Skip to main content
Fig. 3 | Journal of Experimental & Clinical Cancer Research

Fig. 3

From: Extracellular vesicle PD-L1 dynamics predict durable response to immune-checkpoint inhibitors and survival in patients with non-small cell lung cancer

Fig. 3

EV PD-L1 dynamics outperformed tissue PD-L1 as a predictor of ICIs response: (A) Representative axial section computed tomography (CT) images from a responder and a non-responder at baseline and during ICIs treatment. (B) Examples of immunohistochemistry micrographs of positive and negative tissue PD-L1 staining (scale bars 5 µm) and (C) EV PD-L1 blots from a responder with decreasing EV PD-L1 (0.29) and a non-responder showing an increase (1.55). (D) ICIs cohort A (n = 33), non-responders (NR) showed increased EV PD-L1 during treatment in comparison to responders (p = 0.017) (Mann–Whitney U test). (E) In the validation cohort, non-responders undergoing Pembrolizumab + Docetaxel (n = 24) showed a trend towards increased EV PD-L1 in comparison to responders (p = 0.050) while those treated with Docetaxel alone (n = 15) showed no differences (p = 0.794) (F) (Mann–Whitney U test). (G) As observed in the ROC curve, EV PD-L1 dynamics was a better predictor than tissue PD-L1 TPS with an AUC = 74.4% vs. 62.6% for the tissue (binary logistic regression). (H) This was also observed in the validation cohort of patients treated with ICIs with AUC = 75% for the EVs vs. 64.1% for the tissue. (I) In comparison, similar AUCs were observed in the Docetaxel treated group with 54.5% and 59.1%, respectively (binary logistic regression)

Back to article page