Individual variability in RT outcomes is largely attributable to the intrinsic genetic landscape of the tumor (Fig. 1). Compelling evidence has indicated that multiple DNA repair genes modulate the cellular response to RT (24), consistent with the fact that the primary cytotoxic effect of ionizing radiation is the induction of DNA double-strand breaks (25). Genetic variations in radiosensitivity-associated genes contribute to molecular heterogeneity at specific genomic loci, ultimately influencing key cellular processes, including the extent of DNA damage sustained and the efficiency of DNA repair pathways (26).

In NSCLC, KRAS mutations have been strongly linked to radioresistance (27). Accumulating evidence has indicated that KRAS-mutant NSCLC constitutes a heterogeneous group characterized by diverse, tumor-specific mechanisms of resistance (28,29). Consistent with this complexity, a study using heterotopic NSCLC xenograft models has shown that co-mutation of KRAS and TP53 is associated with reduced local tumor control following standard RT regimens. This co-mutation may therefore serve as a predictive biomarker for patient stratification, informing dose escalation strategies or the addition of molecular agents targeting the EGFR signaling pathway. More broadly, these findings underscore the need for biomarker-driven clinical trials that integrate comprehensive genomic profiling with clinically relevant endpoints to refine and personalize RT strategies (30).

The predictive value of EGFR mutations remains controversial. In patients with EGFR-mutant stage III adenocarcinoma, concurrent chemoradiation improved local control but was paradoxically associated with shorter progression-free survival (PFS) compared with EGFR wild-type patients, primarily due to relapse or distant metastasis (31). Conversely, another study reported improved local control following chemoradiation in patients with EGFR-mutant, locally advanced nonsquamous NSCLC (32). These discordant results may reflect differences in patient demographics, smoking status (which influences tumor biology), RT dose and fractionation schemes, sequencing or concurrency of chemotherapy, or the criteria used to define treatment response and failure. Such variability highlights the need for standardized methodologies and better stratified studies to clarify the true predictive value of EGFR mutations in RT outcomes.

Parallel clinical trials have explored combining targeted therapies with RT to improve outcomes in EGFR-mutant NSCLC. Notably, the phase II trial NCT03667820 evaluated osimertinib with consolidative RT for residual lesions in advanced EGFR-mutant NSCLC. Among the 42 patients enrolled, 76% of whom received stereotactic RT, the combination yielded a PFS of 32.3 months and an OS of 45 months, with a favorable safety profile. These results suggest that integrating third-generation tyrosine kinase inhibitors with local RT may prolong disease control, offering a viable alternative to more intensive first-line regimens for this patient subgroup (33). Collectively, these data illustrate how EGFR mutations serve not only as potential predictors of radioresistance but also as guides for incorporating targeted therapies into combined-modality treatment strategies.

Additional mutations have also been implicated in radioresistance. Binkley et al (34) demonstrated that mutations in Kelch-like ECH-associated protein 1 (KEAP1) and nuclear factor erythroid 2-related factor 2 (NFE2L2) are predictive of radioresistance and are associated with high rates of local recurrence in localized NSCLC. Notably, this resistance could be mitigated with CB-839, a small-molecule glutaminase inhibitor that potentiates radiation-induced DNA damage by reducing the free-radical scavenging capacity of cancer cells. Furthermore, sequencing of peripheral blood samples has revealed that mutations in the SWI/SNF chromatin-remodeling complex may serve as stratification biomarkers, as they are enriched in RT-sensitive patient cohorts (35). Table I provides an overview of the key mutations associated with differential radiosensitivity in lung cancer.

SNPs have been extensively investigated for their potential role in modulating radiosensitivity (Table II). Variants in excision repair cross-complementation (ERCC)1 and inhibitor of apoptosis-stimulating protein of p53 (iASPP), both key mediators of DNA repair, have been associated with differential responses to chemotherapy and combined chemoradiation in patients with NSCLC (36). Certain iASPP polymorphisms may result in decreased protein levels, leading to activation of ASPP1 and ASPP2 and enhanced p53-mediated apoptosis, thereby sensitizing tumors to RT (37). A combined SNP signature involving ERCC2 rs238406 and ERCC1 rs11615 has also been notably associated with radiosensitivity in LC (38), consistent with the pivotal role of these genes in the nucleotide excision repair pathway (39).

Additional polymorphisms in DNA repair genes, including X-ray repair cross-complementing protein XRCC1 (rs25487), XRCC2 (rs3218556) and XPD (rs13181), have been linked to RT response and OS in NSCLC. Notably, the XRCC1 rs25487 CC genotype and XRCC2 rs3218556 AG/AA genotypes were more frequently observed among responders than non-responders (40). These findings highlight the potential utility of these SNPs as predictive biomarkers for stratifying patients probably to benefit from RT.

Polymorphisms in the promoter region of TGFβ1 have also been associated with increased radiosensitivity (41), suggesting their clinical potential as predictive markers for favorable RT response. Similarly, heat shock protein β-1 rs2868371 polymorphisms have been shown to influence treatment outcomes: In U.S. patients with NSCLC, the CC genotype was associated with poor OS following RT or chemotherapy (42), whereas in a Chinese cohort, the C allele was linked to increased radiosensitivity and improved survival, probably due to reduced Hsp27 expression and impaired DNA repair capacity (43). These discrepancies underscore the influence of genetic background, ethnic differences in allele frequencies and clinical heterogeneity (such as treatment regimens and disease stage) on study outcomes (42).

Butkiewicz et al (44) reported that EGFR polymorphisms associated with increased gene expression and aberrant protein activity may predict clinical outcomes following DNA-damaging treatments. Specifically, EGFR rs2227983 GG and rs712829 TT genotypes were associated with tumor recurrence and acted as independent predictors of poor prognosis, while rs712830 CC genotype was associated with increased mortality risk. Given the established link between EGFR upregulation, reduced RT response and shortened survival (45), these findings support the utility of EGFR genotyping as both a prognostic tool and a predictor of radioresistance. Incorporating EGFR polymorphism screening into clinical practice could refine risk stratification and guide therapeutic decision-making.

Moreover, SNPs in autophagy-related genes have emerged as promising predictive biomarkers. For example, the autophagy related 16 like 2 gene rs10898880 CC homozygous variant has been reported to be associated with improved local recurrence-free survival, PFS and OS (46).

Despite these advances, most studies have examined individual SNPs in isolation. RT response is a multifactorial phenotype shaped by the interplay of multiple genetic variants and environmental exposures. Interactions between genes regulating DNA repair, apoptosis and cell-cycle checkpoints can synergistically influence radiosensitivity, as can gene-environment interactions such as smoking status, comorbidities and baseline clinical characteristics. The lack of integrative, systems-level analyses remains a key limitation in the field. Well-designed studies incorporating polygenic risk models and environmental data are needed to improve prediction accuracy and enable clinically actionable stratification. Furthermore, the underlying mechanisms by which these SNPs influence RT outcomes remain incompletely characterized, limiting their current utility as robust predictive biomarkers. Reliance solely on previously reported functional variants also risks overlooking rare but clinically meaningful SNPs that could inform precision RT strategies.

Several approaches have been developed to identify differential gene expression patterns between radiosensitive and radioresistant samples. One widely cited method is the radiosensitivity index, proposed by Torres-Roca et al (47), which quantifies intrinsic tumor radiosensitivity by measuring the surviving fraction at 2 Gy. Through gene expression profiling, three genes, RbAp48, RGS19 and R5PIA, were identified the expression levels of which were reported to be associated with radiosensitivity, suggesting their potential as predictive biomarkers for patient selection prior to RT (47). Table III summarizes representative studies identifying gene expression signatures that influence radiotherapy response in lung cancer.

Building on these efforts, Peinado-Serrano et al (48) used data from The Cancer Genome Atlas LC cohort to establish a six-gene predictive signature (APOBEC3B, GOLM1, FAM117A, KCNQ1OT1, PCDHB2 and USP43) associated with OS and PFS. This gene panel displayed differential expression according to irradiation phenotype in both NSCLC cell lines and patients with stage I–III NSCLC treated with RT, supporting its utility for patient stratification and prediction of long-term outcomes. Similarly, an eight-gene risk model (APOBEC3B, DOCK4, IER5L, LBH, LY6K, RERG, RMDN2 and TSPAN2) was found to independently predict prognosis and immune infiltration. Among these, LBH has been experimentally validated as a functional biomarker: Its upregulation enhances the radiosensitivity of A549 cells, whereas its silencing restores radioresistance. These findings suggest that LBH, as well as multi-gene risk scores, hold promise for prognostic stratification and therapeutic decision-making in NSCLC (49).

Transcriptome sequencing of p53 wild-type A549 and p53-deficient H1299 NSCLC cell lines, along with their radioresistant derivatives (A549IR and H1299IR), has revealed shared molecular signatures associated with radioresistant phenotypes (50). Specifically, seven genes (ATRNL1, CA2, CNR1, FAM189A1, GFRA1, RASGRP1 and RGL3) were upregulated, whereas nine genes (ADGRF1, EPHA7, LOX, LY6G5C, NSUN7, SLC22A31, SNAI2, TNFRSF11B and ZNF233) were downregulated. These dysregulated pathways represent potential predictive biomarkers and therapeutic targets.

Pathway-level analyses further highlighted activation of the WNT signaling cascade, including WNT ligand biosynthesis, WNT5A-FZD4 internalization and canonical WNT/β-catenin signaling, together with NF-κB pathway upregulation. Radioresistant H1299IR cells exhibited elevated β-catenin levels due to sustained GSK3 activity downstream of AKT, promoting enhanced cellular motility and invasion. These transcriptomic alterations were accompanied by a senescence-associated secretory phenotype, characterized by increased IL6, CXCL8 and TGFβ2 expression and decreased NRG1 expression, thereby supporting tumor progression (50).

Other studies have identified additional gene panels with potential clinical relevance. Eight genes (FOLR3, SLC6A11, ALPP, IGFN1, KCNJ12, RPS4XP22, HIST1H2BH and BLACAT1) have been shown to be associated with RT response in patients with NSCLC (51). Moreover, profiling of drug-tolerant persister cancer cells has yielded a four-gene prognostic risk index (LYNX1, SYNPO, GADD45B and PDLIM1), the co-expression network of which was revealed to be notably associated with RT response and OS (52).

Given the established link between autophagy and radioresistance, Gao et al (53) developed a three-gene autophagy-related prognostic model (SHC1, NAPSA and AURKA), in which SHC1 and AURKA upregulation, combined with NAPSA downregulation, predicted enhanced radioresistance. Furthermore, strong expression of DNA repair genes (ERCC1, ERCC4, ERCC5, XRCC5, RAD9A, PRKDC and the gene encoding DNA-PK), as well as proliferation- and apoptosis-related genes (MDM2, PIM2, BCL2 and PKCZ), has been observed in radioresistant A549 cells compared with the more radiosensitive NCI-H446 line, reinforcing their potential as biomarkers for treatment stratification and combination therapy development (54).

Emerging evidence has also implicated metabolic pathways in modulating RT response. A recent study identified a three-gene lactate metabolism-related risk model (ADAMTS3, FADS2 and RTBDN) capable of distinguishing radiosensitive from radioresistant patients, with FADS2 emerging as a key predictive biomarker. Further mechanistic validation linked this signature to DNA repair modulation and tumor immune response, highlighting lactate metabolism as a promising source of novel biomarkers for precision RT (55).

Finally, keratin 6A (KRT6A) expression has been associated with improved OS in patients with LC. Functional experiments have shown that KRT6A downregulation reduces proliferation and invasion in vitro, suggesting that while KRT6A may not directly mediate radioresistance, it could serve as a valuable prognostic marker and potential therapeutic target (56).

Despite these advances, translating radiosensitivity gene-expression signatures from preclinical models to clinical practice remains challenging. Cell lines offer a controlled platform for dissecting radiation response independent of clinical confounders, but they lack microenvironmental, stromal and immune interactions, realistic dose-rate effects and intratumoral heterogeneity. Therefore, validation in physiologically relevant models, such as patient-derived xenografts (PDXs), organoids and prospective clinical cohorts, remains essential before these signatures can be reliably implemented in routine clinical decision-making.

ctDNA represents a fraction of cell-free DNA released from tumor cells through necrosis, apoptosis, macrophage-mediated clearance or direct secretion (57). LC exhibits marked intratumoral heterogeneity, both spatially, across primary and metastatic sites and temporally, as molecular characteristics evolve during disease progression. Consequently, liquid biopsy offers a practical and minimally invasive strategy for quantifying ctDNA levels in the bloodstream, providing a dynamic snapshot of tumor burden, stage and treatment response.

This approach holds particular value in radiation oncology, as ctDNA measurement enables real-time monitoring of therapeutic efficacy, detection of resistance emergence and surveillance of disease progression during RT (58). Changes in ctDNA kinetics during treatment closely mirror the cytotoxic effects of ionizing radiation and tumor cell death, positioning ctDNA as a promising biomarker of radiation response (59).

Recent evidence has demonstrated that a decline in ctDNA levels during RT in patients with NSCLC is associated with a reduction in tumor burden and may serve as an early predictor of treatment response, even in cases where complete ctDNA clearance is not achieved (35). These findings support the integration of ctDNA monitoring into clinical practice to guide adaptive RT strategies and enable timely treatment modifications.

Using a proteomics approach combining two-dimensional electrophoresis and matrix-assisted laser desorption ionization-time of flight mass spectrometry, Yun et al (60) identified several proteins with notably increased expression in radioresistant H460 cancer stem cell (CSC) lines compared with radiosensitive parental H460 cells. These candidate biomarkers included fatty acid synthase, vimentin, glucose-regulated protein 78, plasminogen activator inhibitor-2, nodal modulator 2, kinesin light chain 4 and procollagen-lysine, 2-oxoglutarate 5-dioxygenase 3 (PLOD3) (Fig. 2). Their upregulation may serve as predictive indicators of RT response, enable pre-treatment identification of radioresistant tumors and represent promising therapeutic targets to overcome resistance. Table IV summarizes proteomic biomarkers identified in multiple studies that are associated with radioresistance in lung cancer.

CSCs, a subpopulation within tumors with self-renewal, differentiation and tumor-initiating capacity, have been shown to be intrinsically more resistant to both irradiation and systemic therapies than non-CSCs (61). This property may contribute to post-RT recurrence and metastasis, highlighting the need for innovative RT strategies aimed specifically at eradicating CSCs (60).

Hypoxia is a well-recognized driver of radioresistance. In NSCLC, serial measurements of serum hypoxia-inducible factor 1α (HIF-1α) before and after chemoradiotherapy have revealed notable decreases during treatment, associated with tumor shrinkage. These findings suggest that HIF-1α could serve as a dynamic biomarker for monitoring tumor response (62).

Angiopoietin-like 4 protein (ANGPTL4) has also been implicated in NSCLC radioresistance. Under hypoxic conditions, ANGPTL4 upregulation suppresses ferroptosis by upregulating antioxidant proteins, thereby enhancing tumor cell survival after irradiation. Furthermore, ANGPTL4 is enriched in exosomes secreted by hypoxic tumor cells, which transfer radioresistance signals to neighboring normoxic cells through indirect ferroptosis inhibition. These findings position ANGPTL4 as both a predictive biomarker and a potential therapeutic target to enhance radiosensitivity (63).

Other hypoxia-induced molecules include osteopontin (OPN) and heat shock protein 70 (HSP70). Elevated pre-RT plasma OPN levels have been associated with poor OS, whereas post-treatment reductions in HSP70 within 4–6 weeks are associated with tumor regression. Notably, patients with relatively high post-RT HSP70 levels demonstrate improved treatment responses, probably reflecting an immunostimulatory effect of extracellular HSP70 released from dying tumor cells (64). Together, these results support the combined assessment of OPN and HSP70 as predictive and dynamic biomarkers of treatment response.

Additional hypoxia-related biomarkers, VEGF and carbonic anhydrase IX, have been linked to poor OS, with their expression associated with adverse clinical features such as weight loss, reduced lung function and larger tumor size. These markers may help identify highly hypoxic, aggressive tumors, supporting their integration into multi-center validation studies and personalized treatment planning (65).

Incorporation of blood-based biomarkers into prognostic models has been shown to notably improve survival prediction in NSCLC. Among a panel of eight analytes, serum CEA and IL-6 have been identified as independent predictors of poor outcomes (66). Similarly, cytokine profiling of patients undergoing chemoradiotherapy has revealed that elevated post-treatment levels of IL-13, TNF-α, IL-8, CXCL5 and CXCL3, together with reduced levels of intercellular adhesion molecule 1 (ICAM-1), IFN-γ and soluble programmed cell death ligand 1 (PD-L1), are associated with improved objective response. Combinations such as IL-8 with ICAM-1 or soluble PD-L1 with TNF-α may further enhance predictive accuracy and IL-8/ICAM-1 levels are associated with PFS (67). These findings highlight plasma cytokines as valuable tools for stratifying patients and optimizing RT strategies.

Proteomic profiling of H460 LC cells pre- and post-irradiation has identified fibronectin 1 and thrombospondin 1 as key contributors to radioresistance and poor outcomes (68). Yao et al (69) demonstrated that cyclin K may stabilize β-catenin and promote G₂/M checkpoint activation, thereby enabling DNA repair and survival. By contrast, silencing cyclin K increased radiosensitivity in vitro and in vivo by abrogating this checkpoint, suggesting cyclin K as both a predictive biomarker and therapeutic target.

Cheng et al (70) reported that PRMT1-mediated methylation of plakophilin 2 (PKP2) can recruit ubiquitin-specific protease 7, which stabilizes DNA ligase 4 and enhances non-homologous end-joining repair. Pharmacological inhibition of PRMT1 has been shown to disrupt this axis and sensitize NSCLC cells to RT, supporting PKP2 as a novel predictive biomarker and therapeutic target. Similarly, knockdown of PLOD3 triggers ER stress and caspase-3-mediated apoptosis, thereby enhancing radiosensitivity (71).

Ras p21 protein activator, a Ras GTPase-activating protein, has been demonstrated to promote p53 phosphorylation and proteasomal degradation, leading to reduced apoptosis and increased radioresistance. Its expression is associated with poor RT response, underscoring its value as a biomarker and target for therapeutic intervention (72). Upregulation of F-box protein 22 (FBXO22) activates the FOXM1/Rad51 axis to promote homologous recombination repair, whereas its inhibition, particularly with the small molecule deguelin, restores radiosensitivity (73). Conversely, high expression of MDMX has been shown to be associated with enhanced response to RT and longer OS, implicating it as a favorable biomarker in LUAD and LSCC (74).

Cartilage oligomeric matrix protein, an extracellular matrix protein, promotes NSCLC proliferation and invasion following irradiation, indicating its potential as a biomarker for identifying patients likely to benefit from intensified treatment or combination therapy (75).

The TME serves a critical role in RT response. High stromal expression of fibulin-5 (FBLN5) in cancer-associated fibroblasts (CAFs) activates Src-STAT3 signaling, inhibits ferroptosis and promotes post-irradiation tumor survival. Clinically, elevated FBLN5 levels are associated with poor RT outcomes, supporting its role as a predictive biomarker (76). Similarly, flotillin-1 (FLOT1) upregulates PD-L1 and promotes epithelial-mesenchymal transition (EMT), enhancing cell motility and DNA damage tolerance. FLOT1 inhibition sensitizes cells to RT, activates the STING pathway and enhances CD8⁺ T-cell recruitment through chemokine secretion (77).

In a large retrospective study of 328 patients with stage III NSCLC treated with concurrent chemoradiation and durvalumab, high PD-L1 expression (TPS ≥90%) and elevated tumor mutational burden were associated with prolonged disease control, supporting their use for patient stratification prior to RT and immunotherapy (78).

A prior multi-omics analysis comparing parental A549 cells and their radioresistant derivatives has revealed >900 differentially expressed genes and ~700 proteins, including >30,000 sites of lysine 2-hydroxyisobutyrylation (khib). Notably, EGFR was shown to be both upregulated and highly khib-modified in radioresistant cells, suggesting that khib-regulated EGFR may represent a novel mechanism of radioresistance and a therapeutic vulnerability (79). Fig. 2 provides a comprehensive summary of proteomic biomarkers associated with lung cancer radioresistance.

Finally, systemic inflammatory markers such as neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio, derived NLR and systemic immune-inflammation index have shown prognostic and predictive value in locally advanced NSCLC. Low values of these markers have been shown to be independently associated with improved PFS, OS and locoregional control and to be linked to increased radiosensitivity (80). These metrics are inexpensive, widely accessible and may complement molecular biomarkers in patient stratification and treatment planning.

Epigenetic regulation serves a pivotal role in shaping the response to irradiation in LC. Radiosensitive and radioresistant NSCLC cell lines display distinct methylation signatures, underscoring the involvement of epigenetic mechanisms in radioresponse. Genome-wide analysis of CpG methylation patterns has revealed that several key regulators of radiosensitivity are epigenetically controlled by CpG methylation (81). Notably, the presence of an unmethylated promoter in the insulin-like growth factor-binding protein-3 gene has been associated with poor RT response and unfavorable clinical outcomes in patients receiving adjuvant chemoradiotherapy (82). Furthermore, radiation-induced changes in DNA methylation vary in a tissue-specific and dose-dependent manner, highlighting the complexity of epigenetic modulation in RT response (83).

miRNAs are short (~22 nucleotides) non-coding RNAs that post-transcriptionally regulate gene expression by binding to complementary sequences on target mRNAs (84). Their notable stability in tissues, plasma and other biofluids makes them attractive candidates for minimally invasive biomarkers (85). Numerous differentially expressed miRNAs have been implicated in LC radiosensitivity (Fig. 3).

For example, miR-214 is upregulated in radioresistant NSCLC cells, where it enhances radioresistance and inhibits RT-induced apoptosis (86). Similarly, miR-410 promotes EMT and radioresistance by augmenting DNA damage repair capacity, suggesting its potential as a therapeutic target (87). Serum levels of miR-130a, miR-191 and miR-25 have been shown to be increased following 2–4 Gy irradiation in patients with NSCLC, and are associated with heightened invasiveness of A549 cells in vitro and in xenograft models, thus supporting their potential role as biomarkers of RT-induced metastasis (88). Elevated exosomal miR-96 has been associated with vascular invasion and poor OS, indicating its prognostic value for radioresistant NSCLC (20).

Plasma miRNA signatures also hold promise for stratifying patients. Hsa-miR-302e, hsa-miR-98-5p, hsa-miR-613 and hsa-miR-495-3p have all been linked to radiosensitivity in NSCLC (89), while miR-18a-5p levels have been associated with treatment response and could be used to categorize patients into radiosensitive or radioresistant subgroups (90). Zhang and Hu (91) identified miR-148a as a candidate biomarker of radioresistance, whereas miR-26b-5p was found to be notably downregulated in LUAD tissues, serum and radioresistant cell lines. Upregulation of miR-26b-5p increases radiosensitivity in both parental LUAD cells and radiation-resistant LUAD cells (A549R) by promoting DNA damage and apoptosis, partly through direct inhibition of activating transcription factor-2. Notably, exosomal transfer of miR-26b-5p from A549 cells to resistant cells may restore radiosensitivity, underscoring its value as a non-invasive predictive marker (92).

Serum exosomal miR-378 has emerged as a clinically relevant biomarker; its levels are notably decreased after RT and high baseline expression is associated with poor OS, making it an independent prognostic factor (93). Similarly, miR-148a, a tumor suppressor, is downregulated in radioresistant patients but upregulated following RT in radiosensitive ones. Its upregulation inhibits proliferation, migration and invasion by targeting SOS2, thereby improving radiosensitivity and serving as a potential marker for both prediction and monitoring of RT response (91).

LncRNAs (>200 nucleotides) have emerged as key regulators of carcinogenesis, immune modulation, apoptosis and therapy response (94). Several lncRNAs have been implicated in RT outcomes in NSCLC, making them promising prognostic and predictive markers (Fig. 3).

LINC01224 expression has been associated with RT response, suggesting its potential as a prognostic indicator (95). Song et al (96) identified a four-lncRNA signature (CASC19, LINC01977, LINC02471 and MAGI2-AS3) that was shown to be associated with radiological response and could accurately predict patient prognosis.

HNF1A-AS1, a lncRNA upregulated in NSCLC, contributes to advanced disease stages and radioresistance by sponging miR-92a-3p, which results in MAP2K4 upregulation and JNK pathway activation. Silencing HNF1A-AS1 has been reported to reduce proliferation, induce apoptosis and notably sensitize NSCLC cells to RT, positioning the HNF1A-AS1/miR-92a-3p/MAP2K4 axis as a potential biomarker and therapeutic target (97).

Notably, male patients with NSCLC exhibit a unique pattern of radiosensitivity linked to Y-chromosome loss (LOY) and dysregulation of Y-linked lncRNAs. Linc-SPRY3-2/3/4 transcripts are dose-dependently induced after irradiation in radiosensitive cell lines but absent in LOY radioresistant lines. These transcripts interact with IGF2BP3 to destabilize oncogenic mRNAs, including HMGA2 and c-MYC, thereby promoting apoptosis. Loss of linc-SPRY3-2/3/4 or LOY is associated with poorer OS, suggesting a male-specific biomarker signature for predicting radiosensitivity (98).

Fig. 4 provides an integrated overview of the key genetic, epigenetic, circulating and proteomic biomarkers that have been implicated in LC radioresistance, highlighting their potential roles in patient stratification and personalized RT strategies. The figure provides an overview of the main molecular biomarkers implicated in LC radioresistance and their associated pathways. It includes genetic alterations (such as EGFR, KRAS, TP53, KEAP1 and NFE2L2), DNA repair factors (Cyclin K, PRMT1 and FBXO22), non-coding RNAs (miRNAs and lncRNAs) and CSC markers. Contributors to the TME, such as hypoxia, stromal CAFs and key signaling pathways (HIF-1α, VEGF, PD-L1) are also shown. Collectively, these biomarkers and pathways interact to promote tumor survival, treatment resistance and disease progression, underscoring potential targets for improving RT outcomes.