Toward precision oncology: deciphering the circRNA-EMT axis in cancer and its therapeutic implications.

1.
Introduction
This review aims to systematically summarize the current understanding of how Circular RNAs (circRNAs) regulate epithelial-mesenchymal transition (EMT) pathways in cancer and explore their translational potential.
Tumour metastasis is a multi-step dynamic process in which malignant cells detach from the primary tumour, achieve local invasion by degrading the extracellular matrix through proteasomes and other mechanisms [1], enter the circulation and survive at distant sites, and ultimately colonize and grow in distant organs by acquiring induced stem cell-like properties, forming deadly secondary tumours [2]. Cancer metastasis remains the primary cause of cancer-related mortality, accounting for approximately 90% of solid tumour deaths [3]. A critical driver of this lethal process is the EMT, a reversible developmental programme that enables stationary epithelial cells to acquire migratory and invasive mesenchymal properties [4].
EMT is not merely a morphological change but profoundly impacts cancer cell behaviour [5,6]. The cell stemness and enhanced aggressiveness conferred by EMT enable cancer cells to proliferate more aggressively, survive under varied stress conditions, and evade external pressures [7]. Clinical evidence demonstrates that patients whose tumours exhibit activated EMT pathways have a considerably higher mortality rate, accounting for approximately 90% of all cancer deaths [8], underscoring the critical importance of investigating EMT in relation to poor prognosis and cancer progression. Consequently, a detailed dissection of the molecular mechanisms that govern EMT is of paramount importance for deciphering the metastatic cascade and for informing clinical prognostication in cancer therapy.
Under the current pathological classification system, EMT can be divided into three distinct types [9]. They play different roles in embryonic development, tissue repair, and cancer [10]. The EMT phenomenon is a complex biological process involving intricate networks of inducers, core regulators, and effectors that interact in a temporally coordinated manner [11]. EMT inducers mainly include transforming growth factor-β (TGF-β), Wnt/β-catenin, PI3K/AKT, and other signalling pathways. These inducers activate the expression of core EMT regulators, which in turn execute the regulatory programme [12]. The core regulators of EMT include three major families of EMT-activated transcription factors (EMT-TFs): the Snail family [13], the Zeb family [14], and the Twist family [15]. Other EMT-TFs mainly include c-Myc, FOXC2, and HIF-1α. These EMT-TFs regulate the expression of effector molecules through epigenetic mechanisms [14]. After experiencing EMT, cancer cells lose proper target recognition and activate self-sufficient growth signals to achieve metastasis while avoiding apoptosis [16].
Meanwhile, the non-coding genome, particularly circRNAs, has emerged as a key player in post-transcriptional regulation [17]. Unlike linear RNAs, circRNAs form a covalently closed loop structure, conferring high stability and making them ideal regulatory molecules [18]. Current research techniques indicated that circRNAs play widespread and pivotal roles in organisms [19] and exhibit specificity in various tissues and diseases [20,21]. Owing to their characteristic properties of high stability and specificity, circRNAs are increasingly recognized as valuable biomarker targets for assessing disease progression [22].
Previously, EMT research focused predominantly on protein-coding genes and their transcriptional regulators [23]. However, with the development of high-throughput detection technology, existing researches reveal dynamic alterations in circRNA expression during EMT [24], positioning them as novel regulators within the established EMT signalling networks.
Numerous studies have indicated that circRNAs are pervasively dysregulated across a spectrum of human diseases, with their aberrant expression patterns being particularly prominent in malignant tumours [25]. This widespread dysregulation suggests that circRNAs may play a critical role in disease pathogenesis and progression. Notably, the EMT – a fundamental biological process driving cancer invasion and metastasis – has been found to be functionally linked to circRNA expression [26]. High-throughput RNA sequencing analyses investigating the role of circRNAs in EMT have revealed that the expression levels of hundreds of circRNAs are significantly altered during EMT, with the vast majority being specifically upregulated [27]. This finding strongly implies that EMT may represent a crucial pathway through which circRNAs govern cancer development.
Existing research demonstrates that circRNAs indirectly regulate EMT through complex molecular interactions [28]. It is well-established that multiple oncogenic signalling pathwayscan undergo aberrant activation or functional inactivation due to dysregulated miRNA activity, thereby acting as triggers for EMT [29]. Given that circRNAs are widely recognized as competing endogenous RNAs (ceRNAs) that function as efficient ‘molecular sponges’ for miRNAs [30], they indirectly modulate the excessive promotion or suppression of EMT-related genes targeted by these miRNAs through this mechanism [31]. Consequently, circRNAs exert a profound influence on cancer cell migration, invasion, and distant metastatic capability.
CircRNAs, characterized by their notable stability and resistance to degradation, have recently emerged as a prominent area of investigation in cancer research [32]. Current studies in the field of circRNAs in cancer therapeutics are primarily centred around three key directions: (1) serving as therapeutic targets; (2) modulating the tumour microenvironment and immune responses; and (3) functioning as diagnostic and prognostic biomarkers [33]. With the deepening understanding of circRNA biology, research on circRNAs in cancer treatment has progressively advanced from fundamental mechanistic exploration to clinical translation, thereby paving novel pathways for the precise diagnosis and targeted therapy of cancers [34].
This review systematically summarizes the molecular mechanisms by which circRNAs regulate EMT in cancer, with a focus on elucidating their modulatory roles via key signalling pathways – including TGF-β, Wnt/β-catenin, and PI3K/AKT – as well as EMTrelated transcription factors. It further evaluates the potential of circRNAs as highly stable, tissuespecific liquid biopsy biomarkers for cancer diagnosis and prognosis, and examines their feasibility as therapeutic targets along with current circRNAbased interventional strategies. Finally, the review highlights existing challenges in research models, detection methods, and clinical translation, and proposes that technological innovation, standardization, and intelligent design are essential to advance circRNA research towards precision oncology, thereby consolidating the foundation for developing novel diagnostic and therapeutic approaches. (Figure 1)

2.
Overview of EMT
2.1.
Molecular characteristics and types of EMT
EMT is a reversible embryonic cell process in which epithelial cells lose their unique characteristics such as apical polarity, epithelial markers, and intercellular junctions. During this process, cells undergo cytoskeletal structural reorganization and dedifferentiation, returning to a mesenchymal phenotype with enhanced migration and invasion capabilities [35]. This transformation is characterized by coordinated molecular changes: epithelial markers including E-cadherin, claudin, occludin, and cytokeratin are downregulated, while mesenchymal markers such as fibronectin, vimentin, integrin-β6, and N-cadherin are upregulated.
Under the existing pathological classification system, EMT can be divided into three types based on biological context and functional outcomes [9]. Type I EMT plays important physiological roles in embryonic development and organogenesis, such as gastrulation and outward migration of cells diverging from neural ridges, contributing to proper tissue patterning during development. Type II EMT plays important roles in wound healing and tissue repair, such as inducing cell migration and growth. However, when dysregulated, Type II EMT can lead to pathological organ fibrosis in tissues such as liver, kidney, and lung. Type III EMT affects the occurrence and progression of various diseases, most notably cancer. Through Type III EMT, epithelial cancer cells transform into mesenchymal cancer cells and metastasize to distant organs of the human body, representing a critical step in cancer progression and a major clinical challenge.

2.2.
The regulatory cascade of EMT
EMT is a complex process coordinately driven by inducers, core regulators, and effector molecules. EMT inducers, including TGF β, Wnt/β catenin, and PI3K/AKT signalling pathways, initiate the programme and converge on core EMT TFs [36]. These master regulators are categorized into three major families: the Snail family [9], the Zeb family [14], and the Twist family. These EMT TFs regulate the expression of downstream effector molecules through multiple mechanisms, such as direct transcriptional control and epigenetic modifications. The functional outcomes of EMT are executed by effector molecules, characterized by downregulation of epithelial markers and upregulation of mesenchymal markers [37]. These alterations lead to fundamental changes in cell morphology, adhesion, and motility. In cancer, EMT enables loss of targeting accuracy, activation of autonomous growth signalling, metastasis, and evasion of apoptosis [16]. The resulting enhanced survival and migratory capacity facilitate dissemination from primary tumours and the formation of lethal distant metastases.
Beyond direct regulation, circRNAs also indirectly influence EMT and metastasis by modulating interconnected cell fate-determining processes such as autophagy, necroptosis, and pyroptosis [35,38]. Autophagy and EMT engage in extensive mutual regulation, where autophagy-related proteins (e.g. ATG5 [39], ATG12 [40]) promote EMT, and EMT transcription factors like ZEB1 can enhance autophagic activity [41], forming a metastasis-driving feedback loop. CircRNAs can alter autophagic flux by sponging miRNAs or binding RBPs, thereby regulating the EMT phenotype [42]. Similarly, circRNAs can regulate key executors of necroptosis or pyroptosis [43], which are programmed pro-inflammatory cell death forms that remodel the tumour immune microenvironment through cytokine release [44–46]. This remodelling can indirectly induce EMT in neighbouring cells via pathways such as NF-κB [47], suggesting a circRNA-mediated axis of ‘regulating cell death modalities —— influencing the tumour microenvironment —— indirectly modulating EMT’.

2.3.
The association between EMT and cancer stem cells
The EMT is not only a critical step for cancer cells to acquire migratory and invasive capabilities but also a core biological process that remodels cell fate and confers stem-like properties [37]. Cancer stem cells (CSCs), also known as tumour-initiating cells, represent a small subpopulation within tumours that possess self-renewal, multipotent differentiation potential, and high tumorigenic capacity. They are considered the cellular root of tumour initiation, metastasis, recurrence, and therapy resistance [48].
The EMT programme and CSC characteristics exhibit profound convergence and synergy at the molecular level. Studies have found that cells undergoing EMT frequently upregulate a series of core pluripotency transcription factors and cell surface markers, thereby acquiring a typical CSC phenotype [49]. This process is primarily driven by core EMT-transcription factors. For instance, SNAIL and ZEB family proteins can not only suppress epithelial markers such as E-cadherin but also directly transcriptionally activate stemness-related gene networks. TWIST, on the other hand, can maintain a dedifferentiated state and cellular plasticity by regulating epigenetic modifiers like Bmi-1 [50]. This sharing of regulatory networks implies that EMT, while inducing changes in cell morphology and motility, fundamentally reprograms the biological identity of the cell, steering it towards a more aggressive, plastic, and therapy-resistant CSC state [51].

3.
Structural characteristics and biological functions of circRNA
3.1.
Genomic landscape and circRNA discovery
Although around 93% of DNA sequences in the human genome can be transcribed into RNA, less than 2% of nucleic acid sequences are used to encode proteins [52]. The remaining transcripts are non-coding RNAs (ncRNAs) that do not encode proteins. NcRNAs lack open reading frames and cannot encode proteins through conventional mechanisms, yet they function as crucial regulators of various biological processes, including development, proliferation, transcription, and post-transcriptional modification. Among the diverse types of ncRNAs, microRNA (miRNA), long non-coding RNA (lncRNA), and circular RNA (circRNA) have been widely recognized for their roles in innate immune regulation and disease pathogenesis.
Circular RNA was discovered approximately 40 years ago and is widely present in fungi, protozoa, plants, and humans [53]. Initially, circRNAs were considered to be splicing error products with minimal biological significance. However, with the application of high-throughput RNA-seq techniques, circRNAs have been demonstrated to play widespread and pivotal roles in organisms and exhibit remarkable specificity in various tissues and disease states [20,21]. This tissue and disease specificity suggests that circRNAs are precisely regulated molecules with important biological functions rather than merely transcriptional byproducts.

3.2.
Unique structural features of circRNAS
CircRNAs represent a special class of ncRNAs that differ fundamentally from other linear RNAs in their molecular architecture. Unlike linear mRNAs, circRNAs lack the 5’cap structure and 3’ poly(A) tail, and instead exist in a covalently closed loop configuration [54]. This unique circular structure has profound implications for circRNA stability and function. Since the circular structure lacks free ends, circRNAs are not degraded by RNA exonucleases that typically process linear RNAs from their terminal ends. Consequently, circRNAs exhibit stable expression patterns and accumulate to relatively high steady-state levels in cells, making them particularly suitable for long-term regulatory functions [55].
CircRNAs arise via back-splicing of precursor mRNAs, forming covalently closed loops. In this process, RNA polymerase II transcribes precursor mRNA, but instead of conventional forward splicing, the 3’end splice site is joined to an upstream 5’ end splice site via back-splicing to form a closed loop structure [56]. This back-splicing mechanism can be facilitated by complementary sequences in flanking introns that bring distant splice sites into proximity, or by RNA-binding proteins that bridge different regions of the pre-mRNA molecule. Based on their genomic origin and the composition of the circular transcript, circRNAs can be systematically classified into three major categories [57]: exonic circRNAs (EcRNAs), which consist solely of exonic sequences; intronic circRNAs (CiRNAs), which are composed entirely of intronic sequences; and exon-intron circRNAs (ElcRNAs), which contain both exonic and intronic sequences within the circular molecule.

3.3.
Molecular mechanisms of circRNA function
The biological functions of circRNAs are mainly categorized into four distinct molecular mechanisms [58], each contributing to gene regulation in unique ways:(1) MicroRNA sponge function: This represents the most extensively studied mechanism of circRNA action. CircRNAs possess multiple binding sites for miRNAs, functioning as competing endogenous RNAs (competing endogenous RNA, ceRNAs). By sequestering miRNAs through complementary base pairing, circRNAs prevent these miRNAs from binding to their target mRNAs, thereby indirectly regulating gene expression by inhibiting miRNA-mRNA interactions. The circular structure and stability of circRNAs make them particularly effective as miRNA sponges, as they can accumulate to high concentrations and persistently sequester miRNAs over extended periods. (2) Regulatory protein binding: CircRNAs can interact with RNA-binding proteins (RBPs) that are involved in mRNA regulation, thereby modulating various aspects of RNA metabolism. Through these interactions, circRNAs can alter the stability, localization, or splicing patterns of target mRNAs. Some circRNAs function as scaffolds that bring multiple proteins together, while others sequester RBPs away from their normal target RNAs, thereby modulating cellular processes through protein sequestration. (3) Protein coding capacity: Although circRNAs are classified as non-coding RNAs and generally do not participate in translation through conventional cap-dependent mechanisms, a small proportion of circRNAs can be translated into functional peptides. This translation can occur through internal ribosome entry sites (IRES) or be driven by N6-methyladenosine (m6A) modifications. Although this represents a minor fraction of circRNA functions, the peptides produced can regulate important cellular processes including transcription and signalling. (4) Gene transcription regulation: CircRNAs, particularly those localized to the nucleus, can directly or indirectly interact with RNA polymerase II and other transcriptional machinery components to regulate gene transcription. This function allows circRNAs to influence gene expression at the transcriptional level, complementing their more prevalent post-transcriptional regulatory roles.

3.4.
CircRNAs in cancer and EMT
Studies have demonstrated that circRNAs are dysregulated in various diseases, with cancer being among the most extensively studied [25]. Comprehensive expression profiling has revealed widespread circRNA dysregulation across multiple cancer types, with altered circRNA expression often correlating with disease stage, metastatic potential, and patient prognosis. EMT, as an essential molecular process in cancer metastasis, has been shown to have important relationships with circRNA expression patterns.
A pivotal study examining the role of circRNAs in EMT conducted high-throughput RNA sequencing analysis, demonstrating that the expression of hundreds of circRNAs is dynamically regulated during EMT induction [27]. Notably, most of these differentially expressed circRNAs were upregulated during the EMT process, suggesting that circRNAs may predominantly function as pro-EMT regulators, although important EMT-suppressive circRNAs have also been identified. This observation indicates that EMT may be an essential pathway through which circRNAs control cancer development and progression. The functional significance of these expression changes is supported by mechanistic studies demonstrating that circRNAs can modulate EMT through regulation of key signalling pathways and transcription factors, ultimately affecting cancer cell migration, invasion, and metastatic capacity. (Figure 2)

4.
Regulation of EMT by circRNA
CircRNAs regulate EMT through various molecular mechanisms, which can be systematically categorized based on the signalling pathways and molecular targets they modulate. The following sections examine how circRNAs interface with major EMT-regulatory pathways.
4.1.
TGF-β signaling pathway
Studies have demonstrated that dysregulation of the TGF-β pathway is associated with the occurrence of EMT in various cancers, including cervical cancer and lymphoma [59,60]. The TGF-β signalling pathway represents one of the most potent inducers of EMT across diverse cellular contexts. Within this pathway, Smad3, a member of the Smad family, has been identified as one of the most pivotal terminal downstream factors in TGF-β signalling transduction [61].
CircRNA-mediated regulation of Smad3 stability serves as a pivotal mechanism in cancer progression, as exemplified by the axis of hsa_circ_0088036/miR-1343-3p/Bcl-3 in NSCLC, wherein this circRNA sponges miR-1343-3p to derepress the proto-oncogene Bcl-3, whose protein product directly interacts with the MH2 domain of Smad3. This interaction competitively blocks the binding of E3 ubiquitin ligases or adapters to this critical domain, thereby inhibiting Smad3 ubiquitination and proteasomal degradation, ultimately enhancing its stability and amplifying the TGF-β/Smad3/EMT signalling cascade [62]. This conserved paradigm is recapitulated in other malignancies with subtype-specific nuances, such as in the diffuse subtype of gastric cancer where circNRIP1 acts as an oncogene by sponging miR-149-5p to upregulate FOXM1, which in turn facilitates Smad3 phosphorylation and nuclear translocation, thereby potentiating the TGF-β-induced EMT programme [63]. This subtype-specificity, further evidenced by circ-HER2 stabilizing Smad3 and promoting EMT predominantly in hormone receptor-positive breast cancer, underscores how cellular context dictates circRNA network functions [64]. Transitioning to clinical translation, while direct therapeutic targeting of such oncogenic circRNAs remains largely preclinical, strategies like antisose oligonucleotides show promise in animal models. The clinical application is currently more advanced in diagnostics, where circRNAs involved in this axis, such as hsa_circ_0000567 in colorectal cancer, have been validated as promising plasma biomarkers for early detection and metastasis prediction in clinical cohorts [65].
CircRNA-mediated regulation of transcriptional intermediary factor 1γ (TIF1γ) exemplifies a sophisticated network-based mechanism for suppressing EMT. Transcriptional intermediary factor 1γ (TIF1γ) is a protein that inhibits TGF-β/Smad signalling by promoting Smad4 ubiquitination and competing with Smad4 for binding to the Smad2/3 complex [66]. CircPTK2 promotes TIF1γ expression by sponging miR-429 and miR-200b-3p, which normally suppress TIF1γ by targeting its 3’-UTR. By upregulating TIF1γ, circPTK2 inhibits TGF-β/Smad signalling transduction, thereby suppressing EMT [67]. Beyond regulating the TGF-β/Smad signalling axis, TIF1γ also exhibits additional EMT-suppressive functions. TIF1γ competes with TATA box-binding protein (TBP) and TBP-associated factor 15 (TAF15), impeding TAF15/TBP-mediated interleukin-6 (IL-6) transactivation. Mechanistically, TIF1γ drives the nuclear export of TAF15 through multiple monoubiquitination modifications of TAF15 [68]. Functionally, TAF15 accelerates epithelial-mesenchymal transition and metastasis of lung adenocarcinoma cells, acting in an opposite manner to TIF1γ. These findings indicate that the TAF15/TBP complex is essential for IL-6 activation-induced EMT and invasion, while TIF1γ inhibits these processes by antagonizing this complex. This positions circPTK2, through TIF1γ upregulation, as a critical node suppressing both TGF-β/Smad-dependent and IL-6-dependent EMT signalling branches.
In the context of positive feedback loops in circRNA-TGF-β regulation, the downstream regulation of circRNA function is primarily governed by competitive endogenous RNA mechanisms [69], which are relatively well-defined. However, the mechanisms underlying the upstream regulation of circRNA biogenesis remain incompletely understood [70]. It has been reported that alternative splicing factors and RNA-binding proteins (RBPs) may be involved in the upstream regulation of circRNA production [71].
In oral squamous cell carcinoma (OSCC), a sophisticated self-reinforcing regulatory circuit with the circRNA at its core has been elucidated, involving circUHRF1 (hsa_circ_0002185). In this system, miR-526b-5p binds to c-Myc’s 3’-UTR to inhibit its expression, while circUHRF1 can sponge miR-526b-5p, thereby positively regulating c-Myc [72]. The transcription factor c-Myc then accelerates the transcription of two key targets: TGF-β1, which promotes EMT, and epithelial splicing regulatory protein 1 (ESRP1) [73]. Importantly, the splicing factor ESRP1 promotes UHRF1 gene cyclization and biogenesis by targeting the flanking introns of circUHRF1. This completes a closed, autoregulatory loop where circUHRF1 initiates and is itself regenerated by the pathway it activates, solidifying its central regulatory role. This positive feedback loop of circUHRF1/miR-526b-5p/c-Myc/TGF-β1/ESRP1 accelerates OSCC development and EMT through the TGF-β1 pathway. Furthermore, analogous regulatory mechanisms involving ESRP1 and TGF-β1 have been identified in breast cancer. In this context, the mechanism can be targeted by the transcription factor upstream transcription factor 1 (USF1). The circular RNA circANKS1B accelerates EMT formation by targeting miR-148a and miR-152-3p, which activate expression of the transcription factor USF1. USF1 then promotes TGF-β and ESRP1 expression, generating a similar feedback loop [74]. These examples demonstrate that circRNAs can function as pivotal orchestrators within positive feedback loops where the TGF-β pathway synergizes with upstream and downstream factors, creating self-reinforcing circuits that stabilize the mesenchymal phenotype. (Figure 3)

4.2.
Wnt/β-catenin pathway
Carcinogenesis is frequently accelerated by inactivation of tumour suppressor genes such as adenomatous polyposis coli (APC) in the Wnt signalling pathway, or by oncogenic mutations in β-catenin itself. High expression levels of β-catenin signalling can be detected in tumour cells at the invasion front or in cells migrating to adjacent stroma. This differential pattern of Wnt signalling activity indicates distinct functional roles: promoting both the proliferation capacity and the EMT potential of cancer cells [75].
CircRNA-mediated regulation of APC2 represents a pivotal mechanism for controlling the Wnt/β-catenin signalling axis, as exemplified in colorectal cancer (CRC) by the hsa_circ_0009361/miR-582/APC2 regulatory axis [76]. Herein, hsa_circ_0009361 functions as a critical network hub by sponging the oncogenic miR-582, thereby relieving the post-transcriptional repression of APC2—a key negative regulator of the Wnt pathway. The subsequent upregulation of APC2 protein promotes the destabilization and degradation of cytoplasmic β-catenin, inhibiting its nuclear translocation and the transcriptional activation of pro-EMT and proliferative targets such as c-Myc and cyclin D1. Thus, this circRNA-directed network effectively suppresses the Wnt/β-catenin-EMT programme [77]. The clinical relevance of this axis is underscored by the consistent downregulation of hsa_circ_0009361 in CRC tissues, where its expression inversely correlates with advanced tumour stage and metastasis, highlighting its prognostic biomarker potential. Therapeutically, while direct targeting remains preclinical, restoring the function of such tumour-suppressive circRNAs or deploying antisense oligonucleotides against their cognate oncogenic miRNAs (e.g. miR-582) represents a promising network-based strategy to impede Wnt-driven EMT and CRC progression.
CircRNAs function as critical regulatory molecules in radiation-resistant cancers by modulating the Wnt/β-catenin signalling axis to influence EMT. For instance, radiation-resistant oesophageal squamous cell carcinoma (ESCC) cells demonstrate high EMT potential. Acting as a central node in a competing endogenous RNA network, circRNA_100367 can sponge miR-217, thereby de-repressing its target Wnt3. This interaction attenuates the EMT capacity of ESCC radiation-resistant cells through the miR-217/Wnt3 regulatory axis [78]. This finding not only delineates a precise molecular mechanism driven by circRNA_100367 but also suggests that this circRNA may serve as a potential therapeutic target to overcome radiation resistance in ESCC by disrupting a key signalling cascade.
CircRNA regulation by inflammatory cytokines represents a key mechanism through which the tumour microenvironment influences cancer progression, highlighting circRNAs as dynamic mediators of extracellular signals. In CRC cells, circ_0026344 can be downregulated by chemokine (C-C motif) ligand 20 (CCL20) and C-X-C motif chemokine ligand 8 (CXCL8), which are abundant in the tumour microenvironment. This downregulation attenuates the critical sponging effect of circ_0026344 on miR-183 [79]. MiR-183 is considered a typical oncogene [80], and Wnt/β-catenin signalling is believed to be a downstream transduction pathway for miR-183. Therefore, in CRC, when circ_0026344 is inhibited by CCL20 and CXCL8, the loss of this circRNA’s regulatory function promotes EMT through activation of the Wnt/β-catenin signalling pathway [81]. This example illustrates how inflammatory signals modulate circRNA expression to indirectly affect EMT, thereby positioning specific circRNAs as central sensors and integrators of oncogenic inflammatory cues.
CircRNAs play a critical regulatory role in melanoma progression by modulating key oncogenic pathways, as exemplified by circRNA_0082835 [82]. In melanoma, circRNA_0082835 is markedly overexpressed, and its knockdown has been shown to inhibit Wnt/β-catenin pathway activity, directly linking this circRNA to a core EMT-driving signalling axis [83]. Bioinformatic analysis from the ENCORI database indicates that circRNA_0082835 contains binding sites for miRNA-429, a well-characterized tumour suppressor known to block melanoma progression [84]. Therefore, it can be inferred that circRNA_0082835 likely functions as an oncogenic driver by sponging and inhibiting miR-429, thereby derepressing the Wnt/β-catenin pathway to promote EMT. This positions circRNA_0082835 as a central ceRNA network hub that actively sustains a pro-metastatic signalling circuit in melanoma.
Direct protein interactions represent a key mechanism whereby circRNAs regulate the Wnt/β-catenin pathway, highlighting their versatile role as direct molecular scaffolds and modulators of signalling components. Beyond their function as microRNA sponges, some circRNAs can directly interact with proteins in the Wnt/β-catenin pathway. For example, circ-glycogen synthase kinase-3 beta (GSK3β), derived from the GSK3β gene, directly interacts with GSK3β protein, an upstream regulatory factor of GSK3β/β-catenin signalling. This RNA-protein interaction effectively sequesters GSK3β, preventing it from executing its normal inhibitory function and thereby promoting β-catenin activity in ESCC cells [85]. Similarly, a novel 370 amino-acid β-catenin isoform generated from circβ-catenin through translation can directly interact with GSK3β. This interaction inhibits GSK3β-mediated β-catenin phosphorylation and degradation, demonstrating how protein-coding circRNAs can produce functional antagonists of core pathway enzymes [86]. These studies demonstrate that circRNAs can regulate Wnt/β-catenin signalling through multiple mechanisms: they can function as microRNA sponges to indirectly modulate pathway components, or they can serve as core direct regulators by interacting with signalling molecules through either RNA-protein interactions or by encoding regulatory peptides, thereby actively controlling cell function and EMT.

4.3.
Phosphoinositide-3-kinase (PI3K)/protein kinase B (AKT) pathway
The PI3K/AKT pathway is a pivotal kinase cascade that controls essential cell functions, including proliferation, transcription, translation, survival, and growth. Abnormal expression of the PI3K/AKT signalling pathway has been observed in many diseases, including breast cancer, lung cancer, and thyroid cancer [87].
CircRNA-mediated regulation of casein kinase II catalytic subunit alpha (CSNK2A1) represents a direct mechanism for activating oncogenic signalling pathways, as exemplified in thyroid cancer by the interaction between circNDST1 and CSNK2A1 [88]. In this context, circNDST1 functions not merely as a sponge but as a critical protein-binding partner, directly interacting with CSNK2A1 to enhance its kinase activity. This circRNA-protein complex significantly potentiates the phosphorylation of a key downstream substrate, AKT. The hyperphosphorylation of AKT leads to the constitutive activation of the PI3K/AKT signalling axis, a major driver of cell survival, proliferation, and notably, the EMT programme in thyroid cancer cells [89,90]. This positions circNDST1 as a central upstream regulator that amplifies a pro-metastatic signalling network through direct protein modulation. The clinical relevance of this axis is supported by the co-upregulation of circNDST1 and CSNK2A1 in aggressive thyroid cancer subtypes, correlating with advanced disease stages and poorer prognosis, thus highlighting their combined potential as a prognostic biomarker signature [91]. Therapeutically, disrupting the specific circNDST1-CSNK2A1 interaction or downstream AKT activation presents a promising strategy to inhibit PI3K/AKT-driven EMT and tumour progression in thyroid cancer.
The circGRAMD1B/SOX4/MEX3A axis plays a critical role in lung adenocarcinoma (LUAD), initiating a multi-tiered oncogenic signalling cascade. In LUAD, miR-4428 binds to the 3’-UTR region of SOX4 to inhibit SOX4 expression, while circGRAMD1B serves as the upstream master regulator by sponging miR-4428, thus derepressing SRY-Box Transcription Factor 4 (SOX4) expression [92]. SOX4, a member of the group C subfamily of SOX transcription factors, has important regulatory functions in cancer metastasis and EMT development [93]. Studies have shown that elevated SOX4 is strongly correlated with mex-3 RNA binding family member A (MEX3A), as SOX4 promotes MEX3A expression by binding to its promoter. Subsequently, MEX3A, as an mRNA-binding protein, binds to LAMA2 mRNA to reduce its stability and decrease LAMA2 expression. Given that PTEN is a classical inhibitor of the PI3K/AKT pathway and LAMA2 can significantly increase PTEN gene expression, MEX3A ultimately promotes LUAD metastasis and EMT by inhibiting LAMA2 to activate the PI3K/AKT pathway. This delineates a comprehensive circRNA-orchestrated network, forming the circGRAMD1B/miR-4428/SOX4/MEX3A/LAMA2 regulatory axis in LUAD [94].
CircRNAs serve as pivotal upstream regulators of the PI3K/AKT signalling pathway across various cancer types, orchestrating diverse pro- or anti-tumorigenic networks. In OSCC, circRNAHIPK3 acts as a central oncogenic hub by sponging the tumour suppressor miR-637, thereby derepressing nuclear protein 1 (NUPR1) expression and activating the downstream NUPR1/PI3K/AKT pathway [95]. Similarly, circPIP5K1A functions as a critical metastatic driver by serving as a competing endogenous RNA for miR-515-5p, which leads to the upregulation of TCF12 and subsequent activation of the PI3K/AKT pathway to promote cancer metastasis and EMT [96]. Conversely, hsa_circRNA_100269 exemplifies a tumour-suppressive circRNA that inhibits gastric cancer progression; its overexpression induces G0/G1 cell cycle arrest and promotes apoptosis [97], suggesting its core regulatory function is likely mediated through suppressing the oncogenic PI3K/AKT axis to inhibit metastasis and EMT.

4.4.
CircRNA regulation of EMT-transcription factors (EMT-TFs)
The transcription factors of EMT play pivotal roles in the activation and maintenance of the EMT pathway. Their primary functions are to inhibit the transcription of epithelial marker proteins, such as E-cadherin, and promote the translation of mesenchymal marker proteins, such as N-cadherin [98]. The EMT-TFs including TWIST1, TWIST2, SNAIL1, SNAIL2, ZEB1, and ZEB2 have all been extensively studied [99].
CircRNAs function as central regulators of EMT by modulating the stability of key EMT transcription factors like SNAIL through ceRNA networks. Studies have demonstrated that miRNAs can regulate EMT-TFs, and circRNAs, acting as master miRNA sponges within these networks, can indirectly govern EMT-TF activity, thus critically affecting the EMT capacity of cancer cells [100]. In melanoma, circRNA_0084043 was reported to be upregulated. The 3’-UTR of Snail shares the same binding site for the tumour-suppressive miR-153-3p as circRNA_0084043 [82]. By sequestering miR-153-3p, circRNA_0084043 effectively derepresses Snail expression, which suggests that this circRNA promotes melanoma cell growth and metastasis by orchestrating a miR-153-3p/Snail regulatory axis [101]. Similarly, in urothelial carcinoma of the bladder, circPRMT5 serves as an oncogenic hub that competitively upregulates Snail expression by sponging miR-30c, thus inhibiting E-cadherin expression and promoting EMT [102].
CircRNA regulation of ZEB transcription factors represent a critical upstream mechanism within the established double-negative feedback loops that dynamically control EMT status [103]. In clear cell renal cell carcinoma, ZEB2 was observed to be a direct target of miR-153, which inhibits the pro-EMT effect of ZEB2. Functioning as a central regulatory node within this network, circPCNXL2 blocks miR-153’s interaction with ZEB2 by sponging miR-153. The finding that miR-153 inhibitors reversed the effects of circPCNXL2 on RCC cells confirms the axis’s functionality. These findings indicate that circPCNXL2 acts as a pivotal oncogenic circRNA and a master upstream regulator that actively promotes EMT in RCC by coordinating the miR-153/ZEB2 axis, thereby positioning circRNAs as key drivers of this oncogenic switch [104].
A reciprocal regulatory relationship exists between circRNAs and EMT-TFs, forming a complex, integrated network. CircRNAs not only function as upstream regulatory factors of EMT-TFs but can also be regulated by EMT-TFs, thereby acting on EMT. In hepatocellular carcinoma, TWIST1 binds to the Cul2 promoter to activate Cul2 transcription, leading to production of circ-10720 through back-splicing [60]. Circ-10720 subsequently functions as a key downstream effector and an integrative ceRNA node, sequestering multiple microRNAs (miR-1246, miR-578, and miR-490-5p) that normally target vimentin mRNA, thereby de-repressing vimentin expression and promoting EMT [60]. Cul2 is a tumour suppressor protein that degrades ubiquitinated HIFα and regulates the cell cycle [105,106]. In experiments, since circ-10720 sponged miR-1246, miR-578, and miR-490-5p, which are primary miRNAs regulating vimentin, it is hypothesized that TWIST1 promotes circ-10720 production, thereby preventing multiple miRNAs that inhibit vimentin expression from functioning, thus promoting vimentin expression. Vimentin, as a mesenchymal marker protein, promotes EMT in HCC cells. A similar regulatory mechanism appears to be widespread in non-small cell lung cancer, where circ-10720 was observed to be upregulated in tumour tissues, and changes in vimentin levels regulate EMT, thus affecting migration and invasion [107]. These studies suggest that the regulation between circRNA and EMT-TFs is intricate, with functional interactions occurring both upstream and downstream, ultimately establishing a self-reinforcing circuit where circRNAs act not just as effectors but as core components sustaining the pro-EMT feedback loop. (Figure 4)

4.5.
Other regulatory pathways
Beyond the classical signalling pathways intensively studied in the regulatory relationship between EMT and circRNA, researchers have examined additional regulatory routes.
CircRNA regulation of the VEGFA/VEGFR2 pathway represents a critical mechanism in cancer progression, where circRNAs act as upstream modulators of angiogenesis and EMT [108]. Vascular endothelial growth factor A (VEGFA) is a member of the growth factor family with angiogenic properties and is a key regulator of angiogenesis in cancer tumours [109]. In bladder cancer, circRNA-MYLK, functioning as a pivotal ceRNA hub, sponges miR-29a, which normally targets the VEGFA 3’-UTR to inhibit its transcription. The sequestration of miR-29a by this circRNA leads to increased VEGFA levels, promoting the phosphorylation of VEGFR2 and thereby activating the VEGFA/VEGFR2 pathway. This activation subsequently triggers the phosphorylation of the downstream RAS/ERK signalling pathway, a cascade orchestrated by the initial circRNA-mediated event, which ultimately induces EMT [110]. Thus, circRNA-MYLK serves as a central regulator that integrates angiogenic signalling with the EMT programme.
CircRNA-mediated regulation of the SOX4/EZH2 axis plays a key role in pancreatic cancer progression, as exemplified by the oncogenic function of circ_0001666. Circ_0001666 is overexpressed in pancreatic cancer samples, and its high expression is associated with poor patient prognosis. Functioning as a central ceRNA network hub, circ_0001666 can sponge miR-1251, which targets SOX4, thereby derepressing SOX4 expression [111]. SOX4 subsequently functions as a direct transcriptional activator of the EZH2 promoter, inducing EZH2 expression to enhance EMT in pancreatic cancer [112]. Therefore, circ_0001666 acts as the upstream master regulator that initiates the oncogenic cascade, driving EMT through the miR-1251/SOX4/EZH2 axis in PC tissues [113], thereby positioning this circRNA as a critical integrator of transcriptional and epigenetic control within the EMT network.
CircRNA sequestration of RNA-binding proteins such as HuR represents a direct post-transcriptional regulatory mechanism affecting EMT, as demonstrated in cervical cancer by circPABPN1. Derived from the PABPN1 gene, circPABPN1 serves as a central molecular decoy that can bind to and sequester HuR, an RNA-binding protein that regulates protein translation by binding to mRNA, thus acting as a sponge for HuR and inhibiting PABPN1 translation in cervical cancer cells [114]. Simultaneously, HuR can bind to the 3’-UTR of EMT transcription factor mRNA to increase Snail mRNA stability, causing increased Snail expression to promote EMT [115]. This indicates that circPABPN1, through its sequestration of HuR, orchestrates a dual regulatory function: suppressing PABPN1 translation while indirectly enhancing Snail stability, thereby positioning itself as a key upstream regulator that coordinately promotes EMT-related processes. (Table 1)

5.
CircRNAs as diagnostic biomarkers and therapeutic targets
5.1.
Potential of circRNA as biomarkers
Compared with linear RNAs, circRNAs are structurally stable and are not degraded by RNA exonucleases, exhibiting stable expression patterns [116]. This intrinsic stability, coupled with their disease-specific dysregulation, forms the cornerstone of their utility as biomarkers in oncology. Critically, a growing body of clinical cohort studies has begun to translate these theoretical advantages into validated performance metrics, directly supporting their diagnostic and prognostic potential.
The promise of circRNAs as effective biomarkers is underscored by several key attributes, now substantiated by initial clinical evidence: (1) Their central role in pathological processes like EMT and metastasis is reflected in patient outcomes. For instance, clinical studies have validated that specific circRNAs are independent prognostic factors. As shown in Table 2, the aberrantly high expression of hsa_circ_0006006 in NSCLC patient serum is significantly associated with poor survival prognosis and cisplatin resistance [117], while in HCC, low expression of hsa_circ_0003570 correlates with advanced tumour stage and serves as an independent favourable prognostic factor [120]. This positions circRNA levels as dynamic indicators of disease aggression. (2) CircRNAs exhibit high tissue- and disease-specificity [125]. This specificity is crucial for diagnostic accuracy, as demonstrated by studies where circRNA signatures successfully distinguish cancer patients from healthy individuals. (3) Their abundance in stable, accessible body fluids like plasma and saliva [126,127] and their enrichment within exosomes [128] facilitate non-invasive ‘liquid biopsy’ applications. Researchers have successfully differentiated healthy populations from colon cancer patients by analysing serum exosomes [129].
Systematic evaluation of diagnostic/prognostic performance metrics is emerging. Beyond mere detection, studies are quantifying the clinical value of circRNAs using standard biomarker metrics. For example, plasma circ_0004592 shows promise for the early detection of gastric cancer (GC), with its diagnostic power characterized by a receiver operating characteristic (ROC) curve analysis yielding an area under the curve (AUC) value that indicates high sensitivity and specificity [118]. In colorectal cancer (CRC), high serum levels of hsa_circ_0008621 constitute an independent prognostic factor [119], and in ESCC, hsa_circ_0002938 is not only correlated with advanced stage but is considered a potential predictive biomarker [122]. These studies move beyond association to establish quantifiable performance criteria.
Comparisons with established clinical biomarkers highlight both potential and the path forward. While traditional protein markers (e.g. CEA for CRC [130], CA-19–9 for pancreatic cancer [131]) are widely used, they often suffer from limited sensitivity or specificity. Preliminary comparative analyses suggest that certain circRNAs, either alone or in panels, may outperform or complement these conventional markers. In gastric cancer, for instance, the correlation of reduced plasma circKIAA1244 and elevated hsa_circ_0000467 with patient prognosis [132] suggests additive value. The ultimate clinical utility will likely reside in multi-analyte models that integrate circRNAs with existing markers (e.g. proteins, ctDNA) and clinical parameters, potentially enabling earlier detection, more accurate risk stratification, and real-time monitoring of therapeutic response. However, large-scale, prospective, multi-centre validation studies are imperative to rigorously establish their clinical superiority, standardize detection protocols, and define clear diagnostic cut-off values before routine clinical adoption.

5.2.
Therapeutic strategies targeting circRNAs
Numerous studies have demonstrated that the knockdown or elevation of circRNAs has marked effects on the EMT pathway in cancer development. The primary existing therapeutic strategies involve targeted regulation of the cancer process through modulating target circRNA expression. The main methods include [133–135]: (1) CircRNA overexpression strategies: CircRNA vectors and techniques for synthetic-mediated circRNA overexpression can be employed to increase tumour-suppressive circRNA levels. (2) CircRNA delivery approaches: Exosome or nanoparticle-mediated circRNA delivery strategies can be used to introduce therapeutic circRNAs into target cells. (3) CircRNA knockdown strategies: These include approaches mediated by siRNA, short hairpin RNA, antisense oligonucleotides, and CRISPR/Cas9 systems to reduce oncogenic circRNA expression.
Although circRNA-based cancer therapies have opened new avenues for oncology treatment, their clinical translation faces a series of complex challenges that must be urgently addressed. These challenges are first reflected in the delivery aspect: traditional delivery vectors, such as cationic liposomes, encounter bottlenecks when delivering molecules [136]. For instance, the limitations of vector design are highlighted by a study in which an attempt to achieve targeted siRNA delivery to a prostate cancer mouse model resulted in less than 6% tumour accumulation efficiency [137]. While emerging materials such as gold nanoparticles show promise in drug delivery, the in vivo safety and biodistribution profiles for applications involving circular RNAs still require comprehensive evaluation [138,139]. Secondly, at the specificity level, rational design is necessary to achieve precise targeting. For example, one study demonstrated that binding to the AAT promoter forms an apoptosis factor Bax-based tumour-killing switch, thereby enhancing hepatocyte specificity and enabling activation exclusively in specific hepatocellular carcinoma cells [140]. Furthermore, the inherent high stability of circRNA presents a dual challenge. Compared to mRNA, circRNA enables superior stability and sustained gene expression in cells. However, the excessively long half-life poses safety concerns for circRNA. In one study, researchers developed a ‘circRNA switch’ capable of sensing intracellular RNA or proteins to control circRNA-encoded protein expression, thereby ensuring the absence of severe cytotoxicity and immunogenicity, while remaining responsive to target miRNAs or proteins [141]. Finally, the risk of off-target effects is also substantial. Given the global regulatory capacity of circRNAs, the fact that a single circRNA can sponge multiple miRNAs suggests that its role as a ‘molecular sponge’ may disrupt miRNA networks globally [142]. Therefore, translating circRNA into a viable therapy is by no means straightforward. It must rely on the systematic integration of intelligent vector engineering, conditional expression design, and dynamic safety regulation to ultimately realize its therapeutic potential. (Figure 5) (Table 2)

6.
Conclusions and future perspectives
In 2022, there were about 20 million new cancer cases worldwide [143], with 4,820,000 new cancers occurring in China [144], accounting for 24.1% of the global total. The International Union Against Cancer believes that one-third of cancers can be prevented, one-third can be cured if diagnosed early, and for the remaining one-third, pain can be alleviated and life prolonged, leading to the concept of tertiary prevention of malignant tumours. Since cancer represents a major threat and challenge to human health, understanding its pathogenesis, treatment, and prevention is one of the most critical tasks for researchers [145].
It is now evident that EMT in cancer metastasis is closely related to circRNA expression and function. CircRNAs are involved in vital EMT processes, such as initiation, maintenance of stem-like properties, and promotion of invasion. This suggests that circRNA dysregulation may be an essential driver of EMT and that in-depth study of these mechanisms may lead to novel therapeutic solutions [43].
The central position of circRNAs in cellular pathways and their close correlation with the upstream and downstream expression of EMT-TFs illustrates the importance of studying circRNAs for understanding carcinogenesis, metastasis, and EMT mechanisms. Additionally, as cancer biomarkers, circRNAs have unique advantages including stability, tissue specificity, and accessibility in body fluids [146]. CircRNA-based biological treatments may become a novel approach and promising direction for cancer therapy.
However, circRNA-based therapeutic approaches face numerous challenges in the transition from basic research to clinical application. First, there are limitations in research models. In vitro cell models serve as the foundation for circRNA functional studies, but standard cell lines lose many endogenous circRNAs during long-term culture [147], and their expression patterns differ from those in primary tumours. This may lead to functional experimental results based on such models not reflecting the actual in vivo situation. Animal models are crucial bridges for translating circRNA research into clinical practice; however, due to the significant species specificity of circRNAs, their distribution and function may differ markedly between humans and experimental animals [148]. This makes it difficult to replicate circRNA distribution observed in mouse models in humans, and the functions of the same circRNA may vary between species [149], hindering translation from basic research to clinical applications.
Second, methodological limitations significantly hinder their translation from basic research to clinical practice. A key feature of circRNAs is back-splicing, making it essential to confirm that detected signals indeed originate from circRNAs formed through back-splicing and to distinguish different splice isoforms from the same gene locus. Traditional methods based on back-splicing junction detection, such as RNA-seq and RT-qPCR, often struggle to precisely differentiate and accurately identify distinct isoforms, which affects subsequent functional research and the specificity evaluation of biomarkers. Following this is the issue of bias introduced by RNase R treatment. CircRNAs are present in various sample types such as tissues, plasma, and exosomes [150], yet standardized extraction protocols are lacking, leading to difficulties in comparing data across different studies. RNase R digestion is a commonly used method for enriching circRNAs [151], but subtle variations in treatment conditions – such as enzyme concentration – can significantly impact outcomes: excessive enzyme may degrade circRNAs, while insufficient amounts may leave residual linear RNAs, resulting in inaccurate detection results. Challenges also persist in circRNA detection and data analysis methodologies, as different enrichment strategies, sequencing approaches, and identification algorithms can yield substantially divergent results [152]. A comparative study revealed that the number of circRNAs identified in the same cell line could vary by up to 42-fold depending on the tool used [153]. Moreover, qPCR quantification of circRNAs currently lacks universally accepted and reliable internal reference genes comparable to GAPDH or β-actin used for mRNAs. At the same time, the entire workflow – from sample collection and preservation to RNA extraction and reverse transcription – lacks unified standards, making data across laboratories difficult to compare and integrate. This is a key obstacle preventing circRNAs from becoming clinically reliable biomarkers.
Finally, there are challenges in clinical application. Although circRNAs show great promise as biomarkers for cancer metastasis, their accurate detection remains difficult with current technologies. Although circRNAs are structurally stable, their half-life and clearance mechanisms in the blood are not yet fully understood, which affects their reliability as dynamic monitoring indicators. Additionally, many circRNAs that are aberrantly expressed in cancer are also involved in other pathological processes, such as cardiovascular diseases and Alzheimer’s disease [154], leading to insufficient diagnostic specificity.
To systematically overcome the translational bottlenecks from basic research to clinical application for circRNAs, future efforts must integrate cutting-edge technologies, innovative models, and intelligent design strategies through a focused and convergent approach along three core pathways.
First, on the technological and standardization pathway, it is essential to transcend the current limitations of short-read sequencing and RNase R enrichment [155]. The solution involves the development and widespread adoption of single-molecule real-time full-length sequencing technologies, such as those offered by PacBio Sequel IIIe or Oxford Nanopore platforms. These technologies directly decipher the complete circular sequence and internal modifications of circRNAs, thereby providing a fundamental resolution to the challenges of authenticating back-splicing junctions and precisely characterizing isoforms [156]. Concurrently, establishing a globally collaborative reference database and algorithmic benchmarking platform is crucial. Initiatives like a proposed Circ-omics Standardization. Initiative could furnish the community with unified reference materials, standardized protocols, and open-source analytical tools. This framework would enable the systematic calibration and performance assessment of laboratory workflows – from library preparation and sequencing to data analysis – ultimately fostering reproducible and comparable quantitative standards across the field.
Second, on the research and validation model pathway, there is a pressing need to develop experimental systems that more accurately mirror the intricate reality of human tumours. The central strategy is to advance a four-dimensional functional screening platform based on patient-derived tumour organoids integrated with organ-on-a-chip technology. This involves coupling organoids generated from fresh patient tumour tissues with sophisticated microfluidic systems that incorporate essential stromal components like immune cells and fibroblasts [157]. Conducting high-throughput functional screenings on such a platform allows for the identification of circRNAs that genuinely drive tumour progression within a context that preserves native microenvironmental cues, cellular heterogeneity, and dynamic growth patterns [158]. To validate these findings without artefact, endogenous circRNA expression should be precisely modulated directly within this system using advanced in situ editing tools like CRISPR-dCas13 [159], moving beyond conventional overexpression models that may yield misleading results.
Finally, the clinical application design pathway necessitates a paradigm shift from mere biomarker discovery towards the creation of intelligent therapeutics. For diagnostic advancement, the goal is to construct a dynamic, multi-omics-integrated liquid biopsy monitoring model. This approach surpasses the simple combination of a few markers by longitudinally analysing patient plasma samples throughout therapy. It integrates temporal data from circRNA expression profiles, circulating tumour DNA mutation spectra, proteomics, and exosome content. Leveraging advanced temporal deep learning models, such as graph neural networks, this model can unravel complex biosignature networks capable of predicting therapeutic response, the emergence of drug resistance, or early signs of relapse [160]. For therapeutic innovation, the field is expanding along two complementary axes. One axis focuses on engineering modular sense-compute-execute closed-loop circRNA circuits. An example is a designer circRNA encoding a tandem of functional modules: a sensing module featuring a peptide linker specifically cleaved by proteases overactive in the tumour microenvironment; a logic computation module housing a riboswitch activated only by a precise combination of miRNA signals; and an execution-and-self-destruct module that produces a therapeutic protein but contains degradation elements in its 3’UTR triggered by miRNAs abundant in normal tissues. This design ensures spatially and contextually restricted activation followed by safe elimination. The other axis leverages drug repurposing strategies, such as repurposing drugs as GLP-1 based therapy or targeting 20S proteasomes and giving various natural compounds as hinokitiol as prophylactic with immuno-modulatory effect with positive impact on cancer [161]. Together, these strategies form a multidimensional and potentially synergistic new framework for cancer treatment.
In conclusion, the clinical translation of circRNAs represents a collaborative revolution, demanding concerted contributions from nanotechnology for precision delivery, synthetic biology for intelligent circuit design, bioinformatics for complex data deconvolution, and clinical medicine for pioneering trial design. Only through such a deeply integrated and engineering-minded endeavour can the vast theoretical promise of circRNAs be transformed into tangible diagnostic and therapeutic tools for patients.