Impact of the kinetics of circulating anti‐BCMA CAR‐T cells and normal lymphocytes on the outcome of MM patients Authors: Sara Gutiérrez‐Herrero, Lourdes Martín‐Martín, María Herrero‐García, Borja Puertas, Eduarda da Silva Barbosa, María Victoria Mateos, Lucía López‐Corral, Verónica González‐Calle, Beatriz Rey‐Búa, Ana África Martín‐López, Estefanía Pérez‐López, Fermín Sánchez‐Guijo, Miriam López‐Parra, Noemí Puig, Alberto Orfao, on behalf of the INCAR and Euroflow Consortium Published: 2025 ABSTRACT AbstractChimeric antigen receptor (CAR)‐T‐cell therapy has improved response rates in relapsed/refractory multiple myeloma (RRMM), yet long‐term disease control remains limited, with only 20% of patients remaining progression‐free at 5 years. Therefore, identifying early predictors of treatment success is critical. Here, we used next‐generation flow cytometry to analyze T‐cell populations in blood at leukapheresis, before infusion, and post‐anti‐BCMA CAR‐T infusion, together with CAR‐T‐cell kinetics and phenotypes before, during, and after the expansion peak, in 53 RRMM patients. Anti‐BCMA CAR‐T cells peaked at Day +14 (72%), comprising >65 distinct populations, predominantly central memory (CM) T‐helper (Th) 1 and Th1/2 CAR‐TCD4+ and CM CAR‐TCD8+ cells. Multivariate analyses revealed that higher frequencies of TCD4+ naive and Th22 transitional memory (TM) cells at leukapheresis, together with circulating tumor plasma cells (CTPCs) before infusion, but none of the specific CAR‐T‐cell populations, were predictors of progression‐free survival (PFS). Based on these variables, we built a risk score that together with disease stage (International Staging System‐Revised [ISS‐R]) independently predicted, before CAR‐T‐cell infusion, which RRMM patients were most likely to benefit from anti‐BCMA CAR‐T therapy. These findings suggest that long‐term PFS is primarily influenced by the patient's immune landscape and the tumor burden rather than anti‐BCMA CAR‐T‐cell properties. INTRODUCTION Immunotherapies are reshaping the treatment of multiple myeloma (MM) by boosting overall response (OR) and survival rates in patients with relapsed/refractory (RR) disease. 1 , 2 Among those therapies, chimeric antigen receptor (CAR)‐T cells have enabled unprecedented response rates and improved disease control in this heavily pretreated group of patients, 3 , 4 leading to recent approval by the FDA (Food and Drug Administration) and EMA (European Medicines Agency) of BCMA‐targeted CAR‐T‐cell therapies for RRMM.5, 6, 7 However, in contrast to B‐cell precursor acute lymphoblastic leukemia and B‐cell non‐Hodgkin lymphoma, anti‐BCMA CAR‐T therapy in RRMM has not led to equally maintained and durable responses. Thus, despite the impressive OR rates (>95%), 6 many RRMM patients treated with anti‐BCMA CAR‐T cells remain at risk of disease recurrence, with only 20% achieving long‐term disease control after 5 years. 8 Because of the still adverse outcome of many anti‐BCMA CAR‐T‐treated RRMM, early identification of reliable biomarkers to predict durable responses (preferably before CAR‐T infusion or early thereafter) has emerged as an unmet clinical need to improve the outcome of RRMM. Currently, the mechanisms that lead to therapeutic failure to anti‐BCMA CAR‐T cells in RRMM remain poorly understood. Despite this, in several clinical studies, diverse key mechanisms have been identified in a subgroup of patients. These include downregulated expression of the target antigen, 9 biallelic deletion of the BCMA gene, 10 T‐cell exhaustion, 11 and limited persistence of functionally effective CAR‐T cells. 5 , 6 Altogether, these data suggest that a more in‐depth understanding of the composition and functionality of the patient's T cells and the CAR‐T cells derived from them, 12 in relation to the tumor burden, 13 might contribute to improve the outcome of RRMM following CAR‐T therapies. Indeed, recent studies revealed an association between the status of the individual MM patient's immune system and disease control, where an increase in circulating T‐helper (Th) cells and T‐memory stem cells, 14 together with a reduced number of exhausted and regulatory T cells (Tregs), might play a critical role in sustained undetected minimal residual disease (uMRD) after infusion. 15 , 16 Despite these findings, at present, there is still limited data on the impact of specific immune cell populations at leukapheresis and before infusion, as well as of anti‐BCMA CAR‐T‐cell phenotypes and their subsets at the peak of in vivo expansion, on the outcome of RRMM patients and long‐term disease control. This is due, at least in part, to the lack of in‐depth T‐cell subset analyses of in vivo expanded CAR‐T cells and the use in the pivotal trials employed for drug approval of molecular (rather than high‐sensitive flow cytometry) techniques for monitoring CAR‐T cells, as those techniques do not allow detailed analyses of the CAR‐T‐cell composition during ex vivo CAR‐T‐cell monitoring. 5 , 6 , 17 Here, we used next‐generation flow cytometry (NGF) to sequentially monitor and characterize in depth (before and after infusion) both the baseline immune cells and the anti‐BCMA CAR‐T cells circulating in blood of a cohort of 53 consecutive RRMM patients who received the treatment within clinical trials. Our goal was to identify immune cell and anti‐BCMA CAR‐T‐cell profiles that might be associated with response to therapy and help to predict the outcome of anti‐BCMA CAR‐T‐treated RRMM. METHODS Patients A total of 53 adult RRMM patients with progressive disease, treated with anti‐BCMA CAR‐T cells at the University Hospital of Salamanca (Spain) between September 2019 and March 2024, were studied. All patients had MM in relapse or progressive disease, were refractory to the last line of therapy, and were consecutively enrolled in different clinical trials active at our center, their eligibility being determined based on the specific inclusion and exclusion criteria of the ongoing studies evaluating anti‐BCMA CAR‐T therapy following approval by the Expert Committee of the Spanish National Health System (Ministry of Health, Madrid, Spain). Before enrollment, each patient gave his/her written informed consent to participate in this specific study after it had been approved by the local Ethics Committee (approval code: PI 2019 05 314). Anti‐BCMA CAR‐T‐cell therapy consisted of idecabtagene vicleucel (ide‐cel; Abecma, Bristol Myers Squibb, New York, NY) (n = 10), ciltacabtagene autoleucel (cilta‐cel; Carvykti, Janssen, Beerse, Belgium) (n = 12), durcabtagene autoleucel (durca‐cel; PHE885, Novartis, Bâle, Switzerland) (n = 13) or ARI0002h (n = 18), an academic CAR administered within the CARTBCMA‐HCB‐01 multicenter clinical trial (EudraCT No. 2019‐001472‐11). 18 Before leukapheresis and post‐lymphodepletion, patients were monitored for their immune cell status and the presence of circulating tumor plasma cells (CTPCs) in blood. After CAR‐T‐cell infusion, blood monitoring of CAR‐T cells and their subpopulations was additionally performed, as described below. Other residual immune cell subsets and CTPC were investigated before lymphodepletion and CAR‐T‐cell therapy, and in parallel to anti‐BCMA CAR‐T cells at months +1, +3, +6, +12, +18, and +24 following administration of the anti‐BCMA CAR‐T cells. This study was not powered enough to compare different CAR‐T‐cell products but rather to characterize the kinetics of CAR‐T cells and lymphocytes and assess their impact on the outcome of patients with RRMM treated with CAR‐T therapy at our center. Response was assessed by bone marrow (BM) aspiration according to the 2016 International Myeloma Working Group (IMWG) criteria. 19 , 20 Immunophenotypic studies Prospective monitoring of CTPC, CAR‐T cells, and other normal residual immune cells was sequentially performed in the blood of each MM patient by NGF, 20 for a total of 782 samples (median: 13 samples/patient; range: 8–24). For this purpose, 10 mL of fresh (<12 h) EDTA (ethylenediaminetetraacetic acid)‐anticoagulated blood was collected per patient and time point, aliquoted and stained with the EuroFlow MM‐MRD 20 reagent kit (Cytognos, Salamanca, Spain) and both the immune monitoring (IMM) TCD4+ (TCD4) and IMM T/NK‐cell cytotoxic (Cytox) antibody panels, to which the BCMA protein reagent conjugated with fluorescein isothiocyanate (FITC) from ACROBiosystems (Bethesda, MD) was added. Sample staining was performed according to the EuroFlow standard operating procedures (SOP) for staining of cell surface membrane‐only and cell surface‐plus‐cytoplasmic markers available at www.euroflow.org, as previously described. 20 , 21 These latter two antibody combinations allowed the identification of ≥85 subsets of TCD4+ cells and ≥45 subsets of TCD8+, Tγδ+, and/or NK‐cell populations, within both CAR‐T cells and normal residual T/NK lymphocytes for a total of >170 cell subsets. The amount of BCMA‐FITC protein‐associated fluorescence reflecting the (median) number of anti‐BCMA receptors bound to the BCMA‐FITC protein reagent on the surface membrane of individual CAR‐T cells, was semi‐quantitatively evaluated by the stain index (SI) as defined by the difference between the median fluorescence intensity (MFI) of BCMA+ and BCMA− cells, divided by 2 standard deviations (SDs) of the MFI observed for the BCMA− cell population, expressed in arbitrary fluorescence ratio units. 22 Stained samples were measured in either a FACSCanto II (BD) (for the MM‐MRD tubes) or a 3‐laser Aurora (Cytek, Fremont, CA) flow cytometer (for the TCD4 and Cytox IMM tubes) using the FACSDiva (BD) and the SpectroFlo (Cytek) software programs, respectively. For data analysis, the Infinicyt software (BD‐Cytognos, Salamanca, Spain) was used. 20 , 21 Primary and secondary endpoints Primary endpoint was progression‐free survival (PFS). Response rates (determined following the IMWG2016 criteria), 19 duration of complete response (CR) (time from CR to the date of recurrence or the last follow‐up visit), and overall survival (OS) in the treated population were secondary endpoints. PFS was defined as the time from CAR‐T infusion (T 0) to disease progression, relapse, or death from any cause, whichever occurred first; OS was defined as the time from T 0 to death from any cause. Patients were categorized into two groups based on treatment response: (1) Ongoing responders, defined as those who achieved at least a partial response (PR) and remained free of disease progression at the last follow‐up, and (2) non‐responders or relapsed patients, including those who failed to achieve a response as well as those who experienced disease progression or relapse after an initial response. Statistical methods For two‐group comparisons, the (two‐tailed) Mann‐Whitney U and the χ2 tests were used. The potential association between continuous variables was assessed by the Spearman correlation. Receiver operating characteristic (ROC) curves were built using the empirical method, and optimal cutoff values were established based on the Youden index. For time‐fixed covariates, PFS, duration of CR and OS curves were plotted according to the Kaplan–Meier method and compared using the (one‐sided) log‐rank test. CAR‐T‐cell persistence and time to B‐cell and PC recovery were evaluated using the cumulative incidence method, considering relapse or death as competing risks. Those variables obtained before CAR‐T‐cell infusion with a significant impact on patient survival in the univariate analyses were submitted to a multivariate analysis using a Cox stepwise regression model. Statistical significance was set at P values ≤ 0.05. RESULTS Patient characteristics A total of 53 patients (29 men and 24 women; median [range] age: 59 [36–76] years) with RRMM and progressive disease were studied; 34 patients had high‐risk cytogenetics defined by the presence of t[4;14], t[14;16] and/or del(17p) (cutoff level of >10%). At enrollment, the patients' Eastern Cooperative Oncology Group (ECOG) performance status was 0, 1, and 2 in 49%, 40%, and 11% of the cases, respectively. According to the International Staging System (ISS), 55%, 28%, and 17% of the patients were Stage I, II, and III, respectively. Before leukapheresis, patients had received between 1 and 6 previous lines of therapy (median: 3); 38 patients (72%) were triple‐class (drug) refractory and 5 (9%) penta‐class refractory, 20 (38%) had paraskeletal plasmacytomas, and 9 (17%) extramedullary disease, including 2 (4%) cases with extramedullary plasmacytomas alone and 7 (13%) with both bone and extramedullary plasmacytomas. Most patients (81%) had received autologous stem cell transplantation (ASCT), and 4% had undergone allogenic stem cell transplantation (AlloSCT). None of the patients had received T‐cell redirecting therapy. Almost all patients (96%) received bridging therapy. At apheresis, 44 (83%) patients had detectable CTPC in blood by NGF, which were still detected at CAR‐T‐cell infusion in most (n = 32) of them (60%). Forty patients (76%) had circulating normal PC, and 44 (83%) displayed mature B lymphocytes in blood at baseline. At apheresis, median absolute lymphocyte (ALC) and TCD4+ cell counts were 1097.5 (range: 397.9–5355) and 313 cells/μL (113.3–1422), and at pre‐infusion, they had decreased to 30.1 (7.9–617.6) and 17.6 cells/μL (4.4–92), respectively. As expected, ALC correlated positively with TCD4+ cell numbers in blood both at apheresis (ρ = 0.75, P ≤ 0.001) and before infusion (ρ = 0.86, P ≤ 0.001). Before infusion, all patients were MRD+ in BM, MRD levels showing a significant (P = 0.01), but marginal (r 2 = 0.16), correlation with the number of CTPC in blood, with 75.4% concordant BM MRD+/CTPC+ cases and 24.6% BM MRD+/CTPC− discordant cases. The most relevant clinical and biological characteristics of the RRMM patients enrolled in this study, and the prior lines of treatment they had received, are summarized in Table S1. Post‐infusion CAR‐T‐cell kinetics and composition in blood Following infusion, CAR‐T cells systematically expanded in blood of MM patients (53/53, 100%), usually with a peak at Day +14 (38/53, 72%) across all studied products, except for a subset of 15 patients in whom CAR‐T cells reached their peak in blood at Day +7 (4/54, 7%), Day +21 (10/53; 19%), and Day +28 (1/53; 2%). The overall median (range) number of CAR‐T cells at the peak was 1363 (9–60,258) cells/µL (Table 1, Figure 1A, and Table S2) with significant differences (P ≤ 0.002) among distinct anti‐BCMA CAR‐T‐cell products (data not shown). Thereafter, the number of circulating CAR‐T cells decreased in each individual patient, being detectable in blood up to 4 years after infusion (Table 1 and Figure 1A), with a cumulative frequency of persistent CAR‐T cells at 6 and 12 months post‐infusion of 43% (95% confidence interval [CI]: 31%–58%) and 29% (95% CI: 18%–45%), respectively. Table 1Chimeric antigen receptor (CAR)‐T BCMA cells and normal B cell and plasma cell counts in blood and their kinetics: impact on the outcome of CAR‐T BCMA ongoing responders versus non‐responders/relapsed multiple myeloma (MM) patients.TotalNon‐responders/relapsedOngoing respondersP value(n = 53)(n = 28)(n = 25)CAR‐T‐cell kinetics in bloodNo. of CAR‐T cells/µL at the CAR‐T peak*1363 (9−60,258)1434 (9−60,258)1234 (29−14,042)0.76Time to the CAR‐T peak (days)*14 (7−28)14 (7−28)14 (7−28)0.95No. of anti‐BCMA receptors/cell at CAR‐T peak (stain index)*14 (2.2−55)13 (2.2−55)15 (3.6−54)0.78CAR‐T‐cell persistence (days)*112 (21−1128)99 (21−1128)118 (27−742)0.30CAR‐T‐cell AUC 0–28 (days × cells/µL)*12.429 (90−456,673)12.631 (90−456,673)12.247 (319−204,405)0.79Tumor/normal PC and B‐cell kinetics in bloodCirculating tumor PC at infusion32/53 (60%)23/28 (82%)9/25 (36%)<0.001Circulating normal PC at infusion39/53 (74%)20/28 (71%)19/25 (76%)0.93Circulating B cells at infusion45/53 (85%)22/28 (79%)23/25 (92%)0.90Circulating tumor PC at the CAR‐T peak2/53 (4%)2/28 (7%)0/25 (0%)0.49Circulating normal PC at the CAR‐T peak8/53 (15%)5/28 (18%)3/25 (12%)0.71Circulating B cells at the CAR‐T peak16/53 (30%)10/28 (36%)6/25 (24%)0.39Re‐emergence of tumor PC in blood16/53 (30%)16/28 (57%)0/25<0.001Time to re‐emergence of tumor PC in blood (months)*9.6 (4.6−19)9.6 (4.6−19)‐‐Loss of normal PC aplasia31/53 (58%)18/28 (64%)13/25 (52%)0.41Time to normal PC recovery (months)*5 (2−39)4.7 (2−37)8 (2−39)0.46Loss of B‐cell aplasia42/53 (79%)23/28 (82%)19/25 (76%)0.41Time to B‐cell recovery (months)*2 (0.7−12)2 (0.7−12)2 (1.9−8)0.67Response to CAR‐T BCMA therapy and outcomeNeurotoxicity (ICANS)2/53 (4%)1/28 (4%)1/25 (4%)1Overall response rate after CAR‐T therapy52/53 (98%)27/28 (96%)25/25 (100%)1CR with uMRD after CAR‐T therapy31/53 (59%)15/28 (54%)16/25 (64%)0.4Time to complete response (months)*3 (1−19)1 (1−19)3 (1−14)0.08Time to disease recurrence (months)*15 (6−49)15 (6−49)‐‐Duration of response (months)*10 (1−42)7 (2−41)11 (1−42)0.02Follow‐up time (months)*12 (3.3−50)12 (3.3−50)12 (3.7−46)0.17 Note: Results expressed as number of cases/total cases (percentage) or as *median (range).Abbreviations: AUC, area under the curve; ICANS, immune effector cell‐associated neurotoxicity syndrome; PC, plasma cell; uMRD, undetected minimal residual disease. Chimeric antigen receptor (CAR)‐T BCMA cells and normal B cell and plasma cell counts in blood and their kinetics: impact on the outcome of CAR‐T BCMA ongoing responders versus non‐responders/relapsed multiple myeloma (MM) patients. Note: Results expressed as number of cases/total cases (percentage) or as *median (range). Abbreviations: AUC, area under the curve; ICANS, immune effector cell‐associated neurotoxicity syndrome; PC, plasma cell; uMRD, undetected minimal residual disease. Chimeric antigen receptor (CAR)‐T BCMA cell kinetics (A) and composition at the CAR‐T peak (B) in blood of multiple myeloma patients (n = 53). In (A), colored dots illustrate the kinetics of circulating CAR‐T cells (per microliter) in blood of each (color‐coded) individual patient at different time points during follow‐up. In (B), the composition of CAR‐T cells at the CAR‐T peak expressed in number of CAR‐T cells per microliter of blood is shown by color‐coded box plots for the distinct CAR‐T‐cell populations identified: blue (CAR‐TCD4+ cell populations, except regulatory T cells [Tregs]), mauve (Tregs), pink (CAR‐TCD8+ cell populations), violet (CAR‐Tᵞẟ+), and yellow (CAR‐TCD4− CD8−). Box plots extend from the 25th to the 75th percentiles, the line in the middle represents median values, and vertical lines correspond to the minimum and maximum values. CM, central memory; D, day; EE, early effector; EM, effector memory; M, month; TE, terminal effector; TFH, T follicular helper cells; Th, T‐helper cells; TM, transitional memory; Y, year. A total of 65 different populations of CAR‐T cells were identified in blood of ≥4 patients at the CAR‐T‐cell peak, from which 6/65 CAR‐T‐cell subsets were highly prevalent (i.e., detected in 100% of the patients) (Figure 1B and Table S2). Globally, CAR‐TCD8+ cells predominated (median: 417 cells/µL) over CAR‐TCD4+ cells (median: 309 cells/µL), with a balanced median relative distribution of 57% versus 40%, respectively (Table S2). Per stage of maturation, CM CAR‐T cells predominated, both among CAR‐TCD4+ (260 cells/µL) and CAR‐TCD8+ cells (305 cells/µL), followed by CAR‐TCD4+ TM cells (58 cells/µL) and CAR‐TCD8+ effector memory (EM) cells (48 cells/µL). From the functional point of view, CAR‐Th1 cells (222 cells/µL), CAR‐Th1/Th2 cells (61 cells/µL) were the most expanded populations within CAR‐TCD4+ cells, the other CAR‐T‐cell populations typically comprising (median values of) <1 cell/µL (Figure 1B and Table S2). Of note, a similar anti‐BCMA CAR‐T‐cell composition profile was found across the four CAR‐T‐cell products used (data not shown). As expected, along the whole monitoring period, an inverse correlation was observed between the number of CAR‐T cells in blood and that of both normal PC (P < 0.001) and CTPC (P < 0.001), as well as between the CAR‐T cells and the normal residual B lymphocytes (P < 0.001) present in blood (Figure 2A–D). The overall rate of recovery from normal B cell and normal PC aplasia following the decline of CAR‐T cells in blood was of 42/53 (79%) and 31/53 (58%) patients, respectively, with a cumulative frequency of PC aplasia loss at 6‐ and 12‐month post‐infusion of 48% (95% CI: 34%–61%) and 63% (95% CI: 46%–76%), respectively (Figure 2A–C). In turn, CTPC systematically decreased in blood to undetected levels (≤2 × 10−6) at the CAR‐T peak in all but one case, re‐emerging in 13% (95% CI: 5%–24%) and 20% (95% CI: 10%–32%) of the patients at 6 and 12 months, respectively (Figure 2D), all of whom subsequently relapsed. No statistically significant differences in CAR‐T‐cell kinetics and CAR‐T‐cell composition at the peak were observed according to the patient prior ASCT status or the number of treatment lines received, except for high CAR‐Tγδ⁺ counts among patients who had undergone an ASCT versus those who did not—<0.001 (<0.001–0.14) versus 0.04 (<0.001–26.2), P = 0.03, respectively—and greater CAR‐Th1 CM cells among cases with ≥4 (vs. <4) prior lines of therapy—90 (<0.001–4158) versus 674 (3.3–16,196) P = 0.04, respectively. The impact of bridging therapy on CAR‐T‐cell kinetics and composition could not be evaluated due to its administration in (virtually) every patient (51/53). Correlation between the total number of blood circulating chimeric antigen receptor (CAR)‐T BCMA cells post‐infusion and normal B lymphocyte, normal plasma cell, and tumor plasma cell counts detected in blood of multiple myeloma patients (n = 782 samples from 53 patients). (A) (from left to right) The correlation between the absolute number of circulating anti‐BCMA CAR‐T cells and that of normal B lymphocytes, normal plasma cells, and tumor plasma cells at different (color‐coded) time points evaluated after CAR‐T‐cell infusion. (B–D) The percentage of patients with detected normal B lymphocytes (B), normal plasma cells (C), and tumor plasma cells (PCs) (D) coexisting or not with anti‐BCMA CAR‐T cells in blood at different (post‐infusion) time points during follow‐up is shown. *Cumulative frequency of circulating tumor PC surrogate. D, day; M, month; Y, year. Impact of the CAR‐T‐cell expansion and composition and normal residual immune cells on response to therapy Each MM patient was followed for ≥100 days after CAR‐T‐cell infusion, except for one case, who did not respond (NR) to CAR‐T‐cell therapy before that time point. Thus, the OR rate observed was 52/53 (98%) (95% CI: 90%–100%): 32/53 (60%; 95% CI: 47%–72%) patients achieved CR as their best response (Table 1), of which all but one showed uMRD in BM. The median time to CR was 3 months (range: 1–19 months). Additionally, at Day +100, 19/19 (100%) evaluable patients had already achieved uMRD CR (19/53), 20/53 (38%; 95% CI: 26%–51%) showed PR and the remaining patient did NR and had progressive disease with similar (P > 0.05) CR and CR plus uMRD rates among patients treated with different CAR‐T products (data not shown). The median (range) duration of response (DOR) was 10 months (1–42 months). Sustained CR for up to 3 years was documented in 17 patients (9%) after a median follow‐up of 12 months post‐infusion; 9/17 were uMRD at Day +100. Maximum CAR‐T‐cell expansion was attained between Day +7 and Day +28, showing no impact on response with similar patterns among PR, CR, and CR uMRD patients, as well as in ongoing responders versus non‐responders/relapsed patients (Figure 3A,B). Similarly, the median amount of BCMA‐FITC protein‐associated fluorescence (number of anti‐BCMA receptors bound to the BCMA‐FITC protein reagent) on the surface membrane of CAR‐T cells measured at the CAR‐T‐cell peak (Figure 3C), as well as the TCD4/TCD8 ratio or the CAR‐T‐cell composition, was not different in ongoing responders and non‐responder/relapsed patients. By contrast, a greater number of CAR‐TCD4+ Tregs was observed in those patients that showed an ongoing response (median: 0.7 vs. ≤0.001 cells/µL, P = 0.03) (Figure 3D); however, these latter differences did not translate into a significant impact on PFS and OS, as discussed below. In contrast, ongoing responder MM patients displayed higher numbers (vs. non‐responder/relapsed patients) of TCD4+ naive (59 vs. 25 cells/µL, P < 0.005) and CM cells (130 vs. 90 cells/µL, P < 0.05), as well as Th22TM cells (2 vs. 1 cells/µL, P < 0.001) and other minor cell subsets (TFHTH1 like, TFHTH2 like, TFHTH1/2 like, Tregs, Th2, Th17, and Th22) immediately before apheresis (Table S3), together with a lower frequency (<0.0002% vs. 0.05%, P < 0.005) and absolute number (≤0.000 02 vs. 0.36 cells/µL, P < 0.005) of CTPC at infusion, respectively (Figure 4A). Of note, this tendency towards higher TCD4+ naive and greater Th22TM residual T‐cell counts in blood of ongoing responder versus non‐responder/relapsed MM patients was maintained after infusion, particularly at the early phases and the peak of in vivo expansion of CAR‐T cells in blood (Table S4). Differences in chimeric antigen receptor (CAR)‐T‐cell kinetics (A, B), CAR‐T BCMA expression levels (C), and the composition of CAR‐T cells at their peak in blood of responder versus non‐responder/relapsed multiple myeloma patients (D) (n = 53). The kinetics of circulating anti‐BCMA CAR‐T cells in blood of (color‐coded) individual patients grouped according to response to therapy (partial plus non‐responders vs. complete response [CR] patients) studied at predefined time points during follow‐up are shown in (A), while the number of anti‐BCMA CAR‐T cells circulating in blood and the median stain index for the CAR‐T BCMA receptor on CAR‐T cells present at the CAR‐T peak are shown for multiple myeloma (MM) patients grouped according to response to CAR‐T therapy in (B) and (C), respectively. The CAR‐T‐cell composition (expressed in No. of cells/µL), at the CAR‐T peak in blood of individual MM patients grouped according to response to therapy (and compared with two‐tailed Mann–Whitney U‐test) is shown in (D). CM, central memory; D, day; DN, double negative; EM, effector memory; M, month; TE, terminal effector; TFH, T follicular helper cells; Th, T‐helper cells; TM, transitional memory; Tregs, T regulatory cells; Y, year. Blood circulating normal immune cell populations at apheresis and tumor plasma cells measured at infusion before chimeric antigen receptor (CAR)‐T BCMA therapy: association with outcome of multiple myeloma (N = 53). The distribution of normal immune cell populations at apheresis and of tumor plasma cells at infusion in blood of CAR‐T BCMA‐treated individual multiple myeloma (MM) patients grouped according to response to therapy (partial responders plus non‐responders vs. complete response [CR] cases) and their impact on progression‐free survival (PFS) are shown in (A) and (B), respectively. (C) The impact of the above variables on patient overall survival (OS) in the univariate and multivariate analyses is displayed. ISS, International Staging System; R‐ISS, Revised International Staging System; PC, plasma cell; Th, T‐helper cells; TM, transitional memory. Impact of the CAR‐T‐cell expansion and composition and the distribution of normal immune cell populations at apheresis and/or following anti‐BCMA CAR‐T‐cell infusion on MM patient outcome At the study closure and after a median follow up of 12 months (range: 3.3–50), the median (95% CI) PFS in the intent‐to‐treat population was 16 months (12–21 months) (Figure S1A) with PFS rates at 1 year and at 2 years of 67% (95% CI: 55%–82%) and 37% (95% CI: 24%–57%), respectively. Only half of the patients who achieved CR with uMRD showed sustained response (16/31, 52%), resulting in a maintained CR with uMRD rate at 1 year of 87% (95% CI: 76%–100%) and at 2 years of 48% (95% CI: 32%–74%) (Figure S1B). From the 19 patients who reached CR with uMRD at Day +100, 10 (53%) had relapsed at the study closure, and three of them (16%) still had detectable levels of circulating CAR‐T cells (median [range]: 107 cells/µL [0.7–2207 cells/µL]) in blood at the time of relapse. Thirteen patients (25%) deceased, mostly (10/13, 77%) due to disease progression or relapse. Of note, significantly shorter median times to relapse were observed among patients who deceased after progression compared to those who progressed but remain alive (6 vs. 12 months, respectively, P < 0.001). The remaining patients (3/13, 23%) deceased from other causes. Median OS was 4 years (Figure S1C) with OS rates (95% CI) at 1 and 2 years of 89% (82‐99%) and 68% (53%–87%), respectively. No statistically significant differences were observed in response rates or DOR according to the CAR‐T product infused (data not shown). PFS and OS were also similar across products although limited number of patients were analyzed by product (data not shown). TCD4+ counts ≥ 300 cells/µL (P < 0.001) at apheresis, <0.01 CTPC/µL at infusion (P = 0.02), and TCD4+ Th22TM numbers ≥ 0.5 cells/µL (P < 0.001) at the time of the CAR‐T peak were all associated with a significant impact on the DOR among CR patients (Figure S2). Multivariate analysis for PFS revealed that TCD4+ counts ≥300 cells/µL at apheresis was the only independent predictor for disease relapse among CR cases (hazard ratio [HR]: 5.7; 95% CI: 1.7–19; P = 0.004) (Figure S2). In order to further identify biomarkers of a durable response, patients that attained CR with uMRD were stratified according to the DOR (<10 months vs. ≥10 months). Several immune and CAR‐T‐cell populations showed a significantly different distribution at the peak between both patient groups (Table S5); however, none yielded cutoff values with a sensitivity and specificity ≥75%. This limitation may be partly explained by the fact that 14 patients remained uMRD CR but had not yet reached the 10‐month minimum follow‐up criteria required for their inclusion in this analysis. In turn, in the whole RRMM cohort (responder plus non‐responder/relapsed patients), a higher absolute number of total TCD4+ naive cells (≥35.7 cells/µL), together with greater Th22TM cell counts (≥1.3 cells/µL) in blood at apheresis (cutoff defined based on ROC curves analysis shown in Figure S3), in addition to an early disease stage (ISS and ISSR) and CR, all emerged as favorable prognostic factors for PFS in the univariate analyses (P = 0.03, P < 0.001, P = 0.008, P < 0.001, and P < 0.001, respectively) (Table S3 and Figure S3), whereas the presence of CTPC at CAR‐T infusion was a strong predictor for a poorer PFS in our anti‐BCMA CAR‐T‐treated MM cohort (Figures 4A–C and S3). None of these variables showed an impact on the presence versus absence of CRS (cytokine release syndrome). Based on those variables related with CAR‐T‐cell composition and the number of CTPC, a (biological) risk score for PFS was constructed, where MM patients were stratified at CAR‐T‐cell infusion (based on the number of adverse prognostic features) into low‐risk (cases with 0/3 adverse prognostic factors: 14 patients [26%]) and high‐risk MM (cases with ≥1 adverse risk factors: 39 patients [74%]) with PFS and OS rates at 4 years of 82% (95% CI: 59%–93%) and of 100% (95% CI: 80%–100%), versus 14% (95% CI: 0.5%–82%) and 66% (95% CI: 50%–80%), respectively (HR: 14; 95% CI: 2–103 for high‐ vs. low‐risk cases; P < 0.001) (Figure S4). Multivariate Cox regression analysis revealed that the new risk score assessed before infusion (P = 0.01), together with disease stage (ISS‐R) at baseline (P < 0.001) were the best combination of independent variables to predict PFS—HR (95% CI) of 12.3 (1.6–92) and 6.1 (1.9–20)—respectively in our MM cohort (Figure 4B). DISCUSSION CAR‐T‐cell therapy represents a remarkable advance in the treatment of RRMM and other hematological malignancies. 3 , 4 However, although the great majority of all CAR‐T‐cell‐treated RRMM achieve CR, there is still a significant fraction of patients who do not show a durable and sustained response to CAR‐T‐cell treatment. 6 , 8 This points out the (urgent) need for reliable biomarkers that help identifying patients at risk of treatment failure versus response for an improved long‐term disease control and patient outcome. At present, there are few studies which investigated in‐depth the kinetics and composition of circulating CAR‐T cells and the patient's immune system, and their impact on treatment response. 12 Understanding the dynamics of the CAR‐T cell and immune cell populations in peripheral blood might be key for the identification of prognostic biomarkers in this setting. Here, we investigated in an unprecedented depth the kinetics and composition of in vivo expanded CAR‐T cells in blood of 53 anti‐BCMA CAR‐T‐treated RRMM patients, in parallel to their immune cell profile at apheresis, and sequentially after CAR‐T‐cell infusion. Our goal was to better understand the potential mechanisms leading to treatment failure and to identify new candidate biomarkers for disease progression and death after anti‐BCMA CAR‐T therapy. The use of multiple CAR‐T products in the cohort should be noted as a limitation, since the current study is not sufficiently powered to confirm or exclude any product‐specific differences. Despite the relatively high variability observed among individual patients and the different anti‐BCMA CAR‐T cells used, overall similar in vivo CAR‐T‐cell kinetics were observed to those previously reported by others, 12 , 23 with maximum anti‐BCMA CAR‐T‐cell counts being reached between Day +14 and (to a lesser extent) Day +21 post‐infusion, in the absence of significant differences in both the magnitude of the CAR‐T‐cell peak and the CAR‐T‐cell kinetics and rate of persistence between (ongoing) responders versus NR/relapsed MM patients, as well as NR/PR versus CR or CR with uMRD cases. Our results further demonstrate that, despite the preferential expansion of CAR‐T cells with a memory phenotype, the overall composition of in vivo expanded CAR‐T cells is highly complex and consists of up to 65 different CAR‐T‐cell populations, among which CAR‐TCD4+ CM with a Th1 and/or Th1/Th2 functional phenotype and CAR‐TCD8+ CM cells predominate over other multiple less represented CAR‐T‐cell subsets. However, no significant differences were found between patients with ongoing response and NR/relapsed cases, except for a greater number of CAR‐TCD4+ Tregs among the former group, in the absence of a significant impact on patient PFS. An association between increased numbers of TCD4+ cells with a stem cell memory phenotype 24 , 25 and a better outcome has been previously reported in anti‐BCMA CAR‐T‐cell‐treated RRMM. In turn, Coffey et al. 15 demonstrated that increased counts of circulating Th cells 1 year after lenalidomide exposure were associated with sustained MRD negativity. In those studies, the impact of the immune cell composition of the apheresed products used for CAR‐T cell manufacture, but not of the in vivo expanded CAR‐T‐cell populations, on the outcome of MM was reported, higher numbers of both TCD4+ naive and memory stem cells being associated with a better MM patient outcome. 24 , 25 In turn, Th22 cells are pro‐inflammatory cells that primarily secrete IL‐22. 26 Elevated Th22 cells have been detected in RRMM patients in association with more advanced disease stages, higher disease burden, and a poorer prognosis. 27 However, recent literature suggests that Th22 cells may exert both pro‐ and anti‐tumoral effects, 28 , 29 and their interaction with disease burden may be more complex. 26 Paradoxically, this might contribute to explain the association observed in our cohort between greater numbers of Th22TM cells at apheresis and improved patients outcomes, suggesting that in the context of CAR‐T‐cell therapy, Th22 cells may be associated with a better preserved immune system; alternatively, subsets of Th22 cells may exert different effects in MM patients undergoing CAR‐T cells versus other therapies at the time of CAR‐T‐cell apheresis favoring an enhanced functional activation of the anti‐tumoral activity of the harvested T cells, and contribute to an improved therapeutic efficacy. In line with these findings, here we also showed that greater numbers in blood at apheresis of TCD4+ naive cells, in addition to Th22TM cells, were both associated with a significantly prolonged PFS and OS of RRMM patients after anti‐BCMA CAR‐T‐treatment. Taken together, these findings highlight the potential relevance of the patient's immune status at apheresis on the quality of the anti‐BCMA CAR‐T product and the clinical benefit associated with it, in line with previous studies.11, 13, 14, 15, 16 Interestingly, in addition to TCD4+ naive and Th22TM cells, the presence of CTPC before CAR‐T‐cell infusion also showed a strong impact on disease outcome. Higher levels of blood CTPC have recently emerged as one of the most powerful prognostic factors in both smoldering MM and MM 30 , 31 at diagnosis and following (immune and other) conventional MM therapies. 32 , 33 Therefore, in patients with high CTPC levels, delaying CAR‐T‐cell infusion and administering additional chemotherapy to reduce tumor burden may help optimize treatment efficacy. Based on these three variables we built a simple risk score model which could accurately identify before CAR‐T‐cell infusion, which RRMM patients might benefit from subsequent anti‐BCMA CAR‐T therapy (cases with no adverse prognostic features) versus those who will not reach a sustained response (cases with one, two or three adverse prognostic features), proving to be, together with the ISS‐R disease stage at baseline an independent predictor of patient outcome. Altogether, these new data might contribute to influencing future post‐CAR‐T management strategies. In this regard, patients with an unfavorable T‐cell profile might benefit from immunomodulatory agents or maintenance therapy aimed at both controlling the disease and restore the patient's immune functionality. Alternatively, interventions such as stem cell boost or other immune reconstitution strategies might also contribute to modulating T‐cell composition before leukapheresis and improve CAR‐T product quality. However, we only observed a limited impact of ASCT and the number or prior lines of therapy on CAR‐T cell numbers and composition. Collectively, these results support the notion that longer PFS might not directly depend on the CAR‐T‐cell expansion, composition and properties, but mostly on (i) the status of the patient immune system, particularly as regards specific populations of TCD4+ cells, (ii) the tumor burden in blood as a surrogate marker of MM dissemination potential and anti‐BCMA CAR‐T‐cell response potential, and (iii) the stage of disease at baseline.30, 31, 32 Based on these results, the use of CAR‐T‐cell therapy in earlier lines of treatment is expected to improve outcomes 34 given the critical roles of these three factors. Considering this, ongoing clinical trials are evaluating CAR‐T‐cell therapy in the first‐line setting and even in patients with smoldering MM, aiming to leverage a more favorable immune environment and lower disease burden to enhance therapeutic efficacy. Despite BCMA expression being confirmed by Friedman et al. 35 on malignant PC of all MM patients analyzed, variable levels of expression were found among tumor PC within and among individual patients. In addition, previous studies on CTPC pre‐ and post‐anti‐BCMA CAR‐T‐cell treatment have shown that (i) BCMA expression might turn negative in some patients after CAR‐T‐cell infusion, before disease progression, 36 or (ii) it can be consistently downregulated on malignant plasma cells after CAR‐T‐cell exposure 37 mostly due to BCMA internalization by myeloma cells, but also trogocytosis by CAR‐T cells. 37 Whether such a decrease is particularly more pronounced among CTPC (vs. BM clonal PC) and/or due to an increased release of BCMA into a soluble form, remains to be investigated. In turn, prior anti‐MM and ASCT therapies administered to RRMM might also perturb the composition of the T‐cell compartment and affect the fitness of T cells for immunotherapies. 37 Thus, it has been previously suggested that the decrease in naive T cells might have an adverse impact on the production of efficient autologous CAR‐T in MM, 38 and clinical trials further indicate that high levels of exhaustion markers in these cells correlate with poor clinical responses. 36 This, together with the existence of an immunosuppressive BM microenvironment, is considered the potential main cause for MM relapse post CAR‐T‐cell treatment. However, further experimental and clinical studies are needed to better understand the interplay between the immune system and CAR‐T‐cell therapy in RRMM and other cancers, particularly as regards their impact on treatment efficacy, resistance mechanisms, and long‐term immune reconstitution. AUTHOR CONTRIBUTIONS Alberto Orfao contributed to the conceptualization, design, and supervision of the study. Borja Puertas, María Victoria Mateos, Lucía López‐Corral, Verónica González‐Calle, Beatriz Rey‐Búa, Ana África Martín‐López, Estefanía Pérez‐López, Fermín Sánchez‐Guijo, Miriam López‐Parra, and Noemí Puig enrolled patients and contributed to data clinical data collection. Sara Gutiérrez‐Herrero, Lourdes Martín‐Martín, María Herrero‐García, and Eduarda da Silva Barbosa performed the experiments and analyzed data. Sara Gutiérrez‐Herrero and Lourdes Martín‐Martín performed statistic and interpreted data. Alberto Orfao, Sara Gutiérrez‐Herrero, Lourdes Martín‐Martín, and María Herrero‐García wrote the paper. All authors critically reviewed the manuscript and approved the final version of the document. CONFLICT OF INTEREST STATEMENT B.P. has received honoraria from Johnson & Johnson, Amgen, Sanofi, Bristol Myers Squibb, and Aptitude Health. E.S.B. received research support from International Myeloma Society (IMS) and Paula and Rodger Riney Foundation Career Development Award Research. M.V.M. has received honoraria derived from lectures and advisory board activities from Janssen, Bristol Myers Squibb/Celgene, Amgen, Takeda, AbbVie, Sanofi, Oncopeptides, Adaptive, Roche, Pfizer, Regeneron, GlaxoSmithKline, Bluebird Bio, and Sea‐Gen. L.L.‐C. has received honoraria derived from lectures and advisory board activities from Gilead Sciences, Novartis, Roche/Genentech, Celgene/Bristol Myers Squibb, Janssen, AstraZeneca Spain, MSD, Miltenyi Biotec, Autolus, Sandoz, and AbbVie. V.G.‐C has received honoraria derived from lectures and advisory board activities from Johnson & Johnson, GlaxoSmithKline, and Pfizer. F.S.‐G reports research support and/or honoraria from Novartis, Gilead/Kite, and Johnson & Johnson. N.P. has received honoraria derived from lectures and advisory board activities from Amgen, Celgene, Janssen, Takeda, Pfizer, GSK, Sanofi, and The Binding Site. The remaining authors declare no competing financial interests. ETHICS STATEMENT Before enrollment, each patient gave his/her written informed consent to participate in this specific study after it had been approved by the local Ethics Committee (approval code: PI 2019 05 314). FUNDING The experimental work was supported by grants awarded by the Accelerator Award‐Cancer Research UK/AIRC/AECC‐INCAR (GEACC18001ORF). Supporting information Supporting Information. Supplementary Information ACKNOWLEDGMENTS The authors thank the patients treated at Salamanca University Hospital Complex, the Biomedical Research Networking Centre Consortium (CIBERONC) program (CB16/12/00400), CB16/12/00233), and the Advanced Therapies Cooperative Research Network Oriented Towards Health Results (RICORS TERAV and TERAV+, RD21/0017/0006, RD24/0014/0034), both from Instituto de Salud Carlos III (ISCIII), and the Spanish Association against Cancer (AECC, GEACC18001ORF). DATA AVAILABILITY STATEMENT Research data are not shared. For original data, please contact orfao@usal.es. REFERENCES Associated Data Supplementary Materials Supporting Information. Supplementary Information Data Availability Statement Research data are not shared. For original data, please contact orfao@usal.es.