Core Framework & Diagram What Is the Tumor Immune Atlas?
July 23, 2026What Is Immune Microenvironment Engineering?
July 23, 2026Every tumor has its own immune 'fingerprint' — reading it is the prerequisite for truly precise attacks
—— The Tumor Immune Atlas: systematically decoding the global relationship map between cancer and the immune system.
I. TCGA: the data foundation for building the tumor immune atlas
The Tumor Immune Atlas's data foundation largely comes from a project named TCGA (The Cancer Genome Atlas). TCGA started in 2006, systematically conducting multi-omics analysis on over 33 major cancer types and approximately 11,000 tumor samples, including whole-genome sequencing, whole-exome sequencing, RNA sequencing, DNA methylation analysis, proteomics, and clinical data. This dataset is one of the largest, most complete public databases in cancer research history, freely open to global researchers. One of the most important results is the systematic description of tumor immune landscape.
In 2018, a large analysis using TCGA data (Thorsson et al., published in Immunity) systematically integrated immune characteristics of all 33 cancer types, dividing tumor immune landscapes into six major 'immune subtypes' (C1-C6), each with different immune cell composition, inflammation levels, DNA repair deficiency characteristics, and different clinical prognoses. This analysis was an important milestone for tumor immune atlas transitioning from concept to systematic science.
|
TCGA is the largest systematic dataset in human cancer biology history. It made 'cross-cancer-type comparison of immune landscapes' possible — for the first time letting us see what's common and what's specific across different cancers' immune responses. |
2. Six tumor immune subtypes: spectrum from 'hot' to 'cold'
Thorsson et al.'s 2018 pan-cancer analysis divided tumor immune landscapes into six major subtypes (C1-C6): C1 (Wound Healing Type) — rich in pro-angiogenic signals, high TGF-β expression, moderate immune cell infiltration; common in breast cancer and lung adenocarcinoma; C2 (IFN-γ Dominant Type) — strong IFN-γ signaling, high cytotoxic immune activation, but also high PD-L1 expression (the immune system's 'brake' after activation); prognosis relatively good, may respond best to PD-1 inhibitors; C3 (Inflammatory Type) — high Th17 inflammation, but lacks effective cytotoxic responses; C4 (Lymphocyte Depleted Type) — high CD8+ T cell gene signature, but these T cells are in an exhausted state with highly immunosuppressive microenvironment; C5 (Immune Desert Type) — extremely few immune cell infiltrates, 'cold tumor'; poor prognosis, poor response to immunotherapy; C6 (TGF-β Dominant Type) — strong TGF-β signaling, immune cells suppressed; found in head and neck cancer, lung squamous cell carcinoma.
These six subtypes aren't random — they reflect different tumors taking different immune 'equilibrium states' in the process of interacting with the immune system. Understanding which subtype a patient's tumor belongs to directly influences treatment strategy selection: C2 type most likely responds to PD-1 inhibitors; C5 type needs to first 'warm up' the tumor (like combining with radiation or oncolytic virus), then immunotherapy; C1/C6 type may respond better to anti-angiogenesis or TGF-β pathway targeting.
3. TIMER and CIBERSORT: reverse-engineering immune cell composition from gene expression data
Tumor biopsies usually provide mixed tissue RNA sequencing data (bulk RNA-seq), containing mixed signals from multiple cell types. How to calculate each immune cell's proportion from this mixed signal? This is the problem Immune Deconvolution technology solves. CIBERSORT (developed by Stanford University) and TIMER (Tumor Immune Estimation Resource) are the most widely used immune deconvolution tools: they use known gene expression characteristic profiles ('signature matrices') of various immune cell subsets, through mathematical optimization, calculating each cell type's relative proportion in mixed samples. This lets researchers estimate CD8+ T cells, CD4+ T cells, NK cells, macrophages (M1/M2), dendritic cells, neutrophils, and B cells' relative infiltration degrees from any existing tumor RNA sequencing data — without additional experiments.
Using these tools to analyze all tumor samples in TCGA produces large-scale immune cell infiltration maps showing: dramatic differences in immune cell infiltration patterns between different cancer types — renal cell carcinoma and melanoma typically have high T cell infiltration, while glioblastoma and ovarian cancer typically have lower immune infiltration; within the same cancer type, immune cell infiltration degree significantly correlates with prognosis; and some immune cell proportions are independent prognostic predictors, even more predictive than traditional tumor staging.
4. TIMER2.0: a public platform for everyone to query the tumor immune atlas
TIMER2.0 (timer.cistrome.org), developed and maintained by Harvard University's Xiaole Shirley Liu team, integrates immune deconvolution data from over 11,000 tumor samples from TCGA, ICGB, and other sources. It allows anyone to query online: correlation between specific gene expression and immune cell infiltration; comparative distribution of immune cell infiltration across different cancer types; impact of specific gene mutations (like KRAS, TP53) on tumor immune microenvironment; and survival analysis of immune cell infiltration and patient survival. This platform transforms what previously required professional bioinformatics capabilities, into an interactive tool any researcher or even clinician can quickly access — the 'public library' of tumor immune atlas data.
5. Tumor immune atlas and NK cells: the underestimated key
In tumor immune atlas analysis, NK cell infiltration and activity is a long-underestimated, recently increasingly valued key variable. TIMER and CIBERSORT analysis shows: in multiple tumor types (lung cancer, colorectal cancer, gastric cancer), higher NK cell infiltration estimates correlate with better overall survival rates. Imai et al.'s eleven-year tracking study in The Lancet (2000) already epidemiologically proved the relationship between NK cell activity and cancer risk. In 'immune desert' (C5 type) tumors, NK cell absence is often more prominent than T cell absence — some tumors have NK cell early invasion attempts also blocked by the tumor's physical and metabolic barriers. These analyses provide biological rationale for NK cell therapy applications in 'cold tumors' — for tumors where T cell responses are poor, NK cells may be another attack route. As single-cell technology is applied in tumor immune atlas analysis, NK cell subdivision (adaptive NK cells, tissue-resident NK cells, senescent NK cells, etc.) is being integrated into more refined tumor immune atlas versions.
6. Future: dynamic tumor immune atlas and AI-driven real-time prediction
Current tumor immune atlases are mostly based on single time-point static data. The most valuable future upgrade direction is dynamic, longitudinal immune atlases: continuous sampling before-during-after treatment, tracking how tumor immune microenvironment responds to treatment (liquid biopsy + immunomics combination); AI-driven atlas navigation — inputting patient's multidimensional immune data, AI recommends the best-matching tumor immune subtype and corresponding optimal treatment plan; and single-cell spatially resolved real-time updated atlases integrating single-cell and spatial technologies for more refined resolution. The ultimate form of the tumor immune atlas is a living, real-time updated dynamic knowledge base integrating data from millions of global tumor patients — when your patient's tumor data is uploaded, the system immediately finds the most similar cases in the global database, presenting their treatment history and outcomes.
Frequently Asked Questions