Table 3 Key clinical studies assessing extracellular vesicle blood-based predictive biomarkers for immunotherapy in NSCLC and melanoma

NSCLC

Shukuya et al. 2020 [284]

Plasma

(n = 29)

miR-200c-3p,

miR-21-5p,

miR-28-5p

Association with poor response.

AUC for the combination (miR-21-5p, miR-28-5p and miR-199a-3p) = 0.925; AUC for (PD-L1 tissue expression) = 0.575

NSCLC

Peng et al. 2020 [285]

Plasma

(n = 30)

miR-320d,

miR-320c,

miR-320b,

miR-125b-5p

Association with poor response.

Association with progressive disease compared to partial response for baseline levels.

Reduction in miR-125b-5p post-treatment levels when compared to pre-treatment samples among those who achieved a partial response.

NSCLC

Miguel-Perez et al. 2022 [286]

Plasma

(n = 72)

Training (n = 33)

Validation (n = 39)

Δ PD-L1

Association with poor response, shorter PFS and OS for the increase in EV PD-L1 following treatment with immunotherapy.

NSCLC

Brocco et al. 2021 [287]

Whole blood

(n = 59)

CD31+ endothelial-derived EVs

10 proteins enriched for neutrophil degranulation

(e.g., annexin A2 and S100A8/9)

Association with poor response

Low blood concentration of CD31+ endothelial-derived EVs pre-treatment

was associated with longer OS and higher disease control.

EV-associated proteins involved in neutrophil degranulation (e.g., annexin A2 and S100A8/9) decreased during treatment in responders while a positive change was observed among non-responders.

Melanoma

And NSCLC

Del Re et al. 2018 [288]

Plasma

Melanoma cohort (n = 18)

NSCLC cohort

(n = 8)

Δ PD-L1

(mRNA)

Association with poor response for the increase in EV PD-L1 following treatment with immunotherapy.

Association with complete and partial responses for the decrease in EV PD-L1 following treatment with immunotherapy.

Melanoma

Chen et al. 2018 [289]

Plasma

(n = 44)

PD-L1

Association with poor response for

pre-treatment plasma EV PD-L1 protein levels.

Association with improved response for the increase in EV PD-L1 among responders. This observation was not found among non-responders.

Melanoma

Cordonnier et al. 2020 [290]

Plasma

(n = 46)

Δ PD-L1

Association with poor response, PFS and OS especially in an increase of EV PD-L1 > 100 pg/mL post immunotherapy.

EV PD-L1 was detected in all patients (100%) whereas only 67% were PD-L1 positive in tumor biopsies.

AUC for Δ PD-L1 = 0.87 for discriminating between responders vs. non-responders.

Melanoma

(Turiello et al. 2022 [291])

Serum

(n = 41)

PD-L1

CD73

Association with improved response for the increase in EV PD-L1 among responders.

Association with poor response for the increase in EV CD73 among non-responders.

Melanoma

(Shi et al. 2020 [292])

Plasma

Training (n = 50)

Validation (n = 30)

Dynamics of several on-treatment transcripts and enriched pathways based on RNA-seq analysis (e.g., T cell receptor, CD28 costimulatory and CTLA4 signaling)

Association with decreased activity during the receipt of immunotherapy in non-responders.

Association with poor response for  transcripts such as (e.g., CD1A, MAP2K4, TRBV7–2, IGFL1) in pre-treatment samples of non-responders.

Melanoma

Tucci et al. 2017 [293]

Plasma

(n = 59)

EV biomarkers from T-cells (P-D1 and CD28) and dendritic cells (CD80 and CD86) based on flow-cytometry analysis

Association of baseline EV P-D1 and CD28 from T cells with improved PFS and OS.

Upregulated levels of costimulatory molecules (CD80 and CD86) on dendritic cells at the end of immunotherapy treatment in patients who achieved a longer PFS.

Melanoma

Serratì et al. 2022 [294]

Plasma

(n = 71)

EV biomarkers from T-cells (P-D1) and melanoma cells (PD-L1)

Association of higher levels of P-D1+ EVs from CD8+ T cells with poor response, PFS, and OS.

Association of higher levels of PD-L1+ EVs from melanoma cells with poor response, PFS, and OS.

AUC = 0.86 for the combination of (P-D1+ EVs from CD8+ T cells and PD-L1+ EVs from melanoma cells) showing a strong predictive value for poor PFS.