Exploiting DNA Replicative Stress for Novel Small Cell Lung Cancer Therapies

Division Of Basic Sciences - Nci

Investigators

Linked publications & trials

Abstract

Small-cell lung cancer (SCLC) comprises at least two major phenotypic subtypes: neuroendocrine (NE) and non-neuroendocrine (non-NE). NE SCLC represents the canonical, chemosensitive form with high ASCL1 and NEUROD1 expression. In contrast, non-NE SCLC is associated with reduced neuroendocrine markers, greater plasticity, immune infiltration, and chemoresistance. These subtypes exhibit distinct vulnerabilities and cellular states driven by epigenetic remodeling and microenvironmental cues. Our program follows a bed-to-benchside approach, leveraging real-time clinical data to inform discovery and therapeutic development in SCLC. Targeting Replication Stress and Nuclear Architecture: We proposed that replication stress is a hallmark of SCLC, driving genome instability and progression (Sci Transl Med 2016). ATR kinase resolves replication stress, and its inhibition induces fork collapse, DNA breaks, and cell death. We demonstrated that ATR and TOP1 inhibitors synergize to exacerbate replication stress in SCLC (Cancer Cell 2021), and in a Phase I trial, berzosertib (ATR inhibitor) plus topotecan showed promising pharmacodynamics and tolerability (J Clin Oncol 2018). A randomized Phase II trial confirmed improved overall survival and response rates, despite no difference in PFS (JAMA Oncol 2023). NE subtypes, with higher endogenous replication stress, were more likely to respond. To predict response, we developed an 18-gene "repstress score" across >60 SCLC models, identifying tumors sensitive to topotecan, gemcitabine, and ATR inhibition (Cancer Res Commun 2022). We are translating these insights to extrapulmonary small cell cancers (ASCO 2022), antibody-drug conjugates for TOP1 delivery (Clin Cancer Res 2023; NCT04826341), and novel combinations involving lurbinectedin (EMBO Mol Med 2023; NCT04802174). Our recent study (bioRxiv 2025) revealed that LMNA, a nuclear envelope protein, is repressed in NE SCLC and its loss induces R-loop accumulation, transcription-replication conflicts, and DNA damage. LMNA depletion disrupts nuclear pores, impairs RNA export, and increases replication stress. EZH2-mediated activation of LMNA occurs during NE-to-non-NE transition, linking epigenetic reprogramming with nuclear architecture and genome maintenance. Clinically, low LMNA correlates with poor prognosis, identifying it as a key biomarker of genomic instability and plasticity in SCLC. Leveraging Anti-Tumor Immune Responses: Despite high mutational burden, durable responses to chemo-immunotherapy remain rare in SCLC, with only 12-18% of patients progression-free at one year. We hypothesized that combining PARP and immune checkpoint inhibition could enhance immunogenicity. However, in a Phase II trial, durvalumab + olaparib failed to meet efficacy benchmarks. Biopsies revealed that most tumors were immune-excluded (J Thorac Oncol 2018), highlighting tumor immune phenotype as a critical determinant of response. These results mirror broader clinical findings: while PARP inhibitors can boost anti-tumor immunity in preclinical models, they show limited synergy in human trials. We argued that this disconnect reflects limitations of model systems and tumor complexity, including genomic heterogeneity and immune exclusion (Clin Cancer Res 2020), necessitating improved models and biomarker-driven combinations. This translational gap underscores the need for patient-centric research. In our Cancer Cell 2022 commentary, we advocated shifting from model-centric to human biology-driven approaches. SCLC exemplifies this disconnect: >100 cell lines exist, yet <100 genomically annotated metastatic tumors have been characterized. Core biological features, such as NE differentiation, are often distorted in models relative to patient tumors (Nat Commun 2022), contributing to clinical failures, particularly in immunotherapy. Further analysis showed that responders to durvalumab + olaparib exhibited Notch activation, promoting non-NE phenotypes and enhanced immune visibility (Nat Commun 2021). In mismatch repair-deficient tumors, we identified defective JAK1 signaling as a mechanism of resistance to immunotherapy (Cold Spring Harb Mol Case Stud 2020), illustrating how genomic instability shapes immune evasion. Based on these insights, we are testing EZH2 inhibitors to epigenetically reprogram SCLC into more immune-responsive states (NCT05353439) and developing CAR T-cells targeting surface antigens independent of MHC, circumventing immune exclusion and lineage plasticity. SCLC Subtypes and Vulnerabilities: Historically treated as a uniform disease, SCLC comprises molecularly distinct subtypes with unique vulnerabilities. Profiling pre-treatment tumors from patients receiving replication stress-targeted therapy revealed that those with high DDR component expression responded better (Cancer Cell 2021). Immunotherapy responders showed elevated Notch signaling and non-NE features (Nat Commun 2021), suggesting immunogenic subtype differences. We identified an NE-v2-like subtype directly from patient tumors associated with chemoresistance and poor outcomes (Nat Commun 2022). Inherited mutations in DNA repair genes predicted better chemotherapy response and remission (Sci Transl Med 2021), challenging the view that tobacco exposure is the sole driver of SCLC. These discoveries, many from small, well-annotated cohorts, highlight the power of clinically grounded translational research. Recent studies have uncovered tumor-intrinsic and -extrinsic mechanisms of SCLC heterogeneity. Extrachromosomal DNA amplifications drive intra-tumoral heterogeneity and poor prognosis (Cancer Discov 2023). Microenvironmental signals also shape phenotypic states and drug response (Cell Rep Med 2024). To dissect subtype-specific metastatic adaptations, we developed autopsy-derived xenograft models preserving organotropism and transcriptomic diversity. LMNA emerged as a key regulator of SCLC metastasis: its expression is significantly reduced in liver metastases relative to primary or brain lesions, and its loss increases nuclear deformability, confined migration, and clinical aggressiveness (bioRxiv 2025). LMNA thus serves as both a mechanical and molecular marker of plastic, metastatic SCLC subtypes. Given the challenges of serial tumor biopsies in SCLC, we pioneered noninvasive molecular profiling using plasma-derived cell-free DNA (cfDNA). We demonstrated high concordance between cfDNA and matched metastatic tumor genotypes and transcriptional states, including subtype and resistance profiles (Genome Med 2025). To extend resolution, we are applying cfChIP-seq to infer gene activity via histone modifications (bioRxiv 2022), mapping nucleosome positioning and chromatin accessibility from cfDNA architecture (Genome Med 2025), and analyzing cfDNA fragmentome signatures to detect tumor-specific chromatin states (ASCO 2022). These methods enable longitudinal, noninvasive tracking of subtype, treatment response, and disease evolution in SCLC. I place strong emphasis on mentoring the next generation of physician-scientists. My trainees have secured NIH and foundation grants and, for five consecutive years, received ASCO Young Investigator Awards supporting transition to independent research careers.