Traditionally, tumors have been regarded as diseases primarily characterized by uncontrolled proliferation of malignant cells, and therapeutic strategies have predominantly focused on their direct eradication through chemotherapy and radiotherapy. However, this perspective has evolved significantly with the growing recognition of tumor complexity, wherein tumors are increasingly viewed as highly dynamic and heterogeneous ecosystems (1). The TME, in particular, represents a complex cellular and molecular landscape composed not only of malignant cells but also of diverse non-malignant components, including immune cells, cancer-associated fibroblasts (CAFs), vascular endothelial cells, pericytes, and tissue-resident stromal cells, all embedded within the extracellular matrix (ECM) (2). In certain tumor types, non-malignant cells may constitute the majority of the tumor mass (3). The cellular composition and functional states in the TME exhibit significant variability influenced by factors such as the anatomical origin of the tumor, genetic and epigenetic features of cancer cells, disease stage, and host-specific factors (4–6). Understanding the complex cellular interactions and spatial heterogeneity in the TME is crucial for enhancing our understanding comprehension of tumor biology and facilitating the development of more precise and effective anticancer therapies.

Despite its central role in cancer progression and therapeutic response, the TME presents significant analytical challenges. One primary limitation arises from the technical constraints of transcriptomic profiling methods. Conventional bulk RNA sequencing (RNA-seq) captures only average gene expression from heterogeneous cell populations, thereby obscuring intrinsic cellular heterogeneity in the TME and failing to identify rare but functionally critical subpopulations (7, 8). Tumor heterogeneity itself constitutes another substantial barrier (9–11). This heterogeneity exists both across patients (inter-tumor heterogeneity) and within individual tumors (intra-tumor heterogeneity), as cancer cells occupy various differentiation states while exhibiting divergent transcriptional profiles and mutational landscapes. Furthermore, non-malignant cell populations, including immune and stromal cells, exhibit extensive phenotypic and functional diversity. The complexity of mechanisms underlying therapy resistance further highlights the urgent need for deeper insights into the TME (12–15). Increasing evidence suggests that non-malignant cells actively contribute to resistance against chemotherapy, targeted therapies, and immunotherapies through multiple mechanisms. For instance, CAFs secrete ECM components and growth factors, establishing physical and biochemical barriers that hinder drug penetration (16, 17). Immunosuppressive cells such as regulatory T cells (Tregs) and M2-polarized macrophages suppress anti-tumor immunity by expressing immune checkpoint molecules (e.g., PD-1, CTLA-4) and releasing inhibitory cytokines such as IL-10 and TGF-β (18–20). Collectively, these findings underscore the necessity of comprehensively characterizing the TME—encompassing cellular composition, functional phenotypes, and spatial interaction networks—to inform the rational design of combination therapies (21, 22). To address these challenges, the integrating scRNA-seq with ST has emerged as a powerful strategy. This approach facilitates insights into the spatial and functional complexity of the TME.

scRNA-seq is a powerful technique enabling high-resolution gene expression profiling for the individual-cell level, enabling the identification and characterization of distinct cellular subpopulations with specialized functions (23). ST, a rapidly evolving complementary approach, maps gene expression within intact tissue sections, preserving critical spatial context and tissue architecture (24). Given the cellular complexity of the TME, no single technology can fully capture its spatial and functional heterogeneity. Although current ST platforms generally lack true single-cell resolution, their integration with scRNA-seq provides a comprehensive perspective on the TME. Combining scRNA-seq and ST overcomes these limitations by bridging cellular identity with spatial localization. For instance, multimodal intersection analysis (MIA) was introduced in 2020 to integrate scRNA-seq and ST data, aiming to map spatial associations cell-type relationships in pancreatic ductal adenocarcinoma (PDAC) (25). This study revealed that stress-associated cancer cells colocalize with inflammatory fibroblasts, the latter identified as major producers of interleukin-6 (IL-6), underscoring spatially organized tumor-stroma crosstalk in PDAC (25).

The integration of scRNA-seq and ST enables researchers to dissect the complexity and spatial organization of the TME with unprecedented resolution. This synergistic approach not only deepens our understanding of tumor biology but also accelerates the discovery of novel diagnostic and prognostic biomarkers, paving the way for more precise and effective therapeutic strategies. We conducted comprehensive searches in PubMed, Web of Science, and Scopus to ensure broad coverage of relevant studies. We used a combination of keywords related to single-cell sequencing, spatial transcriptomics, tumor microenvironment, cancer heterogeneity, and their respective applications in oncology. We defined explicit criteria for including studies based on relevance, study type (e.g., original research, key reviews), and publication status. Studies were excluded if they were not peer-reviewed, not published in English, or deemed outside the scope of this review. We focused primarily on literature published between January 2010 and June 2025 to capture the most recent and impactful advances in the field. (GA).

ST is an emerging technology that enables spatially resolved gene expression profiling within intact tissue sections, preserving the native histological context. By combining high-resolution imaging and transcriptomic analysis, ST maps gene expression patterns with precise spatial localization, achieving subcellular resolution in some cases. Current ST methodologies can be broadly classified into two categories: image-based (I-B) and barcode-based (B-B) approaches (26, 27). Image-based methods, such as in situ hybridization (ISH) (28) and in situ sequencing (ISS) (29), utilize fluorescently labeled probes to directly detect RNA transcripts within tissues, allowing visualization of gene expression patterns while maintaining spatial integrity. In contrast, barcode-based approaches rely on spatially encoded oligonucleotide barcodes to capture RNA transcripts. In solid-phase transcriptome capture, RNAs hybridize to immobilized barcoded probes on slides before sequencing. Deterministic spatial barcoding assigns unique barcodes to each transcript, retaining positional information throughout sequencing (30–32). These complementary strategies facilitate comprehensive spatial transcriptome profiling, when integrated with single-cell techniques, they yield unprecedented resolution for investigating tissue architecture and tumor heterogeneity.

scRNA-seq of patient-derived tumors has uncovered diverse cellular subpopulations and revealed intricate intercellular communication networks within the TME (33–39). However, scRNA-seq requires tissue dissociation, leading to the loss of spatial context and limiting insights into tissue architecture and cell-cell interactions. To address this, several strategies have been developed to preserve or reconstruct spatial information. For example, combining ISH-based gene expression mapping with scRNA-seq data has proven effective for identifying rare cell types and subpopulations using targeted gene panels (40, 41). Recent advances have evolved ISH into high-plex RNA imaging (HPRI) techniques, including in situ sequencing, multiplexed error-robust fluorescence in situ hybridization (MERFISH) (42), and sequential fluorescence in situ hybridization (seqFISH) (43–45). However, these approaches are often limited to well-defined tissues and remain challenging when applied to heterogeneous solid tumors characterized with complex structures and diverse transcriptomic profiles.

Emerging methods, such as sci-Space, have been developed to address this limitation by generating spatially resolved transcriptomic maps at near-single-cell resolution across extensive tissue areas. In mouse embryonic development studies, sci-Space enabled the simultaneous capture of approximate spatial coordinates and complete transcriptomes from over 120,000 nuclei. However, its spatial resolution is currently limited to approximately 200 micrometers. Although there have been improvements in spot density and size, the resolution remains insufficient for precisely capturing interactions between neighboring cells. As a result, this approach typically yields composite transcriptomic profiles derived from small cell clusters or cellular fragments rather than genuine single-cell resolution. Currently, ST remains one of the most widely adopted approaches for high-throughput spatial gene expression analysis (46, 47). scRNA-seq is a high-throughput method for transcriptomic profiling at individual-cell resolution. By isolating individual cells, capturing their mRNA, and performing high-throughput sequencing, scRNA-seq reveals cellular heterogeneity typically masked in bulk RNA analyses. The advantages of scRNA-seq include: (i) identification of rare cell populations, including tumor stem cells and transitional cellular states, which are undetectable by bulk RNA-seq (48); (ii) classification of cells based on canonical markers, enabling precise identification of immune cell subsets and epithelial cell states (49); (iii) characterization of dynamic biological processes, such as differentiation trajectories and cellular transitions (50); and (iv) integration with multi-omics approaches, including single-cell ATAC-seq (chromatin accessibility) and CITE-seq (surface protein expression), providing multidimensional insights into cell states (51).

Despite these strengths, scRNA-seq also exhibits notable limitations. RNA capture efficiency per cell is relatively low (52). The method remains costly and technically challenging, necessitating careful optimization of sample processing protocols (53, 54). Critically, the mandatory tissue dissociation disrupts native spatial relationships, hindering analysis of cell–cell interactions within intact tissue architectures (55, 56). The comparison of scRNA-seq and ST is shown in Table 1 .

The integration of scRNA-seq and ST confers significant advantages for deciphering complex biological systems: (i) Comprehensive gene expression profiling: scRNA-seq enables high-resolution gene expression analyses, revealing cellular heterogeneity and transcriptional dynamics within tissues (8). It is essential for cell-type identification, developmental tracking, and elucidating disease mechanisms. (ii) Spatial context and tissue architecture: ST preserve native tissue spatial architecture, enabling localization of gene expression patterns, cellular distributions, and intercellular interactions (57). (iii) Complementary strengths: While scRNA-seq lacks spatial information, ST technologies face resolution and throughput limitations. Their integration overcomes their individual limitations, offering a comprehensive understanding of tissue biology (58–60).

Combining scRNA-seq and ST provides deeper insights into cellular interactions with their microenvironment (61–63), with critical implications for both diagnostics and therapeutics. This integrative strategy supports the identification of spatially informed biomarkers and therapeutic targets by linking gene expression patterns to precise tissue regions, thereby advancing personalized medicine and enhancing disease diagnosis (64). Currently, two major computational approaches are used to this integration: deconvolution and mapping. Deconvolution utilize single-cell reference datasets to computationally estimate the cellular composition within each spatial capture spot, determining proportions of various cell types. Mapping approaches assign scRNA-seq-defined cellular subtypes to cells within spatial maps or localize individual scRNA-seq profiles to specific tissue niches. The characteristics of different integration strategies are shown in Table 2 . These analyses provide critical spatial context to inferred ligand–receptor interactions and other forms of intercellular communication derived from scRNA-seq data.

The immunosuppressive tumor microenvironment (ITME) is a specialized ecosystem in tumor tissues. The ITME suppresses anti-tumor immune responses via multiple mechanisms. This promotes immune escape, tumor growth, and therapy resistance. Complex cellular crosstalk drives ITME formation, representing a major challenge to immunotherapy. The interaction between immune-mediated tumor editing and cancer cell immune evasion influences disease progression and therapeutic outcomes (82, 83).

CAFs are a major stromal component in the TME, critically contributing to tumor initiation, progression, invasion, metastasis, and therapeutic resistance. CAFs typically originate from resident fibroblasts or precursor cells activated by tumor-derived signals. They exhibit high heterogeneity and secrete diverse cytokines, growth factors, and ECM components, collectively remodeling the TME to facilitate tumor development. CAFs significantly modulate tumor behavior (123–125). Their functional plasticity and diversity not only promote tumor progression but also represent potential therapeutic targets. SeRNA-seq has revealed substantial CAFs heterogeneity, identifying multiple transcriptionally distinct CAF subtypes within the TME ( Table 3 ) (126–128).

Although scRNA-seq and ST have significantly enhanced our understanding of tumor biology, several challenges remain to be addressed (142). Tumors exhibit extensive somatic genetic heterogeneity (143), and their pathogenesis involves intricate regulatory mechanisms across multiple omics dimensions, including transcriptomics, epigenomics, proteomics, and metabolomics (144). With the rapid advancement of single-cell multi-omics technologies, research has increasingly transitioned from single-omics analyses to integrated approaches combining transcriptomic, genomic, epigenomic, and proteomic data. Such integrated multi-omics strategies have already provided valuable insights into several malignancies, including colorectal cancer (CRC) (145), lung cancer (146), and prostate cancer (147). The combination of single-cell multi-omics with ST is anticipated to offer a more comprehensive and spatially resolved understanding of tumor heterogeneity at single-cell resolution.

However, despite these technological advancements, clinical translation remains challenging. Several practical barriers remain for clinical transformation: (i) Cost-benefit trade-off: these technologies are currently expensive and have long experimental cycles; (ii) High requirements of infrastructure and data analysis capabilities; (iii) Lack of regulations and standardization. This requires collaborative efforts from regulators, industry, and academia (148, 149).

Achieving true single-cell resolution in spatial transcriptomics technologies and the associated computational challenges in analyzing such high-dimensional data. The necessary next step of integrating spatial multi-omics data, particularly spatial proteomics and metabolomics, to build a more comprehensive functional understanding of the tumor microenvironment. The urgent need for standardizing and validating analytical pipelines to ensure robustness, reproducibility, and ultimately, their successful translation into clinical settings for diagnostics and therapeutic decision-making.

In conclusion, the integration of single-cell and spatial transcriptomics technologies has fundamentally expanded our understanding of tumor heterogeneity and microenvironmental organization. However, to translate these insights into clinical impact, future work must focus on three critical frontiers. First, the integration of single-cell and spatial transcriptomics will be essential to move beyond transcriptional data and achieve a functional, multi-layered understanding of cellular phenotypes and interactions within their native context. Second, the prospective clinical validation of spatial biomarkers is urgently needed to establish their utility in patient stratification, prognosis, and therapy guidance. This will require rigorous standardization of analytical and reporting protocols to ensure reproducibility across platforms and cohorts. Finally, the development of advanced computational frameworks capable of unifying multi-omic spatial data—and ultimately enabling real-time mapping—will be crucial for informing diagnostic and even intraoperative decisions. With sustained development, these integrative approaches hold substantial promise for enhancing cancer diagnostics, guiding precision therapeutic strategies, and ultimately improving clinical outcomes for patients.