related gene signaturePrognostic prediction in HCCFacilitates elucidation of the biological role of OXPHOS in HCC and assists in patient survival prognostication141Prognostic assessmentGMAGs signaturePrognostic prediction and therapeutic guidance in HCCIdentified four core genes: RRM1, RRM2, G6PD, and GPX7142Diagnosisccf mtDNAEarly diagnosis of HCC and monitoring of TACE efficacyDemonstrates superior diagnostic performance for early-stage HCC compared with AFP; serves as a risk biomarker for TACE efficacy144,145DiagnosisUV39H1 joint AFPDiagnosis of HBV-associated HCCCombined detection yields superior diagnostic performance compared with AFP alone31DiagnosisSix mitochondrial-related genes: MTHFD1L, NT5DC2, POLQ, RECQL4, TOMM40L, and TXNRD1Potential diagnostic biomarkers for HCCAll six genes are upregulated in HCC tissues and cell lines, suggesting their potential as diagnostic biomarkers146Emerging directionMitochondrial temperatureAdjunctive diagnostic indicator for HCCIn metformin-treated HepG2 cells, cells with lower metabolic activity exhibited higher mitochondrial temperature147Abbreviations: OXPHOS, Oxidative Phosphorylation; HCC, Hepatocellular Carcinoma; GMAGs, Glutamine Metabolism-Associated Genes; ccf mtDNA, Circulating Cell-Free Mitochondrial DNA; TACE, Transarterial Chemoembolization; AFP, Alpha-Fetoprotein; HBV-HCC, Hepatitis B Virus–Associated Hepatocellular Carcinoma. Table 2Summary of Mitochondrial Dysfunction-Related Biomarkers and Platforms in Hepatocellular CarcinomaCategoryBiomarker/PlatformApplicationKey FindingsTreatment sensitivity predictionOXPHOS biosensor combined with AI-based prediction platformPrediction of tumor sensitivity to the OXPHOS inhibitor IACS-010759Enables rapid and efficient drug sensitivity assessment, providing technical support for precision medicine140Prognostic assessmentOXPHOS-related gene signaturePrognostic prediction in HCCFacilitates elucidation of the biological role of OXPHOS in HCC and assists in patient survival prognostication141Prognostic assessmentGMAGs signaturePrognostic prediction and therapeutic guidance in HCCIdentified four core genes: RRM1, RRM2, G6PD, and GPX7142Diagnosisccf mtDNAEarly diagnosis of HCC and monitoring of TACE efficacyDemonstrates superior diagnostic performance for early-stage HCC compared with AFP; serves as a risk biomarker for TACE efficacy144,145DiagnosisUV39H1 joint AFPDiagnosis of HBV-associated HCCCombined detection yields superior diagnostic performance compared with AFP alone31DiagnosisSix mitochondrial-related genes: MTHFD1L, NT5DC2, POLQ, RECQL4, TOMM40L, and TXNRD1Potential diagnostic biomarkers for HCCAll six genes are upregulated in HCC tissues and cell lines, suggesting their potential as diagnostic biomarkers146Emerging directionMitochondrial temperatureAdjunctive diagnostic indicator for HCCIn metformin-treated HepG2 cells, cells with lower metabolic activity exhibited higher mitochondrial temperature147Abbreviations: OXPHOS, Oxidative Phosphorylation; HCC, Hepatocellular Carcinoma; GMAGs, Glutamine Metabolism-Associated Genes; ccf mtDNA, Circulating Cell-Free Mitochondrial DNA; TACE, Transarterial Chemoembolization; AFP, Alpha-Fetoprotein; HBV-HCC, Hepatitis B Virus–Associated Hepatocellular Carcinoma. Summary of Mitochondrial Dysfunction-Related Biomarkers and Platforms in Hepatocellular Carcinoma Abbreviations: OXPHOS, Oxidative Phosphorylation; HCC, Hepatocellular Carcinoma; GMAGs, Glutamine Metabolism-Associated Genes; ccf mtDNA, Circulating Cell-Free Mitochondrial DNA; TACE, Transarterial Chemoembolization; AFP, Alpha-Fetoprotein; HBV-HCC, Hepatitis B Virus–Associated Hepatocellular Carcinoma. To achieve precision in mitochondria-targeted therapy, it is essential to establish a matching system linking “biomarkers to therapeutic strategies.” Based on available evidence, we propose a preliminary stratification logic. Tumors positive for OXPHOS gene signatures (eg, cases with high SNX17 or CRIF1 expression) indicate potential sensitivity to OXPHOS inhibitors such as IACS-010759. For patients with high GLS1 or SLC1A5 expression, glutamine metabolism blockers (eg, CB-839) may confer clinical benefit; however, such strategies require concurrent assessment of ASNS expression levels to circumvent compensatory resistance. In the context of ferroptosis inducers, tumor characteristics such as low SLC7A11 expression or inactivating GPX4 mutations predict better responses to agents like Erastin or RSL3. Furthermore, ccf-mtDNA levels combined with TLR9 activity evaluation may serve as companion diagnostic indicators for immuno-combination therapy, guiding the rational combination of anti-PD-1 antibodies with mitochondrial stress-inducing agents. While these hypotheses require further validation in prospective clinical trials, this framework provides an actionable theoretical basis for the selection of optimal patient populations for mitochondria-targeted therapies. Discussion Mitochondrial dysfunction not only constitutes a critical driver of HCC malignant progression but also represents an important entry point for overcoming therapeutic resistance.5,94 Current standard-of-care treatments for advanced HCC primarily include TKIs and immune checkpoint inhibitors (ICIs).148 However, the majority of patients fail to achieve durable responses to immunotherapy.149 In this context, correcting mitochondrial dysfunction to reverse resistance has gradually emerged as a promising direction for novel therapeutic strategies.22 There exists intricate crosstalk between mitochondrial metabolism and the tumor immune microenvironment. Mitochondrial dysfunction in tumor cells can undermine the efficacy of ICIs through dual mechanisms. On one hand, mitochondrial stress upregulates PD-L1 expression via the cGAS-STING pathway, promoting immune evasion.150 On the other hand, chronic antigen stimulation suppresses OXPHOS function in T cells, leading to aberrant ROS accumulation, accelerating T cell exhaustion, and impairing their antitumor activity.151 These mechanisms suggest that combining mitochondrial metabolic interventions with ICIs may represent a novel strategy to overcome immunotherapy resistance in HCC. Nevertheless, the clinical translation of mitochondria-targeted therapies faces multiple challenges. The OXPHOS inhibitor IACS-010759 was discontinued due to dose-limiting neurotoxicity observed in Phase I clinical trials.137 The GLS1 inhibitor BPTES has been hampered by poor solubility and low oral bioavailability.138 Similarly, the GLS1 inhibitor CB-839 has shown limited efficacy as a monotherapy in HCC and remains at the preclinical stage.119 However, its combination with 5-FU has been shown to enhance drug sensitivity in resistant HCC cells by inducing oxidative stress and ferroptosis.152 To circumvent neurotoxicity, triphenylphosphonium (TPP⁺)-conjugated mitochondria-targeted derivatives (eg, MitoQ, Mito-ATO, Mito-LND, Mito-HNK) have demonstrated antitumor activity across various tumor models without significant neurotoxicity; however, their therapeutic value in HCC remains to be validated.137 In summary, mitochondrial dysfunction plays a pivotal role in HCC progression and immunotherapy resistance, and targeting mitochondrial metabolism holds considerable potential for clinical translation. However, current evidence remains largely confined to the preclinical stage, and its translation to clinical practice faces three major limitations: uncertainty regarding targeting specificity, the confounding effects of HCC heterogeneity on therapeutic efficacy, and insufficient predictive power of preclinical models. These factors collectively constrain the development of mitochondria-targeted intervention agents. Given that the regulatory mechanisms of mitochondrial function in HCC have not yet been fully elucidated,153 future research should focus on the following directions. At the level of model development and patient stratification, patient-derived organoids (PDOs) and patient-derived xenograft (PDX) models should be employed to systematically evaluate the therapeutic window and biomarkers of mitochondria-targeted agents, with patient stratification based on HCC metabolic subtypes to guide subsequent clinical trial design. At the mechanistic level, the regulatory nodes of iron metabolism, mitochondrial lipid peroxidation, and mitochondrial retrograde signaling in sustaining drug resistance warrant in-depth investigation to identify novel targets for combination strategies. At the translational level, the potential of mitochondria-related metabolites and gene expression profiles as diagnostic/prognostic biomarkers should be explored to identify patient populations with high mitochondrial dependency, thereby facilitating individualized therapeutic decision-making.