Table 1 Summary of studies using scRNA-seq to discover novel ICB biomarkers

Melanoma

TCF7 expression of CD8⁺ T cells

N = 48 melanoma tumor tissues

CD8⁺ T cells

6,350

Higher levels of TCF7 protein in CD8⁺ T cells at baseline predict better responses to ICB therapy.

Anti-PD-1,

anti-CTLA-4

Tumor

[108]

TNBC

CXCL13⁺ CD8⁺ T cells, CXCL13⁺ CD4⁺ T cells, and MMP9⁺ macrophages

N = 78 TNBC tumor tissues and blood samples

CD45⁺ cells

489,490

The baseline levels of CXCL13⁺ CD8⁺ T cells, CXCL13⁺ CD4⁺ T cells, and MMP9⁺macrophages are high in patients responding to PD-L1 blockade immunotherapy.

Atezolizumab

Tumor

[109]

NSCLC

PD1⁺ CXCL13⁺ T cells, IgG⁺ plasma cells, and SPP1⁺ macrophages

N = 35 NSCLC samples and N = 29 matched normal tissues

CD45⁺ cells

361,929

The lung cancer activation module (LCAM) includes PD1⁺ CXCL13⁺ T cells, IgG⁺ plasma cells, and SPP1⁺ macrophages. A high baseline LCAM score is a biomarker of a favorable ICB response.

Atezolizumab

Tumor

[110]

GC

CXCL13 expression in CD8⁺/CD4⁺ T cells

N = 19 GC samples and N = 19 normal tissues

T cells

53,768

CXCL13⁺ CD8⁺ and CD4⁺ T cells are important for TLS formation and high CXCL13 expression is associated with improved efficacy of ICB.

Pembrolizumab

Tumor

[111]

NSCLC

Tissue-resident memory CD8⁺ T (T_RM)-like cells

N = 10 NSCLC patients

CD45⁺ cells

31,887

The high expression signature of CD8⁺ ITGAE⁺ T_RM-like cells with high cytotoxicity, effector signature, and memory signature is predictive of favorable anti-PD-1 therapy responses.

Anti-PD-1

Tumor

[129]

Melanoma and NSCLC

TOX expression of T cells

N = 17 melanoma tumors and N = 14 NSCLC tumors

CD8⁺ T cells

3,195

TOX is specifically expressed by T cells and is a key transcription factor regulating T cell exhaustion. Lower expression of TOX in tumor-infiltrating T cells predicts a better response to anti-PD-1 therapy.

Nivolumab,

pembrolizumab

Tumor

[113]

Multiple cancers

T_STR cells

N = 486 samples

T cells

308,048

T_STR cells exhibited high expression of heat-shock genes and stress response signatures. T_STR cells are more abundant in non-responsive tumors.

Anti-PD-1,

anti-PD-L1,

anti-CTLA-4

Tumor

[114]

Bladder cancer

Cytotoxic CD4⁺ T cells

N = 7 bladder cancer patients

CD4⁺ T cells

19,842

The signature genes of cytotoxic CD4⁺ T cells include ABCB1, APBA2, SLAMF7, GPR18, and PEG10. High expression of this signature is associated with a positive response to PD-1 blockade.

Atezolizumab

Tumor

[130]

NSCLC

Plasma cells

N = 44 NSCLC patients

CD79A⁺ B cells

20,362

High levels of intratumoral plasma cells at baseline predict a favorable response to PD-L1 blockade therapy.

Atezolizumab

Tumor

[115]

Melanoma

CD69⁺ B cells and IGLL5⁻ CD69⁺ B cells

N = 43 melanoma tumor tissues

B cells

N.A.

The fractions of CD69⁺ B cells and IGLL5⁻ CD69⁺ B cells at baseline are both higher in responders to ICB therapy. And a higher expression level of the TLS signature predicts a better prognosis after ICB.

Anti-PD-1,

anti-CTLA-4

Tumor

[116]

NSCLC

FCRL4⁺ FCRL5⁺ memory B cells and CD16⁺ CX3CR1⁺ monocytes

N = 15 NSCLC tumor tissues

All the cells

92,330

High signatures of FCRL4⁺ FCRL5⁺ B cells and CD16⁺ CX3CR1⁺ monocytes in the TME predict a positive response to ICB therapy.

Anti-PD-1

Tumor

[131]

Melanoma

TREM2^high macrophages

N = 48 melanoma tumor tissues

CD45⁺ cells

16,291

High levels of TREM2^high macrophages at baseline are associated with poor clinical responses to ICB therapy.

Anti-PD-1,

anti-CTLA-4

Tumor

[117]

NSCLC

TREM2⁺ TAMs

N = 26 NSCLC tumor tissues and blood samples

All the cells

50,483

High levels of TREM2⁺ TAMs at baseline predict poor clinical responses to immunotherapy.

Anti-PD-1

Tumor

[118]

HCC

APOC1

N = 8 HCC tumor and normal tissues and blood samples

CD45⁺ cells

212,494

APOC1 is over-expressed in TAMs of HCC tissues and is negatively correlated with PD-1 and PD-L1 expression. High levels of APOC1 predict unfavorable responses to anti-PD-1 therapy.

Anti-PD-1

Tumor

[119]

PDAC

LRRC15⁺ CAFs

N = 20 normal and PDAC mice, N = 22 PDAC patients

Dataset1: PDPN⁺ stromal cells

Dataset2: CAFs

Dataset1: 13,454

Dataset2: 8,931

TGF-β driven LRRC15⁺ CAFs increase with tumor progression and higher expression of LRRC15⁺ CAFs signature is predictive of worse clinical response to atezolizumab in several human cancers.

Atezolizumab

Tumor

[85]

Breast cancer

ecm-myCAFs and TGFβ-myCAFs

N = 7 breast cancer tissues

FAP^hi CD29^{med − hi} SMA^hi CAFs

18,296

CAF subsets with high expression of genes encoding extracellular matrix proteins and TGF-β signaling pathway are respectively termed ecm-myCAFs and TGFβ-myCAFs. High levels of ecm-myCAFs and TGFβ-myCAFs predict primary resistance to ICB in several cancer types.

Pembrolizumab, nivolumab

Tumor

[86]

PDAC

meCAFs

N = 16 PDAC tumor tissues and normal pancreatic samples

CAFs

8,439

 A new CAF subset exhibiting highly activated metabolic signatures and high expression of PLA2G2A and CRABP2 is termed meCAFs. High levels of meCAFs at baseline predict a positive response to immunotherapy.

Anti-PD-1

Tumor

[120]

HNSCC

tsCAFs

N = 8 HNSCC tumor tissues

CAFs

5,414

Two CAF subsets (HNCAF-0/3) that can prevent exhaustion and enhance the cytotoxicity of CD8⁺ T cells are together termed T-cell stimulating CAF (tsCAF). High levels of tsCAF signatures at baseline predict favorable responses to ICB therapy.

Nivolumab,

pembrolizumab

Tumor

[121]

Multiple cancers

TaNK cells

N = 1,223 samples

NK cells

160,011

An NK cell subset (CD56^dimCD16^hi c6-DNAJB1) with low cytotoxicity and high stress response is termed tumor-associated NK (TaNK) cells. TaNK cell signatures predict resistance to ICB.

N.A.

Tumor

[122]

Melanoma

SIRPA expression of melanoma cells

N.A.

All the cells

N.A.

Melanoma cells interact with CD8⁺ T cells via the SIRPα/CD47 axis to increase CD8⁺ T cell cytotoxicity. Higher expression of SIRPA is predictive of a better response to ICB therapy in melanoma.

Anti-PD-1,

anti-PD-L1

Tumor

[123]

Melanoma

Toxicity and large count number of CD8⁺ T cells

N = 16 blood samples

CD8⁺ T cells

22,445

Patients whose CD8⁺ T cells have lower toxicity and fewer expanded clones may have worse responses to ICB therapy.

Pembrolizumab,

nivolumab,

ipilimumab

Blood

[124]

Melanoma

MAIT cells

N = 183 blood samples from 20 patients

CD3⁺ CD8⁺ T cells

51,701

 A high proportion of MAIT cells in peripheral blood at baseline predicts a favorable response to anti-PD-1 therapy.

Nivolumab,

pembrolizumab

Blood

[125]

Melanoma

S100A9⁺ monocytes

N = 16 blood samples

PBMCs

~ 50,000

High levels of baseline S100A9⁺ monocytes in peripheral blood predict poor clinical responses to ICB therapy.

Nivolumab

Blood

[127]

Melanoma

CD8⁺ T_OXPHOS cells

N = 16 melanoma tumor tissues and blood samples

CD8⁺ T cells

173,061

 A unique CD8⁺ T cell subset with high levels of oxidative phosphorylation (OXPHOS), high exhaustion markers, and high cytotoxicity markers is termed CD8⁺ T_OXPHOS. High CD8⁺ T_OXPHOS cell signatures in tumors and peripheral blood predict ICB resistance.

N.A.

Blood and tumor

[126]

CRC

Inflammatory conditions and neutrophil-to-lymphocyte ratio (NLR)

N = 6 MSI-H CRC tumor samples

All the cells

12,118

Neutrophils participate in immunosuppression via the CD80/CD86-CTLA4 axis. A high NLR is predictive of an unfavorable ICB response in MSI-H CRC patients.

Anti-PD-1

Blood and tumor

[128]