Current data indicate that nearly all SCLC cells express one or more of the four key transcription factors—ASCL1, NEUROD1, POU2F3, and YAP1—or exhibit inflammatory gene expression characteristic of the SCLC-I subtype (8). ASCL1 and NEUROD1, both members of the bHLH transcription factor family, are central to regulating NE differentiation (35). Historically, studies suggested that ASCL1 and NEUROD1 are independently expressed, rarely showing co-expression (13, 16, 23). However, RNA-seq analysis by Gopal et al. revealed that a single SCLC tumor may express one or more transcription factors, including ASCL1, NEUROD1, and YAP1, underscoring the complexity of intratumor heterogeneity (36). Emerging evidence indicates that co-expression of ASCL1 and NEUROD1 is more common than previously thought, occurring in 20% to 40% of SCLC tumors (21, 26). Zhang et al. found that the number of tumors with dual high expression of ASCL1 and NEUROD1 exceeds those with high NEUROD1 expression alone, suggesting the presence of distinct subgroups within the tumor that exhibit differing biological behaviors and therapeutic responses (20). This further blurs the boundaries of existing SCLC subtype classifications. Baine et al. observed that almost all cases of dual high expression occurred within the same tumor cell population, with only one case showing subclonal regions with differential ASCL1 and NEUROD1 expression (21). Additionally, studies have noted the occurrence of ASCL1 and NEUROD1 co-expression in SCLC samples in a mutually exclusive manner, implying that co-expression may be restricted to spatially distinct regions (8). This intratumor heterogeneity has also been confirmed in SCLC CDX models, where spatially distinct regions of ASCL1- and NEUROD1-positive cells were found in co-expressing tumors (26). Patient tumor samples co-expressing ASCL1 and NEUROD1 suggest that SCLC-A cells are derived from highly dedifferentiated normal NE cells, while SCLC-AN cells, which co-express both transcription factors, exhibit a higher degree of differentiation. This differentiation pattern hints that SCLC-AN could be classified as a distinct subtype (37, 38). Further investigation into SCLC-AN revealed that although ASCL1 and NEUROD1 are occasionally co-expressed at the tissue level, their expression is largely mutually exclusive at the cellular level. Induction of co-expression in SCLC cell lines demonstrated that concurrent expression of these two transcription factors results in reciprocal suppression, leading to impaired cell growth and increased apoptosis (39). A newly identified H subtype frequently co-expresses ASCL1 and NEUROD1, and its drug resistance involves various NADPH metabolic enzymes downstream of the KEAP1-NRF2 pathway, suggesting the existence of a gene-transcript-metabolism reprogramming-mediated resistance pathway. Integrated immunohistochemistry and spatial copy number variation analysis of a rare case of pulmonary neuroendocrine carcinoma confirmed the coexistence of both A and H subtypes, revealing their origin from a common ancestral clone followed by divergent evolutionary trajectories (29). The co-expression and mutual exclusivity patterns observed in intratumor heterogeneity emphasize the need for personalized treatment strategies for patients, tailored to their unique expression profiles.

In contrast, non-NE subtypes are defined by a fundamentally different biology. POU2F3 and YAP1 are key markers of non-NE subtypes in SCLC. POU2F3 is a transcription factor selectively expressed in tuft cells, a rare type of chemosensory cell found in the respiratory tract that responds to external stimuli by releasing bioactive substances to regulate local epithelial and immune cell functions (40). Using CRISPR gene-editing technology, Huang et al. identified POU2F3 as being overexpressed in the low-NE subgroup of SCLC, linking it to the growth of various human cancer cells, including SCLC (40). Further studies confirmed that POU2F3 is primarily expressed in SCLC tumors that are negative for both ASCL1 and NEUROD1, representing 90% (9 out of 10) of cases (21). Single-cell studies by Baine et al. supported this finding, showing that POU2F3 expression in SCLC is mutually exclusive with ASCL1 and NEUROD1 (21). However, Gay et al. found that although less than 1% of cells in patient-derived SCLC CDX models expressed POU2F3, all POU2F3-positive cells co-expressed ASCL1, challenging the exclusivity (8). Additionally, research from the Berns lab suggests that different cells within the respiratory epithelium can give rise to SCLC-like tumors, indicating that the cellular origin of various subtypes may differ (41). The expression profile of SCLC-P closely mirrors that of pulmonary tuft cells, implying a distinct cellular origin compared to other subtypes (13). This distinction could account for its intratumor heterogeneity, although the small sample size of studies on POU2F3 necessitates further research into its heterogeneity.

YAP1, a key regulator of malignant transformation in numerous tumors, plays a critical role in controlling cell proliferation and stem cell growth (42). Its function as a transcription factor defining distinct isoforms in SCLC has been the subject of debate. While YAP1 is expressed across all subtypes, it is predominantly found in tumors negative for both ASCL1 and NEUROD1 (21). Moreover, studies indicate that YAP1 expression in SCLC-Y cell lines, enriched with SMARCA4 mutations, exhibits characteristics of SMARCA4-deficient malignancies rather than traditional SCLC (22). The expression patterns and intratumor heterogeneity of POU2F3 and YAP1 are closely tied to non-NE subtype traits (21). Further investigation into the intratumor heterogeneity of non-NE-specific transcription factors like POU2F3 and YAP1 could provide new insights into therapeutic strategies for targeting these subtypes.

From a proteomics standpoint, direct evidence has emerged elucidating the intratumoral heterogeneity of SCLC. In 2024, Liu et al. conducted an integrative multi-omics analysis of 107 SCLC samples, categorizing SCLC into four distinct clusters (nmf1 to nmf4) based on protein and phosphoprotein profiles. Among these, the nmf1 subtype exhibited the highest neuroendocrine (NE) score, whereas the nmf3 subtype was enriched in mesenchymal markers, systematically revealing fundamentally divergent protein functional states across SCLC subtypes (28). This heterogeneity is further manifested in the spatial tumor microenvironment: Baine et al. demonstrated via multiplex immunofluorescence that within the same tumor, NE subpopulations (e.g., SCLC-A) specifically overexpress proteins such as INSM1 and ASCL1, while adjacent non-NE subpopulations (e.g., SCLC-P) are enriched with POU2F3 and Vimentin. These mutually exclusive protein expression domains define significant spatial intratumoral heterogeneity. Such heterogeneous protein distribution directly impacts therapeutic response. A prominent example is the patchy distribution of the targetable protein DLL3 within tumors, where only a subset of cell subpopulations highly expresses DLL3, while others show low expression. This spatial limitation explains why targeted agents like Rova-T can only eliminate a fraction of tumor cells, ultimately leading to treatment resistance—a phenomenon widely confirmed by immunohistochemistry across numerous preclinical and clinical studies (21).

Acknowledging the limitations of classical molecular subtyping in predicting clinical outcomes, another study utilized spatial transcriptomics to classify SCLC into three phenotypes based on intratumoral heterogeneity (ITH) levels: high-ITH (h-ITH), medium-ITH (m-ITH), and low-ITH (l-ITH). GO enrichment analysis indicated that the h-ITH phenotype correlates with cell fate and differentiation processes, m-ITH is enriched for immune-related pathways, and l-ITH is associated with cellular stress responses and organogenesis (43). Research employing CODEX and Visium spatial multi-omics technologies demonstrated that tumor regions with a high MPTC signature, characterized by co-expression of ASCL1 and NEUROD1, were linked to poor patient prognosis. Conversely, regions enriched with the MT2 immune niche, defined by aggregates of M1 macrophages, CD8-positive T cells, and NKT cells, were significantly correlated with favorable clinical outcomes. Further analysis confirmed that the proportion of the MT2 niche is a superior prognostic predictor, independent of both tumor mutational burden (TMB) and PD-L1 expression levels (44). Similarly, spatial transcriptome analysis classified SCLC into Epithelial-Resistant (Epi-I) and Epithelial-Sensitive (Epi-II) subtypes. The Epi-II subtype secretes MIF, which educates M2-type myeloid cells. These cells, in turn, release SPP1, activating the PI3K-AKT pathway and driving the transition from the sensitive Epi-II to the highly proliferative and resistant Epi-I subtype. Targeting the MIF/SPP1 axis may block this transition process, offering a new direction for improving patient outcomes (45). The functional specialization and dynamic interactions among these spatial compartments not only provide a profound explanation for the intricate mechanisms underlying intratumoral heterogeneity in SCLC but also elucidate how the local microenvironment dictates the functions of distinct subpopulations. Thereby, these findings offer critical spatial-dimensional insights for precisely locating high-risk tumor regions and optimizing therapeutic strategies.

Epigenetic reprogramming and dysregulated molecular networks are central mechanisms amplifying intratumoral heterogeneity and driving therapy resistance in SCLC. Pharmacological inhibition of EZH2, the histone methyltransferase for H3K27me3, has been shown to restore T cell-mediated killing and upregulate MHC class I expression. These findings indicate that EZH2 inhibition enhances tumor immunogenicity and may subsequently improve responses to immune checkpoint inhibitors in SCLC patients (46–48). Additionally, preclinical evidence demonstrates that EZH2 suppression facilitates the loss of the neuroendocrine phenotype, leading to upregulated SLFN11—a key factor that induces lethal replication blockade in response to DNA-damaging agents (47). Alternatively, a chromatin accessibility-based framework has been proposed, categorizing SCLC into novel subtypes (SCLC-NE, SCLC-IM, SCLC-SL). Notably, the stem-like SCLC-SL subtype demonstrates a strong correlation with chemotherapy resistance and unfavorable patient outcomes, thereby establishing it as an independent prognostic predictor (49). Another integrative epigenomic and transcriptomic profiling uncovered a NEUROD1-driven regulatory axis in the SCLC-N subtype, involving PDE2A/miR-139-5p as a subtype-specific marker. This study also delineated a cooperative miRNA targeting mechanism directed against genes including NFIB and NOTCH1, which underlies the molecular heterogeneity in SCLC (38). Characterization of super-enhancers, defined by genome-wide histone modifications, serves as a powerful tool for delineating the lineage of unclassified tumors. The study applied this approach and revealed two distinct epigenomic subclusters within the major SCLC-A subtype: SCLC-Aα and SCLC-Aσ. The SCLC-Aα subcluster is characterized by a core regulatory circuitry formed by the super-enhancers of NKX2–1 and SOX1, which function collaboratively to maintain the neuronal lineage state (50). The NE and non-NE dichotomy in SCLC is epigenetically enforced through distinct DNA methylation landscapes. The new classifiers developed based on these profiles—SCLC-DMC (for tissue) and cfDMC (for circulating tumor DNA)—demonstrated perfect concordance with the RNA-seq gold standard. Most importantly, longitudinal tracking via liquid biopsy using cfDMC revealed therapy-driven subtype conversion, such as a shift from SCLC-A to SCLC-I upon disease progression, which was accompanied by promoter methylation alterations in immune-related genes including CXCL12 (51). In addition, EGFR TKI treatment can induce epigenetic reprogramming in EGFR-mutant lung adenocarcinoma (LUAD), facilitating its transformation into SCLC through a process that involves transitional cell states (52). In a separate classification effort, transformed SCLCs (T-SCLCs) have been categorized into a “LUAD-feature retained” subgroup with high NKX2–1 expression and a “LUAD-feature absent” subgroup harboring canonical SCLC genomic features like TP53/RB1 co-inactivation. These two subtypes exhibit significant differences in their transcriptomes, patient outcomes, and responses to therapy (53).

Furthermore, intratumoral heterogeneity in SCLC manifests through distinct, coordinated functional layers. At the cellular differentiation level, diffusion pseudotime (DPT) analysis delineated a continuous differentiation spectrum from basal cells to NE or tuft-like cells, namely the “Basal → Atoh1+ → NE/tuft” lineage. Specifically, basal cells first transition into an early Atoh1+ state, which subsequently differentiates further into NE subtypes (SCLC-A/SCLC-N) or the tuft-like subtype (SCLC-P), forming a “trunk-branch” differentiation model. This implies that therapeutic strategies might need to target the differentiation trunk while eliminating the various branched subtypes (54). Furthermore, recent studies highlight electrical activity as a core driver of SCLC malignancy, enhancing metastasis and drug resistance via calcium-dependent signaling pathways (e.g., CREB/FOS). Metabolic heterogeneity, with NE subpopulations relying on OXPHOS and non-NE subpopulations secreting lactate, is key for inter-subpopulation collaboration and also represents a therapeutic vulnerability. Molecules like SOX1, p-CREB, and nAChR show promise as biomarkers for patient stratification towards precision therapy (55). The interplay between tumors and the neural microenvironment is also gaining attention. SCLC cells co-cultured with neurons exhibit enrichment of synapse-related gene signatures, including key genes such as NRXN1, NLGN1, and HOMER1. In vivo experiments confirmed that activating cortical neurons (using Thy1-ChR2 transgenic mice) or selectively activating GABAergic interneurons alone (using Dlx-ChRmine transgenic mice) significantly promotes the proliferation of intracranial SCLC, providing a rationale for targeting neuron-tumor interactions (56). Further mechanistic investigation suggests that SCLC hijacks neuronal glutamatergic (and partially GABAergic) synaptic signaling, combined with its intrinsic pulmonary neuroendocrine cell (PNEC)-like phenotype to form synapses, establishing a “neural signaling → SCLC proliferation → neural hyperexcitability” positive feedback loop that exacerbates tumor progression (57). In terms of clinical translational exploration, classification based on TMB indicates that high TMB (TMB-h) is associated with better response to immune checkpoint inhibitors in SCLC (58). Trophoblast cell surface antigen 2 (Trop-2), a cell surface protein implicated in the malignant potential of several cancers including SCLC, has emerged as a potential therapeutic target (59). A phase II study (TROPiCS-03) involving 43 patients with advanced SCLC reported that 41.9% of patients experienced significant tumor shrinkage with a Trop-2-targeting antibody-drug conjugate (ADC), demonstrating promise (60). Other molecular features such as MYC family oncogene amplification, PTEN inactivation, and NOTCH pathway mutations are also implicated in SCLC progression (61). Another study used deep learning to analyze HE-stained whole-slide images, quantified SCLC intratumoral heterogeneity by identifying its histomorphological phenotypes (e.g., HIPO subtypes, HPCs), classified SCLC into subtypes with significant prognostic differences (e.g., HIPOS-I, HIPOS-II) to achieve heterogeneity-based precision prognostic assessment (62). However, biomarker-driven classification systems currently face challenges. Firstly, the high heterogeneity of SCLC limits the utility of single biomarkers for accurate classification (63). Secondly, the co-expression or mutual exclusivity patterns of various biomarkers suggest their expression may vary across different tumor regions. Looking forward, future research should focus on elucidating the driving mechanisms behind the dynamic evolution of SCLC subtypes, understanding the impact of spatial architecture on functional states, and developing multi-target combination or sequential strategies to overcome resistance challenges posed by subtype switching. Constructing dynamic subtyping systems integrating multi-omics data and artificial intelligence algorithms will be crucial for achieving precision therapy in SCLC ( Figure 2 ).

The MYC family of genes, including MYC, MYCN, and MYCL, serves as key oncogenic drivers in various tumors. These MYC family proteins, classified as bHLH transcription factors, bind to E-box DNA sequences (CACGTG) and form heterodimers with smaller bHLH proteins like MAX to regulate target gene expression (68). In SCLC, overactivation of MYC family transcription factors influences cell proliferation, controls the cell cycle, and facilitates malignant transformation (69). Notably, the overexpression of MYCN or MYCL in chemotherapy-sensitive PDX models drives the development of chemoresistance in SCLC (70). Studies have demonstrated that MYCL is amplified or highly expressed in the SCLC-A subtype and plays a pivotal role in ASCL1 function, while other subtypes tend to display MYC amplification or overexpression, suggesting that MYC predominates in SCLC with low NE status (13, 69).

Research by Ireland et al. using SCLC genetically engineered mouse models (GEMM) revealed that MYC activates Notch/REST signaling, prompting the transition of SCLC from an ASCL1+ to a NEUROD1+ and eventually to a YAP1+ state. This transformation results in histological features resembling large-cell neuroendocrine carcinoma (LCNEC) (66). Similar patterns were observed in human SCLC cell lines H1963 and H187, providing valuable insights for the evaluation of clinical strategies (71). Ireland et al. demonstrated that loss of ASCL1 (RPMA model) significantly suppresses the neuroendocrine subtype and induces a transition toward SCLC-P and SCLC-Y subtypes. In contrast, PTEN loss (RPP model) activates the PI3K/AKT pathway and upregulates MYC, doubling the proportion of POU2F3+ cells in tumors of basal origin, thereby establishing a molecular mechanism for the cooperative regulation of SCLC-P (54). Additionally, MYC-driven SCLC has shown heightened sensitivity to Aurora kinase inhibitors—serine/threonine protein kinases that, when overexpressed, promote tumor cell proliferation and survival in SCLC (72, 73). Treatment of MYC-driven SCLC mice with the Aurora kinase inhibitor Alisertib, in combination with chemotherapy, significantly extended median survival time compared to controls (74). These findings suggest that MYC family members play a pivotal role in treatment sensitivity in SCLC, positioning them as critical biomarkers for patient stratification and potential targets for therapy (75).

The mammalian Notch signaling pathway comprises four Notch receptors and five DSL family ligands, interacting with multiple proteins (76). This pathway regulates key physiological functions, including cell proliferation, differentiation, organ development, and tissue homeostasis. Emerging evidence indicates that the Notch signaling pathway plays a pivotal role in the transformation between NE and non-NE characteristics in SCLC, potentially influencing 10% to 50% of tumor cells, contingent on the activation threshold required (65, 77). Target genes of the Notch pathway, such as HES1 and HEY1, are known to suppress ASCL1, a key regulator of NE differentiation in SCLC, thereby facilitating the transition toward a non-NE phenotype (78).

In human SCLC cell line H1688, activation of Notch1 resulted in a reduction of NE marker expression and induced a morphological shift to a glandular cell arrangement, with loosely clustered aggregates indicative of a more non-NE state (79). Both in vivo and in vitro studies suggest that, alongside MYC, factors like YAP1 can promote REST gene expression via a Notch-dependent mechanism, further driving the shift from NE to non-NE characteristics (80). The Notch pathway also has implications in SCLC immunotherapy. Transcriptomic analysis of primary SCLC tumors, patient-derived CDX models, and cell lines revealed that Notch activation upregulates MHC-I gene expression and enhances immune cell infiltration, rendering non-NE subgroups more susceptible to immunotherapy (81). Additionally, Notch1 loss is negatively correlated with PD-L1 expression and can drive the polarization of macrophages toward the M1 type. This reduction in NE differentiation corresponds with an increase in tumor-intrinsic immunity, suggesting a link between Notch signaling activation, the NE differentiation state, and the tumor immune microenvironment (82, 83). A more comprehensive understanding of the Notch pathway and its interplay with other signaling cascades may provide valuable insights for developing targeted therapies that leverage Notch-dependent mechanisms in specific SCLC subtypes.

Enhancer of Zeste Homolog 2 (EZH2), the catalytic subunit of Polycomb Repressive Complex 2 (PRC2), induces transcriptional silencing through tri-methylation of histone H3 at lysine 27 (H3K27me3), playing a critical role in cisplatin resistance in SCLC (84). Studies on samples from patients with SCLC, GEMMs, and human cell lines have shown that transient inhibition of EZH2 promotes a transition of SCLC cells from a high-NE state to an inflammatory low-NE state. EZH2 inhibition unblocks the TAP1 gene, a key player in MHC I antigen processing, and activates the STING pathway, a critical component of innate immune signaling⁷¹. This activation enhances the immune microenvironment by inducing cytokine production, such as interferon, thereby improving tumor immunogenicity. Specifically, EZH2 inhibition upregulates TAP1, restores MHC I expression on the tumor cell surface, and increases the presentation of antigenic peptides, making the tumor more likely to be recognized and targeted by the immune system (47). Moreover, immunohistochemical analysis revealed strong nuclear expression of EZH2 in all SCLC cells. In human SCLC cell lines (H146, H345), EZH2 silences the TGF-β type II receptor (TβRII) via epigenetic mechanisms, thereby inhibiting TGF-β-mediated apoptosis. Since TGF-β downregulates ASCL1 through the SMAD pathway, EZH2 promotes SCLC progression by blocking the TGF-β-SMAD-ASCL1 axis, suggesting that reducing EZH2 expression indirectly lowers ASCL1 levels (85). Additionally, EZH2 inhibition has been shown to activate p63 expression, facilitating the transdifferentiation of esophageal neuroendocrine carcinoma cells into squamous cell carcinoma, which improves drug sensitivity and enhances survival rates (86). However, in medulloblastoma (MB), EZH2 inhibition upregulated NEUROD1 and promoted cellular differentiation, highlighting the complexity of EZH2’s role in various cancer contexts (87). This evidence opens the possibility that EZH2 inhibition may facilitate subtype switching in SCLC, potentially influencing therapeutic outcomes. In PDX models, the addition of EZH2 inhibitors to standard chemotherapy restored the expression of key genes, such as SLFN11, which are silenced under chemotherapy pressure via EZH2-mediated methylation. This reactivation sensitizes tumor cells to chemotherapeutic agents, potentially delaying the development of acquired resistance and improving treatment efficacy (48). Several clinical trials are currently investigating EZH2 inhibitors in advanced solid tumors, including SCLC, such as a phase I trial of Mevrometostat (NCT03460977) and a phase II trial of XNW5004 (NCT06022757), though results are still pending (88). The potential of EZH2 inhibitors in overcoming cisplatin resistance and promoting subtype switching in SCLC presents a promising therapeutic avenue, offering a novel strategy for managing this aggressive cancer.

Lysine-specific demethylase 1 (LSD1), encoded by the KDM1A gene, is a lysine demethylase intricately linked to malignant transformation, EMT, cell proliferation, and differentiation across various cancers, making it a critical target for cancer therapies (89). LSD1 can bind to Notch sites, suppressing Notch1 expression and its downstream signaling, thus facilitating subtype transitions through the Notch pathway, which significantly influences SCLC development and progression (90). Inhibition of LSD1 in four human SCLC cell lines (H510A, H1417, H146, and H187) using ladademstat, followed by RNA-seq analysis, revealed activation of Notch signaling and upregulation of REST, leading to decreased expression of ASCL1 and other NE lineage genes (90). Another LSD1 inhibitor, T-3775440, was found to reduce NE markers such as CHGA and GRP in SCLC cell lines H1417 and H510A by directly disrupting the interaction between LSD1 and INSM1/GFI1B, leading to the downregulation of ASCL1 expression (91). Additionally, LSD1 can inhibit the key target gene ZFP36L1, which plays a tumor-suppressive role by regulating hypoxia and cell cycle signaling. ZFP36L1 is an mRNA-binding protein that targets AUUUA/UUAUUUAUU elements in the 3’ UTR of genes, leading to mRNA degradation (92, 93). While ZFP36L1 is typically expressed at low levels in the SCLC-A subtype, it is more highly expressed in the inflammatory subtype, suggesting that restoring ZFP36L1 expression could enhance the plasticity of the inflammatory subtype and inhibit NE differentiation and cell proliferation, offering a promising therapeutic avenue (94). Further research is needed to explore the roles of LSD1 and ZFP36L1 in SCLC-N and SCLC-P subtypes. LSD1 and MYC both activate the Notch pathway and promote NE differentiation in SCLC cells, and they are known to interact. Experimental studies have demonstrated that MYC recruits LSD1 to regulate chromatin and DNA oxidation, while LSD1 activity influences MYC-driven transcription, underscoring the importance of their interaction for MYC-mediated gene expression (95). Although no direct evidence currently indicates an upstream-downstream relationship between LSD1 and MYC in the Notch pathway, the possibility that they jointly activate Notch signaling to drive the transition from ASCL1 to NEUROD1 remains to be investigated. These insights suggest that LSD1 inhibitors, by inducing phenotype transitions, may boost SCLC immunogenicity and offer a novel therapeutic approach.

EMT refers to the process by which epithelial cells lose their defining traits and acquire mesenchymal characteristics, a key mechanism in tumor metastasis, therapy resistance, drug resistance, and embryonic development (96). Single-sample gene set enrichment analysis (ssGSEA) of human SCLC cell lines, RPM mouse tumors, and SCLC CDX tumor samples revealed that the SCLC-A2 subtype is notably enriched with epithelial signature genes, while SCLC-A and SCLC-N display more mesenchymal traits. Other subtypes, such as SCLC-P and SCLC-Y, also tend to exhibit increased mesenchymal features (97). EMT scoring in SCLC samples by Gay et al. further confirmed that the SCLC-A subtype is the most epithelial, while SCLC-I shows the most mesenchymal characteristics, with the remaining subtypes falling between these extremes (8). These observations underscore a strong link between EMT status and SCLC subtype plasticity. EMT is regulated by several signaling pathways within the tumor microenvironment, including TGF-β and Wnt signaling. TGF-β can induce mesenchymal markers like N-cadherin and vimentin while suppressing epithelial markers such as E-cadherin (98). The Wnt pathway, by activating β-catenin and other transcription factors, promotes EMT-related gene expression (99). These pathways interact, facilitating both EMT and SCLC subtype transitions. TGF-β, through SMAD-dependent and SMAD-independent pathways, promotes EMT across multiple cancer types. In human SCLC cell lines H146 and H345, TGF-β downregulates ASCL1 via a SMAD-dependent mechanism, suggesting its potential as a target for influencing SCLC phenotype transitions (85, 100). Additionally, research on SCLC cell lines identified that Wnt11, acting as a Wnt ligand, modulates E-cadherin expression and the NE switch in an ASCL1-dependent manner (101). The reversible transition between NE and non-NE cells in SCLC may be driven by EMT mechanisms, which are influenced by dynamic extracellular signals or intracellular factors, offering new therapeutic perspectives for SCLC (97).

SRY-Box 2 (SOX2), a key transcription factor for maintaining stem cell pluripotency, plays a significant role in sustaining cell self-renewal and proliferation. Its dysregulated expression is strongly linked to SCLC development, differentiation, metastasis, and poor prognosis (102, 103). Studies have demonstrated that knocking down SOX2 in human SCLC cell lines H69 and H889 inhibits the proliferation of SCLC-A subtype cells and significantly reduces the expression of key molecules associated with the SCLC phenotype, such as INSM1 (104). In the SCLC-N subtype cell lines H29 and H82, SOX2 was found to directly bind to the promoter region of NEUROD1, with overexpression leading to NEUROD1 silencing (105). These findings suggest that inhibiting SOX2 could drive a phenotypic transition from the SCLC-A to the SCLC-N subtype. Further research on the human SCLC cell line CORL47 revealed that ZFP36L1 binds to and downregulates mRNA levels of both SOX2 and INSM1, thereby blocking ASCL1-driven NE differentiation and proliferation (94). Several studies indicate that SOX2 influences the transitions between SCLC subtypes by modulating the expression of ASCL1 and NEUROD1, contributing to tumor heterogeneity (94, 106). Consequently, targeting SOX2 may represent a promising therapeutic strategy for patients with SCLC.

Lysine-specific demethylase 6A (KDM6A) is a histone demethylase that removes di- and tri-methylation from lysine 27 on histone H3 (H3K27), playing a key role in regulating gene expression, cell fate determination, and development (107, 108). Studies have shown that KDM6A inactivation facilitates the transition from the SCLC-A subtype to the SCLC-N subtype. In KDM6A-knockout SCLC GEMMs, KDM6A was found to sustain active chromatin states in SCLC-A through its demethylase activity and scaffolding role in the COMPASS complex. Loss of KDM6A leads to decreased H3K4me1 and increased H3K27me3, thereby activating NE gene enhancers and upregulating NEUROD1 and its target gene PAX6, which promotes the transition from ASCL1 to NEUROD1 (109–111). While KDM6A does not directly upregulate NEUROD1, studies in KDM6A-mutant SCLC GEMMs and cells suggest that other epigenetic modifiers and transcription factors, such as KMT2A and MYC, may interact with KDM6A to increase NEUROD1 expression (111). Ongoing research aims to clarify the mechanisms underlying SCLC subtype transitions, which will guide the development of targeted therapies for specific SCLC subtypes.

KDM5A, also known as JARID1A or RBP2, is a member of the KDM5 family, which removes di- and tri-methylation marks on lysine 4 of histone H3 (H3K4me2/3). This family is closely linked to tumor resistance, EMT, and other critical processes (112). Studies using LSL-Cas9 SCLC GEMMs revealed that KDM5A demethylates H3K4me3 to regulate gene expression and inhibit Notch2 and its downstream targets, which are essential for maintaining high ASCL1 levels and NE differentiation. Inactivation of KDM5A promotes non-NE differentiation in SCLC (113). Although some mechanisms underlying KDM5A’s role in SCLC progression have been uncovered, the precise relationship between KDM5A and different SCLC subtypes remains unclear. Future research will focus on whether KDM5A promotes the transition from SCLC-A to SCLC-N, potentially offering therapeutic insights for modulating SCLC subtypes ( Figure 3 ).

Furthermore, several molecules play critical roles in the temporal heterogeneity and plasticity of SCLC subtypes. In SCLC cells, co-culturing with fibroblasts activates the JAK2/STAT3 signaling pathway, which is associated with phenotypic reprogramming (114). Genetic studies using SCLC GEMMs identified TAZ as a pivotal molecular switch that coordinates phenotype transformation and metastasis. The SWI/SNF complex promotes the transition of SCLC cells from non-metastatic to metastatic states through the epigenetic silencing of TAZ (115). VGF promotes the upregulation of the transcription factor ASCL1 through the CREB-dependent pathway, thereby driving neuroendocrine differentiation, and plays a critical role particularly in ASCL1-positive subtypes (116). In SCLC GEMMs and their cells, YAP has been shown to induce REST expression via both Notch-dependent and Notch-independent pathways, facilitating the transformation from NE to non-NE cells (80). In the SCLC cell line H69, CSF2 was found to regulate phenotypic plasticity by phosphorylating the STAT3/MYC pathway, restricting the transition between adherent (H69A) and suspension (H69S) phenotypes and altering drug sensitivity in specific cell clones (117). Additionally, the loss of histone acetyltransferases, including CREBBP (also known as CBP) and its paralog EP300 (also known as p300), leads to a reduction in NE-related epithelial markers and a corresponding rise in non-NE markers, such as ZEB1 and VIM, in SCLC GEMM models (118). Mutations in CREBBP/EP300 have been linked to the activation of Notch signaling by regulating FBXW7 in B-cell lymphoma, though the specific mechanism in SCLC remains unclear (119). These molecules present promising therapeutic targets for SCLC.

While studies confirm that SCLC can transition from an NE to a non-NE phenotype and that subtypes can switch between one another, the precise mechanisms underlying these transformations and their impact on drug resistance remain to be fully elucidated. Exploring how the plasticity mechanisms can be harnessed to shift SCLC subtypes toward SCLC-I, which is associated with longer survival, represents a valuable future research direction.

SCLC-A, the most common subtype, accounts for 40-50% of all SCLC cases. Cells in this subtype exhibit classic morphology and high expression of NE markers such as CD56 and CHGA (21). Key molecular features of SCLC-A include elevated levels of Delta-Like Ligand 3 (DLL3) and INSM1, while CREBBP expression is reduced (8). Potential targets for this subtype include DLL3, a protein downstream of the transcription factor ASCL1, as well as B-cell lymphoma 2 (BCL2) and LSD1, both of which are involved in histone modification. DLL3-targeted therapies, such as antibody-drug conjugates, bispecific T-cell engagers, and chimeric antigen receptor (CAR) T-cell constructs, are currently under clinical development. As DLL3 expression correlates with ASCL1, SCLC-A may be particularly sensitive to DLL3-targeted therapies (128). Tarlatamab showed promising Phase I results, with a disease control rate (DCR) of over 50%, significantly improving patient survival (129). Inhibition of BCL2 has also been effective in controlling SCLC growth in both in vitro and in vivo models (8, 130, 131). Additionally, inhibiting LSD1 activates Notch signaling and suppresses ASCL1 expression. The LSD1 inhibitor Bomedemstat improved SCLC response to PD-1 inhibitors in mouse models, and ongoing clinical trials are evaluating its efficacy (NCT05191797) (90, 132). Another potential therapeutic target in SCLC-A is the CDK2-CyclinA2 complex, which phosphorylates ASCL1 and promotes its degradation during mitosis, suggesting a new therapeutic avenue (133). In summary, DLL3, BCL2, and LSD1 represent promising targets for treating SCLC-A.

SCLC-N is primarily characterized by the expression of NEUROD1 and NE markers such as synaptophysin, chromogranin A, and CD56/NCAM (8). MYC amplification is frequently associated with the this subtype and represents a key mechanism imparting therapeutic resistance in SCLC (134–137), making it more responsive to Aurora kinase inhibitors, CHK1 inhibitors, and other targeted therapies (74, 138–140). Short-term follow-up findings from a phase II trial demonstrated that combining Aurora kinase inhibitors with chemotherapy improved progression-free survival (PFS) and OS (134). SCLC-N also relies on arginine metabolism. A key mechanism find that chronic exposure to ADI-PEG 20 can induce the reexpression of argininosuccinate synthase (ASS1) in tumor cells, thereby restoring endogenous arginine biosynthesis and conferring acquired resistance to arginine deprivation therapy (141, 142). With in vitro studies showing that arginine deiminase (ADI) and arginase (ARG) exhibit cytotoxicity in SCLC cell lines, particularly those lacking argininosuccinate synthase (ASS) (136, 141, 143, 144). Clinical trials are ongoing to evaluate the efficacy of these metabolic inhibitors (NCT05616624). Moreover, SCLC-N shows sensitivity to the SVV oncolytic virus, and combining SVV with immunotherapy holds promise for enhanced therapeutic outcomes (16, 145). Blocking transcriptional co-activator BET proteins also reduces NEUROD1 expression, thereby inhibiting SCLC growth both in vitro and in vivo (146). Furthermore, KRS1, a kinase inhibitor of RAS1, has been identified as a driver of cisplatin resistance and may represent a therapeutic target in SCLC-N (147). These findings highlight the potential of therapies targeting Aurora kinases, arginine metabolism, and SVV oncolytic virus for more effective treatment of SCLC-N.

SCLC-P predominantly expresses POU2F3 with low or absent NE marker expression. CRISPR screening has revealed that the SCLC-P subtype uniquely depends on Insulin-like Growth Factor 1 Receptor (IGF-1R), suggesting that linsitinib, an IGF-1R inhibitor, could be a potential therapeutic option for these patients (40). However, a phase II clinical study of OSI-906 (an oral tyrosine kinase inhibitor [TKI] of IGF-1R) combined with topotecan in relapsed SCLC demonstrated safety but limited clinical activity, indicating the need for further investigation (148). Given the frequent loss of TP53 and RB1 in SCLC, these tumors are particularly vulnerable to DNA damage. Targeting DNA damage response mechanisms, such as Poly ADP-ribose polymerase (PARP) and RAD3-related protein (ATR), has shown promise in patients with SCLC-P (8, 149). PARP inhibitors, which disrupt DNA repair mechanisms, induce genomic instability and cell cycle arrest, exhibiting anti-tumor effects in vitro and in vivo (150, 151). Adding the PARP inhibitor veliparib to cisplatin and etoposide in first-line chemotherapy for ES-SCLC resulted in improved objective response rate (ORR), PFS, and OS (75). In addition, a phase II trial evaluating olaparib (a PARP inhibitor) with ceralasertib (an ATR inhibitor) in relapsed or refractory SCLC showed promising disease stabilization, although it did not meet predefined endpoints (152). In vitro studies have also demonstrated that SCLC-P is particularly sensitive to antimetabolites, including antifolates and nucleoside analogs (8, 153). Additionally, the marine-derived anticancer agent lurbinectedin, an inhibitor of RNA polymerase II, selectively disrupts oncogene transcription processes and has shown the highest efficacy in the SCLC-P subtype, particularly in the human SCLC cell line H526. A phase I trial conducted in China reported an overall response rate of 45.5% for lurbinectedin as a second-line treatment (154, 155). The PTEN-MYC axis drives SCLC-P pathogenesis, revealing a rationale for PI3K/AKT inhibition (54). Concurrently, mSWI/SNF complex activity (SMARCA4/2) is essential for the POU2F3 transcriptional program, and its loss desensitizes this subtype to corresponding inhibitors (156). These findings suggest that PARP inhibitors and IGF-1R inhibitors may offer promising therapeutic options for patients with SCLC-P.

The inflammatory SCLC-I subtype, characterized by high infiltration of CTLs, NK cells, and tumor-associated macrophages, has demonstrated heightened responsiveness to PD-L1 monoclonal antibody treatments. Retrospective studies have shown significantly better OS in patients with SCLC-I treated with chemotherapy plus atezolizumab compared to other subtypes (157). Furthermore, Bruton’s Tyrosine Kinase (BTK) is highly expressed in SCLC-I tumors and may serve as a novel therapeutic target (8). Gene expression analysis and bioinformatics studies suggest that SCLC-I and YAP1-expressing cell lines are more sensitive to inhibitors of mammalian target of rapamycin (mTOR) and Polo-like kinase (PLK) presenting additional opportunities for targeted therapies (25, 158). However, these potential targets still require further clinical validation ( Figure 5 ).

Given the intratumor heterogeneity and plasticity of SCLC subtypes, tailored treatments for patients with mixed types, particularly those with the SCLC-AN subtype, may offer more effective therapeutic outcomes. Kelenis et al. proposed that disrupting the nuclear transport of specific transcription factors by inhibiting Karyopherin subunit beta 1 (KPNB1), a nuclear transport receptor highly expressed in tumors, could represent a promising therapeutic strategy (159). Inhibitors of KPNB1, such as INI-43 and Importazole (IPZ), reduced nuclear levels of ASCL1 and NEUROD1 in several human SCLC cell lines (H2107, H2171, H524), selectively inhibiting the growth of ASCL1-positive and NEUROD1-positive SCLC cells in vitro and suppressing ASCL1-positive tumor growth in mice (159). Studies across multiple in vivo chemoresistance models have established that EZH2 promotes resistance by silencing SLFN11 via H3K27me3, which blunts the sensitivity of tumor cells to DNA-damaging drugs (48). In addition, high EZH2 expression in the SCLC-A subtype suppresses the TGF-β–Smad–ASCL1 pathway and enhances tumor progression, indicating a promising yet unconfirmed role for EZH2 inhibitors in treating this subtype (85). While RB1 loss abrogates the canonical CDK4/6 signaling axis, thereby conferring resistance to CDK4/6 inhibitors, evidence shows that these drugs can still counteract SCLC chemoresistance via non-canonical, RB1-independent mechanisms, such as through the disruption of lysosomal function and autophagy (160). Additionally, in multiple patient-derived SCLC cell lines and SCLC CDX models, Jumonji inhibitors or KDM4A knockdown led to downregulation of key markers like INSM1, ASCL1, or NEUROD1, highlighting their potential as therapeutic targets for SCLC (161). Moreover, the small molecule inhibitor iBET-762, which targets bromodomains and extra-terminal domain (BET) proteins, demonstrated selective efficacy by promoting the growth of suspended cell clusters while inhibiting the growth of adherent, mesenchymal-like cells. In SCLC PDX models, iBET-762 increased ASCL1 expression and reduced NEUROD1 and YAP1 levels, suggesting its potential for manipulating subtype dynamics in SCLC ( Figure 6 ) (36).

Molecular subtype-directed precision therapy holds emerging promise for the management of SCLC. However, the clinical translation of this paradigm faces several fundamental challenges. Firstly, the inclusion of unselected patient populations in clinical trials may obscure the efficacy of targeted agents in specific molecular subsets, potentially leading to the premature abandonment of active compounds. Secondly, the intrinsic dynamic evolution of SCLC presents a major therapeutic obstacle. Intratumoral heterogeneity at diagnosis limits the representativeness of single-site biopsies, as minor, non-dominant subclones may exhibit intrinsic resistance to first-line chemotherapy or radiotherapy. When therapies are precisely designed to target the dominant subtype, they may inadvertently create a niche for the expansion of these resistant subclones, directly contributing to primary resistance and seeding future relapse (162). Under therapeutic pressure, SCLC progression and resistance are primarily driven by two mechanisms. The first is “clonal selection”: the eradication of drug-sensitive cellular populations allows pre-existing resistant subclones to proliferate, rendering treatment strategies based on initial biopsy profiles ineffective against recurrent disease—a classic manifestation of acquired resistance. The second, and more critical mechanism, is “cellular plasticity”—the ability of tumor cells to actively switch their transcriptional identity, beyond mere clonal outgrowth. For instance, under the selective pressure of a therapy targeting the SCLC-A (ASCL1-driven) subtype, cancer cells can rapidly downregulate ASCL1 and concurrently upregulate NEUROD1 or POU2F3, thereby “remodeling” their identity to evade attack and invalidating the original targeted agent. This dynamic plasticity represents a key adaptive survival strategy for SCLC and underscores the vulnerability of monotherapies targeting a single subtype. Consequently, a strategic shift in the therapeutic paradigm is imperative. Future research should prioritize the development of “combination therapies” aimed either at concurrently targeting vulnerabilities shared across multiple subtypes (e.g., DLL3) or directly attacking the core signaling pathways that drive plasticity. Furthermore, implementation of dynamic monitoring technologies such as liquid biopsy, is essential to track subtype evolution in real-time during treatment, enabling adaptive precision medicine. Ultimately, the overarching goal should be to prevent or lock tumor evolution—developing novel agents that can lock tumor cells in a drug-susceptible state or directly deprive them of their plasticity, rather than merely killing cells in their current state. Looking ahead, the systematic integration of molecular subtyping in clinical research is crucial to fully elucidate the spatiotemporal heterogeneity of SCLC before and after therapy and to objectively evaluate the effectiveness of existing regimens. This approach is a necessary pathway to overcome current limitations and ultimately achieve truly personalized treatment for SCLC patients.