Decoding inflammatory regulation in ovarian cancer at single-cell resolution.

OC has the highest mortality rate among gynecological malignancies ( 1 ). Globally, over 300,000 new cases are diagnosed annually, with a 5-year survival rate of less than 50% ( 2 ). The core clinical dilemma lies in the hidden early symptoms, high incidence of peritoneal and ascites metastasis, and frequent occurrence of chemotherapy resistance ( 3 ). Epidemiological studies link chronic inflammation to OC carcinogenesis: women with endometriosis face a 2.5-fold higher OC risk and the risk rises to over 4-fold for ovarian endometriosis ( 4 ), and a history of pelvic inflammatory disease is associated with increased risk of OC ( 5 ), underscoring a close correlation between inflammation and tumorigenesis ( 6 ). Previous studies relied on bulk sequencing, which cannot distinguish specific responses of key cell subpopulations ( 7 )and overlooks OCME heterogeneity, while the tumor microenvironment(TME) can be a strong driver of tumor aggression ( 8 ). Additionally, traditional imaging techniques fail to capture the dynamic changes of inflammatory signals during OC peritoneal metastasis. The technical limitation has long left the molecular mechanism of inflammation regulation of OCME in a “black box”, becoming a critical bottleneck restricting OC mechanism research and clinical translation. The advent of single-cell resolution technologies offers a breakthrough ( 9 ): Single-cell RNA-sequencing(scRNA-seq) enables high-throughput single-cell transcriptome analysis to accurately identify inflammation-responsive cell subpopulations and their gene expression profiles, as well as gene regulatory networks; Spatial transcriptomics further unravels cellular spatial distribution and interactions in OCME to reveal the regional distribution pattern of inflammatory signals; single-cell multi-omics integrates transcriptomic and proteomic analyses to address the key issue of transcriptome-function disconnect, enabling direct detection of inflammatory pathway protein activity to clarify immune cells’ functional status ( 10 – 13 ). Based on these technologies, researchers can track inflammatory signals regulation of OCME cellular components at the single-cell level, analyze immune cell signaling network mediated by inflammation, and capture dynamic interaction patterns between inflammation and immune cells, providing unprecedented opportunities to deepen understanding of OC pathogenesis and develop novel targeted therapies. The article will systematically review the regulatory mechanisms and clinical applications of inflammation in OCME based on the research progress of single-cell technology, aiming to offer new insights for the field of OC research.

Single-cell technologies are introducing a new dimension to the clinical management of OC, extending beyond conventional histological subtyping and genotyping. For instance, the analysis of circulating cells from ascites or peripheral blood enables the non-invasive construction of a “real-time” TME atlas, thereby allowing dynamic monitoring of therapeutic response and the evolution of drug resistance. In the future, clinical trial design will no longer rely solely on tumor histology or a limited set of driver gene alterations. Instead, it will integrate single-cell–resolved “cellular subset signatures”—such as specific tumor-associated macrophage lineages or T cell state ratios—to define molecular subtypes, predict treatment outcomes, and refine patient stratification. Realizing this vision may require a new paradigm capable of systematically integrating and reasoning over these complex features, as demonstrated by the potential of graph-enhanced AI knowledge graphs like RSA-KG in integrating multimodal data to enhance clinical decision-making and ultimately paving the way for authentic precision immunotherapies ( 74 ). However, OC research faces two key challenges: significant TME spatiotemporal heterogeneity hindering precise targeting ( 66 ), and extensive compensatory pathways between inflammation and immunosuppression leading to treatment resistance ( 75 ). Future efforts should use single-cell and spatial transcriptomics to map regional molecular targets, guiding rational drug combinations. Developing organoids mimicking human tumor environments ( 76 ), combined with smart delivery systems, to improve preclinical predictions ( 77 ). We could design synergistic therapies targeting specific immune cells/pathways and validate them through well-designed clinical trials to advance personalized treatment.

Breakthroughs in single-cell and spatial omics technologies have elucidated the pivotal role of inflammatory and immunosuppressive networks within the OCME. Chronic inflammation, mediated by key signaling pathways including NF-κB, JAK/STAT, and PI3K/Akt/mTOR, dynamically remodels the cellular composition and function of the OCME, driving immune suppression, stromal remodeling, and tumor metastasis. Single-cell resolution uncovers heterogeneous cellular responses and interactions under inflammatory signals, as well as the spatiotemporal dynamics of chemotherapy resistance and immune evasion. These findings provide precise molecular targets for developing novel therapies directed at the inflammation-immunity axis. They propel OC treatment toward combination targeted and immunotherapy, along with personalized approaches based on microenvironment molecular subtyping, offering new prospects for improving patient prognosis.

Based on the single-cell analysis of the OCME cell heterogeneity, inflammatory-immune regulatory pathways, and their spatiotemporal dynamic characteristics, a number of molecular targets and cell subpopulation markers with clinical translational value have been identified, providing a basis for targeted treatment of OC. It is known that CD163+M2 TAMs are key immunosuppressive cells in advanced OC. However, a single-cell study revealed that one IRF8+NR1H2 macrophage cluster and another CD274 cluster exhibited similar gene expression and were similar to M1 macrophages during the anti-tumor immune response of early-stage OC, and IRF8 and CD274 were upregulated in an activation-dependent manner. Perhaps the macrophages in the NR1H2+IRF8+ cluster are in a transitional state, stimulating them to become M1 macrophages, conversely, preventing them from becoming M2 TAMs, might be a method of OC immunotherapy ( 24 ). The immunosuppressive phase of OC is characterized by abnormal activation of inflammation-driven signaling pathways. Various JAK/STAT inhibitors have been proven to have anti-tumor properties in OC cell lines. Among them, the natural compound α-Hederin functions as a dual JAK1/2 inhibitor, with microscale thermophoresis analysis confirming its high binding affinity for both JAK1 and JAK2. This direct interaction effectively suppresses STAT3 phosphorylation and nuclear translocation, leading to inhibition of tumor proliferation, migration, and EMT. Furthermore, α-Hederin exhibits synergistic effects with cisplatin in overcoming platinum resistance ( 72 ). Additionally, Wnt/β-catenin inhibitors target pathway dysregulation associated with tumor stemness and chemoresistance. Preclinical evidence confirms that suppression of this signaling cascade reduces tumor growth while enhancing sensitivity to conventional therapies, supporting their integration into combination treatment regimens ( 73 ). During the OC progression process, the signaling pathways related to energy metabolism, such as cell cycle and tumor cell proliferation, have significantly strengthened. PARP inhibitors, particularly niraparib, have shown significant clinical efficacy in heavily pretreated patients. The QUADRA trial demonstrated a 28% objective response rate in HRD-positive, platinum-sensitive populations, with substantial improvements in both duration of response and overall survival ( 73 ). After chemotherapy, the NECTIN2-TIGIT signaling pathway is significantly activated between myeloid cells and CD8+T cells, and it is a key driver of T cell exhaustion. In patient-derived immunocompetent cultures(iPDCs), monotherapy with anti-TIGIT or combination therapy with anti-PD-1(pembrolizumab) can significantly enhance the expression of granzyme B and IFN-γ in CD8+T cells in samples after chemotherapy, activating the anti-tumor function of T cells, and the treatment response is more significant in samples with high expression of NECTIN2-TIGIT ( 65 ). This method may reverse the T-cell exhaustion state in HRD-recurring OC that maintains immunogenicity.

Traditional analyses of the OCME, based on histology and bulk omics, have identified general immune infiltration patterns and categorized them as “inflamed” or “non-inflamed” ( 53 ). These methods detect overall shifts in major immune populations—such as T cells and macrophages—and link markers like Tregs or M2 macrophages to poor prognosis. However, they obscure critical heterogeneity by averaging cellular signals across the tumor. The inflammatory-immune interactions in OCME exhibit dynamic changes across disease stages and spatial regions. Single-cell and spatial transcriptomics decode these spatiotemporal features, providing new insights into OC progression and therapy resistance ( 54 ). A number of studies have used methods such as RNA velocity and pseudo-time reconstruction to conduct single-cell level time-resolved analyses of the OCME to infer the progression path of OC and establish dynamic gene expression maps during the tumor development process. And it was consistently observed that the number of metastatic epithelial cells increased along the trajectory in the later stage, while genes related to the immune response were significantly reduced, on a macro level ( 27 , 34 ). During early-stage OC, the OCME lacks significant immunosuppression and is characterized by predominant inflammation-initiated anti-tumor immunity. Low-level inflammatory factors secreted by OC cells recruit dendritic cells(DCs) into OCME and stimulate their maturation ( 55 ), DCs synthesize high levels of IL-12 to enhance NK cells and B/T cells immunity ( 56 – 58 ) and upregulate co-stimulatory molecules including LFA-3/CD58, ICAM-1/CD54, B7-2/CD86 to enhance adhesion and signal transduction ( 59 , 60 ); Mature DCs migrate to lymphoid tissue, where they recruit T and B cells via chemokine release and sustain the viability of recirculating T lymphocytes ( 61 ). However, as tumors proliferate and breach local tissue barriers, leading to peritoneal metastasis and ascites formation, the dual microenvironment remodeling is completed, marking the transition to advanced disease ( 62 ). And the OCME enters a robust immunosuppressive state. Advanced OC cells secrete abundant inflammatory factors like IL-6, IL-10, TGF-β, IL-10 impairs DCs maturation, reducing co-stimulatory molecule expression and abrogating T cell activation ( 55 ); IL-6 activates the JAK/STAT pathway, inducing monocytes to differentiate into M2-TAMs, whose secreted IL-10/TGF-β further suppresses CD8+T cell cytotoxicity. Meanwhile, TNF-α activates CAFs through the NF-κB pathway and secretes CXCL12 to recruit MDSCs. Long term chronic inflammation induces the transformation of CD8+T cells from the cytotoxic subtype to the exhausted subtype(HAVCR2+LAG3+PD-1+) ( 63 ), as well the immunosuppression in OCME. Although chemotherapy kills some tumor cells, it also induces a drastic remodeling of the OCME. On the one hand, chemotherapy induces apoptosis of tumor cells, releasing a large amount of DAMPs, which abnormally activate the NF-κB pathway and enhance the PI3K/Akt/mTOR pathway ( 64 )to promote OC chemotherapy resistance. On the other hand, a single-cell study of 15 OC patients revealed that neoadjuvant chemotherapy expands CD8+ T cells but concurrently induces immunosuppressive Myelonets that drive T cell exhaustion via NECTIN2-TIGIT signaling ( 65 ). It should be noted that the progression of tumor cells after chemotherapy-induced dormancy can evade immune surveillance ( 31 ). While the initiating cells in recurrent tumors were mainly related to the cell cycle such as MKI67, CKS2, ANLN, CDC20, CDCA8, CENPF ( 31 ), inflammatory signals no longer drive anti-tumor immunity. And single-cell analysis revealed that at OC recurrence, HRD and HRP tumors diverge: HRD tumors maintain the CD8+T cell-DC interaction niche to maintain immunogenicity, secret prostanoid E 2 (PGE 2 ) through the COX/PGE 2 pathway to inhibit T cell proliferation and function and achieve immune evasion, whereas HRP tumors lose the CD8+T cell-DC interaction niche and accumulate TREM2+/ApoE+ TAMs ( 66 ).HRD and HRP recurrent OC have completely different immune escape mechanisms, and therefore require targeted therapeutic approaches. Spatial Transcriptomics(ST) revealed that OC and its microenvironment exhibit significant spatial heterogeneity. For instance, by integrating ST with copy number variation(CNV) analysis, a study constructed clonal evolution trees within a spatial context, visually demonstrating that OC arise from multiple CNV events, the same ancestral clone can be dispersed in different spatial regions, and the similarity of pathway activity is related to the spatial location of the clones, rather than the CNV pattern. They spatially validated the demarcation of tumor regions, characterized by high expression of epithelial markers, and stromal regions, marked by elevated fibroblast markers and revealed distinct pathway activities specific to different spatial niches: tumor regions were enriched for cell proliferation pathways such as E2F and MYC targets, whereas stromal regions were enriched for inflammatory response and EMT ( 67 ). Similarly, four immune phenotypes were constructed based on the distribution density of CD8+T cells in the tumor and stroma: the purely inflamed phenotype with CD8+ T cell enrichment in over 70% of tissue regions correlates with the highest 5-year survival rate; the mixed inflamed phenotype with enrichment in 50% to 70% of regions is associated with favorable prognosis; the excluded phenotype with scarce CD8+ T cells in under 50% of tumor regions but enriched in over 10% of stromal regions corresponds to intermediate prognosis; and the desert phenotype with sparse CD8+ T cells in under 10% of both tumor and stromal regions shows the worst prognosis.This immune phenotype classifier significantly correlated with clinical outcome, as patients with purely and mixed-inflamed phenotypes had a statistically significant longer OS compared to those with excluded and desert tumors ( 66 ). Together, these multi-faceted approaches unveil the highly organized and complex spatial and functional landscape of the OCME. Further research has shown that the inflammatory-immune regulation in the OCME exhibits significant regional specificity, forming unique spatial functional niches at the tumor-stroma interface(TSI) and the ascites-peritoneum junction. After chemotherapy, TSI exhibits maximal cellular activity, with Myelonets differing in size and composition ( 65 , 66 ). The infiltrating CD8+ T cells and IBA1+ macrophages were co-localized here, the T cell exhaustion pathway and the M1 polarization pathway were significantly enriched, and CD8+ GZMK+ T cell interactions with M2 macrophages at TSI predict poor survival ( 65 , 68 ). At the ascites-peritoneum junction, ascites constructs the pre-metastatic microenvironment through two mechanisms: one is to disrupt the integrity of the mesothelial barrier, and the other is to regulate the behavior of ECM and tumor spheres, promoting the peritoneal migration of OC cells. Moreover, within ascites, tumor cells, mesothelial-derived CAFs, and infiltrating leukocytes produce a multitude of factors, including but not limited to cytokines, chemokines, and growth factors. These autocrine and paracrine soluble molecules form complex signaling networks that govern, in part, tumor-peritoneum interactions ( 69 ). These findings inform clinical translation. For example, targeted depletion of M2 TAMs to restore CD8+T cell cytotoxicity in advanced OC ( 70 ); inhibiting the IL-6 pathway to attenuate STAT3-mediated tumor growth promotion ( 71 ); and modulating DCs migration to increase intratumoral distribution, which may enhance anti-tumor immunity.