Selective autophagy in hepatocellular carcinoma: Mechanisms, roles and therapeutic implications (Review)

1. Introduction

Hepatocellular carcinoma (HCC) was the third leading cause of cancer-related mortality in 2020, with ~905,000 new cases and 830,000 deaths worldwide (1). The incidence remains the highest in East Asia and Africa, but is increasing in Western nations. In total, ~54% cases are caused by chronic hepatitis B infection, whereas 15% are caused by hepatitis C. Alcohol abuse, aflatoxin, non-alcoholic steatohepatitis and metabolic syndrome-related diabetes are also contributing factors (2,3). The male predominance (2-3:1) in hepatocellular carcinoma (HCC) suggests gender-specific differences in exposure to risk factors like chronic viral infections and alcohol consumption, which may contribute to the higher incidence in men (2).

Autophagy is a cellular self-digestion process that serves to maintain homeostasis by degrading damaged organelles and macromolecules through the lysosomes. In cancer, autophagy serves complex, context-dependent roles, suppressing tumor initiation by removing harmful cellular components, but can also promote the survival of established tumors under stress (4,5). Starvation-induced non-selective macroautophagy randomly engulfs the cytoplasm and generates nutrients through indiscriminate lysosomal breakdown (5). Selective autophagy uses cargo receptors containing LC3-interacting regions, such as sequestosome 1 (p62), nuclear dot protein 52 kDa or nuclear receptor coactivator 4 (NCOA4), to bind ubiquitin-tagged substrates (such as damaged mitochondria, ferritin and aggregates) and recruit phagophores exclusively to them. This receptor-guided precision allows for organelle-specific quality control whilst minimizing biomass loss and can be adjusted independently of bulk autophagy (6,7). HCC frequently exhibits dysregulated autophagy (4,5). Selective autophagy involves the autophagic targeting of specific intracellular organelles or materials. Key forms of autophagy (Table I) include mitophagy (mitochondria) (8-11), lysophagy (lysosomes) (10,12), reticulophagy [endoplasmic reticulum (ER)] (9,13), pexophagy (peroxisomes) (14), nucleophagy (nucleus) (15), ribophagy (ribosomes) (16,17), lipophagy (lipid droplets) (11,17-19), glycophagy (glycogen) (19,20), ferritinophagy (ferritin/iron) (13,21) and fluidophagy (autophagy of phase-separated liquid condensates (18-22). Each of the aforementioned processes has been implicated in the pathogenesis and therapeutic response in HCC. Target regulators and therapeutic agents unique to each type of selective autophagy are shown in Fig. 1.

Figure 1 Target regulators and therapeutic agent unique to each selective autophagy type in hepatocellular carcinoma.

Table I Experimental findings in selective autophagy in HCC.

Table I

Experimental findings in selective autophagy in HCC.

Autophagy type	Key experimental findings	(Refs.)
Mitophagy	• PINK1/Parkin inhibition leads to dysfunctional mitochondria and increased apoptosis	(8)
	• High mitophagy gene expression correlates with worse survival	(8)
	• Parkin-mediated mitophagy supports cancer stem cells and neutralizes p53	(8)
Lysophagy	• Sorafenib-resistant HCC cells rely on lysophagy; polyphyllin D disrupts this process, causing cell death	(12)
	• UBE2QL1 knockdown leads to damaged lysosome accumulation, sensitizing cells to lysosomal death	(12)
Reticulophagy	• Sorafenib activates FAM134B-mediated ER-phagy, protecting against ferroptosis	(34)
	• FAM134B knockdown enhances ferroptotic cell death	(9)
	• circFAM134B silencing enhances lenvatinib-induced ferroptosis	(34)
Pexophagy	• Oxidative stress triggers peroxisomal protein ubiquitination and p62/NBR1-mediated clearance	(14)
	• Blocking autophagy leads to dysfunctional peroxisome accumulation and DNA oxidation	(14)
	• HCC cells may use pexophagy to shift metabolism toward mitochondria	(14)
Nucleophagy	• Autophagy degrades micronuclear DNA via cGAS-LC3 interaction, reducing STING signaling	(15)
	• Autophagy inhibition increases cytosolic DNA and interferon signaling	(15)
	• cGAS-STING can trigger autophagy-dependent cell death under severe telomere damage	(15)
Ribophagy	• Nutrient starvation/mTORC1 inhibition relocates NUFIP1 to ribosomes, driving degradation	(17)
	• Blocking autophagy during nutrient deprivation reduces HCC cell survival	(17)
	• Ribophagy may help cells remove stalled ribosomes and adapt to stress	(17)
Lipophagy	• Autophagy dysfunction leads to lipid accumulation and HCC development	(18)
	• Blocking autophagy causes lipid droplet buildup and energy depletion	(19)
	• BNIP3-driven lipophagy restrains tumor growth; low BNIP3 yields lipid-rich tumors	(30)
Glycophagy	• STBD1/PTG loss sensitizes HCC cells to glucose withdrawal	(19)
	• Autophagy inhibition reduces glucose output during fasting	(20)
	• G6PC deficiency impairs glycophagy, leading to adenomas/HCC	(64)
Ferritinophagy	• NCOA4 releases iron from ferritin, enabling ferroptosis	(13)
	• NCOA4 overexpression increases labile iron and slows tumor growth	(49)
	• PTBP1 silencing upregulates NCOA4 and sensitizes tumors to sorafenib	(13)
Fluidophagy	• Optineurin/NDP52 form droplets on damaged mitochondria to facilitate clearance	(22)
	• p62 forms liquid droplets that sequester ubiquitinated proteins	(22)
	• In autophagy-deficient tumors, p62 aggregates activate NRF2, promoting tumorigenesis	(18)

[i] AMPK, AMP-activated protein kinase; BNIP3, BCL2 interacting protein 3; CCDC50, coiled-coil domain containing 50; CQ, chloroquine; DRP1, dynamin-related protein 1; ER, endoplasmic reticulum; FAM134B, family with sequence similarity 134 member B; HCC, hepatocellular carcinoma; NCOA4, nuclear receptor coactivator 4; NRF2, nuclear factor erythroid 2-related factor 2; PINK1, phosphatase and tensin homolog-induced kinase 1; PPT1, palmitoyl-protein thioesterase 1; PTBP1, polypyrimidine tract-binding protein 1; ROS, reactive oxygen species; SQSTM1, sequestosome 1 (p62); STBD1, starch-binding domain-containing protein 1.

In this review work, PubMed and Google Scholar were searched for entries added between January 2015 and March 2025 with Boolean strings combining ‘selective autophagy’, ‘hepatocellular carcinoma’ and each pathway name (such as ‘mitophagy’ and ‘ferritinophagy’) for a narrative-review approach. In addition, the reference lists of the retrieved articles were hand-screened. The inclusion criteria were as follows: English-language primary data or systematic reviews detailing mechanisms, biomarkers or therapeutic manipulation of selective autophagy in HCC or liver models. The exclusion criteria were non-HCC cancers, conference abstracts without the full text, studies lacking mechanistic details and commentaries. Two authors independently screened titles/abstracts, and any disagreements were resolved by a third reviewer. Data on the pathway effectors, experimental design and translational relevance were extracted and tabulated to ensure consistency.

2. Mitophagy in HCC

Mitophagy is the selective autophagic removal of mitochondria that is critical for mitochondrial quality control. This is primarily mediated by the phosphatase and tensin homolog-induced kinase 1 (PINK1)/Parkin pathway, wherein PINK1 accumulates in depolarized mitochondria and recruits Parkin to tag the mitochondria for autophagy (23,24). In HCC, Parkin expression is frequently lost or reduced (downregulated in ~50% HCC tumors) (25). Previous studies have identified siah E3 ubiquitin protein ligase 1 (SIAH1) as a key Parkin substitute in HCC. Specifically, SIAH1 can cooperate with PINK1 to ubiquitinate mitochondrial proteins and initiate mitophagy (26,27).

Sorafenib-induced mitophagy serves as a cytoprotective mechanism, clearing drug-damaged mitochondria and preventing the excess accumulation of reactive oxygen species (ROS) (26). Sorafenib actively induces mitophagy as a defense mechanism against HCC, by inducing PINK1 accumulation and Parkin recruitment to damaged mitochondria, in turn activating mitophagy in HCC cells. Inhibition or silencing of PINK1 or SIAH1 has been found to suppress mitophagy and increase sorafenib-induced cell death, confirming their role as a cytoprotective component (28,29). In orthotopic Huh7 xenografts, combining sorafenib with the dynamin-related protein 1 (DRP1) inhibitor mdivi-1 (50 mg/kg) was observed to cut tumor volume by 63 vs. 41% for sorafenib monotherapy, indicating that blocking mitophagy can magnify sorafenib lethality (30). Consequently, HCC cells can survive more readily after sorafenib treatment when mitophagy remains intact. Although basal mitophagy can prevent liver tumor initiation by removing dysfunctional ROS-producing mitochondria in established HCC, it frequently promotes tumor cell metabolism and therapy resistance (5). Chemotherapy (such as cisplatin) has been reported to induce mitophagy in surviving HCC cells through DRP1-mediated mitochondrial fission (31). Inhibiting DRP1 with mdivi-1 can block this mitophagy and increase the apoptosis of cisplatin-treated HCC cells (31).

Excessive mitophagy activation has been explored as a therapeutic strategy for inducing HCC cell death (Table II). The iron chelator deferiprone triggers PINK1/Parkin-independent mitophagy in liver cells and slows HCC tumor growth in mice, presumably by forcing mitochondrial turnover and compromising tumor bioenergetics (8). Chronic hepatitis B virus infection, a major etiological factor of HCC, can influence mitophagy (32,33). Hepatitis B virus X protein has been shown to induce NIP3-like protein X-dependent mitophagy in liver cells, which reprograms metabolism towards glycolysis to enhance cancer stem cell properties (34).

Table II Therapeutic strategies for selective autophagy in HCC.

Table II

Therapeutic strategies for selective autophagy in HCC.

Autophagy type	Therapeutic targets/strategies	Mechanism/rationale	(Refs.)
Mitophagy	• Parkin/PINK1 pathway inhibition	• Increases dysfunctional mitochondria, ROS and apoptosis	(8)
	• Chloroquine/Hydroxychloroquine	• Prevents autolysosome formation, increases mitochondrial ROS	(10)
	• AMPK activators (Metformin)	• Removes damaged mitochondria, reduces steatosis	(11)
	• BNIP3 upregulation	• Triggers excessive mitophagy and cell death	(9)
Lysophagy	• ASM inhibitors (Desipramine)	• Destabilizes lysosomes, mimicking polyphyllin D effects	(10,12)
	• Lysosomal BH3 mimetics (BMH-21)	• Activates autophagy to overwhelm lysosomes	(10)
	• TFEB activators (Trehalose)	• Increases lysosome biogenesis, generates autophagy trap	(10)
	• Lysosomal pH alkalinizers	• Impairs degradation, causing substrate accumulation	(10)
Reticulophagy	• FAM134B inhibition	• Enhances sorafenib-induced ferroptosis	(9)
	• PABPC1 inhibitors	• Reduces FAM134B translation and ER-phagy	(9)
	• Ferroptosis inducers + autophagy inhibitors	• Prevents ferroptosis mitigation via ER-phagy	(9)
Pexophagy	• USP30 inhibitors	• Increases peroxisome ubiquitination and clearance	(14)
	• ROS inducers + autophagy inhibitors	• Forces peroxisome damage accumulation	(14)
	• PPARα agonists + autophagy inhibitors	• Increases peroxisome load then blocks turnover	(14)
Nucleophagy	• STING agonists + autophagy inhibitors	• Amplifies innate immune activation by sustaining STING signals	(15)
	• Immune checkpoint inhibitors + autophagy inhibitors	• Promotes immunogenic cell death via DNA release	(15)
	• p53 reactivators + mTOR inhibitors	• Induces death in genomically unstable cells	(15)
Ribophagy	• mTORC1 inhibitors	• Induces autophagy including ribophagy	(16)
	• Dual PI3K/mTOR inhibitors + autophagy blockers	• Traps non-functional ribosomes in autophagosomes	(16)
	• NUFIP1/ZNHITE3 disruptors	• Prevents ribosome clearance during stress	(17)
Lipophagy	• Autophagy inhibitors in NAFLD-HCC	• Blocks fat mobilization, causing energy deficit	(17-19)
	• AMPK activators + autophagy inhibitors	• Creates autophagy trap during peak lipid mobilization	(11)
	• DGAT1/2 inhibitors + autophagy blockers	• Prevents triglyceride formation, increasing toxic free fatty acids	(11)
Glycophagy	• Glycogen phosphorylase inhibitors + autophagy blockers	• Redirects to lysosomal route then blocks it, causing energy crisis	(19)
	• STBD1 stabilizers	• Increases glycogen autophagy beyond handling capacity	(19)
	• Glucose restriction/SGLT2 inhibitors	• Forces tumors to consume internal glycogen	(19,20)
Ferritinophagy/	• NCOA4 upregulation	• Increases iron release from ferritin	(13)
Ferroptosis	• Ferroptosis inducers	• Promotes iron-dependent lipid peroxidation	(13)
	• Iron mobilizers	• Primes cells for ferroptosis	(13,21)

[i] AMPK, AMP-activated protein kinase; ASM, acid sphingomyelinase; BNIP3, BCL2 interacting protein 3; DGAT1/2, diacylglycerol O-acyltransferase 1/2; HCC, hepatocellular carcinoma; mTORC1, mechanistic target of rapamycin complex 1; NAFLD, non-alcoholic fatty liver disease; NUFIP1, nuclear FMR1 interacting protein 1; PABPC1, poly(A) binding protein cytoplasmic 1; PI3K, phosphatidylinositol 3-kinase; PINK1, phosphatase and tensin homolog-induced kinase 1; PPARα, peroxisome proliferator-activated receptor α; ROS, reactive oxygen species; SGLT2, sodium-glucose cotransporter 2; STBD1, starch-binding domain-containing protein 1; STING, stimulator of interferon genes; TFEB, transcription factor EB; USP30, ubiquitin specific peptidase 30; ZNHITE3, zinc finger HIT domain-containing protein 3.

In summary, mitophagy in HCC appears to have a dual role. It can either protect cells by clearing damaged mitochondria (limiting oxidative stress and cell death) or enable tumor survival under treatments, such as kinase inhibitors or chemotherapy (5,31). Therapeutically, mitophagy can be manipulated by inhibition (to prevent tumor cells from escaping apoptosis during therapy) or overactivation (to induce metabolic crisis in tumors) (8,31).

3. Lysophagy in HCC

Lysophagy is the selective removal of lysosomes damaged by autophagy. When lysosomal membranes are compromised, cells can target these leaky lysosomes for autophagic degradation to prevent the release of hydrolytic enzymes into the cytosol (9). The process involves damage recognition by cytosolic galectins (whereby galectin-3/8 binds to exposed luminal glycans on ruptured lysosomes) and the ubiquitination of lysosomal membrane proteins, followed by the recruitment of autophagy receptors (35,36).

Recently, coiled-coil domain containing 50 (CCDC50) was identified as a dedicated lysophagy receptor in human melanoma cells (9). CCDC50 can bind to galectin-3-decorated K63-ubiquitinated lysosomes and facilitate their clearance through autophagy (9). In HCC, CCDC50 serves a tumor-promoting role by sustaining lysosomal homeostasis. It was previously shown that CCDC50 expression is upregulated in HCC tumors and cell lines, where its high expression associates with more aggressive tumor features and worse patient prognosis (9).

Functionally, CCDC50 loss led to the accumulation of broken lysosomes in HCC cells, impaired autophagic flux, elevated ROS levels and cell death (9). In mouse models, CCDC50 deficiency was found to significantly suppress HCC tumor growth, presumably due to the cytotoxic buildup of dysfunctional lysosomes and oxidative damage (9). Therapeutically, inhibiting lysophagy could harm cancer cells by allowing lysosomal damage to accumulate. Polyphyllin D has been observed to cause lysosomal membrane permeabilization in HCC cells, resulting in the loss of lysophagy and spillage of enzymes to trigger cell death (37). Lysosome-targeting agents, such as chloroquine or the palmitoyl-protein thioesterase 1 (PPT1) inhibitor GNS561, can effectively kill HCC cells by overwhelming the lysosomal system (38). GNS561 (ezurpimtrostat) showed potent anti-tumor activity in preclinical HCC models and has since entered clinical trials (39).

Overall, these aforementioned findings suggest that lysophagy allows HCC cells to maintain the functionality of their digestive organelles. Disruption of lysophagy (either genetically or with lysosomotropic drugs) will likely cause catastrophic lysosomal failure and tumor suppression (9,37).

4. Reticulophagy (ER-phagy) in HCC

Reticulophagy is the selective autophagic degradation of portions of the ER. Several ER-resident receptor proteins [family with sequence similarity 134 member B (FAM134B), SEC62 homolog, preprotein translocation factor (SEC62), reticulon 3 (RTN3) and cell cycle progression 1 (CCPG1)] can mediate reticulophagy by binding LC3/γ-aminobutyric acid receptor-associated protein (GABARAP), a type B autophagy-related gene 8 family protein receptor, to form autophagosomes and capture ER fragments (40). FAM134B is a critical ER-phagy receptor that binds to LC3 through a conserved LIR motif and selectively fragments the ER membrane for autophagic engulfment (40).

Sorafenib induces ER stress and activates FAM134B-dependent reticulophagy (41). Liu et al (41) reported that sorafenib can stimulate ER-phagy through FAM134B, potentially mitigating the cytotoxicity of sorafenib. FAM134B knockdown blocked ER-phagy and enhanced sorafenib-induced ferroptosis in HCC cells (40). Intact FAM134B-mediated reticulophagy was protective because it contains ER components, including misfolded proteins, calcium-overloaded ER fragments, and damaged ER membranes, along with oxidized lipids and proteins generating ROS, that contribute to oxidative stress and ferroptosis (40).

A recent study on the circular (circ)RNA of the FAM134B gene (circFAM134B) showed that this circRNA increases FAM134B expression and promotes ER-phagy, leading to the suppression of ferroptosis in HCC cells treated with lenvatinib. Silencing circFAM134B increased the sensitivity of HCC cells to Lenvatinib (40). From a therapeutic perspective, inhibiting FAM134B-mediated ER-phagy may be beneficial by eliminating ferroptosis in tyrosine kinase inhibitor-treated tumors (40). This hypothesis is based on the observation that FAM134B knockdown potentiated sorafenib and lenvatinib lethality in preclinical models (40).

5. Pexophagy in HCC

Pexophagy is the selective autophagic degradation of peroxisomes. Peroxisomes are vital organelles involved in lipid metabolism, ROS detoxification, biosynthesis of bile acids and plasmalogens (42). Quality control of peroxisomes is essential because dysfunctional peroxisomes can produce excessive ROS and toxic byproducts. Pexophagy removes surplus or damaged peroxisomes to maintain the cellular redox balance (5).

Pexophagy malfunction leads to peroxisomal accumulation, which is associated with oxidative stress and inflammation in the liver (43). In catalase-knockout mice, extended fasting triggered excessive pexophagy and hepatocyte cell death (44). This suggests that under conditions of high oxidative stress burden, pexophagy may be activated to the point of cell death, a mechanism potentially relevant in chronic liver disease and HCC development (5).

To the best of our knowledge, direct studies of pexophagy in HCC are sparse. However, analyses of HCC tissues have revealed the dysregulation of peroxisomal proteins and redox imbalance, which could be partly due to altered pexophagic activity (5). Pexophagy in HCC remains underexplored therapeutically. Peroxisome proliferator-activated receptor-α agonists (such as clofibrate) can induce peroxisome proliferation and subsequent pexophagy in the liver, such that they trigger pexophagy in fatty liver contexts (45). Another approach is to target autophagy adapters, such as neighbor of BRCA1 gene 1 protein which, along with p62, can act as receptors for ubiquitinated peroxisomes to enhance the efficiency of pexophagy (46).

6. Nucleophagy in HCC

Nucleophagy is the selective autophagic degradation of nuclear material. In mammalian cells, nucleophagy is less common but has been observed during senescence and after DNA damage (21). One key mechanism described in cancer cells is DNA damage-induced nucleophagy through lamin tagging (21). Nucleophagy excises and degrades damaged nuclear constituents when chromatin breaks or lamin scaffolds become irreparable. During this process, SUMO-modified lamin A/C binds to LC3, allowing autophagosomes to sequester micronuclei or ruptured envelope fragments for lysosomal digestion (47). In hepatocytes, inhibiting lamin A/C SUMOylation increases γ-H2AX foci and micronuclei, reflecting elevated DNA damage. Activation of nucleophagy mitigates this damage by removing damaged nuclear components, thereby preserving genome stability. However, this protective process may allow tumor cells to survive genotoxic stress. Cells do not necessarily die immediately, because nucleophagy selectively engulfs and digests damaged nuclear material, maintaining nuclear integrity and cellular viability despite persistent DNA lesions. Consequently, nucleophagy functions as a double-edged sword, limiting DNA-damage signaling while potentially promoting tumor cell survival (21).

Li et al (21) demonstrated that when DNA damage causes small fragments of DNA to leak from the nucleus, the cell responds by SUMOylating nuclear lamin A/C through the E2 enzyme ubiquitin-conjugating enzyme E2 I. SUMOylation creates a signal that recruits LC3, effectively bridging the nuclear lamina to the autophagosome (21). In the context of HCC, direct evidence of nucleophagy is limited. However, the liver tumor environment such as DNA-damaging chemotherapeutic agents, metabolic stress or lamina deformations, trigger nucleophagy in HCC. HCC cells are frequently exposed to genotoxic stress, which can result in double-stranded DNA breaks and micronucleus formation (21,47).

From a therapeutic standpoint, targeting nucleophagy is challenging because it is not well defined and can harm normal cells. One potential angle is that inhibiting SUMOylation can block nucleophagy and perhaps make HCC cells more prone to accumulating nuclear damage, leading to cell death.

7. Ribophagy in HCC

Ribophagy involves the selective autophagic degradation of ribosomes. During nutrient starvation, cells degrade their ribosomes to recycle amino acids and nucleotides (48). Autophagy-dependent degradation of ribosomal RNA has been shown to be crucial for maintaining nucleotide pools during development (13). During metabolic stress, ribophagy dismantles superfluous or stalled ribosomes to recycle nucleotides and amino acids. Starvation or mTOR1 inhibition drives nuclear FMR1-interacting protein 1 (NUFIP1) from the nucleoplasm to the 60S subunits, where its LC3-interacting region, aided by zinc finger HIT domain-containing protein 3, targets ribosomes to autophagosomes, whereas concurrent p62 recruitment clears ubiquitinated ribosomal proteins (49). Suppressing NUFIP1 in nutrient-deprived HCC cells heightens proteotoxic stress, ATP depletion and apoptosis, underscoring the role of ribophagy in balancing translation load and energy supply (49). Direct studies on the role of ribophagy in HCC are lacking. However, given that solid tumors frequently experience nutrient-poor and hypoxic microenvironments, it is conceivable that HCC cells can invoke ribophagy as an adaptive mechanism. A study has previously shown that blocking late-stage autophagy with chloroquine similarly impaired ribophagic recycling in HCC cells (22). From a functional standpoint, ribophagy in HCC could contribute to therapy resistance, where when exposed to metabolic stress (such as glycolysis inhibitors or anti-angiogenic therapy), HCC cells may digest ribosomes and survive longer.

8. Lipophagy in HCC

Lipophagy is the autophagic degradation of lipid droplets. In liver cells, lipophagy is crucial for mobilizing stored triglycerides and cholesterol esters, delivering them to lysosomes where lipases break them down to free fatty acids (50). This process is central to energy homeostasis. In the context of liver disease and HCC, lipophagy has significant implications in counteracting lipid accumulation (51).

In HCC tumors, lipophagy appears to have a context-dependent role. Cancer cells can use lipophagy to tap into lipid droplet reserves to meet energy demands (4). During starvation or therapy-induced nutrient stress, HCC cells can upregulate factors, such as CCAAT/enhancer binding protein α, enhancing lipophagy to release fatty acids from lipid droplets to sustain ATP production (4).

Conversely, lipophagy can act as a tumor-suppressive mechanism by preventing excessive lipid accumulation and lipotoxicity. Preventing lipotoxicity suppresses tumors because BCL2-interacting protein 3 (BNIP3)-mediated lipophagy tethers lipid droplets to autophagosomes, ensuring their degradation. By clearing excess lipids, this process avoids lipid-driven energy supply and toxic lipid accumulation, restraining tumor growth. Loss of BNIP3 leads to lipid-rich tumors and poorer survival, underscoring lipophagy's tumor-suppressive role. Chen et al (34) previously showed that BNIP3 can tether lipid droplets to autophagosomes through LC3 binding in a process termed ‘mitolipophagy’, which restrained HCC development. In a genetic HCC mouse model, loss of BNIP3 led to the earlier development of larger tumors that contained markedly more lipid droplets (51).

In human HCC specimens, low BNIP3 expression has been associated with higher lipid content in tumors and poor patient survival (51). These findings indicate that insufficient lipophagy (at least in part due to BNIP3 inactivation) can promote HCC progression by allowing lipids to accumulate and fuel tumor growth (51). Therapeutically, there are two potential strategies: Inhibition of lipophagy to starve HCC cells or enhancement of lipophagy to induce lipotoxic stress in tumors. It has been previously demonstrated that combining an autophagy inhibitor (e.g., chloroquine or its analog hydroxychloroquine) with a metabolic stressor increased lipid accumulation and cell death in HCC (22,38).

9. Glycophagy in HCC

Glycophagy is the selective autophagy of glycogen. The liver is the main storage site for glycogen, where in addition to canonical cytosolic glycogenolysis, hepatocytes possess a lysosomal route to degrade glycogen. This pathway is mediated by starch-binding domain-containing protein 1, which binds glycogen and recruits it to the autophagosome by interacting with γ-aminobutyric acid receptor-associated protein-like 1(52). Inside lysosomes, glycogen is broken down by acid α-glucosidase (52).

In HCC, the role of glycophagy is not well characterized but is likely relevant, given that numerous HCC cells exhibit metabolic flexibility and the ability to store glucose as glycogen (53). HCC cells may invoke glycophagy when extracellular glucose is scarce. During transarterial embolization therapy, which induces acute nutrient starvation in the tumor region, cancer cells can survive by consuming their glycogen through autophagy (53). Previous studies have shown that depriving HCC cells of glucose triggers autophagy activation (26,53). Therapeutically, targeting glycophagy is challenging. A conceivable approach during transarterial embolization or fasting-mimicking diet therapy for HCC is to add an agent that stimulates glycophagy to rapidly deplete tumor glycogen, thereby rendering the nutrient cut-off more lethal to cancer cells.

10. Ferritinophagy in HCC

Ferritinophagy is the selective autophagy of ferritin, a cellular iron storage complex. This process is mainly mediated by the cargo receptor NCOA4, which binds to ferritin and delivers it to autophagosomes (54). Through ferritinophagy, iron stored in ferritin can be released upon lysosomal degradation, increasing the labile iron pool within the cell (54). Ferritinophagy links autophagy to iron homeostasis and ferroptosis, which is a form of cell death caused by iron-dependent lipid peroxidation.

Aberrant iron metabolism is common in HCC. Previous studies have indicated that ferritinophagy serves a role in HCC cell death and survival (10,55). The induction of ferritinophagy can kill HCC cells by ferroptosis. NCOA4 mediates ferritinophagy, targeting ferritin for lysosomal degradation to release stored iron. This NCOA4-driven iron liberation expands the cellular labile iron pool, providing redox-active Fe²+ that catalyzes ROS generation through Fenton chemistry (56). The resulting oxidative stress triggers the peroxidation of polyunsaturated fatty acids in membranes, which is a hallmark of ferroptosis (10). Mechanistically, NCOA4-mediated ferritinophagy sensitizes cells to ferroptosis by elevating catalytic iron and accelerating lethal lipid peroxide accumulation. Consistently, inducing ferritinophagy (such as by caryophyllene oxide) increases free Fe²+ and malondialdehyde levels (a lipid peroxidation product), whereas NCOA4 knockdown or interfering with NCOA4-ferritin binding prevents iron release and protects cells from ferroptotic damage (42). In addition, suppression of NCOA4 or its regulator, polypyrimidine tract-binding protein 1 (PTBP1), can reduce the labile iron pool and lipid peroxidation, blunting ferroptotic death (43). Therefore, NCOA4-driven ferritinophagic iron release is a pivotal trigger of ferroptosis through labile iron accumulation and subsequent ROS-mediated lipid peroxidation. In an in vitro study, caryophyllene-oxide treatment (20 µM; 6 h) doubled the extent of NCOA4-ferritin co-localization in Hep3B cells, raised calcein-quenchable Fe²+ by 42% and increased malondialdehyde by 2.5X. By contrast, NCOA4 knockdown abolished iron release and rescued 60% cells from cell death (54).

Another study by Yang et al (57) identified a pathway involving PTBP1 that can regulate ferritinophagy and ferroptosis in HCC. Specifically, PTBP1 can bind to NCOA4 mRNA to enhance its translation, where the knocking down PTBP1 expression in sorafenib-treated HCC cells led to lower NCOA4 levels, reducing free iron and malondialdehyde levels whilst increasing glutathione levels, suppressing ferroptosis (57). Simultaneously, PTBP1 depletion lowered NCOA4 by 48 %, shrank the labile iron pool by 35 % and suppressed erastin-induced ferroptosis in MHCC-97H xenografts (57). However, ferritinophagy may also help HCC cells survive under certain conditions by providing iron for essential enzymes. A recent report on pancreatic cancer has shown that NCOA4-mediated ferritinophagy is required for tumor growth by supplying iron to the proliferating cells (58).

Induction of excessive ferritinophagy appears to be a promising therapeutic approach for HCC, essentially pushing cells into ferroptosis, which is a lethal oxidative form of cell death that cancer cells cannot easily reverse. Various compounds, such as artesunate, sorafenib and caryophyllene oxide, can induce ferroptosis, at least partly by increasing free iron levels either through ferritinophagy or direct iron import (54).

11. Fluidophagy in HCC

Fluidophagy is a relatively novel term, describing the autophagic degradation of fluid-like, phase-separated intracellular condensates (59). Cells contain numerous non-membrane-bound organelles that behave as liquid droplets. A number of biophysical studies have previously shown that the ‘wetting’ properties of droplets on autophagosome membranes are crucial (60-62). This process of droplet autophagy has been named ‘fluidophagy’ (59,63). At present, fluidophagy is a mechanistic insight into selective autophagy, rather than a distinct pathway with unique regulators.

For HCC, the biophysical state of the cargo (solid aggregates vs. liquid droplets) can affect autophagic clearance. In particular, the NCOA4-ferritin complex, which can form liquid condensates in the cytosol, helps concentrate ferritin into such droplets, which are then autophagically consumed (63). In addition, the autophagic degradation of p62/sequestosome 1 (SQSTM1) bodies is another possible example of fluidophagy (59).

In HCC, fluidophagy likely underlies the removal of certain protein aggregates or storage complexes from the cells. HCC cells frequently accumulate aggregates of p62 and ubiquitinated proteins (p62 is commonly elevated in HCC and is used as a marker of autophagic flux) (11).

12. Therapeutic targeting

Dysregulated selective autophagy contributes to HCC progression and resistance to therapy, making the components of these pathways attractive therapeutic targets. Strategies to modulate autophagy in cancer are two-fold: Inhibition (to prevent tumor cells from recycling organelles and evading death) and activation (to drive cancer cells into lethal self-digestion).

Inhibiting pro-survival autophagy

HCC has been a prime candidate for autophagy inhibition trials. Chloroquine (CQ) and hydroxychloroquine, which inhibit lysosomal acidification and late-stage autophagy, have been tested in preclinical HCC models and in early clinical studies (38,39). Combining CQ with chemotherapeutic oxaliplatin led to more pronounced tumor suppression in HCC xenografts compared with chemotherapy alone (38). Newer lysosome-targeted agents, such as GNS561, show even greater potency. GNS561, a PPT1 inhibitor, accumulates in the lysosomes and blocks autophagy at a late stage, causing cancer cell death (12). In HCC models, GNS561 not only killed bulk tumor cells but was also active against cancer stem cell populations (56). One previous clinical trial tested hydroxychloroquine with the mTOR inhibitor everolimus in HCC according to the ‘simultaneous activation and blockade’ strategy, with disease control reaching 67% (with 6% partial responses) and 45% of patients had ≥6-month progression-free survival (14). Additionally, nutrient modulation, such as ketogenic or fasting-mimicking diet to induce autophagy in tumors, followed by autophagy inhibitor administration, is another reported experimental approach aiming to catch cancer cells in a vulnerable state (‘autophagy trap’) (15).

Targeting specific autophagy pathways

A more tailored approach is to target regulators unique to the selective autophagy types on which HCC depends. The role of DRP1 in mitophagy in HCC has been proposed. DRP1-driven mitophagy allows HCC cells to survive chemotherapy, since inhibiting DRP1 with mdivi-1 was found to sensitize tumors to cisplatin in mice (31). Another proposed lysophagy target is CCDC50. Since CCDC50 knockdown causes HCC cell death and tumor suppression by blocking lysophagy (9), a drug that inhibits CCDC50 may mimic this tumor-suppressive effect. In the ER-phagy arena, FAM134B and its regulator circFAM134B are potential targets to enhance ferroptosis in tumors resistant to tyrosine kinase inhibitors. An antisense oligonucleotide against circFAM134B was found to decrease FAM134B levels, impairing ER-phagy and promoting lenvatinib-induced ferroptosis (40). For lipophagy, one approach is to restore BNIP3 function in HCC. BNIP3 is frequently silenced epigenetically in cancers. Therefore, demethylating the BNIP3 promoter or delivering BNIP3 by gene therapy may reinstate autophagic clearance of lipid droplets, potentially slowing tumor growth in steatotic HCC (51).

Exploiting ferritinophagy and ferroptosis

This is a promising novel direction. Ferroptosis inducers (including sorafenib, sulfasalazine or experimental agents, such as erastin and RSL3) can be combined with strategies to increase ferritinophagy. Furthermore, various drugs, such as caryophyllene oxide, may upregulate NCOA4 and drive ferritinophagy, since they have shown efficacy in HCC models (54).

Precision considerations

Whether a given autophagy process is inhibited or induced depends on the tumor context. Early-stage, well-differentiated HCC may be more vulnerable to autophagy induction to trigger cell death, whereas late-stage, therapy-resistant HCC may require autophagy inhibition to remove its survival crutch (4,5). Biomarkers are being investigated: For instance, a ‘mitophagy gene signature’ stratifies patients with HCC and correlates with immune microenvironment differences (16). The mitophagy gene signature is an eight-gene score derived from core mitophagy regulators (e.g., optineurin and ATG12). By calculating a composite expression score and splitting at the median, it divides patients with HCC into high-risk and low-risk groups for survival. High scores indicate aggressive disease and are associated with immunosuppressive tumour microenvironments-characterised by increased infiltration of certain immune cells (B cells, CD4/8 T cells, dendritic cells, NK cells and macrophages) and altered cytokine gene expression. Thus, the signature not only predicts prognosis but also reflects differences in immune cell infiltration patterns (16).

13. Biomarkers of autophagy in HCC

Several autophagy-related biomarkers have been documented to serve prognostic and predictive values for HCC (Table III).

Table III Autophagy-related biomarkers in HCC.

Table III

Autophagy-related biomarkers in HCC.

Biomarker	Clinical findings	(Refs.)
p62/SQSTM1	• Accumulates in autophagy-defective HCC	(18,65,66)
	• High levels associate with larger tumors, venous invasion and poor survival
	• Acts as oncoprotein through NRF2 activation
LC3B	• Increased expression associates with superior differentiation and smaller tumors	(16,67)
	• Positive staining predicts longer survival
	• Context-dependent: Sometimes indicates high autophagic flux in advanced HCC
Beclin-1	• Frequently downregulated in HCC (17q loss or low transcription)	(68-70)
	• Preserved expression associates with favorable OS and DFS
	• Low levels associate with poor differentiation and vascular invasion
BNIP3	• Frequently silenced in HCC, especially in fatty liver-associated cases	(44)
	• Low expression associates with high lipid content and worse prognosis
	• Higher expression associated with lower microvascular invasion
NCOA4	• Emerging biomarker with higher expression in some HCC tissues	(13,49)
	• May predict response to ferroptosis-inducing therapy
	• Prognostic value unclear; correlation with iron load needed
p62/LC3 ratio	• High p62/low LC3 indicates autophagy-deficient tumors with poor RFS	(66,67)
	• Low p62/high LC3 associated with less aggressive tumors
	• Dual staining used to gauge autophagy flux
Autophagy gene	• 8-gene mitophagy score (such as optineurin and ATG12) stratifies HCC prognosis	(72,74)
signatures	• Multiple predictive signatures (6-13 genes) correlate with outcomes
	• High autophagy gene expression indicates aggressive disease
Immune-autophagy markers	• High LC3 in surrounding tissue but low in tumor correlates with recurrence	(10,15)
	• Circulating HMGB1 may predict immunotherapy response
	• Links autophagy status to tumor microenvironment

[i] ATG12, autophagy related 12; BNIP3, BCL2 interacting protein 3; DFS, disease-free survival; HCC, hepatocellular carcinoma; HMGB1, high mobility group box 1; LC3B, microtubule-associated protein 1A/1B-light chain 3B; NCOA4, nuclear receptor coactivator 4; NRF2, nuclear factor erythroid 2-related factor 2; OS, overall survival; RFS, recurrence-free survival; SQSTM1, sequestosome 1.

p62/SQSTM1

p62/SQSTM1 frequently accumulates in HCC cells with defective autophagy. High p62 levels are associated with larger tumor size, venous invasion and poorer overall survival (17). In addition, p62 can serve as an oncoprotein through nuclear factor erythroid 2-related factor 2 activation (18).

LC3B

Increased LC3B expression is associated with increased differentiation and smaller tumor size. A meta-analysis has previously indicated that positive LC3B immunostaining predicts longer survival in patients with HCC (19).

Beclin-1

This protein is downregulated in HCCs due to the monoallelic loss of 17q or low transcription. Patients with preserved Beclin-1 expression tend to have a more favorable overall survival and disease-free survival (20).

BNIP3

It is frequently silenced or have its expression reduced in HCC, particularly in fatty liver-associated HCC. Low BNIP3 expression is associated with high intracellular lipid levels and poor prognosis (51).

NCOA4

Initial data have suggested higher NCOA4 mRNA levels in HCC tissues compared with those in adjacent tissues. It has been hypothesized that NCOA4-high tumors respond with superior efficacy to ferroptosis-inducing therapies (51).

p62/LC3 ratio

The combination of high p62 and low LC3 levels in HCC tumor tissues indicates autophagy-deficient tumors, which are associated with shorter recurrence-free survival (64).

Autophagy gene signatures

Several prognostic signatures using autophagy-related genes have been developed to predict recurrence and overall survival (65). An eight-gene mitophagy score containing optineurin, ATG12 and related genes that stratifies patients into high- and low-risk groups and a series of predictive signatures using 6-13 autophagy genes or gene pairs (e.g., LC3B/p62/BNIP3). These signatures correlate with recurrence or overall survival, and higher autophagy-gene expression generally signifies more aggressive disease (65).

14. Clinical implications and future directions

Understanding selective autophagy in HCC has important clinical implications because HCC cells frequently co-opt organelle-specific recycling to survive hypoxia, metabolic stress and chemotherapy (51,66). The autophagy status may serve as a stratification tool to identify patients who can benefit from autophagy modulation. Tumors with high autophagic flux may be more vulnerable to autophagy inhibition, whereas those with impaired autophagy may benefit from strategies that overcome their compromised degradation systems. Furthermore, late-stage agents, such as hydroxychloroquine or the lysosomotropic inhibitor GNS561, can be used to exploit this vulnerability by arresting lysophagy and bulk flux. GNS561 was found to achieve durable disease control in a previous phase I clinical study and is advancing to phase II (30). Targeted inhibition of pro-survival mitophagy may enhance drug efficacy. Mdivi-1, genetic PINK1 or PRKN (the Parkin gene) knockdown was found to boost sorafenib-induced apoptosis by >30 % in xenografts (67).

Conversely, activating death-linked pathways is equally promising. Agents that stimulate NCOA4-dependent ferritinophagy (such as caryophyllene oxide) can enlarge the labile-iron pool, intensifying lipid peroxidation to synergize with ferroptosis inducers, achieving tumor shrinkage without added systemic toxicity (54,68). ER-phagy modulation offers precision, where circFAM134B silencing can disable reticulophagy, magnifying lenvatinib-triggered ferroptosis in resistant HCC models (40). Finally, autophagy gene signatures (LC3B, p62 and BNIP3) are emerging prognostic tools for stratifying patients according to combination regimens (69). Collectively, these data support a future in which pathway-specific autophagy modulators, paired with existing systemic or locoregional therapies can be used to drive personalized HCC management whilst sparing normal hepatocytes (70,71).

The concept of ‘synthetic lethality’ is particularly relevant, identifying genetic or metabolic vulnerabilities in HCC that, when coupled with autophagy manipulation, can lead to tumor cell death (53,55,72,73). HCC tumors with BNIP3 deficiency (and thus impaired mitophagy/lipophagy) may be particularly sensitive to treatments that increase lipid or oxidative stress (41,51,74).

Future directions in this field include the following: i) Development of selective autophagy inhibitors, moving beyond chloroquine to more targeted agents against specific autophagy receptors or regulators (4); ii) biomarker-guided therapy, where autophagy signatures can be used to guide treatment decisions (such as which patients should receive autophagy inhibitors in addition to conventional therapy) (75); iii) Combined metabolic and autophagic targeting, by exploiting the dependence of HCC on autophagy for metabolic adaptation by simultaneously targeting metabolic pathways and autophagy (76); iv) local delivery technologies, by developing liver-targeted delivery systems for autophagy modulators to minimize systemic effects (77); and v) integration with immunotherapy, by understanding how selective autophagy influences tumor immune microenvironments and exploring combinations with immune checkpoint inhibitors (78).

15. Limitations of the review

The present narrative review has several limitations that readers should consider. No formal systematic review methodology with structured database queries or predefined inclusion/exclusion criteria was applied, which introduced potential selection bias in the literature citations. Given the breadth of the field, certain relevant studies may have been inadvertently omitted, particularly since the primary focus was on English-language publications in common databases, such as PubMed. In addition, research on certain selective autophagy pathways in HCC, such as nucleophagy and ribophagy, is limited, requiring analogies to be drawn from other contexts or rely on preliminary data. the majority of the evidence discussed is also only preclinical, derived from cell lines or animal models rather than from clinical trials. Although ongoing trials were mentioned, the proposed therapeutic strategies using autophagy modulators lack proven clinical efficacy for HCC treatment. These limitations establish appropriate interpretive boundaries, ensuring the scope of the present review is within the current preclinical landscape whilst highlighting areas that require further investigation.

16. Conclusion

The selective autophagy pathways in HCC are emerging as druggable nodes. By understanding whether a given type of autophagy supports or hinders tumor growth, interventions can be designed to tip the balance against cancer. The complexity of autophagy in HCC highlights the need for personalized approaches. Some tumors rely on specific autophagy pathways for survival, whereas others may be suppressed by the same pathways. Therefore, characterizing the autophagy dependence of individual tumors is crucial for effective therapeutic interventions. The bidirectional nature of autophagy, both tumor-promoting and tumor-suppressing, necessitates careful consideration of the context, timing and specific pathways when designing therapeutic strategies. As understanding deepens, selective autophagy modulation promises to become an important component of personalized HCC treatment. As research progresses, a promising approach may be to first stress cancer cells with conventional treatments and then disable their selective autophagy defenses, thereby driving them into irrecoverable failure.

Acknowledgements

Not applicable.

Funding

Funding: The present review was supported by a grant from Chang Gung Memorial Hospital (grant no. CORPG8N0241).

Availability of data and materials

Not applicable.

Authors' contributions

CHH was involved in the conceptualization of the review, literature search, writing of the manuscript and funding acquisition. Data authentication is not applicable. The author has read and agreed to the final version of the manuscript.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The author declares that they have no competing interests.

References