3. TIME Subtype Classification Based on Analysis of Immunogenomic Data from the Cancer Genome Atlas (TCGA)
To further understand the cancer immune landscape, researchers used various immunogenomic methods to classify the TIME across 33 cancers into the wound-healing, IFN-γ-dominant, inflammatory, lymphocyte-depleted, immunologically quiet, and TGF-β-dominant subtypes on the basis of the distinct distribution of five immune-oncologic gene signatures (macrophages/monocytes, lymphocyte infiltrate, TGF-β response, IFN-γ response, and wound healing)
[41]. Their characteristics are summarized in .
Table 2. Characteristics the TCGA TIME subtype classification.
TIME Subtypes |
Wound Healing ǂ |
IFN-γ Dominant |
Inflammatory |
Lymphocyte Depleted |
Immunologically Quiet |
TGF-β Dominant |
Leukocyte fraction * |
Intermed. |
High |
Intermed. |
Low |
Low |
Highest |
Lymphocyte fraction (25–55%) |
High |
Highest |
High |
Intermed. low |
Lowest |
Intermed. |
TIL (H and E) |
High |
Highest |
Intermed. low |
Low |
Lowest |
Intermed. |
Immune cell composition |
|
|
|
|
|
|
T cells |
|
|
|
|
|
|
CD8 T cells (<15%) |
Intermed. high |
Highest |
High |
Intermed. low |
Lowest |
Intermed. |
CD4 T cells (<35%) |
|
|
|
|
|
|
Th1 |
Lowest |
|
Elevated |
|
Elevated |
Elevated |
Th2 |
Highest |
Highest |
Lowest |
Intermed. |
Low |
Intermed. high |
Tfh (<10%) |
High |
Highest |
Intermed. |
Low |
Lowest |
Intermed. low |
Tregs (<5%) |
High |
Highest |
Intermed. high |
Low |
Lowest |
High |
Macrophages (38–60%) |
|
|
|
Elevated |
Most elevated |
Elevated |
M0 (<15%) |
Highest |
High |
Intermed. low |
Intermed. |
Lowest |
High |
M1 (<10%) |
Intermed. |
Highest |
Intermed. |
Intermed. low |
Lowest |
Intermed. |
M2 (>20%) |
Intermed. low |
Lowest |
Intermed. |
High |
Highest |
High |
Tumor proliferation rate |
Highest |
Highest |
Low |
High |
Lowest |
High |
Survival |
|
|
|
|
|
|
OS |
Intermediate |
Intermediate |
Best |
Worst |
Worse |
Worst |
PFI |
Intermediate |
Intermediate |
Best |
Worst |
Worse |
Worst |
NSCLC subtype |
Predom. in LUSC; third common in LUAD ** |
Second most common in LUAD and LUSC |
Predom. In LUAD *** |
LUSC ** |
|
|
Factors of immunogenecity |
|
|
|
|
|
|
DNA damage |
|
|
|
|
|
|
Tumor neoantigen load |
|
|
|
|
|
|
SNVs |
Highest |
Second highest |
|
|
Lowest |
|
Indels |
Highest |
Second highest |
|
|
Lowest |
|
ITH |
Elevated |
Elevated |
Lowest |
|
|
|
Enriched oncogenic driver mutations |
APC, JAK1, PIK3CA, FGFR3 |
PIK3CA, FGFR3 |
CDH1, PIK3CA, FGFR3 |
EGFR |
|
|
TCR diversity |
Intermediate |
Highest |
Intermediate |
Low |
Lowest |
Highest |
Immunomodulators |
|
|
|
|
|
|
Expression |
|
|
|
|
|
|
CXCL10 |
|
Highest |
|
|
Lowest |
Second Highest |
EDNRB |
Low |
Lowest |
|
|
Highest |
|
BTLA |
|
|
|
High |
High |
|
Networks modulating the immune response |
|
|
|
|
|
|
Predominant immune cells |
|
CD8 T cells |
CD8 T cells, CD4 T cells |
CD4 T cells |
|
CD4 T cells |
Intracellular regulatory networks |
|
|
|
|
|
|
TGF-β (somatic mut+) |
|
↓Leuk Fract. |
↑Leuk Fract. |
|
|
↓Leuk Fract. |
|
↑r DC, M0, M1, M2, r NK, plasma cells |
↑E, a Mast, M0/2, a DC, r NK, TγΔ |
↑M1, M2, N, CD4, Treg |
↑M0,M1, a DC |
↑M0, Treg, mr CD4 |
↑r DC |
|
↓a NK, Treg, Tfh, CD8 |
↓CD8, Treg, Tfh, a NK |
↓DC, M0, Tfh, m B cells, plamsa cells |
↓monocytes |
↓n CD4, CD8 |
|
Extracellular comm. networks |
|
|
|
|
|
|
|
|
IFN-γ (+) |
IFN-γ (+) |
|
|
|
|
|
TGF-β (+) |
TGF-β, TGF-βR(+) |
|
|
TGF-β, TGF-βR(+) |
T cell and macrophage-related signaling |
CD80-CTLA4 |
LAG-3, CD27/28 |
CD27, PD-1 |
TLR4, VEGFB |
TLR4 |
TLR4 |
|
CD70-CD27 |
TIGIT, ICOS, CTLA, PD-1 |
CCR4, 5; CXCR3 DARC |
|
EDN3-EDNRB, CX3CL1-CX3CR1 |
ITGB2 |
|
IL1A/1B-IL1R2 |
CXCR3, CCR1,4,5 |
|
|
|
CD276 |
|
CXCL9-CXCR3 |
BTLA |
|
|
|
|
The wound healing, IFN-γ-dominant, and inflammatory subtypes are associated with relatively higher lymphocyte fractions (LF), which is the highest in the IFN-γ-dominant TIME. Type II helper T cells (Th2) and regulatory T cells (Tregs) were also elevated in the wound healing and IFN-γ-dominant TIME subtypes, as observed in a TGF-β-dominant TIME. The lymphocyte-depleted, immunologically quiet, and TGF-β-dominant TIME subtypes are associated with noticeably higher fractions of M2 macrophages and lower fractions of M1 macrophages. The highest and lowest M1/M2 ratios were observed in the IFN-γ-dominant and the immunologically quiet subtypes, respectively. Overall, the inflammatory subtype was associated with the best overall survival (OS). Only increased LF in the wound healing and IFN-γ dominant TIMEs significantly correlated with increased OS. This was likely related to the lower tumor proliferation rate associated with the inflammatory TIME. The lymphocyte-depleted, immunologically quiet, and TGF-β-dominant subtypes were associated with lower LF, worse survival, and higher incidence of progression. Factors associated with increased immune activation, such as lymphocyte infiltration, TCR richness, and increased fractions of Th17 and Th1 cells are associated with improved survival, while features of immune suppression, such as the wound healing (high angiogenic gene expression), macrophage regulation, and TGF-β signatures are associated with shortened survival
[41].
The proportions of different TIME subtypes vary substantially among different cancers. The inflammatory, IFN-γ-dominant, and wound-healing subtypes are most common in lung adenocarcinoma (LUAD), while wound-healing and IFN-γ-dominant subtypes predominate in lung squamous cell carcinoma (LUSC). The immunologically quiet TIME is absent in both LUAD and LUSC. Consistent with their predominant TIME subtypes, LUAD and LUSC have the highest leukocyte fractions among all solid tumors analyzed, which partially explains their response to ICIs
[9][41][42][43]. Increases in lymphocyte and macrophage signatures are associated with increased OS for LUAD and prolonged progression-free interval (PFI) for both LUAD and LUSC. This is most likely related to the increased fractions of CD8
+ T cells and M1 macrophages in their predominant TIME subtypes. When broken down to specific immune cells, monocytes, mast cells (resting), dendritic cells (DCs), and memory B cells are prominently associated with prolonged OS for LUAD, whereas Tfh cells, γδ T cells, CD8
+ T cells, activated NK cells, and M1 macrophages are associated with prolonged OS for LUSC. Tregs, CD8
+ T cells, CD4 T cells, resting mast cells, M1 macrophages, DCs (resting), and memory B cells are associated with prolonged PFI for both LUAD and LUSC, thus suggesting the importance of an overall active immune infiltrate for achieving a durable response and prolonged survival after ICB in lung cancer patients.
The tumor neo-antigen load is highest in the wound healing and IFN-γ dominant TIMEs and lowest in the immunologically quiet TIME. Higher tumor neo-antigen loads in the first two types of TIMEs are associated with increased PFI, but the opposite has been observed in the inflammatory, lymphocyte-depleted, and immunologically quiet TIME subtypes
[41]. This finding may relate to the presence of a normal adaptive antitumor immune response to increased tumor neo-antigens in the first two TIME subtypes but the presence of immune tolerance and immunological ignorance/exclusion in the latter three TIME subtypes. The way in which the level of tumor neoantigens associates with the level of TILs in each TIME subtype remains to be further investigated. Among all factors of immunogenicity, elevated SNV neoantigen load, non-silent mutations, and intra-tumoral heterogeneity (ITH) generally correlate with increased leukocyte fraction within the TIME. This usually represents elevated CD8
+ T cells, M1 macrophages, and CD4
+ memory T cells, and decreased Treg, mast, DC, and memory B cells. These correlations are strongest for in an inflammatory TIME, with weaker correlations observed in the wound healing, IFN-γ dominant, and the lymphocyte depleted TIMEs.
Different levels of driver mutation enrichment are found in different TIME subtypes, with most of them identified in the wound healing and IFN-γ dominant TIMEs, which are also predominant TIME subtypes in LUSC and LUAD. These alterations are associated with different levels of tumor neoantigens and/or the expression of various immunomodulators (IMs) ().
Table 3. Mutations associated with the most common neoantigens and enriched in different TIME subtypes based on TCGA data.
TIME Subtype |
Neoantigen-Related Driver Mutations |
Enrichment |
Wound healing |
KRAS, KRAS G12, PIKC3A, TP53 |
APC (OM), JAK1 (OM), TP53 *, FAT1, PPP2R1A, BRCA1, RB1, PIK3CA (OM), PTPRD, SPTA1, CTNNB1 *, FGFR3 * (OM), SMARCA4, KRAS G12, DACH1, PTEN *, SMARCA1, JAK1, KRAS *, MSH3 |
IFN-γ-dominant |
PIKC3A, TP53 |
CASP8, HLA-A, HLA-B, ZNF750, TP53 *, MLH1, NF1 *, FAT1, PPP2R1A, BRCA1, RB1 *, PIK3CA(OM), PTPRD, SPTA1, DACH1 |
Inflammatory |
BRAF |
BRAF, CDH1 (OM), PBRM1 * |
Lymphocyte-depleted |
IDH1 |
EGFR (OM), CTNNB1 * |
Immunologically quiet |
TP53, IDH1 |
IDH1 R132H, ATRX, CIC *, TP53 * |
TGF-β-dominant |
KRAS G12 |
KRAS G12 |
Some are associated with increased leukocyte fraction (TP53, HLA-B, BRAF, PTEN, NF1, APC, and CASP8), while others are associated with decreased leukocyte fraction (IDH1 R132H, GATA3, KRAS, NRAS, CTNNB1, and NOTCH1). Their association with tumor neoantigen generation, IM expression, and ultimately leukocyte fraction provides further evidence for tumor intrinsic gene alterations’ role in the sculpting of the TIME, which warrants further exploration to guide the treatment of NSCLC and other solid tumors
[41].
The pattern of IM expression varies in different TIME subtypes. Stimulatory modulator CXCL10 is most highly expressed in the IFN-γ-dominant TIME, while inhibitory modulators, such as EDNRB and BTLA, are most highly expressed in the more immune-suppressive TIME subtypes. A balance between T cell activation and suppression is found in more immune-stimulatory TIME subtypes, which is evidenced by the expression of both stimulatory and inhibitory IM genes, such as SLAMF7, TNFSF4 (OX40L), IL10, CD40, and IDO1. On the contrary, modulators associated with immune infiltration are more frequently deleted in the immunologically quiet TIME (e.g., TGFB1, KIR2DL1, KIR2DL3), which is consistent with a lack of TILs in this TIME subtype. Overall, TIME subtypes with increased CD8
+ T cell infiltration have been associated with the expression of stimulatory IMs, while those with increased infiltration by CD4 T cells and macrophages were associated with increased TGF-β signaling (). This pattern of IM expression reflects the predominance of different extracellular signaling networks associated with the fraction of different immune cells in the TIME
[41].
Intrinsic tumor mutations interact with external signaling networks in a particular TIME with different driver mutations modulating IM expression in a TIME subtype-specific manner through common transcription factors (TFs). For example, ATM mutations and co-occurring STK11 and SMARCA4 mutations may drive wound healing TIME-specific gene expression through STAT5A in LUAD, while KEAP1 mutations, which often co-occur with STK11 and SMARCA4 mutations, drive the expression of genes specific to the immunologically quiet and TGF-β-dominant TIMEs through IRF8 in LUAD
[41][44]. In LUSC, NFE2L2 mutation may drive the expression of wound healing and IFN-γ-dominant TIME-specific genes through IRF4, as well as the TGF-β dominant TIME specific gene expression through NFKB2
[41]. TIME characterization may be further enhanced with identifying T cell associated receptors and ligands that are uniquely present or absent in particular TIME subtypes, such as the absence of CTLA, LAG-3, TIM-3, TIGIT, ICOS, and IL2A in the inflammatory TIME, or the presence of IL1B and VEGFB in the TGF-β dominant TIME
[41].