All procedures were approved by the Ethics committee of Grenoble University Hospital (CHUGA) and the French Blood Agency’s Institutional Review Board Committee (IRB), and accredited by the Ministry of Education and Research under the reference #AC-2020–3959 and approval #2023-A01722-43. Mandatory written informed consent was obtained from all patients prior to their participation in this study and samples were analyzed anonymously. All enrolled participants finished the study, i.e., there was no attrition. Sex is not a biological variable. Randomization was not used.

Blood samples were obtained from patients with unresectable (stage III) or metastatic (stage IV) melanoma (n=30) starting a first-line treatment with KEYTRUDA^® (anti-PD1; pembrolizumab) as monotherapy in curative setting in the Dermatology Department of CHUGA. Treatment was administrated by intravenous (IV) infusion at 200 mg every 3 weeks (wk) (n=21) or 400 mg every 6 wk (n=9), until progression or unacceptable toxicity. At each visit, a physical exam, vital signs, ECOG performance status, weight, concomitant medications, laboratory assessments and adverse events evaluation were performed. Tumor response assessments for efficacy were performed every 12 wk by a complete physical exam and tumor imaging by CT scan or MRI, and determined by clinicians using the Response Evaluation Criteria in Solid Tumors (RECIST 1.1 criteria). Patients were classified into two groups according to their clinical response after 48 weeks of treatment: responder (R): patients with a complete response (CR; disappearance of all target and non-target lesions) or a partial response (PR; at least a 30% decrease in the sum of all target lesions, taking reference at baseline and/or persistence of one or more non-target lesions); non-responder (NR): patients with a progressive disease (PD; at least a 20% increase in the sum of target lesions, taking as reference the smallest sum on study (including baseline) and an absolute increase of sum ≥ 5 mm or the appearance of one or more new lesions and/or the unequivocal progression of existing non-target lesions) or stable disease (SD; all patients that do no present a sufficient decrease to qualify for CR or PR nor a sufficient increase to qualify for PD).

Blood samples were collected from patients before treatment initiation (T0) and at each infusion (T3/6, T12 and T24 for weeks 3/6, 12 and 24 respectively) when possible. Plasmas were collected from blood samples upon centrifugation and stored at -80°C whereas, peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll-Hypaque density gradient centrifugation (Eurobio) and stored frozen at -150°C. Clinical features of melanoma patients are summarized in Supplementary Table 1 . For this cohort, the mean age was 72 years, median age 74 years, with min = 36 and max = 93 years. Experimenters were unaware (blinded) of group assignment and outcome assessment.

Frozen PBMCs from four different time points for each patient (T0, T3/6, T12, T24) were thawed and stained in PBS 2% fetal calf serum (FCS) for 20 minutes at room temperature with multiple fluorochrome-labelled anti-human antibodies allowing the study of distinct immune cells (detailed gating strategy in Supplementary Figure 1 ). Dead cells were excluded with Live&Dead staining and Brilliant Stain buffer was used to limit staining artifacts caused by the utilization of several antibodies conjugated with fluorescent polymer dyes. After washing and fixation using BD FACS™ lysing solution, stained cells were analyzed using LSRII Flow Cytometer and FACSDiva software 9. Isotype controls were used to differentiate positive cells from nonspecific background staining (CD45⁺ cells were also used to determine the positivity threshold). To ensure quality control, standardization of the fluorescence intensities was performed using cytometer setup and tracking beads (CST).

Cultures were performed in RPMI 1640 medium supplemented with 1% nonessential amino acids, 100 μg mL^-1 gentamicin, 10% FCS and 1 mmol L^-1 sodium pyruvate at 37°C, 5% CO2. For cytokine production assessment, PBMCs were cultured whether at 2x10⁶ cells mL^-1 for DC subsets or at 0.5x10⁶ cells mL^-1 for effector immune cells. DC subsets were stimulated with either Polyinosine-polycytidylic acid (PolyI:C; 30 µg mL^-1), Resiquimod (R848; 1µg mL^-1), Class A CpG oligonucleotide (CpGA ODN 2336; 1,5 µM) or the mixture of the three TLR ligands (TLR-L, a combination of polyI:C, R848 and CpGA). Effector cells were cultured in the presence of IL-2 (0.1 IU mL^-1) and stimulated with either Phorbol 12-myristate 13-acetate (PMA; 20 ng mL^-1) and Ionomycin calcium salt (Iono; 1 µg mL^-1), or anti-CD3/CD28 antibodies, (E)-1-Hydroxy-2-methyl-2-butenyl-4-pyrophosphate lithium salt (HMB-PP; 200 nM), IL-12/IL-18 (50 ng mL^-1), α-Galactosylceramide (αGalCer; 100 ng mL^-1) or a mixture of several activators (CD3/CD8, HMB-PP, IL-12/IL-18 and αGalCer). Immune cells’ functionality was then assessed by flow cytometry (intracellular cytokine staining) and/or cytokine secretion measurements by LUMINEX technology.

To evaluate patient’s cytokine and chemokine basal status, collected plasmas were thawed and the following soluble factors were measured for each patient: Arginase-1, B7-H6, BTLA, CD27, CD28, CD47 (IAP), CD48 (BLAST-1), CD73 (NT5E), CD80, CD96 (Tactile), CD134 (OX40), CD137 (4-1BB), CD152 (CTLA4), CD276 (B7-H3), E-Cadherin, GITR, HVEM, ICOS Ligand (B7-H2), IDO, LAG-3, MICA, MICB, Nectin-2, PD1, PD-L1, PD-L2, PVR, Perforin, S100A8/A9, Siglec-7, Siglec-9, TIM-3, TIMD-4, ULBP-1, ULBP-3, ULBP-4 and VISTA (B7-H5). The dosages were conducted following manufacturer’s instructions and measured by LUMINEX technology using MAGPIX^®200 Instrument with xPONENT^® software.

Statistical analyses were done using GraphPad Prism software and applying non-parametric Kruskal-Wallis test for inter-group comparisons and/or Wilcoxon matched-pairs signed rank test with Bonferroni correction for intra-group analyses, or Spearman test for correlations. Data are shown as medians, and significance threshold was set at P < 0.0167. Euclidean distance-based hierarchical clustering (heat maps), PCA, correlation matrix and radar plots were performed using the RColorBrewer (to select color palettes), gplots (function heatmap.2), missMDA (function imputePCA), FactoMineR (function PCA), ggbiplot (function ggbiplot), factoextra (functions fviz_pca_var and fviz_contrib), corrplot (function corrplot), psych (function corr.test) and fmsb (function radarchart) packages of the R software.

Blood samples were collected from melanoma patients with unresectable (stage III) or metastatic (stage IV) melanoma (n=30) starting a first-line treatment with KEYTRUDA^® (anti-PD1; pembrolizumab) as monotherapy in a curative setting, at treatment initiation (T0), and when possible at several infusions (T3/6, T12 and T24 for weeks 3/6, 12 and 24, respectively). PBMC and plasma were retrieved and stored until use. Patients were classified into different groups according to their clinical response to immunotherapy after 48 weeks of treatment. Their clinical features are reported in Supplementary Table 1 . Patients with complete response (CR; n=6) had a disappearance of all target and non-target lesions, while patients with a partial response (PR; n=9) showed at least a 30% decrease in the sum of all target lesions when taking the baseline as reference and/or persistence of non-target lesions. Patients with a progressive disease (PD; n=14) had at least a 20% increase in the sum of target lesions when taking as reference the smallest sum on study (including baseline), and an absolute increase of sum ≥ 5mm or the appearance of new lesions. Patients who did not present a sufficient decrease or an increase of target lesions were grouped and classified as stable disease (SD; n=1).

We explored in each patient sample ten distinct immune cell types for which features were previously shown to be pivotal in dictating the clinical outcome of melanoma patients using a specific multi-parametric flow cytometry approach ( Supplementary Figure 1A ). On one hand, DC subsets were defined amongst alive CD45⁺Lin^–HLA-DR⁺ cells as CD11c^dimBDCA3⁺ cDC1s, CD11c⁺BDCA1⁺ cDC2s and CD11c^–BDCA2⁺ pDCs. On the other hand, effector cells were depicted amongst alive CD45⁺ cells as follows: CD3^–CD56⁺ for NK cells (NK^bright and NK^dim depending on CD56 expression levels); CD3⁺Vδ2TCR⁺ for γδ2⁺T cells; CD3⁺TCRγδ⁺Vδ2TCR^– for γδ2^–T cells (γδ2⁺T cells appears TCRγδ negative due to competition between the two antibodies targeting the δ chain, Supplementary Figure 1B ); CD3⁺TCRγδ^–TCRVα24-Jα18⁺ for iNKT cells; CD3⁺TCRγδ^–TCRVα24-Jα18^–CD8⁺ for CD8⁺ T cells; and CD3⁺TCRγδ^–TCRVα24-Jα18^–CD8^– for CD4⁺ T cells. We examined cell proportions, basal activation status, ICP and/or NCR expression and functionality for every immune cell subset studied. Memory T-cell differentiation stage and Th profile were also studied for γδ2⁺T, γδ2^–T, CD8⁺ T and CD4⁺ T cells. In addition, soluble factors were explored in cell supernatants after external stimulation (cytokines, chemokines) and in plasma collected from patients (sICP, immune regulators, adhesion molecules, lectin receptors) by ProcartaPlex^® dosages using Luminex.

To investigate the relevance of circulating DC subsets (cDC1s, cDC2s and pDCs) and effector cells (γδ2⁺T, γδ2^–T, iNKT, CD4⁺ T, CD8⁺ T, NK^bright and NK^dim cells) in the response to anti-PD1, we first assessed the frequencies of these immune cells in the blood of melanoma patients before and during the treatment (T0, T3/6, T12, T24). Given the limited number of patients in some groups, we regrouped patients as non-responders (NR; PD + SD; n=15) or responders (R; PR + CR; n=15) throughout the study. Heat map visualization and PCA analysis based on median proportions of circulating immune cells of NR or R patients brought forward different cell frequency patterns and allowed to distinguish the two response groups ( Figures 1A, B ), mostly by cDC1 and CD8⁺ T cell frequencies ( Supplementary Figure 2A ). Radar plot illustrations also highlighted variations of circulating immune cell proportions both between the response groups (inter-group) at different cures (also referred as time points) and within each group (intra-group) during the duration of the treatment ( Figure 1C ; Supplementary Figure 2B ). For circulating DC subsets, the frequency of cDC1s was higher in R when compared to NR patients at T0 and T12, and these levels were maintained throughout the duration of the treatment for responders. Interestingly, in both groups, we noticed a drop in pDC frequency after the first cure, which reached significance only in responders ( Figure 1D ). The cDC1/pDC ratio was also higher in R when compared to NR patients at T12, and we noticed an increase of the cDC2/pDC ratio between T0 and T3/6 in responders ( Figure 1E ), probably due to the decrease of pDCs since cDC2s’ proportions remained unchanged ( Supplementary Figure 2F ). Regarding immune effector cells, the frequency of CD8⁺ T cells as well as the CD8⁺/CD4⁺ T cells ratio increased, while the proportion of NK^bright cells decreased in R when compared to NR patients at T12 ( Figures 1F, G ). The proportion of iNKT cells also decreased after the beginning of the treatment in non-responders (comparison between T0 and T3/6; Figure 1F ). Besides, proportions of circulating CD4⁺ T, γδ2⁺T, γδ2^-T and NK^dim cells remained unchanged both between groups and within each group during the course of the treatment ( Supplementary Figure 2C ). To assess interrelations between immune cell populations, we performed statistical Spearman’s correlations between the frequencies of the immune subpopulations studied, and found no significant differences between NR and R patients ( Supplementary Figure 2D ). Thus, these results indicate that responder melanoma patients had higher frequencies of circulating cDC1s and CD8⁺ T cells before and/or during anti-PD1 treatment when compared to non-responders.

To examine the importance of DC phenotype in the response to anti-PD1, we assessed the basal activation state and ICP expression profile of circulating DC subsets from immunotherapy-treated melanoma patients using flow cytometry ( Supplementary Figure 3A ). Radar plot underlined distinct phenotypic profiles of circulating DC subsets both inter-group ( Supplementary Figure 3B ) and intra-group during the duration of the treatment ( Figure 2A ). In addition, heat map and PCA analysis based on the median proportions of DC subsets expressing activation markers and ICP of patients (NR, R) also highlighted different patterns according to response to treatment. This distinguished the two response groups ( Figures 2B, C ) mostly based on PD-L2 and CD70 expression by DC subsets ( Supplementary Figure 3C ). Overall, we observed tendencies toward increased proportions of PD-L2-expressing DC subsets ( Figure 2D ), together with cDC1s and pDCs expressing CD70 and/or TIM3 in non-responders during the course of anti-PD1 therapy, contrary to responders whose frequencies remained stable ( Figure 2E ; Supplementary Figure 3D ). Such evolution contributed to significantly higher circulating frequencies of PD-L2-expressing cDC1s, cDC2s and pDCs, TIM3-expressing cDC1s and pDCs, and of CD70-expressing cDC1s in NR compared to R patients at T3/6, T12 and/or T24 ( Figure 2E ; Supplementary Figure 3D ). In addition, intra-group comparisons revealed decreased frequencies of ICOS-L⁺ cDC2s after the beginning of the treatment in non-responders at T3/6, as well as of proportions of LAG3- or CD40-expressing pDCs during treatment in responders at T24 compared to baseline ( Supplementary Figure 3E ). Furthermore, we performed Spearman’s correlations to assess the association between PD-L1/-L2 expression and other ICPs expressed by DC subsets in patients (NR, R) before the start of anti-PD1 treatment. The expression of PD-L1 and/or PD-L2 on DC subsets positively correlated with the expression of LAG3, CD70, OX40L and/or TIM3 in NR patients, while only positive correlations between PD-L1⁺ cDC2s and LAG3⁺ or OX40L⁺ cDC2s were preserved in responders before anti-PD1 ( Supplementary Figures 4A–C ). Thus, these results show that non-responders had higher proportions of circulating PD-L2⁺, CD70⁺ and/or TIM3⁺ DC subsets during anti-PD1 treatment than responders.

To investigate the potential differences between NR and R patients to anti-PD1, the activation status, ICP and NCR expression profiles of circulating effector cells from melanoma patients were investigated using flow cytometry ( Supplementary Figure 5A ). PCA analysis based on phenotypic features of effector cells allowed the clustering of patient groups (NR versus R) at different times of the treatment (T0, T3/6, T12, T24), mostly driven by the status of γδ2⁺T, γδ2^-T and NK^dim cells ( Figure 3A ; Supplementary Figure 5B ). Heat maps also illustrated distinct patterns of activation status and NCR expression ( Supplementary Figure 5C ) as well as ICP expression profile ( Figure 3B ) on effector cells in R compared to NR patients before and during anti-PD1 treatment. We first focused on PD1, as this molecule is targeted by the therapy and expression of its ligands (PD-L1, and mostly PD-L2) was perturbed on DC subsets. Before anti-PD1 treatment initiation, higher frequencies of PD1-expressing CD4⁺ T cells (and potentially γδ2⁺T) were found in the blood of R compared to NR patients ( Figure 3C ). During treatment, several effector cell subsets (γδ2⁺T, γδ2^–T and T_conv cells, the latter divided into CD8⁺ or CD4⁺ T cells) from R patients endured a significant decrease in PD1 expression ( Figure 3D ; Supplementary Figure 5D ). When assessing other ICPs, activation and NCR markers on circulating effector cells, our study revealed both inter-group differences according to response to treatment and modulations of patterns during the course of the treatment ( Figure 3E ; Supplementary Figures 5E, F ). Before anti-PD1 treatment, proportions of circulating 41BB-expressing γδ2⁺T and TIGIT-expressing γδ2^–T cells were higher in R compared to NR patients, whereas proportions of GITR⁺ γδ2^–T cells were decreased ( Supplementary Figure 5G ). During treatment, higher proportions of 41BB⁺ T_conv cells, LAG3⁺ γδ2^–T and TIGIT⁺ iNKT cells were observed in R compared to NR patients, while frequencies of GITR-expressing γδ2^–T and iNKT, 41BB-expressing NK^bright and CD27-expressing NK^dim cells were decreased in responder melanoma patients ( Supplementary Figure 5G ). Regarding intra-group comparisons, we observed in NR patients decreased proportions of TIGIT⁺ NK^dim and CD27⁺ NK^bright, while proportions of 41BB⁺ γδ2⁺T and NK^bright cells and CD27⁺ NK^dim cells significantly decreased in responders ( Supplementary Figure 5G ). Correlations of PD1 expression with other ICPs on effector cells before treatment initiation (T0) showed a positive correlation between PD1 and GITR expression on γδ2^–T cells derived from non-responders, while a negative correlation was observed between PD1 and TIGIT expression on circulating CD8⁺ T cells in responders ( Supplementary Figure 5H ). Concerning activation status, the levels of CD40-, CD86- and/or CD69-expressing γδ2⁺T, γδ2^–T and iNKT cells were lower in the blood of R when compared to NR patients before and/or during anti-PD1 treatment ( Supplementary Figure 5I ). Regarding NCR expression, we observed a decreased level of NKG2D on γδ2⁺T and lower proportions of circulating NKp46-expressing NK^bright cells in R compared to NR patients before treatment ( Supplementary Figure 5J ). Furthermore, there were lower levels of NKG2D or NKG2A on γδ2⁺T and NK^bright cells respectively during treatment, and of NKp46-expressing NK^dim cells, whereas NKG2C-expressing iNKT increased in R compared to NR patients ( Supplementary Figures 5J ). Regarding intra-group comparisons, we observed during anti-PD1 treatment a decrease of NKp46-expressing NK^bright cells in NR patients, while there was an increase of NKG2C⁺ γδ2^–T cells specifically in R patients ( Supplementary Figure 5J ). Thus, these data highlight distinct activation, ICP and NCR patterns on circulating effector cells in responder and non-responder melanoma patients during anti-PD1 treatment.

Soluble isoforms of ICP are generated by alternative splicing or cleavage of extracellular parts by metalloproteinases. Such molecules can behave as decoy receptors or adjuvants and potentially interfere with ICB efficacy. In this context, we analyzed 37 soluble factors (especially sICP) in the plasma of melanoma patients before and during the course of anti-PD1 treatment by using Luminex. Heat map visualization based on median levels of soluble factors found on the plasma of NR and R patients did not show major inter-group differences according to the patient response to the treatment ( Supplementary Figure 6A ), except for an increase in soluble CD73 in R compared to NR patients at T24 ( Supplementary Figure 6B ). Besides, intra-group analyses highlighted increases of soluble CD27, PD-L2 and perforin, and a decrease of soluble GITR during anti-PD1 treatment in R patients exclusively ( Supplementary Figure 6B ).

To examine whether T-cell differentiation stage fluctuated depending on the patient’s clinical response to immunotherapy, the frequencies of differentiation stages (N: naive; CM: central memory; EM: effector memory; EMRA: effector memory re-expressing CD45RA) of γδ2⁺T, γδ2^–T, CD8⁺ T and CD4⁺ T cells were evaluated in melanoma patients undergoing anti-PD1 treatment using flow cytometry. Heat map visualization and PCA analysis based on the median frequencies of T-cell differentiation stages derived from the blood of NR and R patients at different time points (T0, T3/6, T12, T24) illustrated distinct patterns of differentiation stage allowing the clustering of both response groups, mostly driven by differences on γδ2⁺T cells ( Figure 4A ; Supplementary Figure 7 ). Before and during anti-PD1 treatment, frequencies of circulating γδ2⁺T cells in the CM stage were higher in R compared to NR patients, while there were fewer in the EMRA stage ( Figure 4B ). Such picture was also depicted for CD8⁺ and CD4⁺ T cells, even though it was not significant.

To decipher the functionality of DC subsets in melanoma patients following anti-PD1 treatment, cytokine production (IL-12p70, IFN-λ1, IFN-α and/or TNF-α) by DC subsets after TLR stimulation was assessed in NR and R patients at several time points (T0, T3/6, T12, T24) by intracellular cytokine staining using multi-parametric flow cytometry ( Supplementary Figure 8A ). After mix stimulation (polyI:C, R848 and CpGA), higher proportions of TNF-α⁺ pDCs were observed in R compared to NR patients at T24 upon treatment ( Figure 5A ). To assess the link between PD-L1/-L2 expression by DC subsets and their functionality before anti-PD1 treatment, we performed Spearman’s correlations and found that PD-L2 expression negatively correlated with TNF-α production by circulating cDC2s in non-responders ( Figure 5B ). We also explored a large panel of cytokine/chemokine secretion by DCs after TLR stimulation. This was assessed by ProcartaPlex dosages of culture supernatants using Luminex and the median levels of each factor were illustrated in the heat map in Figure 5C . Without TLR stimulation, higher levels of secreted MIG and MCP-1 were found in R when compared to NR patients before or during anti-PD1 treatment ( Figure 5D ). After single or mix TLR stimulation, we observed higher levels of IL-12p70 and lower levels of TGF-β1 in R patients before and/or during immunotherapy ( Figure 5E ; Supplementary Figure 8B ). Furthermore, after single TLR stimulation, inter-group comparisons revealed higher levels of MIP-1β and RANTES at T0, as well as higher levels of TNF-α and MIP-1α during anti-PD1 treatment in responders when compared to non-responders ( Figure 5F ). Furthermore, higher levels of IL-1β were specifically found in non-responders during immunotherapy ( Supplementary Figure 8C ). Thus, these results demonstrate that upon DC-specific stimulation higher levels of secreted IL-12p70, TNF-α and MIP-1β and lower levels of TGF-β1 were found in melanoma patients responding to anti-PD1 treatment when compared to non-responders.

To explore the function of circulating effector cells in melanoma patients following anti-PD1 treatment, cytokine production by γδ2⁺T, γδ2^–T, iNKT, CD8⁺ T, CD4⁺ T, NK^bright and NK^dim cells was assessed by intracellular cytokine staining after PBMC stimulation by PMA/Iono using flow cytometry ( Supplementary Figure 9A ). Heat map illustration based on cytokine production by effector cells upon PMA/Iono stimulation highlighted differences between patient groups (NR, R) before and/or during treatment ( Figure 6A ). Without any external stimulation, IL-17- and TNF-α-producing γδ2⁺T cells were more abundant in NR compared to R patients before and/or during anti-PD1 treatment ( Figure 6B ), whereas proportions of IL-13⁺ γδ2⁺T cells decreased in responders during immunotherapy ( Supplementary Figure 9B ). After PMA/Iono stimulation, higher levels of TNF-α- or IFN-γ-producing iNKT were found in NR compared to R patients before and during anti-PD1 treatment, while higher frequencies of TNF-α-secreting CD8⁺ T and NK^dim cells were found in responders during immunotherapy ( Figure 6C ). Higher frequencies of IL-13-producing NK^bright cells were also found in R when compared to NR patients before treatment ( Figure 6D ). In addition, proportions of TNF-α-producing iNKT cells, IFN-γ-producing γδ2^–T cells and IL-13⁺ NK^bright cells decreased in responders during immunotherapy ( Figures 6C, D ). To study the link between PD1 expression and T-cell functionality before treatment, we performed Spearman’s correlations and found a positive correlation between PD1 expression and IFN-γ production by CD4⁺ T cells in responders ( Figure 6E ).

We further explored Th profiles of T_conv and γδT cells according to patient’s response to anti-PD1 treatment through the assessment of transcription factors specific for functional orientation. Tbet, GATA3, RORγt, AhR and FoxP3 (corresponding to Th1, Th2, Th17, Th22 and Treg profiles, respectively) were studied within γδ2⁺T, γδ2^–T, CD8⁺ T and CD4⁺ T cells before and during the course of the treatment following intra-nuclear staining and flow cytometry analysis ( Supplementary Figure 10A ). Even though Tbet-expressing γδ2⁺T cells were elevated in NR compared to R patients before anti-PD1, intra-group comparisons showed tendencies to decrease in this population during treatment in non-responders and to increase in responders, which was also the case for Tbet-expressing CD8⁺ T and CD4⁺ T cells ( Figure 7A ; Supplementary Figure 10B ). Furthermore, non-responders displayed elevated proportions of γδ2^–T cells with a Th17 or Th22 profile compared to responders before anti-PD1 treatment, while γδ2^–T cells with a T_reg profile decreased in non-responders after the beginning of the treatment ( Figure 7A ; Supplementary figure 10B ). In addition, by comparing transcription factor ratios, distinct Th profiles were found between NR and R patients before and/or during anti-PD1 treatment ( Figure 7B ). Before treatment, we noticed an increase of the ratios T_reg/Th17 γδ2⁺T cells, Th2/Th22 and Th2/Th17 γδ2^–T cells, and a decrease in T_reg/Th22 CD8⁺ T cells in R when compared to NR patients. During the treatment, responders displayed an increase in the ratio Th2/Th22 γδ2⁺T and CD4⁺ T cells compared to non-responders ( Figure 7B ). Thus, these data pinpoint distinct Th profiles of conventional and γδ T cells in R when compared to NR melanoma patients before and/or during immunotherapy.

Lastly, to further decipher the combined functionality of circulating effector cells in melanoma patients following anti-PD1 treatment, cytokine secretion profiles were analyzed by ProcartaPlex dosages of PBMC supernatants after specific stimulation of each subset using alone or combined anti-CD3/CD28 antibodies, HMB-PP, IL-12/IL-18 and αGalCer. PCA analysis based on the median levels of secreted molecules in PBMC supernatants allowed the clustering of patients depending on their clinical response, which seemed mostly driven by TGF-β1 secretion ( Supplementary Figure 10C ). Heat map visualizations also illustrated distinct patterns of cytokine/chemokine secretion by NR and R patients upon effector-specific stimulation ( Figure 7C ). Prior to stimulation, TNF-α levels were higher in responders before and during anti-PD1 treatment ( Figure 7D ). Notably, following effector-specific stimulation, TNF-α and IL-5 secretions were higher in R compared to NR patients during anti-PD1 treatment, while TGF-β1 levels were higher in non-responders before and/or during immunotherapy ( Figure 7D ; Supplementary Figure 10D ). Altogether, these results highlighted that anti-PD1 treatment triggers deep changes in the functional orientation of effector cells, and mostly T_conv and γδ T cells.