Monitoring of kinetics and exhaustion markers of circulating CAR-T cells as early predictive factors in patients with B-cell malignancies

Monitoring of kinetics and exhaustion markers of circulating CAR-T cells as early predictive factors in patients with B-cell malignancies Clara Beatriz Garcı ´a-Caldero ´n 1 † , Bele ´n Sierro-Martı ´nez 1 † , Estefanı ´aGarcı ´a-Guerrero 1 *, Luzalba Sanoja-Flores 1 , Raquel Muñoz-Garcı ´a 2 , Victoria Ruiz-Maldonado 1 , Marı ´a Reyes Jimenez-Leon 1 , Javier Delgado-Serrano 1 , A ´ gueda Molinos-Quintana 1 , Beatriz Guijarro-Albaladejo 1 , Inmaculada Carrasco-Brocal 1 , Jose ´-Manuel Lucena 2 , Jose ´-Rau ´lGarcı ´a-Lozano 2 , Cristina Bla ´zquez-Goñi 1 , Juan Luis Reguera-Ortega 1 , Marı ´a-Francisca Gonza ´lez-Escribano 2 , Marta Reinoso-Segura 1 , Javier Briones 3 , Jose ´Antonio Pe ´rez-Simo ´n 1 and Teresa Caballero-Vela ´zquez 1 1 Servicio de Hematologı ´a, Hospital Universitario Virgen del Rocı ´o, Instituto de Biomedicina de Sevilla, (IBIS/CSIC), Universidad de Sevilla, Sevilla, Spain, 2 Servicio de Inmunologı ´a, Instituto de Biomedicina de Sevilla (IBiS), Hospital Universitario Virgen del Rocı ´o, Centro Superior de Investigaciones Cientı ´ficas (CSIC), Universidad de Sevilla, Sevilla, Spain, 3 Hematology Service, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain Purpose: CAR-T cell therapy has proven to be a disruptive treatment in the hematology field, however, less than 50% of patients maintain long-term response and early predictors of outcome are still inconsistently defined. Here, we aimed to optimize the detection of CD19 CAR-T cells in blood and to identify phenotypic features as early biomarkers associated with toxicity and outcomes. Experimental design: In this study, monitoring by flow cytometry and digital PCR (dPCR), and immunophenotypic characterization of circulating CAR-T cells from 48 patients treated with Tisa-cel or Axi-cel was performed. Results: Validation of the flow cytometry reagent for the detection of CAR-T cells in blood revealed CD19 protein conjugated with streptavidin as the optimal detection method. Kinetics of CAR-T cell expansion in blood confirmed median day of peak expansion at seven days post-infusion by both flow cytometry and digital PCR. Circulating CAR-T cells showed an activated, proliferative, and exhausted phenotype at the time of peak expansion. Patients with increased Frontiers in Immunology frontiersin.org01 OPEN ACCESS EDITED BY Stephen Goldrick, University College London, United Kingdom REVIEWED BY Enrico Maffini, University of Bologna, Italy Li Xiaoyang, Shanghai Jiao Tong University, China *CORRESPONDENCE Estefanı ´a Garcı ´a-Guerrero [email protected] † These authors contributed equally to this work and share first authorship SPECIALTY SECTION This article was submitted to Cancer Immunity and Immunotherapy, a section of the journal Frontiers in Immunology RECEIVED 27 January 2023 ACCEPTED 31 March 2023 PUBLISHED 14 April 2023 CITATION Garcı ´a-Caldero ´n CB, Sierro-Martı ´nez B, Garcı ´a-Guerrero E, Sanoja-Flores L, Muñoz-Garcı ´aR,Ruiz-Maldonado V, Jimenez-Leon MR, Delgado-Serrano J, Molinos-Quintana A ´ , Guijarro-Albaladejo B, Carrasco-Brocal I, Lucena J-M, Garcı ´a-Lozano J-R, Bla ´zquez-Goñi C, Reguera-Ortega JL, Gonza ´lez-Escribano M-F, Reinoso-Segura M, Briones J, Pe ´rez-Simo ´n JA and Caballero-Vela ´zquez T (2023) Monitoring of kinetics and exhaustion markers of circulating CAR-T cells as early predictive factors in patients with B-cell malignancies. Front. Immunol. 14:1152498. doi: 10.3389/fimmu.2023.1152498 TYPE Original Research PUBLISHED 14 April 2023 DOI 10.3389/fimmu.2023.1152498 expansion showed more severe CRS and ICANs. Immunophenotypic characterization of CAR-T cells at the peak expansion identified the increased expression of co-inhibitory molecules PD1 and LAG3 and reduced levels of the cytotoxicity marker CD107a as predictors of a better long-term disease control. Conclusions: These data show the importance of CAR-T cells in vivo monitoring and identify the expression of PD1LAG3 and CD107a as early biomarkers of longterm disease control after CAR-T cell therapy. KEYWORDS CAR-T, flow cytometry, dPCR (digital PCR), monitoring, biomarkers, B-ALL, Lymphoma 1 Introduction Adoptive immunotherapy using engineered chimeric antigen receptor (CAR)-T cells has stood out as a disruptive treatment in relapsed/refractory B cell leukemia and lymphoma. Despite the outstanding results of CAR-T therapy in patients with B-cell malignancies, less than 50% of patients experience long-term disease response (1). Therefore, intensive efforts are being made to improve the outcomes of these patients and to identify biomarkers capable to early predict patient outcomes (2). Among other factors, the efficacy of the treatment has been related to specific features of the CAR-T cell product as well as CAR-T expansion and persistence after infusion. Accordingly, some trials have shown a correlation between in vivo expansion and persistence of CAR-T cells in peripheral blood of pediatric B-ALL patients with response, event free-survival and the incidence of cytokine release syndrome (CRS) or immune effector cell-associated neurotoxicity syndrome (ICANS) (3–7). However, some other trials have shown inconsistent or weak correlation between CAR-T expansion and clinical outcomes (8–10). Therefore, this topic is still an area of debate and further studies are required to gain more insight into the role of CAR-T monitoring as a predictive biomarker. Furthermore, the analysis of leukapheresis products has shown that an elevated frequency of CD27+CD45RO-CD8+ T cells and a higher CD4+ to CD8+ T cells ratio correlated with a higher complete response (CR) (6,11). In addition, RNA-seq analyses of CD19 CAR-T cell products has revealed differences in gene expression in T cell differentiation and metabolism from responder and non-responder patients (6,12). Increased expression of the co-inhibitory receptors PD1+, TIM3+ and LAG3+ in the leukapheresis and the infusion product has also been associated with treatment failure in several studies (6,12,13). On the other side, scanty information is available regarding the functional and/or immunophenotypic characteristics of the circulating CAR-T cells after infusion. Recently,Good,etal.,havediscovereda correlation between increased levels of CAR-Treg population at day 7 post-infusion and progression at 6 months in lymphoma patients treated with Axicabtagene ciloleucel. Besides, this population has also been associated with reduced neurotoxicity (8). Nevertheless, the information according to immunophenotypic characteristics of CAR- T cells post-infusion is limited and whether these features might have any predictive value in the long term has not been clearly elucidated. The persistence and monitoring of circulating CAR-T cells in peripheral blood of patients can be followed by molecular techniques such as quantitative or digital polymerase chain reaction (PCR), being the digital PCR more sensitive and precise to detect CAR-T cells (14–16), and/or by flow cytometry. The identification and characterization of CAR-T cells by flow cytometry can be very challenging since its detection may depend on the cell product. Several reagents are commercially available for the detection of CAR-T cells (11,17). Also, indirect methods have also been tested for the detection of CAR-T cells such as Protein L that binds to kappa light chain of immunoglobulins (18). Additionally, some CAR constructs have incorporated detection markers to facilitate in vivo monitoring (2). Thus, validation of all these detection methods should be done for each CAR-T cell product prior in vivo monitoring. In this study, we propose the validation of a sensitive and specific method for the identification of CD19 CAR-T cells by flow cytometry. Samples from forty-eight patients receiving commercial CD19 CAR-T cells for B-ALL or DLBCL treatment were analyzed by flow cytometry and by digital PCR up to 20 days post-infusion. Finally, correlation studies between the cell kinetics and immunophenotype characterization of circulating CAR-T cells post-infusion were performed in an attempt to search for early biomarkers associated with a better response, thus allowing to identify those groups of patients with worse prognosis at very early time points after infusion, when therapeutic maneuvers might have a higher probability of success before relapse does occur. 2 Materials and methods 2.1 Ex vivo PBMCs isolation and manufacturing of academic CD19 CAR-T cells Peripheral blood mononuclear cells (PBMCs) were obtained from buffy coats of healthy donors kindly donated by the Regional Garcı ´a-Caldero ´n et al. 10.3389/fimmu.2023.1152498 Frontiers in Immunology frontiersin.org02 Centre for Blood Transfusions at Virgen del Rocı o University Hospital, Seville (Spain) after written informed consent. For academic CD19 CAR-T cell generation, CD4+ and CD8+ cells were isolated from human PBMCs, activated with anti-CD3/anti- CD28 (Gibco, Ref: 11131D) and transduced with a lentiviral vector containing CD19 CAR-T cell construct co-expressing EGFRt as a detection marker. Transduced cells were expanded with IL-2 (Miltenyi, Ref: 130-097-743) and irradiated feeder cells. 2.2 Patient samples Samples were obtained from patients with relapsed/refractory acute lymphoblastic leukemia or lymphoma treated with CD19 CAR-T therapy (Tisagenlecleucel or Axicabtagene ciloleucel) at different time points: leukapheresis, days 5, 7, 11, 14 and 20 post CAR-T infusion. 2.3 Flow cytometry Markers expression was evaluated by multicolor flow cytometry. For the validation of CAR-T detection of academic CD19 CAR-T, cells were thawed and washed with phosphate buffer saline (PBS) (Gibco, Ref: 18912-014). A quantity of 500,000 cells were used for each staining. To differentiate live/dead cells 7- AAD (BD Biosciences, Ref: 559925) was added to each sample. Samples were stained according to the antibodies and dyes shown in Supplementary Table 1. For the monitoring of commercial CD19 CAR-T cells postinfusion, a quantity of 200,000 PBMCs was added for staining according to blood count. Samples were stained according to the antibodies and dyes shown in Supplementary Table 1 and incubated 20 min at room temperature in the dark. Red blood cells were lysed using BD FACS Lysing solution (BD Biosciences, Ref: 349202) following mixing and 10 min incubation in the dark. Lysed cells were centrifuged at 2000 rpm 5 min and resuspended in FACs Flow solution (BD Biosciences, Ref: 12756528). For the immunophenotypic characterization of the CAR-T, a bulk lysis of the sample was carried out according to the Euroflow protocol. Whole blood sample was lysed using ammonium chloride 1x (Cytognos, Ref: CYT-BCP-ALL-MRD-BL) and incubated in rotation for 15 min. Then, samples were washed twice with washing solution (PBS+0.2%BSA+ 1mM EDTA), centrifuged 10 min at 2000 rpm and resuspended in 100 uL FACs Flow solution (BD Biosciences, Ref: 12756528). Finally, cells were stained according to the antibodies and dyes shown in Supplementary Table 1 and incubated 20 min at room temperature in the dark. 2.4 Digital PCR To isolate peripheral blood mononuclear cells (PBMCs), Ficoll gradient with PD Vacutainer ®CPT™(Becton Dickinson, NY, USA) was used. To extract genomic DNA from PBMCs, the QIAmp DNA Mini Kit (Qiagen, Hilden, Germany) was used according to manufacturer’s recommendations. The purity and concentration of the DNA samples were measured using a Q3000 UV Spectrophotometer (Quawell, San Jose, CA, USA). Only samples with concentrations ≥20 ng/µl and an A260/280 absorbance ratio >1.8 were included in the study. For Digital PCR (dPCR) reactions, the QuantStudio™Absolute Q™Digital PCR System (Applied Biosystems, CA, USA) was employed. To quantify CAR-T, primers and TaqMan probes designed to amplify PCR fragments in the FMC63 region of the CAR-T construct were used. The primers used were 5’-TGA AAC TGC AGG AGT CAG GA-3’(forward) and 5’-CTG AGA CAG TGC ATG TGA CG-3’(reverse), and the FAM TaqMan MGB probe was 5’-CTG GCC TGG TGG CGC CCT CA-3’. TaqMan® Copy Number Reference Assay RNase P labelled with VIC (Applied Biosystems) was used as gen reference (REF). Each sample was simultaneously amplified to CAR-T and REF sequences using differently labelled probes. A 10 mLfinal volume mix reaction consisted of 80 ng of DNA, 2 mL of Absolute Q™DNA dPCR Mix (5X), 250 nM of FMC63-probe, 900 nM of each FMC63- primer, and 0.5 mL primer-probe set for RNase P (20X). Cycle conditions were 96°C for 10 min, followed by 40 cycles at 96°C for 5 sec and 60°C for 15 sec. 2.5 Data analysis No normalization was implemented in the flow cytometry data analysis. All raw data was analyzed with either FlowJo™v10.7 Software (BD Life Sciences) or Infinicyt 2.0 (Cytognos) software. Statistical analyses were performed with GraphPad Prism v9 and SPSS 26. We applied unpaired two-sided Mann–Whitney U-tests to assess statistical significance between two groups of unpaired samples. We applied paired two-sided Wilcoxcon-matched-paired test to asses statistical significance between two groups of paired samples. AUC was calculated on days 5, 7, 11, 14 and 20 and peak CAR-T expansion was calculated only when data was obtained from at least three timepoints. For dPCR data analysis, “QuantStudio Absolute Q Digital PCR Software 6”(Applied Biosystems) with automatic Poisson correction was used. Normalization across samples was implemented by assuming there were two copies of the reference RNase P gen per cell. Results were expressed in CAR- T copies per cell. OS and EFS were estimated with the Kaplan-Meier method using survival in GraphPad Prism v9. EFS was computed from the date of infusion to the date of event, with both progression and death scored as event. To assess correlation, we calculated Spearman’s rank correlation coefficient. 3 Results 3.1 Validation of flow cytometric CD19 CAR staining with different reagents For the validation of CAR T- cell monitoring, distinct CAR T- cell detection methods using different reagents were performed. The Garcı ´a-Caldero ´n et al. 10.3389/fimmu.2023.1152498 Frontiers in Immunology frontiersin.org03 binding mechanisms of flow cytometry detection reagents is shown in Figure 1A. Firstly, since there was no limitation with the available sample volume, we evaluated the identification of our academic CD19 CAR-T cell product to determine optimal detection reagent (Figure 1A). We were able to compare four different reagents: CD19 CAR detection reagent from Miltenyi, CD19 protein and Protein L from Acrobiosystems and Anti-EGFR from Biolegend. T-cells from four healthy donors were transduced and academic CD19 CAR-T cell products were analyzed by flow cytometry. In Figure 1B, the percentage of detection is depicted for each reagent. For our academic CD19 CAR-T cell, the maximum percentage of detection is achieved with the anti-EGFR antibody with a similar detection level of the CD19 CAR detection reagent both for CD4+ and CD8+ cells (p=0.1142, p=0.6857). Besides, the detection with anti-EGFR allowed the highest resolution in CD4+ population (Figure 1C) (p=0.0283). In contrast, Protein L and CD19 protein from Acrobiosystems did not allow an optimal detection of our academic CD19 CAR-T cell (Figure 1B). Finally, for the evaluation A B DE C FIGURE 1 Validation of the detection reagent for academic and commercial CD19 CAR-T cells by flow cytometry. (A) Illustration of the different detection methods available for the identification of CAR-T cells by flow cytometry. (B) Percentage of detection of academic CD19 CAR-T cells in CD4 and CD8 compartments with different reagents. (C) Stain index of detection of academic CD19 CAR-T cells in CD4 and CD8 compartments with different reagents. (D) Example of flow cytometry plot of the identification of commercial CD19 CAR-T cells with two recombinant CD19 proteins. (E) Percentage (left) and stain index (right) of the detection of commercial CD19 CAR-T cells with two different reagents. Depicted are median and individual values of four independent experiments. P-values between the indicated groups were calculated using unpaired Mann-Whitney U-t tests. ns: non-significant, *p<0.05. Garcı ´a-Caldero ´n et al. 10.3389/fimmu.2023.1152498 Frontiers in Immunology frontiersin.org04 of unspecific binding, untransduced T-cells (UTD) from the same donors were stained with the respective detection reagents to measure background staining in negative controls that presented no signal except for the CD8+ UTD stained with Protein L-FITC that showed some background signal (Supplementary Figure 1). Once we validated the identification of our academic CD19 CAR-T cell, samples from three different patients infused with either Tisagenlecleucel or Axicabtagene ciloleucel were analyzed by flow cytometry using the commercially available CD19 detection reagents from Acrobiosystems and from Miltenyi. In Figure 1D, representative plot of one patient stained with both antibodies is shown. Leftover samples from randomly selected healthy donors who had no history of CAR-T cell treatment served as negative controls. Assay specificity was measured with no background staining observed in any of the healthy donors (Supplementary Figure 2). Similar results according to CAR+ percentage of detection were obtained with both reagents. However, the CAR detection reagent showed the highest resolution in contrast to CD19 protein both in CD4+ and CD8+ cells (stain index of 36.82 vs. 6.18, p=0.0286 and stain index of 16.49 vs. 3.42, p=0.0286, respectively) (Figure 1E). Accordingly, we validated CD19 CAR detection reagent as the optimal reagent for the following monitoring of commercial CAR-T cell expansion in vivo. 3.2 CAR-T cell kinetics and phenotype characterization Monitoring of peripheral blood commercial CAR-T cells and non-modified T cells post-infusion was performed in forty-eight patients with lymphoma and leukemia. The characteristics of patients included in this study are described in Table 1.The overall response rate (ORR) at 1-month post-infusion was 68.8% (Table 1). In Figure 2A, representative flow cytometry of the expansion of CAR-T cells in blood from patient 32 at different time points post-infusion is illustrated. Cellular kinetics of all patients show the same pattern of expansion with an exponential increase until days 7-11 followed by a rapid loss of CAR-T cell absolute count until day 20 post-infusion. The median day of peak expansion of CAR-T cells was day +7 (range of 5-14 days) after infusion, reaching a median of 112.2 CAR+ cells/uL (range of 0.7- 1019 CAR+ cells/uL) in blood (Figure 2B). CAR-T cell expansion of patients with lymphoma, this is, excluding patients with LLA, is also shown (Supplementary Figure 3A). The median percentage of CAR + cells within the CD3+ compartment in the peak expansion was 23.9% (n=48, range 0.3 to 82) (Supplementary Figure 3C). CAR-T cell expansion was also monitored by digital PCR. Similarly to flow cytometry data, kinetics of all patients shows the same pattern of expansion with an exponential increase until days 7-11 followed by a rapid loss of CAR-T cell absolute count until day 20 post-infusion. The median day of peak expansion of CAR-T cells was also day +7 (range of 5-14 days) after infusion, reaching a median of 0.245 CAR copies/cell (range of 0.03-3.23 CAR copies/cell) in blood (Figure 2C and Supplementary Figure 3B). Significant correlation was observed between flow cytometry and dPCR monitoring of CAR-T cells in blood (Spearman’s rho=0.61, p<0.0001) (Figure 2D). CAR-T cell expansion by flow cytometry and dPCR was compared at the different time points post-infusion with similar kinetics observed (Figure 2E). The expansion dynamics by flow cytometry and dPCR of patients infused with either Tisagenlecleucel (Tisa-cel) or Axicabtagene ciloleucel (Axi-cel) is depicted in Supplementary Figure 4.Nosignificant differences were obtained by flow cytometry between both products in median peak expansion (p=0.4684) or median day of expansion (p=0.33) (Supplementary Figure 4E). However, significant increased CAR copies/cell at peak expansion were observed in patients infused with Axi-cel (p=0.0006) (Supplementary Figure 4F). Immunophenotyping of CAR+ cells and non-modified circulating T cells was performed at peak of expansion in each patient. The CD4/CD8 ratio was analyzed both in the CAR+ population and the circulating T-cells at the day of peak expansion. An increase in CD8+ population can be observed in the CAR+ population with a median CD4/CD8 ratio of 0.5 (Figure 3A). T cell subpopulations were identified as follows: naïve (CD62L+/CD45RO-/CD45RA+), central memory (CD62L +/CD45RO+/CD45RA), peripheral memory (CD62L-/CD45RO +/CD45RA-), effector (CD62L-/CD45RO-/CD45RA+). No significant differences were observed between CAR+ and nonmodified circulating T-cells at the time of expansion in the CD4+ compartment. Among the CD8+ T cells, there was a decrease of effector cells in the CAR+ compared to the non-modified circulating T-cells accompanied by a trend towards an increase in memory phenotype in CAR+ cells (p=0.0056 and p=0.0533, respectively) (Figure 3B). Besides, CAR+ cells showed a significant increase in the expression of the activation marker CD69 compared to non-modified circulating T-cells both in CD4 + and CD8+ cells (p=0.0004, p=0.0008) (Figure 3Ci). The proliferation marker Ki67 and the cytotoxic marker CD107a were also significantly enhanced in CAR+ cells at the time of expansion in both CD4+ and CD8+ cells (p<0.01) (Figures 3Cii,iii). Regarding the exhaustion markers, CAR+ cells showed significant increased expression of the markers FasL, PD1+/TIM3+ and PD1+/LAG3+ compared to non-modified circulating T-cells both in CD4+ and CD8+ cells (p<0.05) (Figures 3Civ,v,vi). All these analyses were also performed excluding patients with B-ALL and we confirmed the same results (Supplementary Figure 5). All together, these data suggest a more activated, proliferative, cytotoxic, and exhausted phenotype of CAR+ cells at the time of peak expansion compared to non-modified circulating T-cells. Immunophenotyping of CAR+ cells and non-modified circulating T cells was also compared between patients receiving Tisa-cel or Axi-cel products. Median CD4/CD8 ratio of CAR+ cells showed no significant differences between the two different products (median 0.4121 vs. 0.6100, respectively, p=0.7802). However, non-modified T cells from patients receiving Tisa-cel showed a tendency to an increased median CD4/CD8 ratio compared to patients receiving Axi-cel (median 0.9903 vs. 0.3375, respectively, p=0.1333) (Supplementary Figure 6A). T cell subsets were also analyzed considering both products. Non-modified T cells from patients infused with Tisa-cel showed a significantly increased proportion of effector CD4+ cells (p=0.0343). Besides, CD8+CAR+ cells from patients receiving Axi-cel showed a trend towards a Garcı ´a-Caldero ´n et al. 10.3389/fimmu.2023.1152498 Frontiers in Immunology frontiersin.org05 reduced proportion of a central memory phenotype as compared to patients infused with Tisa-cel (p=0.0789) (Supplementary Figure 6B). Regarding the activation status, no differences were observed in the expression of CD69 between patients receiving Tisa-cel or Axi-cel products either in CAR+ or non-modified T cells CD4+ or CD8+ (Supplementary Figure 6Ci). A trend towards an increased Ki67 expression was observed in CAR+ cells of patients infused with Axi-cel products both in CD4+ and CD8+ compartments (p=0.0789) (Supplementary Figure 6Cii). No significant differences were observed between the different products regarding CD107a expression (Supplementary Figure 6Ciii). As far as the expression of co-inhibitory markers is concerned, we observed a trend towards an increased expression of FASL in CAR+ and non-modified T cells of patients infused with Axi-cel compared to Tisa-cel products both in CD4+ and CD8+ compartments (p=0.0263, p=0.1895, p=0.1128 and p=0.0653, respectively) (Supplementary Figure 6Civ). In line with this, a significantly increased expression of PD1+TIM3+ and PD1 +LAG3+ was observed in CAR+ cells of patients infused with Axi-cel in both CD4+ and CD8+ cells (p=0.0279, p=0.0010, p=0.0220 and p=0.0789, respectively) (Supplementary Figures 6Cv,6Cvi). In summary, CAR+ cells from patients infused with Axi-cel products showed an increased CD4+ effector non-modified T cells, reduced CAR+CD8+ central memory, and increased proliferation and exhaustion at the time of peak expansion compared to patients receiving Tisa-cel products. CD4/CD8 ratio was also compared between leukapheresis samples and non-modified T or CAR-T cells at the time of peak expansion. A tendency to a progressive reduction in CD4/CD8 ratio was observed from leukapheresis to CAR-T cells at the time of peak expansion (median 0.95 vs. 0.42) (Supplementary Figure 7A). T cell memory subsets were analyzed in CD4+ and CD8+ compartments comparing leukapheresis samples with non-modified T and CAR-T cells at peak expansion. A significant increase in effector memory and a reduction in naïve cells was observed both in CD4+ nonmodified and CAR-T cells (p=0.0005, p=0.0047, p=0.0004, p=0.0013, respectively) (Supplementary Figure 7B). A tendency to reduced proportion in central memory cells was also observed in both non-modified T and CAR-T cells compared to leukapheresis in CD4+ compartment (p=0.08 and p=0.07, respectively) (Supplementary Figure 7B). CD8+ CAR-T cells showed a significant reduction in naïve and effector cells and an increase in effector memory cells compared to leukapheresis (p=0.05, p=0.0446 and p=0.0046, respectively) (Supplementary Figure 7B). TCR repertoire was also evaluated in 10 patients to study clonality of CAR+ cells and non-transduced circulating T-cells (Supplementary Figure 8). Of these 10 patients, three of them relapsed (patients 30, 33 and 35). Focusing on patient 30, relapse occurred one month after CAR-T cell infusion. In this patient, the clones with the higher frequency in CAR+ cells compartment were Vb7.1, Vb17 and Vb21.3. In non-transduced circulating T-cells compartment, the clones observed with a higher frequency were Vb3, Vb13.6 and Vb8. A higher number of patients and longer follow up are required to evaluate the impact of TCR clonality after CAR-T cell infusion. TABLE 1 Characteristics of the patients included in the study. Patients (n) 48 PMBCL 8 DLBCL 26 FL 1 tFL 3 B-ALL 10 Median age (range) 51.5 (5-73) Sex (%) Female 20 (41.7) Male 28 (58.3) Product (%) Tisagenlecleucel 18 (37.5) Axicabtagene-ciloleucel 30 (62.5) Previous lines (range) 3 (2-6) Bridging therapy (%) YES 42 (87.5) RT 2 CT 35 Inotuzumab 4 Steroids 1 NO 6 (12.5) CRS grade (%) 97.9 01 124 221 32 ICANs grade (%) 45.8 026 112 24 34 42 Response D30 (%) 68.8 CR 19 PR 14 PD 9 SD 5 PMBCL (primary mediastinal B-cell lymphoma), DLBCL (diffuse large B-cell lymphoma), FL (follicular lymphoma), tFL (transformed follicular lymphoma), B-ALL (B-cell acute lymphoblastic leukemia), RT (radiotherapy), CT (chemotherapy), CRS (cytokine release syndrome), ICANs (immune effector cell-associated neurotoxicity syndrome), CR (complete response), PR (partial response), PD (progressive disease), SD (stable disease). Garcı ´a-Caldero ´n et al. 10.3389/fimmu.2023.1152498 Frontiers in Immunology frontiersin.org06 3.3 Correlation of CAR-T cell expansion in vivo with toxicity and event-free survival Summary outcomes of each patient during the months postinfusion is depicted in Figure 4 including toxicity, follow up, B-cell aplasia, relapse or death. Median peak expansion of patients was correlated with the grade of CRS or ICANs. Patients who developed grades 2-3 CRS showed a significantly increased peak of CAR-T cell absolute count in vivo (p=0.0261) as compared to those who did not develop it (Figure 5A). In addition, significant increased CAR copies/cell at the peak expansion were observed in patients with CRS grade 2-3 compared to patients with less severe CRS (p=0.0224) (Figure 5A). No significant differences were observed regarding the area under the curve (AUC) by flow cytometry between patients with CRS grades 2-3 compared to those with grades 0-1 (p=0.2758) but increased AUC by dPCR was observed in patients with grades 2-3 CRS (p=0.0299) (Supplementary Figure 8A). Besides, correlation A B DE C FIGURE 2 Commercial CD19 CAR-T cell expansion in the blood of patients post-infusion. (A) Flow cytometry plot of the expansion of CAR-T cells in the blood of patient 32 at different time points post-infusion. (B) Absolute count of CAR-T cells/uL by flow cytometry in the blood of patients with B-ALL and lymphoma. (C) CAR copies/cell measured by digital PCR in the blood of patients with B-ALL and lymphoma. (D) Spearman correlation in peripheral CAR-T cell expansion between flow cytometry and dPCR assays (n=48 patients and 185 observations). (E) Normalized peripheral CAR-T cell expansion by flow cytometry and dPCR assays (n=48 patients). Correlation was calculated using non-parametric Spearmans rank correlation coefficient. Garcı ´a-Caldero ´n et al. 10.3389/fimmu.2023.1152498 Frontiers in Immunology frontiersin.org07 between CAR-T cell expansion by flow cytometry and the appearance of ICANs was also performed but no significant differences were observed in median peak expansion (p=0.1787) (Figure 5B) or AUC (p=0.4212) (Supplementary Figure 8B). However, a trend towards an increased peak CAR copies/cell (p=0.08) and a significantly higher AUC by dPCR (p=0.0224) were observed in patients with some grade of ICANs (Figure 5B and Supplementary Figure 9B). Same analyses were performed including only those patients with lymphoma, but no significant differences were observed (Figures 5C,D,andSupplementary Figures 9C,D), although a trend towards an increased expansion in patients suffering from ICANs was observed (p=0.09). Correlation studies between CAR-T cell expansion in vivo and efficacy of the therapy were performed. However, no significant impact of CAR-T cells/uL on event-free survival was observed, neither for the peak expansion (p=0.5579, Figure 5E) or the area under the curve (p=0.2236, Supplementary Figure 9E). No significant correlation was observed between event-free survival and CAR copies/cell at the time of peak expansion or AUC (Figure 5F and Supplementary Figure 9F). Same analyses were performed including only those patients with lymphoma, but no significant differences were observed (Figures 5G,Hand Supplementary Figures 9G,H). Toxicity and efficacy of Tisa-cel and Axi-cel products were also compared. No significant differences were observed in the incidence of grades 2-3 CRS comparing patients infused with Tisa-cel (38.89%) or Axi-cel (53.33%) (p=0.3831) (Supplementary Figure 10A). However, an increased incidence of ICANs was observed in patients infused with Axi-cel (63.33%) as compared to patients receiving Tisa-cel (16.67%) (p=0.0025) AB C FIGURE 3 Immunophenotype characterization of non-modified T cells and CAR-T cells at the time of peak expansion in blood. (A) CD4/CD8 ratio of nonmodified T cells and CAR-T cells at the time of peak expansion. (B) Memory subsets of CD4+ and CD8+ compartments comparing non-modified and CAR-T cells. (C) Levels of CD69 (i), Ki67 (ii), CD107a (iii), FasL (iv), PD1 and TIM3 (v) and PD1 and LAG3 (vi) in the CD4 and CD8 non-modified and CAR-T cells at the time of peak expansion. Depicted are median and individual values of the nine-teen samples. P-values between the indicated groups were calculated using unpaired Mann-Whitney U-t tests. ns: non-significant, **p<0.01, ***p<0.001. Garcı ´a-Caldero ´n et al. 10.3389/fimmu.2023.1152498 Frontiers in Immunology frontiersin.org08 (Supplementary Figure 10B). As far as the efficacy is concerned, no significant differences were observed in event-free survival comparing patients infused with Tisa-cel (EFS=63%) or Axi-cel (EFS=44%) (p=0.2539) (Supplementary Figure 10C). 3.4 Increased exhaustion markers at peak expansion correlate with increased eventfree survival Characterization of the immunophenotype of both CAR+ cells and non-transduced circulating T-lymphocytes was performed at day of peak expansion in nineteen patients with lymphoma and acute lymphoblastic leukemia. We found that patients with increased levels of PD1+LAG3+ cells (>5.2%) among the CD4+ CAR-T cell compartment correlated with an increased event-free survival (p=0.0203) (Figure 6A). Upon excluding those patients with acute lymphoblastic leukemia (n=8), the same result was confirmed with a significantly increased event-free survival of those patients with a higher exhaustion phenotype at the peak expansion (p=0.0126) (Figure 6B). Besides, reduced levels (<6%) of the cytotoxic marker CD107a+ among the CD8+CAR+ population was also associated with an increased event-free survival (p=0.0365) (Figure 6C). We also confirmed this result after excluding those patients with acute lymphoblastic leukemia (p=0.0537) (Figure 6D). Together these data show how the monitoring and characterization of the phenotype of CAR-T cells in vivo may contribute to the identification of patients with worse prognosis very early after CAR-T cell infusion. Our results reveal that an increased exhaustion phenotype in the CAR-T cells at the peak expansion in vivo correlates with an increased event-free survival in patients with both lymphoma and leukemia. 4 Discussion The correct detection, monitoring and characterization of CAR- T cells in vivo might be a fundamental tool for the identification of biomarkers associated with response or toxicity at very early timepoints post-infusion which would serve to provide earlier interventions. In line with this, blood samples from 48 patients treated with CD19 CAR-T cells were analyzed by flow cytometry and digital PCR to count CAR+ cells in vivo and characterize their immunophenotype. Validation of the detection method for the identification of CAR-T cells by flow cytometry is essential. Therefore, several commercially available reagents were tested for the identification of both academic and commercial CD19 CAR-T cells. These methods can be directed against the CAR itself or to gene markers incorporated into the construct. Besides, flow cytometry staining can be direct or indirect using secondary detection reagents. According to this, there are advantages and disadvantages that need to be considered to choose the optimal detection reagent. Our academic CD19 CAR-T cells incorporate the transduction marker EGFRt and we correctly identified these cells with both anti-EGFR and CD19 CAR detection reagent. However, neither CD19 protein nor protein L were able to properly detect our academic CAR-T cells. On the other hand, commercial CD19 CAR- T cells were correctly identified with both CD19 CAR detection reagent and CD19 protein. Of note, although our academic CD19 CAR contains the heavy variable (VH) and light variable (VL) chain of anti-CD19 monoclonal antibody FMC63, a codon optimization was performed. Moreover, different linker, hinge and transmembrane domains were used compared to commercial CD19 CARs (19,20); therefore, these specific characteristics of our academic CD19 CAR could explain the observed differences in FIGURE 4 Swimmer plots of 48 patients with B-ALL or lymphoma after CD19 CAR-T cell infusion. BCA (B-cell aplasia). CRS and ICANs grades are depicted from 1-4 according to the intensity of the color. Garcı ´a-Caldero ´n et al. 10.3389/fimmu.2023.1152498 Frontiers in Immunology frontiersin.org09