Introduction

CSCs were firstly identified in acute myeloid leukemia (AML), where they exhibited stem cell-like and cancer cell-like properties and were found to be the sole cause of the initiation and progression of the corresponding cancer [1]. The tumorigenicity and self-renewal ability are indispensable properties of CSCs. To identify potential populations of CSCs, an important test was developed to determine the ability of CSCs to form tumors at low cell densities. This test, known as Extreme Limiting Dilution Assays (ELDA) became widely used as a gold standard to estimate active CSCs frequencies [2]. Studies have shown that as few as 100 cells exhibiting the CSCs phenotype were capable of forming tumors in mice, while tens of thousands of cells with an alternative phenotype were unable to do so [3]. Furthermore, CSCs must have the ability to sustain themselves and continue to generate cells with the same tumorigenicity and primitive tumor-forming capabilities [4]. A study provided evidence by using human cells with CD45 marker from the bone marrow of AML transplant recipients. These cells were found to have the same capacity to induce most subtypes of AML in secondary recipients, highlighting the self-renewal property of CSCs [5]. Unfortunately, CSCs are highly resistant to systemic anti-cancer therapies due to their complicated drug resistance mechanisms and active DNA repair capacity. Even after successfully resecting the primary tumor, the dormant disseminated CSCs can contribute tumor relapse and metastasis due to their long-term capacity for self-renewal [6].

Based on CSC theory, only a small subset of cells sustain tumorigenesis and contribute to cellular heterogeneity in primary tumors. These CSCs share certain properties of stem cells although they are not necessarily derived from stem cells found in normal tissues [7]. This suggests CSCs are a particular cell state that could initiate tumor growth. One widely accepted hypothesis is that cells can be transformed from more specialized, non-stem cells into stem-like cells through a process called epithelial-mesenchymal transition (EMT) [8]. During EMT, cancer cells undergo heritable phenotypic changes that are brought about by epigenetic modifications, rather than the introduction of new genetic alterations. As a result, the cancer cells lose their epithelial characteristics, such as cell–cell junctions and apical-basal polarity and gain mesenchymal features such as elongated, fibroblast-like morphology. These EMT-activated cancer cells exhibit CSC-like characteristics, including the expression of cell markers associated with stemness, as well as the ability to form tumors [9, 10]. The coexistence of both epithelial and mesenchymal characteristics allows cancer cells to survive, metastasize and colonize distal organs. Current research on CSCs focuses on understanding how these cells interact with the surrounding TME. This interaction involves various bidirectional cellular mechanisms, such as direct cell-to-cell contact, ligand-receptor interactions, and interactions with non-tumor cells that are present in the TME [11]. These interactions play a role in driving tumor progression. The chronic inflammatory and immunosuppressive TME is thought to be the primary factor contributing to EMT and the high stemness of CSCs [12]. Signals from the TME activate intracellular signaling pathways, leading to changes in biomarker expression, and promote immune evasion, which ultimately converge to maintain CSCs properties [13, 14].

Evidence suggests that conventional cancer therapies often fail to completely eliminate cancer cells that have undergone a switch to the CSC state. This switch is made possible through the activation of the EMT program, which can lead to CSC-related clinical recurrence. In light of these findings, eradicating or differentiating the CSC subpopulation appears to be a potential strategy for cancer treatment. However, despite these promising prospects, there is still a long way to go in comprehensively demonstrating the formation and development of CSCs, as well as in develo** targeted therapies to counter them. This review focuses on current status of research and development in eradicating CSCs. The strategies discussed include cell biomarker-/pathway-targeting strategies, TME-targeting strategies, immune modification strategies, and agent-induced differentiation strategies. By examining these different approaches, this review aims to provide insight into the progress made thus far in the field of CSC-targeted therapies and to highlight the potential for future advancements in this area.

CSC biomarkers, stemness-associated pathways targeting therapies

Cell markers that distinguish CSCs from normal cells have potential applications in the diagnosis, treatment, and prognosis of cancer.

CSCs biomarkers

Different biomarkers have been utilized to identify CSCs, with cancer type-specific biomarkers being well-reviewed and showed in Fig. 1 [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33]. This revealed common biomarkers for CSCs including CD44, CD133, aldehyde dehydrogenase (ALDH), and epithelial cell adhesion molecule (EpCAM).

Fig. 1
figure 1

Specific biomarkers for CSCs in different types of cancer

Full size image

CD44 expression is upregulated in cancer cell subpopulations and serves as a molecular marker of CSCs. CD44+ cells isolated from cancer patients were capable of initiating tumor growth when transplanted into immunocompromised mice and showed increased resistance to radiochemotherapy [34]. CD44 is a crucial receptor for hyaluronic acid (HA) and extracellular matrix (ECM). Binding of HA or osteopontin to CD44 results in the activation of the STAT3, Oct4-Sox2-Nanog or c-Src kinase signaling pathways, which are known to promote anti-apoptosis and chemoresistance in CSCs [106]. Furthermore, the mechanistic Target of Rapamycin (mTOR) is a downstream component of the KI3K/Akt pathway and has been reported to be involved in programmed cell death protein 1 (PD-1) expression through Hh signaling cascade, independent of SMO [107, 108].

Abnormal activation of Wnt, Notch, Hh, and Hippo pathways has been shown to facilitate the maintenance of CSC properties and promote tumor progression. Moreover, these pathways interact with other oncogenic cascades, such as RAF/MEK/ERK, PI3K/Akt, TGF-β, EGFR, STAT3, and NF-κB to further enhance the tumorigenicity of CSCs. Therefore, targeting the WNT, Notch, Hh, and Hippo pathways and related oncogenic pathways is crucial for the eradication of CSCs and the prevention of tumor progression.

Biomarker- and pathway-targeting strategies

Biomarker-targeting strategies

Selective targeting of CSCs is an effective strategy to inhibit cancer progression and reduce risk of tumor relapse. Clinical trials have shown that suppressing the expression of CSCs biomarkers can reduce CSCs stemness [109,110,111,112,113,114,115,116,117,118] (Table 1). The interaction between CD44 and HA is associated with poor prognosis, and CD44 has been shown to be an important biomarker for CSCs. RG7356, for example, is a recombinant anti-CD44 immunoglobulin G1 humanized monoclonal antibody. It binds specifically to the HA-binding region of the extracellular domain of all CD44 heterodimers, to inhibit the interaction between HA and CD44, and has exhibited growth inhibition effect on several tumors in vitro [109]. A randomized phase II trial of HA-irinotecan on CD44-expressing SCLC showed significant clinical benefit in patients, suggesting that delivery of chemotherapeutic agents to activated CD44, thereby abrogating the interaction of CD44 with HA, is a compelling approach to cancer treatment [110]. CD133 has been reported to play a role in tumor spread and is considered a good candidate for targeting CSCs. A phase II clinical study provides preliminary evidence that CART-133 cells have antitumor activity with a manageable safety profile in advanced HCC [118]. High expression of EpCAM has been observed on CSCs and targeting EpCAM is an effective strategy for cancer treatment. A phase I clinical trial indicated that VB4-845 successful blocks tumor growth in patients with high EpCAM expressing non-muscle-invasive bladder cancer [115]. However, although in vitro experiments have shown significant anticancer effects, clinical trials of adecatumumab, huKS-IL-2, and catumaxomab targeting EpCAM exhibited limited clinical benefits for cancer patients [114, 116, 117]. ALDHs have been evaluated as potential prognostic markers of cancer. In a clinical trial of curcumin and curcumin combined with 5-fluorouracil/oxaliplatin in patients with colorectal liver metastases (CRLM), curcumin alone and in combination significantly reduced the number of spheroids and ALDH-active cells. Curcumin enhanced anti-proliferation and apoptosis effects of 5-fluorouracil/oxaliplatin and reduced the expression of stem cell associated markers ALDH and CD133 [111]. Nevertheless, no clinical benefits were observed in phase II trials of paclitaxel (in combination with reparixin) and disulfiram [112, 113]. In summary, targeting CSC biomarkers to eradicate CSCs selectively is expected to be an effective strategy to inhibit cancer progression and reduce the risk of tumor relapse.

Table 1 Clinical trials of different strategies for eradicating CSCs
Full size table

Pathway-targeting strategies

Developmental signaling pathways that regulate the maintenance and survival of CSCs are potential targets to eradicate CSCs, such as Wnt, Notch, Hh, and Hippo (Table 1, Fig. 3).

Fig. 3
figure 3

Illustration of abnormal pathways and potential targets in CSCs. FZD antagonists target either the Wnt proteins or FZD receptor complexes to inhibit the ligand-receptor interactions in both canonical and non-canonical Wnt pathway. DVL inhibitors block the DVL-PDZ interaction, resulting in subsequently inhibition of the signal transduction pathway. Tankyrase inhibitors stabilize Axin via inhibition of its proteasomal degradation, conversely resulting in increased degradation of β-catenin. CK1 agonists selectively potentiate CK1 kinase activity and stabilize the β-catenin destruction complex that decreasing Wnt signaling. β-catenin/TCF regulators inhibit Wnt-mediated transcriptional activity. LRP5/6 inhibitors competitively bind to the LRP5/6-sclerostin complex thus reverse the activation of Wnt/β-catenin signaling. ROR1-inhibitors ameliorate the access activated Wnt-signaling-mediated cancer cell proliferation, invasion, and therapy resistance. Potential anticancer therapeutic agents targeting the Notch pathway include targeting Notch ligands or receptors, inhibitors of the γ-secretase complex, and inhibitors of NICD-interacting transcriptional complex. SMO inhibitors block the Hh signaling by cyclopamine-competitively binding to SMO. GLI inhibitors prevent the transportation of GLI protein to nucleus thus decreased tumorigenesis gene expression. Inhibition of STAT3 and PI3K blocks their interactions with self-renewal pathways to facilitate CSCs eradication

Full size image

Inhibition of Wnt signaling involves several critical steps, including blocking the secretion of Wnt ligands, interference with ligand-receptor binding, and modulation of intracellular signal transduction. Targeting these steps is crucial to inhibit Wnt signaling and its effects on CSCs. Porcupine inhibitors, which block the secretion of Wnt ligands, hold promise as ideal drugs for eliminating proliferative CSCs induced by canonical Wnt signaling and dormant CSCs induced by non-canonical Wnt signaling. Clinically, porcupine inhibitors have shown their therapeutic potential in various tumors. Currently, only 4 molecules, namely LGK974, ETC159, CGX1321, and RXC004, are undergoing Phase I clinical trials [119, 190]. However, the clinical use of these inhibitors may have a relatively narrow therapeutic window due to the involvement of the Wnt signaling cascade in the homeostasis of key organs and tissue [191]. Interfering with the binding of Wnt ligands to their receptors is another effective strategy to inhibit Wnt signaling. Wnt/FZD antagonists such as ipafricept (IPA; OMP54F28) and vantictumab (OMP-18R5) can directly bind to Wnt ligands or FZD receptors, competing with Wnt ligands for binding to FZD receptors [120, 121]. This competition inhibits Wnt regulatory processes in both the canonical and non-canonical Wnt pathways [192, 193]. Alternatively, blocking intracellular signal transduction may help to inhibit Wnt signaling. DVL inhibitors disrupt the interaction between DVL and PDZ, resulting in the subsequent inhibition of the signaling pathway [194, 195]. In canonical Wnt signaling, the LRP5/6 inhibitor BMD4503-2 competitively binds to the LRP5/6-sclerostin complex, thereby reversing Wnt/β-catenin pathway activation [196]. Stabilization of the β-catenin structural complex prevents the localization of β-catenin in the nucleus, making it an attractive therapeutic target. Tankyrases inhibitor, E7449, regulates the stability of AXIN by directing its ubiquitylation and proteasomal degradation, thereby increasing the activity of the destruction complex and reducing free β-catenin [122, 197]. CK1 agonists selectively potentiate CK1 kinase activity and stabilize the β-catenin destruction complex that decreasing Wnt signaling [198]. Finally, targeting the downstream effectors, such as β-catenin/TCF and β-catenin-dependent transcriptional activators, is another feasible strategy to inhibit Wnt-mediated transcriptional activity [199, 200]. Previous studies have suggested that compounds like CWP232291 and acylhydrazones block or disrupt the interaction between β-catenin and the TCF complex, suppressing Wnt target genes [123, 124]. In non-canonical Wnt signaling, ROR1-inhibitors ameliorate the access of a ctivated Wnt-signaling-mediated cancer cell proliferation, invasion, and therapy resistance [201]. Cirmtuzumab has shown obvious inhibitory effects on ROR1 expression and tumor progression in chronic lymphocytic leukemia patients [125].

Various potential anticancer therapeutics targeting the Notch pathway have been explored to eliminate CSCs including inhibiting the release of Notch ligands, blocking the proteolytic cleavage of Notch receptors, interrupting the Notch signaling transduction and inhibiting the expression of target genes associated with Notch signaling. Targeting Notch ligands or receptors can inhibit aberrant signaling initiation and decrease tumorigenesis in CSCs [135,136,137,138, 202]. Phase I Clinical trials have demonstrated the antitumor activity of DLL4-targeted agents, such as demcizumab and enotizumab, either as monotherapy or in combination therapy [134, 136]. Brontictuzumab, a Notch 1 inhibitor, has shown efficacy signal in patients with Notch 1 activation and combination therapy with other anticancer agents has improved clinical benefits [137]. However, Notch 2/3 receptor inhibitor tarextumab had limited additional effects when combined with gemcitabine and nab-paclitaxel in untreated advanced pancreatic adenocarcinoma [138]. ADAMs-catalyzed S2 cleavage occurs in the ligand-receptor binding domain, which mediates the release of the ectodermal structural domain and regulates the rate of Notch signaling [203]. Notch signaling from extracellular to intracellular relies heavily on the γ-secretase complex-mediated final cleavage. γ-secretase inhibitors (GSIs) block the S3 cleavage of Notch receptors, preventing the release of NICO and subsequent activation of Notch signaling [204]. Several GSIs including PF-03084014, OMP-59R5, BMS-986115, RO2929097, MK-0752, RO4929097, and LY900009, have been investigated in clinical trials [127,128,129,130,131,132,133]. However, dose-limiting toxicities have been observed with most GSIs. A potentially superior option to GSIs is the use of γ-secretase modulators (GSMs). GSMs modify the catalytic activity of γ-secretase rather than non-selectively inhibiting it, thereby preserving some Notch signaling function and theoretically reducing side effects [205]. Activating target gene transcription is the final step in Notch signaling transduction. Disrupting the Notch transcriptional complex downstream of abnormal Notch activation has advantages in repressing the expression of Notch targeted genes [206]. Although inhibition of the Notch pathway has shown significant anti-tumor efficacy in preclinical research, these results have not been consistently identified in clinical trials.

Inhibiting the expression of Hh ligands, SMO and GLI transcription factors is thought to be an effective way to suppress the over activation of the Hh signaling in CSCs. Clinical trials are currently underway for agents targeting these components [152, 153]. EGF/EGFR interact with the MEK-ERK and AKT-PI3K signaling pathways, which leads to cancer cell proliferation [227]. Afatinib or olmutinib in combination with conventional chemotherapeutic agents improve CSC eradicating efficacy by inhibiting EGFR tyrosine kinase and ATP-binding cassette subfamily G member 2 (ABCG2) [228, 229]. VEGF/VEGFR interaction involves the activation of downstream pathways, including Ras-Raf-MAPK, AKT-mTOR, and Scr-FAK. Activation of these pathways promotes cell survival, proliferation, migration, and differentiation [227]. A randomized clinical trial indicated that fruquintinib, a VEGFR inhibitor, increased overall survival (OS) in patients with metastatic CRC [155]. Ramucirumab increased the sensitivity of cancer cells to immune checkpoint inhibitors (ICIs) and improved OS by inhibiting VEGF/VEGFR [156]. HGF from TME has been shown to regulate the innate resistance of BRAF-mutant cancer cells to RAF inhibitors by activating the MAPK and PI3K/Akt signaling pathways in cancer cells [230]. A phase I/II clinical trial evaluating rilotumumab, an anti-HGF antibody, in combination with erlotinib in patients with metastatic NSCLC showed that the combination of rilotumumab and erlotinib was more effective than erlotinib alone [157]. Suppressing TME-induced cytokines production can help restore antitumor immunity and sensitize tumors to immunotherapy. IL-6 plays a critical role in inflammation and cancer development. It has been reported that high levels of IL-6 confer resistance to cisplatin in patients with non-small-cell lung cancer (NSCLC) [231]. However, siltuximab, an anti-IL-6 monoclonal antibody, showed limited efficacy in patients with advanced solid tumors [158].

CSCs and immunoevasion

Tumor immunosurveillance is orchestrated by tumor immunogenicity and immunoevasion, immune cell infiltration, and T cell checkpoints [232]. Stromal cells and recruited immune cells in TME collectively formed an immunosuppressive niche that facilitates the maintenance and proliferation of CSCs. Furthermore, dysregulated cellular antigen processing and presentation machinery, as well as upregulated expression of immune checkpoint molecules allow CSCs to escape from immune surveillance (Fig. 5).

Fig. 5
figure 5

DC and T cell mediated tumor immunology and immunotherapy. In MHC-I antigen presentation pathway, oligopeptides degraded from cytosolic and nuclear protein are taken up and translocated into ER by TAP and further trimmed by ERAPs. The modified peptides bind to MHC-I and are transported to the cell surface for exposure to CD8 + T cells [233]. DC produce CXCL9, CXCL10 and IL-2 to recruit effective T cells therefore increase immune response. Activation of CTLA-4 and PD1/PDL1 reprogrammed immune homeostasis and induced cancer cells to eliminate T cell function. Immunotherapies that inhibit CTLA-4 and PD1/PDL1 and enhance MHC-I expression can effectively modulate DC function and improve cancer immunotherapy

Full size image

CD8+ T lymphocytes play an indispensable role in regulating adaptive immune responses. They detect antigenic peptides bound to MHC-I perceptively and efficiently eliminate abnormal cells, thereby preventing cancer cell colonization. Stimulation with interferon-gamma (IFN-γ) can upregulate the expression of MHC-I antigen presentation components [233]. While these compositions (TAP, ERAPs, IFN-γ) are not strictly required for cell proliferation, their loss results in a reduction of pathway function and decreased cell surface levels of MHC-I molecules [234]. The presentation of cancer-associated peptide antigen by MHC-I is an important step for antitumor CD8 + T cell responses. However, CSCs have mapped out strategies to reduce antigen presentation thus to escape immune recognition, including inhibition of DC function and downregulation of MHC-I expression [235]. A growing number of studies demonstrated that CSCs alter DC phenotypes and impair their recruitment to limit them to activate T cells [236, 237]. On the other hand, activation of immune checkpoint pathways in the TME, including CTLA-4 and PD1/PDL1, reprograms immune homeostasis and enables cancer cells to evade immune attack [238]. ICIs that block CTLA-4 and PD1/PDL1 have demonstrated promising therapeutic efficacy in various human cancers. However, the effective antitumor responses of ICIs often require the secretion of IL-12 by DCs [239]. Additionally, ICIs induce, instead of relieve, T cell dysfunction in situation of low MHC-I levels [240]. In this basis, enhancement of antigen processing and presentation machinery and/or combination with ICIs may be an attractive strategy for CSCs eradication.

Type 1 conventional DCs (cDC1s) are a major subset of DCs that respond to invasive pathogens and antitumor immunity via antigen presentation to cytotoxic CD8+ T cells [241]. The cDC1s facilitate the differentiation and recruitment of tissue-resident memory CD8+ T cells within the TME by producing CXCL9 and CXCL10 to activate STING pathway. In addition, cDC1s-derived IL-12 increases the sensitivity of cancer cells to ICIs by augmenting CD8+ T cell activation. Adequate CD8+ T cell activation, in turn, induces the maturation and migration of cDC1s to the draining lymph nodes [235]. Therefore, DCs serve as a potent tool for motivating antitumor responses to eradicate CSCs effectively. DC vaccines are currently under clinical investigation and have shown promising therapeutic modality. DC vaccines, including Neo-DCVac, autologous DCs in combination with doxorubicin and cyclophosphamide (NAC-AC) or toll-like receptor agonist have demonstrated efficacy in restraining tumor progression by promoting T cell mediated immunity and sensitizing cancer cells to ICIs (Table 1) [159,160,161,162].

The binding of MHC-I with specific peptides and their presentation on the cell surface is an indispensable step in antitumor immunity. Loss expression of MHC-I renders CSCs invisible to the immune system. Clinical evidence has confirmed that advanced melanoma patients with low levels of MHC-I on the cancer cell surface derived limited benefit from ICIs therapy [242]. The expression level of MHC-I is dependent on NOD-like receptor family CARD domain containing 5 (NLRC5), which is regulated by IFNγ-activated STAT1 signaling [243]. Yet no clinical trial records regarding the regulation of MHC-I by NLRC5 in the PubMed database. The downregulated expression of MHC-I is associated with repressive histone modifications of Lys-27 in histone 3 (H3K27m3), including hypermethylation, histone deacetylation and trimethylation. These modifications are partly regulated by enhancer of zeste homolog 2 (EZH2). Accordingly, inhibition of DNA Methyltransferases (DNMTi), Histone deacetylases (HDACi), and EZH2 (EZH2i) theoretically has the potential to increase the expression of MHC-I [244]. The efficacy of these drugs, either as monotherapy or in combination with other treatments, has been evaluated in clinical trials. Some of these trials have demonstrated improved sensitivity of cancer cells to drug administration (Table 1) [163,164,165,166,167,168,169]. For instance, SHR2554, an EZH2 inhibitor, has shown promising antitumor activity in patients with relapsed or refractory follicular lymphoma, peripheral T-cell lymphoma, and classical Hodgkin lymphoma [171]. EZH2 inhibitor tazemetostat showed encouraging efficacy in patients with R/R EZH2 mutation-positive follicular lymphoma with a manageable safety profile in the overall population [169]. HDAC inhibitors including abexinostat, panobinostat, and entinostat have shown antitumor efficacy in clinical trials [166,167,168]. However, DNMT inhibitors, such as decitabine, azacitidine, and guadecitabine, have shown limited clinical activities on tested cancers [163,164,165].

T cells are activated through a complex interplay involving antigen specific T cell receptor recognition of peptides presented by MHC molecules and interactions between membrane proteins on antigen-presenting cells (APCs, including CD80 and CD86) and CD28 on T cells [245]. CTLA-4 inhibits the activation of T cells by competing with CD28 for binding to CD80 and CD86 on APCs, thereby attenuating CD8+ T cell responses. Another important immune checkpoint pathway involves the PD-1 and its ligand PD-L1. The expression of PD-1 on T cells and PD-L1 on tumor cells or other immune cells is considered a hallmark of T cell dysfunction and promotes T cell exhaustion [246]. Blockade of the expression of CTLA-4 and the interaction between PD-1 and PD-L1, has been shown to restore T cell function and enhance antitumor responses during chronic infections and in the TME [247]. ICIs specifically target these immune checkpoint molecules and modify the immune environment, have demonstrated the effectiveness of ICIs in various cancers (Table 1) [172,173,174,175,176,177,178,179]. Nivolumab and pembrolizumab, the PD1 inhibitors, have shown longer OS/FPS in multiple cancers compared to chemotherapy in clinical trials [172, 173]. Balstilimab (PD-1 inhibitor) and zalifrelimab (CTLA-4 inhibitor) are checkpoint inhibitors emerging as promising investigational agents for the treatment of advanced cervical cancer. A phase II clinical trial indicated that the combination of balstilimab and zalifrelimab had a high proportion of complete responses and efficacy in patients with recurrent and/or metastatic cervical cancer [174]. Tremelimumab is a fully monoclonal antibody that binds to CTLA-4 on the surface of activated T cells, which triggered the accumulation of intratumoral CD8 + cells in patients with advanced HCC [175]. PD-L1 inhibitors, such as atezolizumab, pembrolizumab, tiragolumab, improve objective response rates and are associated with significantly longer PFS and OS [177,178,179]. However, LY3415244, a TIM-3/PD-L1 inhibitor designed to overcome primary and acquired anti-PD-(L)1 resistance, was terminated early due to unexpected immunogenicity [176].

Agent-induced CSC differentiation

CSC-induced poorly differentiated cancers are more malignant, differentiation therapy is therefore a strategy to inhibit tumorigenesis by inducing the conversion of highly malignant undifferentiated cancer cells into benign differentiated cells. Several agents, including retinoic acid (RA), cAMP, sodium butyrate and cytokines, have been proved to induce cell differentiation in specific types of cancer. ATRA induces terminal differentiation and exhibits significant anticancer effect in patients with AML and acute promyelocytic leukemia (APL) [248]. Ongoing research is exploring the potential of ATRA and its derivatives for differentiation therapy in non-AML/APL [180, 181, 183]. New differentiation-inducing drugs that inhibit mutant isocitrate dehydrogenase (IDH) 1 and IDH2 have shown differentiating potential clinically, which have been approved for AML therapy. Mutations in IDH1 and IDH2 render their functional activity in differentiation regulation, particularly by increasing histone and DNA methylation [249]. Furthermore, epigenetic regulatory inhibitors such as DNMTi and HDACi, which promote MHC-I expression, have also demonstrated their involvement in cell differentiation. IDH1/2 mutations have been observed in solid tumors, and clinical trials are underway to evaluate the effectiveness of IDH1/2 inhibitors in inducing CSC differentiation in solid tumors. Continued research in differentiation therapy holds promise for expanding its application across various cancer types (Table 1) [184,185,186,187,188,189]. IDH1-Vaccine or IDH1-targeting agents including ivosidenib and olutasidenib showed clinical benefit in cancer patients in terms of reduced tumor burden and increased PFS [184,185,186]. Vorasidenib, a dual IDH1/2 inhibitor, showed preliminary antitumor efficacy in patients with recurrent or progressive nonenhancing mIDH lower grade gliomas [187].

Other strategies of target CSCs

Cancer is a complex disease that affects the health of people around the world. Chemotherapy remains an important treatment for cancer, despite advances in surgery and radiotherapy. Current treatments are expensive and are associated with many adverse events. In addition, cancer cells become resistant to chemotherapy as treatment progresses, making it difficult for patients to benefit from unmodified chemotherapy. The development of new drugs remains a top priority, but with the financial burden of drug research, drug repurposing is an innovative way to update the chemotherapy arsenal [250]. Metformin is the first choice for treating type 2 diabetes because of its robust glucose-lowering effects, well-established safety profile and relatively low cost [251]. However, metformin is repurposed as an anti-cancer agent. A clinical trial indicated that metformin has showed anticancer efficacy by inhibiting CD133 [252]. In addition, a phase II clinical trial showed that metformin treatment resulted in a significant reduction in the CSC population and alteration of DNA methylation of chemoresistance carcinoma-associated mesenchymal stem cells (CA-MSCs), which eliminated CA-MSC–driven increases in chemoresistance [253]. The drug repurposing strategy provides an alternative for cancer treatment.

Conclusion and future perspectives

Despite ongoing debates regarding the origin and specific characteristics of CSCs, it is widely acknowledged that these cells have exhibited stemness properties such as self-renewal, proliferation, differentiation, and therapy resistance. Therefore, targeting CSCs with different therapeutic agents holds great promise for future antitumor treatments. Currently, most cancer therapies only control the growth and proliferation of CSCs instead of completely eradicating the tumor bulk. In this context, researchers are exploring the modulation of abnormal signaling pathways, inhibition of CSC specific proteins, and regulation of the immune environment to gain new insights into cancer treatment strategies. However, several challenges remain a significant hurdle, as it is crucial to specifically target CSCs while minimizing damage to normal cells. Based on the plasticity and heterogeneity of CSCs, precision oncology will be the future trend of tumor therapy. Selecting the right combination of drugs for each patient and using them at the right stage of the disease require a comprehensive understanding of the biomarkers, stemness-associated pathways, TME, and immune mechanisms in CSCs. Additionally, efforts are ongoing to mitigate adverse effects associated with treatment, and explore innovative approaches for delivering therapeutic agents and maintaining effective drug concentrations. Continued research and development in these areas hold the potential to revolutionize cancer treatment by specifically targeting CSCs, overcoming therapy resistance, and achieving more comprehensive and durable therapeutic outcomes.