Targeting cancer stem cells as the most aggressive and tumor-initiating cells

Highlights

• •Cancer stem cells (CSCs), as the most aggressive and tumor-initiating cells, play a fundamental role in driving key malignant behaviors in cancer.
• •CSCs exhibit low immunogenicity, enabling them to evade recognition and destruction by the immune system. CSCs also contribute to an immunosuppressive tumor microenvironment.
• •Conventional cancer therapies have significant limitations in effectively targeting CSCs.
• •Since CSCs are responsible for tumor repopulation after tumor cell destruction, targeted elimination of CSCs is critical for complete tumor eradication.
• •A combination of immunotherapy and cytotoxic treatments to eliminate CSCs and non-CSCs simultaneously will lead to complete eradication of cancer.

Abstract

Cancer stem cells (CSCs) are a small subpopulation of tumor cells characterized by their self-renewal capacity and the ability to differentiate into different cell types. These partially differentiated cells exhibit properties of both stem cells and cancer cells. CSCs drive tumor initiation and progression by generating additional stem cells through self-renewal and differentiation into heterogeneous populations of tumor cell. They are among the most aggressive tumor cells that contribute to the development of key features of malignancy such as increased proliferation, metastasis, tumor growth, multidrug resistance (MDR), and resistance to radiotherapy and chemotherapy. CSCs are also associated with relapse and minimal residual disease, highlighting their critical role in cancer persistence. Therefore, targeting CSCs is essential to achieve complete tumor eradication. Available evidence suggests that combination therapies that integrate immunotherapy with cytotoxic therapies to concurrently eliminate CSCs and non-CSCs offer a promising approach to completely eradicate cancer. This review summarizes the current strategies employed to target CSCs and improve cancer treatment outcomes.

Graphical abstract

Cancer stem cell concept

Cancer stem cells (CSCs) are partially differentiated cells distinguished by their capacity for self-renewal and differentiation into various cell types. Alongside embryonic stem cells (ESCs) and adult stem cells, this concept has extended to include CSCs and induced pluripotent stem (iPS) cells. These undifferentiated cells generate additional stem cells through self-renewal (1). The CSC concept, introduced about four decades ago, is based on the idea that tumor growth, similar to the renewal of healthy tissues, is driven by a small population of specialized stem cells (2,3). While the concept is not new, it has recently gained renewed attention due to advances in defining the hierarchies of healthy tissue, improved understanding of the multi-step nature of tumorigenesis, and enhanced methods for inducing human cancer in immunodeficient mice (3).

Cancer stem cell definition and characteristics

CSCs are a small subpopulation of cells within tumors that exhibit characteristics of both stem cells and cancer cells. They possess the ability to self-renew, differentiate, and initiate tumor formation when transplanted into animal hosts. These cells can be distinguished from other tumor cells by their symmetrical cell division and altered gene expression (1,4). CSCs generate more stem cells through self-renewal and give rise to differentiated progeny. This process typically involves symmetric cell division, where one daughter cell retains stem cell properties while the other undergoes multiple rounds of division and differentiation. There is currently no consensus on the precise criteria for defining CSCs, making it difficult to determine their proportion within tumors, their clinical significance, or their exact origins. Initially, CSCs were believed to represent only a small fraction of the tumor cell population in solid cancers, but some studies suggest they may constitute up to 25 % of tumor cells (1,5). Various cell surface markers including CD24, CD29, CD44, CD90, CD133, epithelial-specific antigen (ESA), and aldehyde dehydrogenase1 (ALDH1) have been used to isolate and enrich CSCs (Fig. 1) (1).

Fig. 1.

Characteristics of cancer stem cells. CSCs possess the ability to self-renew, differentiate, and initiate tumor formation. They contribute to key malignant features such as tumor growth, metastasis, and MDR. In addition, CSCs can suppress the immune system and express various cell surface markers, including CD24, CD44, CD90, CD133, and ESA, which are commonly used for their isolation and enrichment. Key signaling pathways also regulate CSC maintenance. Abbreviations: ESA, epithelial-specific antigen; MDR, multidrug resistance.

CSCs can also proliferate rapidly to replenish the progeny pool in fast-growing tumors; however, they can enter a quiescent state. Quiescence or dormancy, halts proliferation while they the cells remain viable. CSCs can remain dormant and non-dividing until the conditions become favorable for proliferation to resume (6,7). Quiescence in CSCs is reversible, as these cells retain the ability to return to a proliferative state. This adaptability suggests that during invasion and metastasis, CSCs may enter a quiescent phase while migrating, traveling through blood or lymphatic vessels, and invading distant target sites. This ability enables CSCs to adapt to the microenvironment of a distant organ, which may explain why tumors can recur years after initial diagnosis at the original or distant site (8,9). Additionally, quiescence represents a crucial aspect of CSCs’ resistance to therapies; cells may reside in the G0/G1 phase of the cell cycle to survive and undergo cellular reprogramming to adapt to tumor microenvironment (TME) conditions, such as hypoxia (10). CSCs maintain their capacity to proliferate, metastasize, and drive cancer recurrence and progression after multiple rounds of chemotherapy and radiotherapy. There is evidence of an overlap between quiescent, slow-cycling cells and CSCs in various solid tumors, including pancreatic cancer, colorectal cancer (CRC), melanoma, and glioblastoma (11,12). CSCs emerge as highly aggressive, tumor-initiating cells (TICs) that contribute significantly to malignant behaviors such as elevated proliferation, metastasis, growth, multidrug resistance (MDR), resistance to radiotherapy and chemotherapy, relapse, and minimal residual disease (9,13). CSCs have been detected in various tumor types, including both solid and non-solid malignancies, such as colon cancer (14), breast cancer (BC) (15), brain tumors (16), lung cancer (17), and melanoma (18).

Cancer stem cell origin

Several theories have been proposed regarding the origin of CSCs. The first theory posits that CSCs arise from natural stem cells or progenitor cells that acquire tumor-forming capabilities after specific genetic mutations or environmental changes. Some CSCs resemble natural stem cells or progenitor cells in cellular characteristics, phenotype, function, and cell-surface biomarkers. For example, CD44+/CD24−/low mammary gland progenitor cells resemble the CD44+CD24−/low lineage cells used to identify CSCs in breast cancer patients. Another theory suggests that CSCs originate from natural somatic cells that acquire stem-like features and malignant behavior through genetic and/or epigenetic alternations. For instance, cancer cells can exhibit stem-like traits through epithelial-mesenchymal transition (EMT). EMT induction in immortalized human mammary epithelial cells (HMECs) led to mesenchymal characteristics and the expression of stem cell biomarkers. EMT is driven by transcription factors such as ZEB1/2, SNAI1/2, and TWIST1/2, which enhance epithelial cell invasiveness. Studies have shown that EMT induction can increase self-renewal and acquisition of CSC properties ([19], [20], [21]). In contrast, some studies indicate that cancer cells with an epithelial phenotype can survive in the bloodstream and contribute to distant metastases ([22], [23], [24], [25]). In prostate cancer (PCa) cell lines, the epithelial gene program-high subpopulation expands in high-grade metastatic TICs, whereas mesenchymal subpopulations show reduced TICs. These findings suggest that plasticity and cell type-specific properties determine mesenchymal self-renewal and gene interactions (1).

Stem-like characteristics can emerge at any stage of tumor development owing to epigenetic and genetic changes (26). Within the TME, interactions with the stem cell niche, neighboring cells, and signaling pathways enable these cells to maintain stem cell characteristics despite their cancerous state. Dysregulation of certain signaling pathways plays a critical role in sustaining the self-renewal and stem-like properties of CSCs. These pathways, normally responsible for controlling differentiation and self-renewal in normal stem cells, can be abnormally activated or suppressed in CSCs. Dysregulation of signaling networks is a central mechanism by which CSCs acquire and maintain their distinctive traits, thereby promoting tumor heterogeneity, progression, and resistance to therapies (27). Understanding the intricate interactions among these signaling pathways and their significance in CSC biology is essential for developing targeted therapies that effectively eliminate CSCs and improve cancer treatment outcomes.

Cancer stem cells and immune evasion

CSCs are characterized by low immunogenicity, enabling them to escape immune surveillance despite significant immune cell infiltration within the TME (Fig. 2). CSCs orchestrate an immunosuppressive TME by secreting factors that inhibit antitumor immunity and foster their own persistence. Notably, they disrupt the recruitment and maturation of dendritic cells (DCs), driving the expansion of immunoregulatory DC subsets that suppress antitumor responses. Through secretion of TGF-β, CSCs further diminish mature DC populations and enhance tolerance-inducing cell subsets (28,29).

Fig. 2.

Mechanisms of cancer stem cell immune evasion. CSCs impair DC recruitment and maturation, promoting immunoregulatory DC subsets. They downregulate MHC expression, induce Tregs and MDSCs, and recruit TAMs via chemokine secretion, polarizing them toward the M2 phenotype. Elevated expression of immune checkpoint molecules, including PD-L1 and CTLA-4, suppresses cytotoxic T-cell activity, while reduced NKG2DLexpression limits NK cell activation. Abbreviations: CSCs, cancer stem cell; DC, dendritic cell; Tregs, regulatory T cells; MDSCs, myeloid-derived suppressor cells; TAMs, tumor-associated macrophages; NKG2DL, NKG2D ligand.

Impaired antigen processing is a hallmark of CSCs, as evidenced by the low major histocompatibility complex class I (MHC-I) expression and downregulation of antigen-processing machinery observed in glioblastoma neurospheres, which results in reduced T-cell activation (30). In head and neck squamous cell carcinoma (HNSCC), CSCs further downregulate MHC expression and promote the induction of regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs), facilitating immune evasion (31,32). In breast cancer, ALDH+ CSCs demonstrate tumorigenic capacity while suppressing transporter associated with antigen processing (TAP) and CD80, further exemplifying the diverse immune-evasive roles of CSC-associated markers (33). Additionally, CSCs downregulate co-stimulatory molecules, including CD80 and CD86, via epigenetic mechanisms such as DNA methylation, thereby impairing T-cell activation (28).

Moreover, CSCs frequently exhibit reduced expression of tumor-associated antigens (TAAs), including key stem cell markers (CD133, CD44, EpCAM), limiting recognition by antigen-presenting cells (APCs) and promoting immune escape, as demonstrated in melanoma (34).

Elevated expression of immune checkpoint molecules, particularly programmed death-ligand 1 (PD-L1) and cytotoxic T lymphocyte-associated antigen 4 (CTLA-4), on CSCs attenuates cytotoxic T-cell activity, restricts T-cell infiltration, and promotes Treg accumulation within the TME (35,36). In HNSCC, CD276 (B7-H3)–expressing CSCs act as barriers to CD8+ T-cell infiltration, reinforcing immune evasion during progression and metastasis (37). Galectin-3, a β-galactoside-binding lectin, has emerged as a key mediator of CSC-driven immune suppression (38). CSCs in lung cancer expressing galectin-3 exhibit enhanced stemness and inhibit T-cell activation by disrupting T-cell receptor clustering (39). Hypoxia further augments the immunosuppressive phenotype of CSCs by upregulating VEGF and PD-L1, thereby impairing T-cell function (40).

CSC-derived cytokines and growth factors, such as TGF-β, result in the differentiation of naïve T cells into Tregs and inhibit both cytotoxic and helper T-cell activity in cancers including breast and gastric malignancies ([41], [42], [43]). Tregs, in turn, perpetuate the immunosuppressive milieu by secreting IL-17 and TGF-β, supporting CSC maintenance and tumor progression. Within the TME, Treg-produced VEGF also facilitates angiogenesis to sustain CSC self-renewal (44).

Through secretion of chemokines (e.g., CCL3), CSCs recruit and polarize tumor-associated macrophages (TAMs) toward the immunosuppressive M2 phenotype, which reciprocally reinforces the CSC phenotype via activation of diverse signaling cascades (45). In breast cancer, TAM–CSC crosstalk activates NF-κB and Src pathways, promoting the release of GM-CSF, IL-6, and IL-8 to stimulate CSC regeneration (46,47).

TAM-derived IL-6 enhances CSC proliferation in hepatocellular carcinoma, while TGF-β maintains CSC self-renewal and drives epithelial–mesenchymal transition (EMT), facilitating metastatic dissemination (48,49). Pleiotrophin (PTN) secreted by TAMs further augments CSC properties via PTPRZ1(50). Upregulation of CD47 on CSCs—induced by TAMs—promotes immune evasion across malignancies including HCC, leukemia, and pancreatic cancers ([51], [52], [53], [54]).

CD47 engages signal regulatory protein α (SIRPα) on phagocytic cells to inhibit phagocytosis. TAMs further reinforce immune evasion by upregulating programmed cell death protein-1 (PD-1) on T cells and PD-L1 on CSCs, thereby suppressing cytotoxic T-cell function. The reciprocal crosstalk between CSCs and TAMs shapes an immunosuppressive TME that supports CSC maintenance and limits the therapeutic efficacy of immunotherapies (55). MDSCs, identified by markers such as CD11b+CD14–CD33+, infiltrate the tumor milieu and amplify immunosuppressive signaling (56). These cells promote CSC survival and stemness through multiple pathways, notably by inducing the expression of stemness-associated transcription factors OCT4, NANOG, and SOX2 via elevated piRNA-823 levels, and by activating STAT3/NF-κB signaling through exosomal S100A9 (57,58).

The abundance of MDSCs in tumors correlates with poor clinical outcome and diminished responsiveness to immune checkpoint blockade, particularly in melanoma ([59], [60], [61]). In pancreatic cancer, MDSCs further facilitate CSC expansion and EMT, accelerating malignant progression (62).

Natural killer (NK) cells normally target cells displaying low MHC-I levels. However, CSCs manipulate this recognition system to escape NK cell-mediated cytotoxicity. While NK cells can eliminate MHC-I-deficient CSCs through activating receptors such as NKG2D, NKp30, and NKp44, some CSC subsets restore MHC-I expression, rendering themselves resistant to NK attack (63).

CSCs also downregulate NKG2D ligands (NKG2DL), weakening NK cell activation. Although CSCs express molecules capable of engaging NK receptors, including CD24, CD44, and CD133, they counterbalance this by expressing inhibitory ligands such as NKG2A and KIR2DL4, thereby attenuating NK cell activation (64). For instance, breast CSCs downregulate the NK-activating ligands MICB and NICA through oncogenic miR-20a, evading immune elimination (65). Likewise, glioma CSCs exhibit low expression of NK receptor ligands, which can be restored by interferon-γ (IFN-γ) treatment, sensitizing CSCs to NK cell cytotoxicity in vitro (66).

CSCs can also enter a reversible dormant state through autocrine activation of the Wnt signaling pathway by Dickkopf-1 (DKK1), suppressing proliferation and reducing UL16-binding protein (ULBP) expression, a key ligand required for NK cell activation. CSCs expressing SOX2 or SOX9 secrete DKK1 to induce dormancy-associated downregulation of NKG2DL, conferring resistance to NK-mediated clearance (67). Moreover, CSCs protect themselves from cytolytic lymphocyte attack by upregulating protease inhibitor 9 (PI9), a granzyme B inhibitor that promotes resistance to both CTL- and NK cell-induced apoptosis (68).

Tumor microenvironment and cancer stem cells

Hypoxia

Hypoxia enhances CSC behavior, driving resistance to immunotherapy and chemotherapy. Many CSCs reside in tumor regions distant from blood vessels, creating a hypoxic microenvironment (69).

Under hypoxia, hypoxia-inducible factors (HIFs), particularly HIF-1α and HIF-2α, are activated and maintain CSC stemness by upregulating Nanog, Sox2, and Oct4. HIF activation also promotes aberrant angiogenesis via VEGF secretion, facilitating tumor progression (70,71).

Additionally, hypoxia modulates Notch and Wnt signaling, promoting EMT and enhancing CSC invasiveness (72).

CSCs in hypoxic conditions display resistance to DNA-damaging therapies through TGF-β signaling and ROS-mediated stress responses (73,74). Hypoxic stress, often accompanied by acidosis and nutrient deprivation, drives metabolic reprogramming toward glutamine-based fatty acid synthesis and aerobic glycolysis, allowing CSCs to adapt to nutrient-poor environments (75,76).

Activation of AKT/β-catenin and HIF-1α increases CSC populations in breast cancer under hypoxia, while HIF-1α inhibition reduces these cells in xenograft models ([77], [78], [79], [80]). In lung cancer, hypoxia induces IGF1R-mediated accumulation of gefitinib-resistant CSCs (80). Similarly, glioma CSCs exhibit enhanced activity under hypoxia, which declines when HIF-1α or HIF-2α is suppressed (81).

HIF-1α also promotes breast CSC activity through Hippo pathway induction and TAZ activation (82). In acute myeloid leukemia (AML), HIF-1 overexpression in CD34+CD38− cells reinforces stem-like traits and expands leukemia stem cell populations (83). HIF-2α supports CSC formation via c-Myc activation and OCT-4 regulation, highlighting the central role of the hypoxic tumor microenvironment in sustaining CSC features (84).

Cancer-Associated fibroblasts

Cancer-associated fibroblasts (CAFs) are a major component of the TME and comprise approximately 50 % of tumor tissue cells (85). They play a critical role in the maintenance of CSCs and enhance their function through various mechanisms. CAFs are stimulated by signaling molecules such as IL-6 and growth factors such as TGF-β, which are released by immune cells or nearby cancer cells and activate signaling pathways such as NF-κB and JAK-STAT (86,87).

CAFs promote tumor proliferation, invasion, drug resistance, and metastasis by secreting metabolites and cytokines that affect CSCs properties, including chemoresistance and self-renewal (88). Key factors secreted by CAFs, such as IL-6 and IL-8, are critical for maintaining the stem-like features of cancer cells, whereas hepatocyte growth factor (HGF) activates the MET/FRA1/HEY1 signaling pathway and sustains stemness in hepatocellular carcinoma ([89], [90], [91]).

Additionally, CAFs secrete chemokines, such as SDF-1, which interact with its receptor CXCR4 on CSCs to activate Wnt/β-catenin and PI3K/AKT signaling pathways that regulate the stem phenotype (92,93). This interaction enhances the proliferation of CD24- and CD44+ breast cancer cells (94). CAFs also contribute to the extracellular matrix (ECM) and induce EMT, facilitating CSC invasion and metastasis (95).

In gastric cancer, CAFs promote angiogenesis and EMT through the secretion of factors such as FGF, PDGF, VEGF, TGF-β, TNF-α, and IL-1β that drive tumor progression ([96], [97], [98]). Furthermore, a novel subpopulation of CAFs, termed GPR77+ and CD10+, is associated with reduced patient survival and maintenance of a CSC reservoir, enhancing self-renewal and expression of stem markers (99).

CAFs also regulate CSC metabolism by utilizing aerobic glycolysis and autophagy to support processes such as migration and cytokine secretion (100,101). In pancreatic ductal adenocarcinoma, CAFs fulfill metabolic requirements through autophagy, whereas in breast cancer, they increase glucose uptake by expressing the GLUT-1 transporter (102,103). The “reverse Warburg effect” allows CAF-derived metabolites to boost tumorigenicity, and CSCs utilize the lactate released by CAFs to promote self-renewal and metastasis (104,105).

Overall, targeting CAFs and their interactions with CSCs represents a promising strategy for improving therapeutic outcomes and patient survival in cancer treatment.

Endothelial cells

Endothelial cells (ECs) are crucial components of the TME, contributing to angiogenesis and supplying essential nutrients that support tumor growth and metastasis. ECs initiate angiogenesis by secreting pro-angiogenic factors such as HIF-1, VEGF, and SDF-1/CXCL12, which activate and recruit additional endothelial cells within the local TME (106,107).

In co-culture systems, ECs enhance CSC maintenance by providing factors that sustain stemness and self-renewal. For example, ECs secrete soluble mediators such as basic fibroblast growth factor (bFGF), which activates signaling pathways like NANOGP8 in CRC and Hedgehog in glioma (108). Under oxidative stress, CSCs may transdifferentiate into tumor endothelial-like cells by employing autophagy and the pentose phosphate pathway to fulfill their oxygen and nutrient requirements (109).

ECs also promote CSC stemness through the Notch signaling pathway. In glioblastoma (GBM), Notch inhibition reduces CD133+ CSC populations and impairs self-renewal capacity. In vivo administration of DAPT, a γ-secretase inhibitor, downregulates endothelial markers including CD31, CD146, and CD105, indicating a depletion of the endothelial-associated CSC compartment (110,111).

In gliomas, the Jagged-1–Notch axis, activated by platelet-derived growth factor–nitric oxide synthase signaling, mediates crosstalk between glioma-initiating cells (GICs) and ECs (112). Similarly, co-culturing ECs with CRC cells increases CD133+ and ALDH+ CSC populations. This effect depends on Notch signaling, which is sustained by soluble Jagged-1 released by ECs in the TME (113). Collectively, these findings highlight the key role of ECs in reinforcing CSC properties and promoting tumor growth and progression.

ECM

The ECM functions as a regulatory niche for CSCs, providing structural and biochemical cues that govern self-renewal, proliferation, and differentiation. Collagens, the predominant fibrous ECM proteins, modulate cell adhesion and migration. Several collagen types, including COL4A2, COL3A1, and COL17A1, are overexpressed in CSCs and enhance their tumor-initiating potential (114).

Type I collagen reinforces CD133+ glioma stem cell maintenance via α2β1 integrin signaling (115), while type IV collagen promotes CSC proliferation in head and neck cancer, and COL11A1 induces EMT and stemness in pancreatic ductal adenocarcinoma (PDAC) (116).

ECM glycoproteins such as laminin, fibronectin, and tenascin-C facilitate adhesion and signaling that sustain CSC traits. Fibronectin activates integrin-mediated pathways to promote EMT and self-renewal, whereas laminins support basement membrane structure. Among proteoglycans, decorin suppresses metastasis, while versican enhances drug resistance, underlining their dual regulatory functions (114).

Hyaluronic acid (HA), a major ECM polysaccharide, supports tissue hydration and modulates mechanical stress. Its interaction with CD44 receptors activates signaling cascades that strengthen CSC phenotypes, with HA accumulation correlating with increased chemoresistance in ovarian and breast cancers (114).

ECM physical properties, including stiffness and porosity, also influence CSC behavior. A rigid ECM supports symmetric division and self-renewal, whereas matrix density impedes drug penetration; for example, cisplatin binds to collagen fibers, limiting chemotherapy efficacy (117).

Mechanotransduction via integrin, YAP/TAZ, and Rho/ROCK pathways links ECM stiffness to enhanced CSC proliferation, metastasis, and drug tolerance (118). The Rho and Hippo/YAP axes are particularly important for maintaining CSC self-renewal under mechanical stress (119).

Finally, ECM-driven biochemical cues promote EMT and the dedifferentiation of tumor cells into CSCs. Collagen I activates β-catenin signaling, inducing the expression of NANOG and SOX2 (120,121). Matrix metalloproteinases (MMPs) remodel the ECM, releasing growth factors that further sustain CSC survival (114). Hypoxia intensifies such remodeling by elevating collagen deposition and HIF1α activity, fostering a microenvironment that enhances CSC maintenance and correlates with poor clinical outcomes.

Targeting cancer stem cells: limitation of conventional therapies

Conventional cancer therapies, including chemotherapy and radiotherapy, remain largely ineffective against CSCs, which drive tumor progression, recurrence, and metastasis. The quiescent or slow-cycling phenotype of CSCs enables them to evade cytotoxic agents targeting rapidly dividing cells (122,123).

Approaches to overcome this dormancy include reactivating quiescent CSCs to sensitize them to therapy or maintaining dormancy to prevent tumor regrowth. Key regulators, including promyelocytic leukemia protein (PML) and Fbxw7, mediate stem cell quiescence, and their inhibition has been shown to enhance the susceptibility of leukemia stem cells (LSCs) to chemotherapy (123). Agents such as granulocyte colony-stimulating factor (G-CSF) and interferon-α (IFN-α) can promote LSC cycling and drug sensitivity, though patient responses remain inconsistent (124). Skp2 inhibitors, which target cell cycle regulators, also show promise in restraining CSC proliferation but require further preclinical and clinical validation (125).

CSCs commonly exhibit high expression of ATP-binding cassette (ABC) transporters such as P-glycoprotein (P-gp) and ABCG2, which facilitate drug efflux and confer multidrug resistance. Anticancer drug resistance is a complex challenge resulting from intrinsic and acquired mechanisms that hinder the effectiveness of treatment. Tumor heterogeneity enables resistant subpopulations to persist through adaptive mechanisms, including altered drug transport, metabolic reprogramming, and epigenetic remodeling. Mutations in driver oncogenes such as EGFR and BCR-ABL1 impede drug binding, while deficiencies in DNA repair pathways (e.g., BRCA and MMR defects) modify chemosensitivity. Additionally, activation of anti-apoptotic proteins including BCL-2 and IAPs, under the regulation of NF-κB and STAT3, contributes to survival under treatment pressure, highlighting BH3 and SMAC mimetics as emerging therapeutic agents (126).

EMT augments invasiveness and confers therapy resistance, particularly in EGFR inhibitor-treated non-small cell lung cancer (NSCLC), where AXL and MED12 loss plays a decisive role. The tumor microenvironment further supports resistance via ECM interactions, cytokine signaling, and stromal-derived growth factors such as IL-6 and HGF. Oncogenic bypass pathways, including MET amplification and alternative kinase activation, sustain tumor viability in KRAS-mutant colorectal cancer and BRAF inhibitor-resistant melanoma. The metalloproteinase ADAM17, which activates receptor tyrosine kinases, also contributes to therapy resistance; its inhibition may enhance the efficacy of combined regimens. Despite genomic and proteomic advances identifying predictive biomarkers and potential targets, adaptive resistance remains a major clinical barrier, underscoring the need for multi-targeted and combination-based therapeutic strategies (126).

CSCs display enhanced DNA damage repair capacity, rendering them resistant to radiotherapy. In glioblastoma, CD133+ stem-like cells exhibit superior DNA repair efficiency, attenuated apoptosis, and higher tumorigenicity following ionizing radiation. These cells preferentially activate DNA damage checkpoints, promoting survival and repopulation after treatment (127). Targeting checkpoint kinases in these populations has emerged as a rational strategy to sensitize glioblastoma CSCs and overcome radioresistance.

Furthermore, CSCs rely on developmental signaling pathways—including Wnt, Hedgehog, and Notch—to maintain stemness, plasticity, and tumorigenic potential. These pathways operate within a coordinated regulatory network that enables CSC adaptation and persistence under therapeutic stress. Effective CSC-directed interventions must therefore integrate pathway crosstalk and phenotypic heterogeneity to disrupt the underlying self-renewal circuitry. Identifying critical regulatory nodes within these networks is essential for designing combination therapies aimed at overcoming CSC-mediated drug and radiation resistance (128).

Immunotherapy approaches to target cancer stem cell

Having identified CSCs as central drivers of tumor self-renewal, researchers hypothesize that selectively targeting CSCs could achieve tumor eradication (129,130). Consequently, CSCs represent attractive targets for cancer therapy. The functions of CSCs are regulated by numerous endogenous and extracellular factors, each of which may serve as a therapeutic target (Fig. 3) (131).

Fig. 3.

Strategies for targeting CSCs. CSCs are regulated by various intrinsic and extrinsic factors, each representing a potential therapeutic target. Conventional chemotherapy and radiotherapy remain largely ineffective against CSCs. Emerging strategies include immunotherapy, virotherapy, and small-molecule approaches involving nanoparticles, miRNAs, RNAi, and exosomes. Abbreviations: CSCs, cancer stem cells; RNAi, RNA interference; miRNAs, microRNAs.

Targeting cancer stem cells by T or car-t cells

T cells, particularly CD8+ cytotoxic T cells, are the principal effector cells of the adaptive immune response (132). Solid tumor cells and their CSCs express MHC-I but generally lack MHC-II, so effective CD8+ T-cell–mediated killing depends on adequate MHC-I expression and a functional antigen-presenting machinery within tumor cells. The susceptibility of CSCs to T-cell–mediated lysis varies by tumor type, cell origin, and culture conditions (133).

To elicit CD8+ T-cell responses, ALDH-high cells from human cancer cell lines (e.g., pancreatic) can be isolated by fluorescence-activated cell sorting (FACS) and co-cultured with CD8+ T cells and dendritic cells from HLA-A2–restricted healthy volunteers. In some experiments, an artificial antigen-presenting cell is used. Adoptive transfer of these activated CD8+ T cells into tumor-bearing mice has been shown to suppress tumor progression and metastasis and to prolong survival, indicating that CSCs, particularly ALDH1A1-expressing cells, are viable targets for T-cell–based immunotherapy in solid tumors (132).

γδ T cells offer a complementary, MHC-independent route to tumor killing. Activated γδ T cells, stimulated with phosphoantigens or aminobisphosphonates, can efficiently lyse CSCs in colon, ovarian, and breast cancer models ([134], [135], [136]).

TAAs shared between CSCs and non-CSCs, as well as CSC-specific TAAs linked to stemness, present challenges for immunotherapy. Targeting shared TAAs can yield transient control but risks immune escape via immunoediting. Incorporating CSC-specific antigens and concurrently targeting multiple antigens may improve durability, especially when combined with strategies that disrupt the immunosuppressive tumor microenvironment. Effective CSC-directed immunotherapy likely requires multi-antigen targeting and functional infiltration by T cells with diverse specificities (133).

Modified immune cells, including CAR-T, CAR-NK, and CAR-macrophages (CAR-M), offer another therapeutic avenue. CAR constructs typically comprise an extracellular antigen-binding domain, a spacer, a transmembrane domain, and an intracellular signaling domain. (137,138).

CAR-T cells have achieved notable success in hematologic malignancies and show promise in solid tumors, though their efficacy against CSCs across solid cancers remains variable (139). Ongoing work seeks biomarkers that distinguish CSCs from non-CSCs and to optimize CAR designs for CSC targeting (133,140).

Common CSC surface markers used to isolate CSCs across tumor types include CD133, CD44, IL-6R, CD24, EpCAM, Lgr5, CD166, and CD29, used alone or in combinations (133).

CD133, in particular, is a widely used CSC marker in glioblastoma, prostate, pancreatic, ovarian, colorectal, lung, and liver cancers (141). Therapeutic CD133-targeted approaches have been explored in several contexts, including a phase II trial in advanced cholangiocarcinoma (NCT02541370) using CD133 CAR-T cells combined with anti-EGFR therapy. Although the use of CAR-T and CAR-EGFR has been proposed for the treatment of cholangiocarcinoma (CCA) and other solid tumors, further investigation is needed to overcome the limitations related to toxicity (142).

Safety concerns for CAR-T therapy in solid tumors include cytokine release syndrome (CRS) and off-tumor toxicity, underscoring the need for careful target selection and engineering (143).

Although CAR-T strategies show promise, limitations such as toxicity, limited efficacy in solid tumors, manufacturing complexity, and cost remain. Nonetheless, leveraging well-characterized CSC markers to guide CAR-T and related therapies remains a compelling direction, pending further preclinical and clinical validation to minimize damage to normal stem cells and to translate these approaches broadly into the clinic (139,143,144).

Targeting cancer stem cells by NK or car-nk cells

Immunotherapy based on NK cells provides an alternative to overcome several limitations associated with T cell–based approaches (145). Although early studies suggested that CSCs exhibit lower immunogenicity than non-CSCs due to reduced MHC-I expression, growing evidence indicates that CSCs may actually be more susceptible to NK cell–mediated recognition and cytotoxicity (146,147).

In vitro studies have shown that CSCs from various solid tumors express ligands that activate NK cell receptors. Upregulation of activating natural cytotoxicity receptors, particularly NKp30 and NKp44, has been correlated with enhanced susceptibility of CSCs to NK cell attack (146,147). Moreover, NK cells demonstrate superior anti-CSC function compared to T cells by effectively recognizing and killing cells expressing low levels of MHC-I (148). Elevated expression of stress-induced antigens, including MICA and MICB, which serve as ligands for the activating NKG2D receptor, facilitates immune surveillance by enabling NK cells to target CSCs and other non-proliferating tumor cells. This suggests that NK-mediated cytotoxicity after depletion of proliferating non-CSCs through conventional therapies represents a promising approach for CSC elimination (149). NK cells can recognize and destroy “stem-like” cells, as evidenced by their ability to reject allogeneic hematopoietic stem cells. CSC elimination by NK cells occurs through direct cytotoxicity, NKG2D ligand recognition, and induction of CSC differentiation ([146], [147], [148]).

NK cell–based immunotherapy offers several advantages. It is not restricted by HLA type or tumor-specific antigens, unlike vaccine- or monoclonal antibody-based methods. Because NK cell activity does not depend on a defined antigen, tumor escape through antigen loss or shedding is less likely. In addition, NK cells are relatively easy to isolate and expand ex vivo, facilitating their use in both autologous and adoptive cell therapies. Their short lifespan compared to clonally expanded T cells also reduces the risk of off-tumor activity, cytokine overactivation, and the need for suicide switches (147).

Further studies are necessary to characterize the immune effects of CSC-directed NK cell activity in vivo, particularly using immunocompetent animal models. Since NK cell activation and persistence are cytokine-dependent, optimizing cytokine support is critical for achieving durable antitumor efficacy. Strategies such as antibody-mediated targeting of NK cells toward CSCs, or combining NK therapy with chemotherapy, radiotherapy, or T cell–based therapies, may further enhance therapeutic outcomes (146,148).

Clinical evidence to date suggests that autologous NK cells show limited efficacy in solid tumors, prompting growing interest in allogeneic NK cells as more effective adoptive immunotherapeutic agents (150).

Although the field remains in early development, combining NK cell therapy with conventional treatments or gene-modified NK platforms has yielded encouraging preclinical results. Nonetheless, CSCs can evade NK-mediated cytotoxicity by altering the expression of NK receptor ligands, reshaping the tumor microenvironment, and acquiring intrinsic resistance to NK cell killing (148).

CAR-NK cells represent an emerging alternative to CAR-T therapy and offer several distinct advantages (151).

Compared with CAR-T cells, CAR-NK cells show a markedly lower risk of graft-versus-host disease (GVHD) due to their allogeneic compatibility, lower production of pro-inflammatory cytokines, and reduced likelihood of CRS. They can also recognize and eliminate tumor cells with downregulated MHC-I expression—a common immune evasion mechanism. Early clinical results demonstrate a favorable safety profile and potential therapeutic efficacy of CAR-NK therapy in solid tumors (152).

Preclinical studies highlight the potential of CAR-NK platforms for targeting CSCs. Rüdiger Klapdor and colleagues developed third-generation anti-CD133 CAR-NK92 cells that effectively inhibited ovarian cancer growth (153).

Another dual-target CAR-NK, specific for CD24 and mesothelin, demonstrated simultaneous cytotoxicity against ovarian CSCs and bulk tumor cells (154). CAR-NK92 cells directed against EpCAM also significantly suppressed colorectal CSC proliferation (155). These results underscore the capacity of NK cells to selectively eradicate CSCs associated with resistance, recurrence, and metastasis (156,157). Consequently, NK cells represent a promising tool for durable cancer therapy and metastasis prevention (158).

Both autologous and allogeneic NK cell therapies have shown safety and partial efficacy in hematologic malignancies and solid tumors (159). However, responses in solid cancers remain suboptimal. Recent studies favor allogeneic NK cells derived from healthy donors due to their consistent therapeutic performance and cost-effectiveness (145). Differentiated NK cells from stem cell sources (160,161) and cytokine-activated NK cells from peripheral blood (162) are being explored as carriers for CAR constructs. Despite these advances, CAR-NK production is complex, time-intensive, and technically demanding (158). Similar to CAR-T therapies, CAR-NK efficacy is also hindered by tumor heterogeneity, limited infiltration, and the immunosuppressive tumor microenvironment (163).

Targeting cancer stem cells with car-m cells

The therapeutic efficacy of CAR-T therapy in solid tumors is significantly affected by the immunosuppressive characteristics of TME, impediments to effective tumor infiltration, the paucity of tumor-specific antigens, the phenomenon of antigen escape, and the occurrence of substantial adverse effects (164).

Targeting CSCs via chimeric antigen receptor macrophages (CAR-M) has emerged as an innovative and promising strategy in the field of cancer immunotherapy. CAR-M therapy exploits the intrinsic phagocytic and immunomodulatory properties of macrophages, which are genetically engineered to express CARs capable of recognizing CSC-associated antigens, thus facilitating the selective elimination of CSCs through both phagocytosis and cytotoxic mechanisms.

CAR-M cells possess unique advantages for targeting CSCs within the solid tumor context. Owing to their myeloid lineage, CAR-M cells inherently display superior tumor-homing and infiltration capabilities, enabling their recruitment into the TME, which is often refractory to conventional immune effector cell penetration. Upon entry into the TME, CAR-M cells can actively modulate the tumor milieu by secreting pro-inflammatory cytokines such as TNF-α and IL-12, as well as MMPs that degrade extracellular matrix components. These activities collectively facilitate immune cell infiltration and disrupt the protective niche that sustains CSCs.

In addition to these functions, CAR-M cells have demonstrated the capacity to repolarize TAMs from a pro-tumorigenic M2 phenotype towards an anti-tumorigenic M1 phenotype, thereby enhancing local anti-tumor immune responses and mitigating immunosuppressive signaling ([165], [166], [167], [168]).

Moreover, as professional antigen-presenting cells, CAR-M cells are able to process and present tumor antigens derived from CSCs to T cells, thereby potentiating adaptive immune responses. This dual mechanism of action—direct CSC elimination and comprehensive TME remodeling—underscores the transformative therapeutic potential of CAR-M approaches against CSC-driven neoplasms, which are otherwise characterized by robust immune evasion and niche protection (166,167).

CAR-M equipped with specific antibodies for antigens expressed on CSCs (such as HER2, CD47, or CD133) display improved phagocytic activity and production of pro-inflammatory cytokines, leading to effective elimination of both differentiated tumor cells and CSCs ([169], [170], [171]).

Through secretion of TNF-α, IL-1, IL-6, and IL-12, CAR-M polarize the TME toward a pro-inflammatory state, disrupt stromal barriers (e.g., through fibroblast activation protein (FAP)-targeting CAR-M constructs that deplete cancer-associated fibroblasts), and increase cytotoxic T cell infiltration. This remodeling is crucial for overcoming the protective niche often exploited by CSCs (171).

Notably, early-phase investigations of HER2-targeted CAR-M therapies have demonstrated favorable safety and preliminary efficacy in preclinical models, and are now advancing in clinical studies encompassing solid tumors with substantial CSC populations (170).

Dual- or multi-targeted CAR-M designs, such as HER2/CD47 or quadrivalent constructs, help reduce relapse by eliminating both tumor cells and the supportive microenvironment, thereby addressing heterogeneity and antigen loss—a common feature of CSC-driven recurrence (171).

New generations of CAR-M are engineered with dominant-negative TGF-β receptors, enabling these macrophages to retain their function in immunosuppressive settings typically enriched for CSCs (171).

Animal models using CAR-M have reported decreased CSC frequencies, reduced tumor growth, and improved survival in models of ovarian, gastric, and hepatocellular carcinoma ([169], [170], [171]).

Clinical trial data on CAR-M therapies specifically targeting CSCs remain limited, however, the field is progressing rapidly toward clinical application. First-in-human studies of HER2-targeted CAR-M in solid tumors reported safety and preliminary efficacy in treating relapsed/refractory malignancies, although explicit data on CSC targeting await publication (172).

Clinical trial protocols now often include CSC markers as exploratory endpoints, assessing how effectively engineered macrophages can deplete functional tumor-initiating cell populations. Despite these promising attributes, challenges remain, including macrophage phenotypic plasticity, restricted in vivo persistence, optimization of dosing regimens, and the necessity for thorough safety profiling.

In summary, CAR-M therapy directed against CSCs constitutes a compelling paradigm in contemporary cancer treatment, offering the potential for simultaneous eradication of CSCs and reprogramming of the immunosuppressive TME, with significant implications for surmounting resistance mechanisms that limit the efficacy of existing cellular immunotherapies.

Targeting cancer stem cells by antibodies

Recent progress in monoclonal antibody (mAb) development for targeting CSCs has opened new therapeutic avenues (173). CD44, a transmembrane glycoprotein that binds extracellular matrix components such as collagen, osteopontin, and hyaluronic acid, induces apoptosis and inhibits tumor growth in AML, highlighting its therapeutic potential (174). Anti-CD44 therapy remains a leading strategy against CSCs (175). The mouse IgG1 antibody H90, the first mAb shown to effectively target CSCs, binds human CD44 and triggers terminal differentiation, apoptosis, and proliferation arrest in myeloid leukemia cell lines (176).

CD133 (prominin-1), another well-established CSC surface marker, is expressed across multiple tumor types and validated as an antibody target, especially in malignancies enriched with CD133⁺ subpopulations. Other FDA-approved antibodies, including cetuximab (anti-EGFR) and rituximab (anti-CD20), can also target antigens shared with CSCs (177,178). Immune checkpoint inhibitors against CTLA-4 and PD-1/PD-L1 have achieved notable clinical responses in advanced cancers and may complement CSC-targeted therapy. However, antibody monotherapy often provides limited efficacy, suggesting that combinatorial approaches—such as integrating checkpoint blockade with other treatment modalities—could enhance antitumor immunity (174).

The CD47–SIRPα axis offers another promising target. CD47 overexpression delivers a “don’t eat me” signal that prevents phagocytic clearance by macrophages, promoting tumor survival and metastasis. Notably, CD47 is highly expressed on AML stem cells compared with normal hematopoietic stem or progenitor cells, reinforcing its relevance as a CSC-selective marker (179).

EpCAM overexpression, particularly alongside CD44, characterizes CSCs from colorectal, pancreatic, and breast carcinomas, identifying it as a universal epithelial CSC marker. The bispecific T-cell–engaging antibody MT110 (Solitomab), which binds CD3 on T cells and EpCAM on tumor cells, eliminates EpCAM⁺ CSCs in ex vivo models of human liver and breast cancers (176,180).

IL-6 signaling also plays a direct role in CSC self-renewal by modulating OCT-4 expression through the IL-6–JAK1–STAT3 axis, promoting stem-like properties in non-CSCs. Blocking IL-6R signaling thus represents a potential immunotherapeutic strategy to inhibit CSC maintenance (132,181).

Despite these advances, CSCs often exist at levels below the detection threshold of most antibody-based therapies, reducing clinical efficacy. The overlap between CSC and normal stem cell markers further complicates selective targeting, and some CSCs lack canonical markers altogether. Additionally, non-CSC-specific factors within the tumor microenvironment indirectly support CSC survival and remain potential therapeutic targets (175).

Protecting normal stem cells while eradicating CSCs remains the core challenge. Achieving this balance requires a detailed understanding of the signaling networks that distinguish CSCs from their normal counterparts. Combination strategies—such as dual antibodies or antibody–drug conjugates—show promise. Given the conservation of self-renewal and survival pathways across cancers, developing pathway-oriented rather than tumor-specific antibodies could yield broader therapeutic benefits (175).

Current antibody-based therapies under investigation target key CSC markers such as CD44, CD133, ESA/EpCAM, ALDH1, ABC transporters, CD24, CD20, IL-4, HER2, and signaling mediators including Notch, Wnt, and IL-6. Additional targets under exploration include CD123, CD200, tetraspanins (CD9), L1CAM (CD171), prostate stem cell antigen, angiogenesis-related molecules, integrins, and chemokine receptors (175).

Targeting cancer stem cells by vaccines

CSC-based immunizations—including cell-, DNA-, and mRNA-based vaccines—have shown efficacy in the targeted treatment of CSCs (182). In cell-based vaccines, the conditioned medium of pluripotent stem cells (PSCs) can mimic the embryonic niche and facilitate cancer treatment through differentiation therapy. Moreover, PSCs provide a valuable source of CSC-specific antigens. This approach shows great promise for developing both prophylactic and therapeutic cancer vaccines (183).

DNA vaccines for cancer immunotherapy deliver one or more genes encoding tumor antigens to elicit or enhance antigen-specific immune responses against key molecules involved in tumor initiation, progression, and metastasis (184). Nishizawa et al. reported that immunization with CSC-specific DNAJB8 expression plasmids induced a robust antitumor immune response (185). Thus, CSC-targeted DNA vaccination represents a promising cancer immunotherapy strategy due to its simplicity and capacity for precise antigen targeting (186). mRNA-based vaccines deliver synthetic mRNA molecules encoding cancer antigens or proteins. Once inside host cells, the mRNA is translated into proteins that activate the immune system to recognize and destroy cancer cells (187). In an in vitro study, Sumransub et al. (188) demonstrated that an mRNA-based DC vaccine using mRNA derived from CD44+/CD24– breast CSCs elicited a stronger cytotoxic T-cell response than mRNA from bulk tumor cells, underscoring its potential for enhanced cancer immunotherapy. Nonetheless, several limitations remain: (1) mRNA-based vaccines have a half-life of <24 h, limiting their stability. (2) DNA-based vaccines require integration into the genome, posing mutagenic risks. (3) Cell- and RNA-based vaccines may trigger autoimmune reactions. CircRNAs, however, may overcome some of these issues. Compared with conventional mRNA vaccines, circRNAs offer distinct advantages: (1) Their circular structure confers greater stability, allowing prolonged protein translation and effective DC sensitization from minimal circRNA amounts. (2) CircRNAs do not integrate into the genome and are translated in the cytoplasm, minimizing safety concerns. (3) CircRNAs encoding purified CSC-associated antigens can transfect DCs to minimize autoimmune risks relative to whole-cell or total RNA vaccines (182).

In 2010, Jian-Cong Sun observed that mature DCs loaded with RNA from CD133+ hepatocellular carcinoma CSCs could induce CD8+ cytotoxic T lymphocyte responses against CSCs in vitro (189).

In a malignant melanoma model, DCs pulsed with CSC lysates promoted IFN-γ and IL-4 secretion, resulting in tumor suppression and prolonged survival in vaccinated mice (190). Combining CSC-loaded DC vaccination with CTLA-4 and PD-L1 blockade enhanced melanoma stem cell eradication in a mouse model (191).

Several clinical trials have examined CSC-targeted DC vaccines in GBM. In one phase I study (NCT02010606), patients received autologous DCs extracted, processed, and administered subcutaneously to stimulate immune responses against CSCs, though some treatment-related toxicities occurred. Another phase I trial (ICT-121, NCT02049489) evaluated CD133 peptide-loaded DCs in patients with GBM; results are pending publication. A phase II study (NCT01567202) is assessing autologous DCs loaded with A2B5+ glioma stem-like cells, showing enhanced IFN-γ and p53 expression indicative of a strong antitumor immune response (192,193).

In preclinical models, DCs loaded with lysates of Panc-1 CSCs (pancreatic cancer) induced IFN-γ and IL-2 production (194), while in ovarian cancer, DCs primed with NANOG peptides stimulated specific anti-CSC T-cell responses (195). Similarly, DCs pulsed with ALDH-high CSCs from melanoma and squamous carcinoma elicited strong IgG-mediated and cytotoxic T-cell responses, effectively targeting CSCs (196,197).

Although CSC phenotypes are often plastic, their self-renewal and stemness depend on a limited set of genes that can serve as vaccine targets. Despite tumor-type diversity, overlap among CSC-related pathways suggests that vaccines targeting stem cell-associated genes could act as universal cancer vaccines for prevention and therapy (198).

However, targeting CSC-associated genes presents challenges, as many pathways are shared with normal adult stem cells, potentially inducing immune tolerance or collateral damage. Genes expressed in CSCs and ESCs—but not in adult stem cells—represent the most suitable immunogenic targets. ES-related genes, such as OCT4, are less tolerogenic but functionally essential, presenting a delicate balance between efficacy and safety (198).

Most CSC-targeting immunotherapies under clinical investigation use isolated CSCs from solid tumors as antigens for DC-based vaccines. Trials include pancreatic, nasopharyngeal, lung, liver, ovarian, colorectal, and brain cancers. One study tested DCs transfected with human telomerase reverse transcriptase (hTERT) and survivin mRNAs amplified from ovarian CSCs (NCT01334047). The antigens used in these trials are often patient-specific. Because monotherapy seldom yields durable effects, combination strategies—particularly with immune checkpoint inhibitors or chemo/radiotherapy—are being explored. Multi-antigen targeting, including CSC-specific epitopes, is essential for optimal therapeutic efficacy (133).

Oncolytic virotherapy

Oncolytic viruses (OVs), including adenoviruses, herpesviruses, reoviruses, and vaccinia viruses, represent an emerging branch of cancer therapy offering distinct advantages over conventional approaches. These include selective replication within tumor cells, induction of immunogenic cell death leading to enhanced anti-tumor immunity, delivery of eukaryotic transgenes, and a generally favorable safety profile compared with other cancer treatments (199).

OVs eliminate cancer cells through multiple mechanisms, such as direct cell lysis of both highly proliferating non-cancer stem cells (non-CSCs) and quiescent CSCs, as well as through induction of anti-tumor immune responses that destroy uninfected tumor cells, including by targeting tumor vasculature. Since CSCs drive treatment resistance and recurrence, it is significant that OVs have demonstrated potent anti-CSC effects (200,201).

By identifying and eradicating CSCs, OVs have the potential to prevent cancer relapse, offering a novel strategy that departs from conventional therapies and uniquely targets drug-resistant CSCs ([202], [203], [204]).

Genetically engineered OVs can be designed to target not only specific biomarkers on CSC surfaces but also the TME and tumor-promoting genes. Biomarkers such as CD markers that differentiate CSCs from normal stem cells can enhance OV specificity toward CSCs. Among these, CD133—a membrane protein highly expressed on several CSC subtypes—has emerged as a promising target (205).

Several OVs have been modified to express immune-stimulatory transgenes that augment selective tumor destruction. Examples include adenoviruses expressing OX-40, vaccinia viruses expressing IL-24, herpes simplex viruses expressing IL-12, and Tanapoxviruses expressing IL-2. The anti-tumor effects of OVs occur through multiple mechanisms: (i) viral replication leads to direct oncolysis; (ii) tumor vasculature is damaged, indirectly causing tumor cell death; and (iii) specific viral proteins exert cytotoxicity. Induction of IFN production within tumors is also critical to OV efficacy (206).

While the sensitivity of CSCs to viral oncolysis compared to non-tumor-initiating cancer cells remains incompletely defined, several studies confirm that OVs can effectively infect and replicate within CSC populations (207). OVs are appealing cancer therapeutics because they bypass mechanisms driving chemo- and radioresistance and can be engineered to deliver transgenes that disrupt CSC-specific pathways governing self-renewal and differentiation (199). Unlike chemotherapeutic agents—often ineffective against CSCs due to multidrug resistance, DNA repair efficiency, and quiescence—OVs can selectively infect and eliminate CSCs independent of these defenses (206).

Defective intrinsic IFN signaling facilitates selective OV replication, particularly in engineered viruses such as rhabdoviruses. OVs trigger IFN responses both through immune-cell detection and intrinsic tumor-cell sensing. Type I IFN signaling in dendritic cells is crucial for OV-mediated immune activation and tumor clearance; thus, enhancing IFN production within the tumor microenvironment can improve efficacy (207).

To augment anti-tumor effects, many OVs are armed with therapeutic genes encoding pro-apoptotic (e.g., p53, TRAIL) or immune-stimulatory molecules (e.g., IL-2, IL-12, GM-CSF) (208). However, rapid immune clearance limits their effectiveness. Nanomedicine-based strategies—such as liposomal encapsulation—can protect virions from immune recognition, preserve infectivity, and enable tumor- or CSC-targeted delivery through surface ligand modification (205,209).

Although native OVs can target malignant cells, engineered modifications are often required to prevent off-target toxicity. Despite varying susceptibility across tumor types and viral species, many OVs demonstrate potent CSC-directed activity in preclinical and clinical models (210). While concerns remain regarding normal stem-cell injury, studies increasingly support selective OV-mediated CSC eradication. Next-generation “multi-armed” OVs designed to enhance anti-tumor immunity and overcome CSC-associated resistance represent a promising future direction for targeted virotherapy (207,208,211).

Targeting cancer stem cells by other molecules

To effectively eliminate CSCs, Landen et al. explored the potential of targeting ALDH1A1 expression to sensitize resistant tumor cells to chemotherapy. Their study demonstrated that using ALDH1A1-specific siRNA encapsulated in nanoliposomal particles to suppress ALDH1A1 expression significantly increased the sensitivity of platinum- and taxane-resistant cell lines to chemotherapy and reduced tumor growth in an ovarian cancer mouse model. Tumor growth was markedly lower compared with chemotherapy alone. These preclinical findings suggest that RNA interference (RNAi)-mediated silencing of CSC-associated genes may represent a promising strategy to eradicate CSCs, thereby reducing disease recurrence and improving patient survival (212).

MicroRNAs (miRNAs) play key roles in the regulation of numerous biological processes, including cell division, apoptosis, developmental timing, stem cell self-renewal and differentiation, aging, DNA methylation, and chromatin remodeling. Yu et al. analyzed miRNA expression profiles in primary breast tumors, differentiated breast cancer cells, and self-renewing breast CSCs. Compared to non-tumorigenic cancer cells, the self-renewing, tumor-initiating cells exhibited downregulation of the let-7 miRNA. Moreover, overexpression of let-7 in vivo reduced tumorigenicity and metastatic potential in NOD/SCID mice. Since let-7 functions as a tumor suppressor and is downregulated in several malignancies, it likely serves as a crucial regulator of stem cell-like characteristics in breast CSCs and provides an attractive therapeutic target for cancer treatment (213).

In hepatocellular carcinoma, highly tumorigenic and metastatic CSCs express the cell surface marker CD44. Liposomes coated with anti-CD44 antibodies can deliver doxorubicin directly to CSCs that express this marker (105).

Similarly, exosomes coated with anti-CD44 antibodies can mediate not only targeted drug delivery but also direct cytotoxicity toward CSCs. To enhance the selectivity and efficiency of exosome-mediated CSC targeting, additional surface markers such as CD133, CD24, EpCAM, and CD200 can serve as potential targeting ligands (214).

Paclitaxel, one of the most widely used antimitotic chemotherapeutic agents, can be encapsulated into exosomes via sonication. In drug-resistant cancer cells in vitro, these exosome-loaded paclitaxel formulations exhibit cytotoxicity up to 50 times greater than that of free drug. In vivo, they markedly reduce tumor size and inhibit lung metastases in murine Lewis lung cancer models. These findings suggest that paclitaxel-loaded exosomes may effectively target and eliminate drug-resistant CSCs (214).

Beyond targeting surface markers, exosomes can also modulate key signaling pathways that regulate CSC maintenance, including Wnt, Notch, Hippo, Hedgehog, NF-κB, and TGF-β pathways. An emerging strategy involves loading exosomes with inhibitors, miRNAs, or siRNAs that interfere with these critical signaling cascades, thereby suppressing CSC self-renewal and tumorigenic potential (215).

The glycolytic phenotype of CSCs also plays a pivotal role in their metabolic adaptation. CSCs (also termed TICs) typically exhibit fewer mitochondria and rely more heavily on glycolysis than their differentiated tumor counterparts, as observed in melanoma, breast, lung, and liver cancers. This metabolic reprogramming is marked by diminished mitochondrial activity, perinuclear localization of mitochondria, reduced intracellular levels of reactive oxygen species (ROS) and ATP, and lowered mitochondrial DNA content. However, therapies that target CSC/TIC self-renewal pathways, surface markers, or metabolic enzymes may also affect non-CSC populations, underscoring the need to carefully evaluate potential off-target effects (133).

Diagnostic methods used to identify cancer stem cells

The accurate identification of CSCs has been advanced through the development of specialized diagnostic techniques that exploit their unique biomarkers and biological characteristics. Detecting and characterizing CSCs is essential for the design of effective anti-cancer therapies. Various diagnostic methods have been established to specifically recognize CSCs based on their distinct surface markers, enzymatic activities, and phenotypic traits.

Identification through surface markers

One of the most important approaches for CSC identification relies on the analysis of specific surface markers. CSCs can be distinguished from other tumor cells by the expression of unique surface antigens. Markers such as CD133, CD44, and EpCAM are frequently employed for this purpose. Flow cytometry, using fluorescently labeled antibodies against these markers, is the most common method for their detection and characterization (216). Flow cytometry enables the efficient isolation of CSCs from heterogeneous tumor cell populations and facilitates in-depth study of their biological properties and behavior (217).

Detection of aldehyde dehydrogenase activity

CSCs typically exhibit elevated ALDH activity, which correlates with their stemness features and resistance to chemotherapy (218). The ALDEFLUOR assay is a widely used method to identify cells with high ALDH activity. This assay employs a fluorescent ALDH substrate that emits fluorescence upon enzymatic conversion (219). Cells showing strong ALDH activity and CD133 positivity can be effectively detected and separated (218). This method has been successfully applied for CSC identification in multiple cancer types, including lung, colon, and breast cancers (219).

Employing the sphere formation assay

The sphere formation assay (SFA) utilizes the ability of CSCs to proliferate in non-adherent conditions, forming spheroids or tumor spheres. This robust in vitro technique is widely used to evaluate the presence and self-renewal capacity of CSCs in various tumor models, including PCa. SFA also serves as a functional assay to test the impact of conventional and experimental therapeutic agents on CSC growth dynamics. Outcomes can be directly assessed by measuring the sphere formation efficiency (SFE) across successive generations or by examining the resulting spheres using immunohistochemistry and molecular profiling (220).

Molecular imaging techniques

Molecular imaging (MI) provides a powerful in vivo approach for tracking CSC behavior, therapeutic response, and metastatic potential with high spatial resolution. Imaging CSCs within living tissues has considerable implications for improving therapeutic monitoring and radiotherapy-based interventions. Techniques such as magnetic resonance imaging (MRI), optical imaging (OI), and positron emission tomography (PET) have demonstrated substantial potential for noninvasive detection of CSCs and for assessing tumor dynamics and plasticity (221).

Notably, OI allows real-time evaluation of tumor growth and recurrence, while PET and MRI provide quantitative and anatomical insights using advanced tracers and targeted probes.

Collectively, these diagnostic approaches—including surface marker profiling, ALDH activity assays, side population analysis, SFA, molecular imaging, and genomic or epigenomic profiling—are instrumental in expanding our understanding of CSC biology. The integration of these methods enhances the precision of CSC detection, informs targeted therapy development, and contributes to improved clinical outcomes by addressing treatment resistance and preventing recurrence.

Challenges and opportunities

Isolation and identification of CSCs

The total tumor mass contains a small fraction (0.01–2 %) of CSCs that share several molecular and functional features with normal stem cells, including common transcription factors and signaling pathways. This similarity makes the isolation and identification of CSCs challenging. However, ongoing technological and methodological advances are gradually improving the ability to distinguish CSCs from their normal counterparts (222).

Biomarkers

Although biomarkers are valuable tools for identifying and targeting CSCs, their practical application remains limited due to variable expression patterns and insufficient biological characterization. A major limitation is the lack of comprehensive understanding of how specific biomarkers contribute to tumor progression and influence drug responses (223). Moreover, the heterogeneity in biomarker expression can reduce the efficacy of therapies targeting single markers, thereby contributing to drug resistance (224). These challenges underscore the need for individualized and adaptive therapeutic strategies rather than universally applied treatments.

Financial, technological, and clinical trials

The development and evaluation of novel cancer therapies are both costly and time-intensive. High drug costs, long trial durations, and the requirement for advanced technologies significantly contribute to the challenges reported in clinical studies and faced by patients. Ethical considerations and the risk of adverse effects further complicate clinical trials, slowing the introduction of new therapeutic approaches (225). Overcoming the barriers to CSC-targeted therapy requires multidimensional strategies, including increased research investment, technological innovation, and the design of tailored treatment modalities. A deeper understanding of CSC biology and its interaction with the tumor microenvironment remains essential for the development of effective CSC-directed therapies despite the remarkable progress achieved to date.

Conclusions

CSCs are partially differentiated cells that constitute a small subpopulation within tumors. They represent the most aggressive and tumor-initiating cells, driving key malignant behaviors such as uncontrolled proliferation, metastasis, multidrug resistance, resistance to radiotherapy and chemotherapy, relapse, and minimal residual disease. Due to their low immunogenicity, CSCs can evade immune surveillance even in the presence of tumor-infiltrating immune cells. Moreover, they actively shape an immunosuppressive TME by secreting factors that inhibit immune activation and promote their survival. CSCs further manipulate the TME by recruiting pro-tumor immune cells, such as Tregs and TAMs, and by shifting the balance from a pro-inflammatory to an immunosuppressive phenotype. This creates a lethal interplay between CSCs and pro-tumor immune cells. Because CSCs are primarily responsible for tumor repopulation following conventional therapies, their targeted elimination is essential for complete tumor eradication. It is believed that a combinatorial strategy involving immunotherapeutic and cytotoxic approaches, aimed at simultaneously destroying CSCs and non-CSCs, offers the most promise for achieving lasting cancer remission. Importantly, effective immunotherapies should target multiple antigens, including CSC-specific ones, and disrupt the reciprocal interactions between CSCs and immune cells to ensure comprehensive cancer eradication. Future research should focus on comprehensive characterization of CSCs and their immune evasion pathways to identify broadly applicable, specific therapeutic targets. Overcoming challenges such as intratumoral plasticity, dynamic antigen expression, and immune suppression within the TME will be critical for translating CSC-targeted immunotherapies into durable clinical responses.

Data availability

No datasets were generated or analyzed during the current study.

Ethics declarations

Ethics approval

This manuscript is a review article and does not involve a research protocol requiring approval by the relevant institutional review board or ethics committee.

Patients consent statement

Not applicable.

Funding

This study was not funded.

Permission to reproduce material from other sources

None of the figures were taken from other people’s work. The figures were prepared using BioRender software.

Clinical trial registration

Not applicable.

CRediT authorship contribution statement

Maryam Sadri: Writing – original draft. Zahra Shafaghat: Writing – original draft. Mona Roozbehani: Writing – review & editing, Writing – original draft. Maryam Dorfaki: Writing – original draft. Fatemeh Kheiri: Writing – original draft. Sahel Heidari: Writing – original draft. Ali Mahmoudi: Writing – original draft. Fatemeh Faraji: Supervision, Conceptualization.

Declaration of competing interest

All of the authors have read and approved the contents of the final version of the manuscript. The authors declare no conflict of interest.