Monitoring Pseudoprogression Using Circulating Small Extracellular Vesicles Expressing PD‐L1 in a Melanoma Patient Treated With Immune Checkpoint Inhibitors

ABSTRACT

Immune checkpoint inhibitors (ICI) have revolutionized the treatment of metastatic malignancy. However, unique immune response patterns can occur, such as pseudoprogression, which corresponds to new lesion development or temporary tumour growth followed by regression. Misidentifying pseudoprogression may halt ICI therapy, due to the absence of biomarkers to distinguish progression from pseudoprogression. In 2020, our team proposed small extracellular vesicles expressing PD‐L1 (sEV‐PD‐L1) as a predictor of melanoma treatment response. We report a case of pseudoprogression in a patient treated with nivolumab and ipilimumab for metastatic melanoma, and showing reduced circulating sEV‐PD‐L1. To our knowledge, this is the first report of PD‐L1 monitoring in circulating sEV during pseudoprogression under ICI. A decrease in PD‐L1 in circulating sEV might be an early sign of disease response to ICI, and may help to diagnose pseudoprogression. This case supports further evaluation of sEV‐PD‐L1 to identify responder patients to ICI, especially in case of pseudoprogression.

Trial Registration: EXOMEL1 P/2018/40 1 AC‐2015‐2496/DC‐2014‐2086. NCT05744076.

Abbreviations

Anti‐PD‐1

anti‐programmed cell death 1 protein

ICI

immune checkpoint inhibitors

iRECIST

immune response evaluation criteria in solid tumours

NSCLC

non‐small cell lung cancer

PD‐L1

programmed cell death ligand 1 protein

RECIST

response evaluation criteria in solid tumours

SEC

chromatographic methods like size‐exclusion chromatography

sEV

small extracellular vesicle

sEV‐PD‐L1

small extracellular vesicles expressing PD‐L1

1. Introduction

Although immune checkpoint inhibitors (ICI) have revolutionized the management of metastatic malignancies, distinct immune‐related patterns of morphologic tumour response have been observed. These may vary from tumour shrinkage (partial or complete response) or stable disease, as defined by the response evaluation criteria in solid tumours (RECIST), to an increase in the size of existing lesions or the occurrence of new lesions later followed by tumour regression. This latter phenomenon is called pseudoprogression (Wolchok et al. 2009; Di Giacomo et al. 2009) and is reported in up to 10% of patients with melanoma. Pseudoprogression represents a clinical challenge, since actual tumour response could be misinterpreted as disease progression, and lead to ICI interruption (Wolchok et al. 2009). However, most patients with initial progression will display progressive disease and not ICI‐related pseudoprogression. Therefore, biomarkers to distinguish pseudoprogression from progression under ICI would be helpful, but are lacking in clinical practice.

Tumour‐derived small extracellular vesicles (sEV) have been identified as a promising new class of biomarkers for monitoring tumour response to treatment. sEVs are 40–200 nm spherical lipid bilayer particles that are generated and released through a well‐defined intracellular trafficking pathway (Chiou and Burotto 2015).

In the setting of cancer, sEV are released by all cancer cells, and can be isolated from the plasma of the circulating peripheral blood. Moreover, the literature supports the finding that sEV‐based diagnostics provide higher sensitivity and specificity than conventional biopsy or liquid biomarkers, due to their stability in biofluids (Colombo et al. 2014). sEV markers are readily available from most biofluids, and recent technological advances in sEV isolation make sEV‐based diagnostics both cost‐ and labour‐effective (Sheridan 2016).

In 2020, our team highlighted that monitoring PD‐L1 expression in circulating sEV could predict tumour response to ICI in melanoma (Cordonnier et al. 2020). This study was carried out on the EXOMEL cohort (“Analysis of circulating EXOsomes in MELanoma patients”; NCT05744076); a prospective cohort of patients with advanced melanoma for which we analysed the circulating sEV and PD‐L1 expression of sEV (sEV‐PD‐L1). The EXOMEL cohort comprised 150 patients, of whom 96 patients were selected to undergo sEV‐PD‐L1 testing, and 46 of these patients had repeated tests. That study concluded that sEV‐PD‐L1 represents a useful biomarker to help in decisions concerning the pursuit and relevance of anti‐PD‐1 therapies, by informing about the chances of response to treatment. Indeed, a decrease in sEV‐PD‐L1 was found to be associated with tumour response, while an increase of sEV‐PD‐L1 > 100 pg/mL was found to be associated with progressive disease (Cordonnier et al. 2020).

We report here the case of a patient from the EXOMEL cohort who was treated with nivolumab and ipilimumab for metastatic melanoma and who had pseudoprogression. The aim of the present study was to describe monitoring of sEV‐PD‐L1 during pseudoprogression.

2. Material and Methods

2.1. Patient Description

A patient in his late 50s with NRAS‐mutant stage IV metastatic melanoma with liver metastases was treated by a protocol consisting of four infusions of nivolumab (1 mg/kg) and ipilimumab (3 mg/kg) every 3 weeks, then followed by nivolumab monotherapy (3 mg/kg) every 2 weeks. sEV‐PD‐L1 was monitored at each imaging exam, as part of a prospective institutional research protocol, called the EXOMEL study (NCT05744076). Tumour response was based on immune‐related iRECIST using unidimensional measurements on contrast‐enhanced computed tomography scan or CT (Computed Tomography). Imaging re‐evaluation was performed in blind‐coded samples. The EXOMEL study was conducted between January 2016 and December 2018 in the Department of Dermatology, University Hospital of Besançon, France.

2.2. Cell Culture

SK‐MEL‐2 human melanoma cells were purchased from the ATCC. SK‐MEL‐2 were cultured at 37°C, 5% CO2 in RPMI 1640 medium (Dutscher) supplemented with 10% foetal bovine serum (Dutscher) depleted in EVs, and was tested weekly for mycoplasma contamination.

2.3. Plasma Collection and sEV Isolation

Peripheral blood was drawn into sodium heparin tubes. A maximum period of 2 h at 4°C was mandatory between the collection and the final freezing of the samples. To ensure sEV integrity, collected blood was centrifuged at 2400 g for 15 min at 4°C to remove cell debris and dead cells. The supernatant was draw from the top down with a pipette, leaving a specific volume of plasma above the pellet to prevent disturbance and releasing platelets. Directly thereafter, sampling and plasma specimens were frozen and stored at −80°C. Four blood samples were collected during patient follow‐up, namely, T1: at the beginning of treatment; T2: 3 months after treatment initiation (four ipilimumab and nivolumab infusions), corresponding to pseudoprogression; T3: after three more infusions (nivolumab alone) corresponding to confirmed pseudoprogression; T4: 3 months later, after six infusions (nivolumab alone), corresponding to disease progression. All analyses of sEV‐PD‐L1 from thawed plasma, including isolation, characterization and quantification were performed as previously reported (Cordonnier et al. 2020). Briefly, sEV‐PD‐L1 from 5 mL of were isolated by differential centrifugation (300 g for 5 min, 17,000 g for 10 min, and 200,000 g for 1 h at 4°C with no brake (Beckman Coulter, Optima XPN‐100)). Supernatants were carefully removed and sEV pellets were resuspended in 50 µL of 1% RIPA lysis buffer to 1X final concentration (for Western blot and ELISA analysis) or in 50 µL of PBS 1X (for NTA analysis). This method was selected due to its simplicity, low cost, and suitability for working with the small plasma volumes available in this proof‐of‐concept study. We acknowledge its limitations in terms of EV purity and clinical transfer and accept this trade‐off in the context of our exploratory approach

2.4. sEV Characterization

Native sEV were evaluated for their size and concentration by nanoparticle tracking analysis using a NS300 Instrument (Malvern, Amesbury, UK). Expression of sEV biomarkers was characterized by Western blot. sEV were lysed and separated on SDS/PAGE gels. Proteins were transferred onto a polyvinylidene fluoride membrane (Amersham GE Healthcare Life Sciences) for Western blotting analysis. After transferring, membranes were blocked with 5% bovine serum albumin for 1 h and incubated overnight at 4°C with specific antibodies (1/1000): CD9 (sc‐13118, Santa Cruz Biotechnology), ALIX (NB100‐65678, Novus Bio), TSG101 (sc‐7964, Santa Cruz Biotechnology), CD63 (NBP2‐4225, BioTechne), PD‐L1 (sc‐50298, Santa Cruz Biotechnology), GRP94 (ADI‐SPA‐850, Enzo Life Sciences) and β‐actin (A3854, Sigma Aldrich). Following incubation with secondary antibodies (JacksonImmunoResearch), immunoreactions were revealed using ECL detection reagents (34095, ThermoFisher Scientific). Band intensities were captured using Chemidoc XRS+ system and images were analysed using Image Lab software (Bio‐Rad Laboratories).

2.5. PD‐L1 Expression Analysis

PD‐L1 in plasma‐derived sEV was measured using an enzyme‐linked immunosorbent assay (PD‐L1 Human ELISA Kit, Invitrogen) according to the manufacturer's instructions. Protein concentrations were determined according to standard curves. The blank value (lysis solution alone) was subtracted from the sample value.

2.6. Transmission Electronic Microscopy (TEM)

For TEM, EVs pellets were resuspended in 50 µL 1X PBS, and an aliquot was diluted 10 times in sterile water. Ten microlitres of diluted sample was deposited on a formvar/carbon‐coated copper effluved grid and left for 4 min. After adding a drop of UranyLess solution (Delta Microscopies) for 60 s for contrast staining, excess liquid was absorbed on filter paper. Images were acquired with a Hitachi 7800 electron microscope (Hitachi high technologies, Tokyo, Japan).

3. Results

The mean size and expression of markers enriched in sEV were verified by Western blot (Figure 1A) and by nanoparticle tracking analysis (Figure 1B). We observed expression of sEV‐specific markers (TSG101, Alix, CD9 and CD63) for sEV from blood. On Western blot analysis, the Grp94 protein (endoplasmin), mainly expressed in endoplasmic reticulum, served as a negative control for sEV purity, and actin as a loading control. The mean size observed was compatible with the size of sEV, at 113.22 ± 16.23 nm (Figure 1B). Isolated exosomes were also observed by TEM (Figure 1C). The expression of PD‐L1 was analysed in sEV by Western blot (Figure 1A).

Figure 1.

Small extracellular vesicle (sEV) characterization. (A) Representative immunoblots showing expression of sEV markers (Alix, TSG101, CD63 and CD9) and PD‐L1 in plasma‐derived sEVs from a melanoma patient or lysis cell line SK‐MEL‐2 (human melanoma). Grp94 is used as negative control and actin as a loading control. (B) Representative size distribution of particles isolated from plasma of melanoma patient, obtained by nanoparticle tracking analysis (NS300) with a sample image from the video used for the tracking analysis. (C) Transmission electron microscopy images of isolated sEV. Red Scale bar = 100 nm.

Next, we determined the concentration of sEV‐PD‐L1 at each time points (T1–T4) by ELISA.

At the beginning of treatment, the patient's circulating sEV‐PD‐L1 level was 177 pg/mL (T1 – Figure 2A). Three months after treatment initiation (four ipilimumab and nivolumab infusions), evaluation by computed tomography (CT) scan showed the appearance of new lesions in both lungs, while the disease remained stable in the liver, suggesting disease progression according to the RECIST criteria, and immune unconfirmed progressive disease (iUPD) according to the immune RECIST (iRECIST) criteria (T2 – Figure 2A). At that time, sEV‐PD‐L1 decreased by 100 pg/mL (T1 to T2 – Figure 2B) to reach 77 pg/mL (T2 – Figure 2A).

Figure 2.

CT scan images and PD‐L1 levels in circulating small extracellular vesicles (sEV‐PD‐L1) of the patient over the course of the disease under immune checkpoint inhibitors. (A) Upper panel: CT images of the liver and the lungs at several follow‐up timepoints (T1–T4) showing tumour progression. Lower panel: Corresponding levels of PD‐L1 detected by ELISA in sEVs isolated from the plasma of the patient during this follow‐up. (B) Change in sEV‐PD‐L1 between each follow‐up correlated with disease status of the patient. The time periods associated with pseudoprogression and progression are indicated below the figure.

Nivolumab was then pursued alone, as foreseen by the protocol, at a dose of 3 mg/kg every 2 weeks. Because of iUPD, CT scan was repeated after 6 weeks (three infusions) and showed partial response in the liver and lungs according to the RECIST criteria (T3 – Figure 2A), suggesting radiologic pseudoprogression in the lungs under ICI according to the iRECIST criteria. Meanwhile, sEV‐PD‐L1 increased by 34 pg/mL (T2 to T3 – Figure 2B) to reach 111 pg/mL (T3 – Figure 2A) but remained below the baseline level observed at T1 (177 pg/mL).

The patient was followed up 3 months later, by which time he had received six infusions of nivolumab. At this point (T4), CT scan revealed disease progression. Concerning sEV‐PD‐L1, we observed a substantial increase of 423 pg/mL (T3 to T4 – Figure 2B) to reach a value of 534 pg/mL (T4 – Figure 2A). The patient eventually died of the disease 6 months later.

4. Discussion

Pseudoprogression is a phenomenon observed in patients undergoing ICI therapy, and is characterized by an initial increase in tumour size or the emergence of new lesions, followed by a subsequent decrease in tumour burden. This response is hypothesized to be linked to transient lymphocyte infiltration into tumours, which may precede actual tumour shrinkage (Martin‐Romano et al. 2020). Despite being a recognized entity, pseudoprogression remains a clinical challenge, particularly due to the difficulty of distinguishing it from true disease progression, with prevalence across solid tumours reported to be below 10% (Park et al. 2023). The iRECIST criteria have been developed to better assess tumour response under ICI. In this regard, prompt discontinuation of therapy in these patients presenting early progressive disease at first evaluation could run the risk of discontinuing a potentially effective drug. Moreover, pseudoprogression is mainly diagnosed a posteriori as suggested by our case. Consequently, there is a compelling need for the development of new biomarkers to enable timely diagnosis of pseudoprogression under ICI.

Small EVs stand out as promising markers carrying contents that are inherently abundant and stable. sEVs are abundant nanovesicles that are non‐invasively available in the bloodstream. More importantly, they are produced at all stages of the disease, unlike other circulating markers, such as tumour DNA, which is mainly derived from apoptosis and necrosis of tumour cells (Jahr et al. 2001). Consequently, sEV are constantly involved in intercellular communication and modulate the tumour microenvironment by delivering tumour‐specific cargo, such as DNA, RNA and proteins.

In metastatic melanoma, the expression of PD‐L1 in circulating sEV represents a potential surrogate biomarker of tumour response to ICIs. In fact, a decrease in sEV‐PD‐L1 may serve as early indicator of tumour response (Cordonnier et al. 2020; Lim et al. 2018; Chen et al. 2018). For instance, Chen et al. recently reported that the level of sEV‐PD‐L1 positively correlated with overall tumour burden, and that an early, on‐treatment increase in sEV‐PD‐L1 correlated with tumour response (Chen et al. 2018). In contrast, we reported a correlation between the increase in sEV‐PD‐L1 and disease progression (Cordonnier et al. 2020). Furthermore, in metastatic melanoma, circulating sEVs expressing PD‐L1 have been identified as mediators of resistance to ICIs, with higher levels of sEV‐PD‐L1 correlating with poorer progression‐free and overall survival. This indicates that sEV‐PD‐L1 can serve as an independent biomarker of response, particularly when released from melanoma (Serratì et al. 2022). Beyond melanoma, the prognostic and predictive value of sEV‐PD‐L1 has been explored in other solid tumours. In a cohort of head and neck cancer patients, sEV‐PD‐L1 was significantly higher in patients with active disease, as compared to patients without evidence of disease (Theodoraki et al. 2018). Similarly, in non‐small cell lung cancer (NSCLC), dynamic change in sEV PD‐L1 expression have demonstrated predictive value, with an area under the curve (AUC) of 0.85, suggesting its potential as a reliable biomarker for guiding treatment decisions in real‐time (de Miguel‐Perez et al. 2022).

Collectively, these findings align with our current observations, where a decrease in sEV‐PD‐L1 correlates with tumour response, while an increase is associated with tumour progression under ICI therapy. In this study, we demonstrated that during pseudoprogression, ∆sEV‐PD‐L1 remained below 100 pg/mL, whereas in true progression, it exceeded 100 pg/mL. This finding corroborates previously observed variations in metastatic melanoma (Cordonnier et al. 2020).

In future studies, extracellular vesicle (EV) isolation protocols should be further refined to optimize both purity and scalability, particularly for clinical applications. Combining physical isolation techniques—such as ultracentrifugation with ultrafiltration—or incorporating chromatographic methods like size‐exclusion chromatography (SEC), could significantly enhance the separation of EVs from soluble protein contaminants and other non‐vesicular particles. From a translational perspective, commercially available precipitation‐based kits (e.g., ExoQuick, Total Exosome Isolation) may also represent a viable alternative. Although these methods typically yield EVs of lower purity compared to ultracentrifugation, they offer practical advantages including speed, ease of use, and compatibility with routine clinical workflows. However, a notable limitation is that precipitation methods often co‐isolate abundant serum/plasma proteins, which can lead to high background noise in omics analyses, potentially complicating downstream molecular characterization. Despite this drawback, these kits may still be well‐suited for the analysis of pre‐identified EV‐associated targets, especially in standardized clinical settings focused on initial diagnostics or patient monitoring, where rapid and reproducible processing is prioritized over comprehensive vesicle profiling.

Beyond this work, it would be interesting to explore the biological significance of this decrease in PD‐L1 expression on sEV. What underlying mechanisms does it reflect? Although, the precise mechanism remains unclear we can hypothesize that sEV‐PD‐L1 may be as efficient as the cancer cells from which they derive in rendering T‐cells anergic. This suggests a key role in immunosuppression, by targeting T lymphocytes in secondary lymphoid organs via the PD‐1/PD‐L1 axis (Cordonnier et al. 2020; Chen et al. 2018). Several studies have reported that the expression of sEV‐PD‐L1 is modulated by various biological and environmental factors. For instance, tumour‐derived sEVs can upregulate PD‐L1 expression in macrophages through the IL‐6/STAT3 and TLR4/NF‐κB signalling pathways, contributing to an immunosuppressive microenvironment (Pucci et al. 2021). Additionally, inflammatory cytokines such as IL‐1β and IFN‐γ serve as critical regulators of PD‐L1 expression. IL‐1β enhances PD‐L1 expression via the NF‐κB signalling cascade, creating an immunosuppressive feedback loop between tumour‐associated macrophages and tumour cells (Xu et al. 2023; Hirayama et al. 2023). Similarly, IFN‐γ commonly produced by tumour‐infiltrating lymphocyte can upregulate PD‐L1 expression through the JAK/STAT3 and PI3K/AKT pathways, further promoting tumour immune evasion (Zhang et al. 2017; Pistillo et al. 2020). More recently, in melanoma, interleukin‐2 (IL‐2) has been shown to reduce sEV secretion and PD‐L1 expression in cancer cells by modulating the IL‐2 receptor signalling pathway. This, in turn, impacts the tumour microenvironment and immune response (Noh et al. 2024). These findings suggest that distinct biological behaviours may occur simultaneously, highlighting the need for further investigation to fully understand these mechanisms.

In this study, we focused on the potential of sEV‐associated PD‐L1 as a biomarker for identifying pseudoprogression in melanoma. Several studies have explored potential biomarkers for distinguishing pseudoprogression, including clinical factors and laboratory parameters such as peripheral blood counts, circulating tumour DNA, and cytokine levels. Notably, fluctuations in serum IL‐8 levels have been identified as a potential biomarker for monitoring and predicting clinical benefit from immune checkpoint blockade in melanoma and NSCLC patients (Sanmamed et al. 2017). Additionally, research in metastatic melanoma has indicated that combining blood biomarkers specifically lactate dehydrogenase (LDH) and S100 with radiomic features from PET/CT imaging can effectively differentiate pseudoprogression from true progression. A combined model achieved an AUC of 0.82, highlighting its potential utility (Basler et al. 2020). However, most of these findings remain at the preclinical stage, likely due to small sample sizes and the heterogeneity in defining atypical responses. Future investigations should consider an integrative approach combining these potential biomarkers, including sEV‐PD‐L1, to enhance the accuracy of pseudoprogression identification and improve patient management.

Thus, in this work, we propose that the analysis of sEV‐PD‐L1 could help to predict tumour response before observing response on medical imaging. Pairing sEV‐PD‐L1 assessments with imaging may make it possible to refine tumour burden by distinguishing pseudoprogression from true progression, or by detecting biological disease progression, which remains unseen on imaging.

Overall, this observation provides a strong rationale to further investigate the role of sEV‐PD‐L1 as a predictor of ICI treatment response in melanoma patients.

Author Contributions

Charlée Nardin analysed and wrote the article. Valentin Vautrot performed analysis of PD‐L1 in sEV and wrote the methodology section of the article. Isen Naiken and Alexandre Doussot participated in writing the article. Eve Puzenat and Celia de‐girval reviewed the article and wrote the discussion. Carmen Garrido reviewed the article for important intellectual content. Jessica Gobbo and François Aubin work Conceptualization designed the methodology, wrote the discussion and reviewed the article for important intellectual content.

Ethics Statement

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Consent

Informed consent was obtained from the patient.

Conflicts of Interest

The authors report no conflicts of interest.

Data Availability Statement

The authors confirm that the data supporting the findings of the case report are available within the article.