Phenotypic Composition of Commercial Anti-CD19 CAR T Cells Affects In Vivo Expansion and Disease Response in Patients with Large B-cell Lymphoma

Following promising results in clinical trials, the CD19-targeted chimeric antigen receptor (CAR) T-cell products tisagenlecleucel (Tisa-cel) and axicabtagene ciloleucel (Axi-cel) have been approved for third-line treatment of relapsed/refractory (R/R) large B-cell lymphoma (LBCL; refs. 1–10).

A number of pretreatment patient clinical features including mainly parameters of lymphoma burden and levels of inflammatory cytokines were used to predict clinical response to CAR T-cell therapy (5, 11–13) but were not able to discriminate the outcome at the single patient level. On the other hand, in vivo CAR T-cell proliferation after infusion was found of central importance to clinical response and toxicities in acute lymphoblastic leukemia (ALL) and within clinical trials (4, 14) but has been less investigated in patients with lymphoma receiving Tisa-cel and Axi-cel as standard-of-care therapy. In addition, there are no data correlating properties of Tisa-cel and Axi-cel infusion bag content and CAR T-cell expansion.

These infusion products differ in several aspects including the type of CAR construct used, the manufacturing process, the dose, the primary container and the fill volume. These characteristics, together with the quality of the T cells harvested from patients for genetic manipulation, the tumor burden and tumor microenvironment might all contribute to CAR T cells’ fate in vivo. Recent evidence suggest that cellular and molecular features of infused CAR T-cell products are major factors accounting for the variability in efficacy among Axi-cel–treated lymphoma patients of the ZUMA-1 trial (15) and among patients with chronic lymphocytic leukemia (CLL) receiving CTL019 therapy (16). Of note, these data have been generated mainly as part of exploratory analysis of clinical trials using comprehensive and challenging methodologic approaches to identify predictive biomarkers in patients treated with specific products. Real-life evaluations assessing the phenotypic composition of the CAR T-cell infusion products and analyzing how these features affect CAR T-cell fate in vivo and outcome are scanty. Besides, no studies directly comparing Tisa-cel and Axi-cel characteristics and their impact on expansion kinetics have been reported yet.

In this single-center prospective study, we aimed at identifying key phenotypic Tisa-cel and Axi-cel CAR T-cell product characteristics associated with improved in vivo expansion and therefore tumor responses.

This single-center, prospective observational study conducted at the Hematology Division of the Fondazione IRCCS Istituto Nazionale dei Tumori (Milano, Italy) included 61 patients with R/R LBCL who received standard-of-care Tisa-cel or Axi-cel between December 2019 and October 2021. All patients received lymphodepletion according to manufacturers (3, 17). The study (INT180/19) was approved by the local ethics committee and was carried out in accordance with the Helsinki Declaration and patients provided written informed consent. Patients were assigned to either Tisa-cel or Axi-cel based on slot production availability and histology, because patients with primary mediastinal B-cell lymphoma (PMBCL) can receive Axi-cel only. Disease status was assessed as per Lugano 2014 classification (18) by CT scan and PET/CT scan at day 30 (PET1) and at day 90 (PET3), or when clinically indicated. Total metabolic tumor volume (TMTV) was computed both by the 41% maximum standardized uptake value threshold and by an automatic delineation with the estimated threshold algorithm supplied with PET-Volume Computer Assisted Reading (VCAR) software (GE Healthcare). Immune effector cell associated neurotoxicity syndrome (ICANS) and cytokine release syndrome (CRS) related to CAR T cells were graded according to the American Society for Transplantation and Cellular Therapy Consensus Grading (19).

For CAR T-cell identification, cells were stained with the CD19 CAR detection reagent (Miltenyi Biotec) following the manufacturer's instructions. For infusion products, 50 μL from bag leftovers were used. For in vivo longitudinal CAR T-cell monitoring, 100 μL of peripheral blood samples (PB; refs. 20–22) collected at day 5, 7, 10, 14, 21 and monthly after infusion were used. After staining with the CAR detection reagent, for PB only, red blood cells (RBC) were lysed using the Ammonium Chloride Lysing solution (BD Biosciences). Then, for both PB and infusion products, cells were labeled with the following set of antibodies: biotin-PE (clone REA746, catalog no. 130–110–951), CD45-VioBlue (clone REA747, catalog no. 130–110–637, RRID:AB_2658243), CD3-FITC (clone REA613, catalog no. 130–113–138), CD4-VioGreen (clone REA623, catalog no. 0130–113–230), CD8-APC-Vio770 (clone REA734, catalog no. 130–110–681), CD14-APC (clone REA599, catalog no. 130–110–520), and 7-AAD (catalog no. 130–111–568) staining solution (all from Miltenyi Biotec). Additional information on CAR T-cell identification strategy are reported in Supplementary Fig. S1. For PB CAR T-cell detection, the protocol was optimized to eliminate the background signals generated by the biotin–PE antibody used to label the CD19 CAR Detection Reagent, which affected absolute CAR⁺ cell enumeration. To assess the level of nonspecific background, 100 μL of PB were processed with RBC lysis, washed, and stained with: biotin-PE, CD45-VioBlue, CD3-FITC and 7-AAD staining solution. The amount of nonspecific signal detected was substracted to avoid the possible overestimation of circulating CAR⁺ cells. Control samples were obtained from healthy volunteers who provided written informed consent.(Supplementary Figs. S2–S4). Data acquisition was performed with a MACSQuant Analyzer or a MACSQuant Analyzer MQ10 (Miltenyi Biotec); data were analyzed with the MACSQuantify software for absolute quantification of cells with custom analysis protocols (Express Mode Master Package, Miltenyi Biotec) or with predefined optimized settings for CAR T-cell analysis.

For CAR T-cell differentiation, at least 10⁶ cells from infusion product leftovers were used. Cells were incubated with the CD19 CAR detection reagent (Miltenyi Biotec). After wash, the following antibodies were added: biotin-PE, CD3-FITC, CD4-VioGreen, CD8-APC-Vio770, CD45RO-APC (clone REA611, catalog no. 130–115–556), CD62L-Pe-Vio700 (clone 145/15, catalog no. 130–113–621) and CD197-VioBlue (clone REA546, catalog no. 130–117–353; all from Miltenyi Biotec). Data were acquired on BD FACSCanto II (BD Biosciences) or MACSQuant Analyzer MQ10 (Miltenyi Biotec) and analyzed using FlowJo software (RRID:SCR_008520), version 10. Additional information on differentiation status of CAR T cells within infusion products are reported in Supplementary Fig. S5.

Time of CAR T-cell administration was used as the origin in all time-to-event analyses. For progression-free survival, overall survival, and duration of response (PFS, OS, and DoR), Kaplan–Meier curves and the log-rank test were used. For group comparison of categorical data Fisher exact test or χ² test were used. Mann–Whitney test was performed for comparison of continuous variables. Multivariable logistic regression was used to study the association between different covariates to disease response. All the reported P values (P) are two-sided and were considered statistically significant when P < 0.05. Plots and statistical analysis were performed with the use of GraphPad Prism v.9.00 (RRID:SCR_002798), RAW Graph website and Adobe Illustrator (RRID:SCR_010279).

The data generated in this study are available upon request from the corresponding author.

To rule out the possibility that the two groups of patients receiving either Tisa-cel or Axi-cel were significantly different, patient characteristics have been analyzed and are listed in Table 1. Axi-cel and Tisa-cel patients are comparable with the only difference being the presence of PMBCLs among Axi-cel recipients that explains a higher proportion of patients treated with check-point inhibitors (CPI) and a lower International Prognostic Index (IPI). As a whole, the population studied is similar to those described in other published real-life CAR T cells experiences except for a better performance status as a consequence of the fact that in Italy, Axi-cel and Tisa-cel can only be prescribed in performance status-Eastern Cooperative Oncology Group (PS-ECOG) 0 and 1 patients (5, 11, 23, 24).

Any grade and grade ≥3 CRS occurred, respectively, in 74% (95% CI, 62%–83%) and 2% (95% CI, 0.1%–9%) of patients. Any grade and grade ≥3 neurotoxicity occurred in 13% (95% CI, 7%–24%) and 2% (95% CI, 0.1%–9%) of patients, respectively. Tocilizumab and steroid were used in 39% and 28% patients, respectively. No difference in terms of CRS and ICANS between Axi-cel and Tisa-cel reached statistical significance although a trend to higher neurotoxicity can be highlighted in Axi-cel recipients who in fact were characterized by higher use of steroids (Supplementary Table S1).

All 61 patients were valuable for response at day 30 after infusion and 58 were valuable at day 90. Median follow-up from infusion was 9 months. The best overall response rate (ORR) and CR rate were 69% (95% CI, 56%–79%) and 52% (95% CI, 40%–64%) respectively. Median time to response was 30 days, and no patients achieved a response after day 90. Although there was a trend in response rates in favor of Axi-cel, we did not observed statistically significant differences in terms of best ORR and CR rates between Axi-cel and Tisa-cel (Supplementary Table S1).

Of the 61 recipients, the median PFS was 10.1 months (not reached for Axi-cel and 6.4 months for Tisa-cel). PFS at 1 year was 49% (56% for Axi-cel and 41% for Tisa-cel). Relapses after 6 months were rare. The median OS was not reached, OS at 1 year was 80%. According to the log-rank (Mantel–Cox) test, PFS and OS curves are not significantly different between Axi-cel and Tisa-cel (Fig. 1).

Collectively, there were no significant differences in terms of safety, response to therapy and outcomes between patients treated with Axi-cel and Tisa-cel (Supplementary Table S1).

Left-over cells collected after washing 61 infusion product bags [29 Tisa-cel (47.5%) and 32 Axi-cel (52.5%)] were analyzed by flow cytometry (FCM). In the infusion products analyzed, CD3⁺ cells were 73.2% of CD45⁺ cells (median; range 25.2%–99.7%), CAR⁺ cells were 55.2% of CD3⁺ cells (median; range 5.8%–92.7%) and the CD4⁺/CD8⁺ ratio among CAR⁺ cells was 1.94 (median; range 0.38–15.8).

When comparing Tisa-cel and Axi-cel infusion products, Tisa-cel was characterized by a lower fraction of CD3⁺ among CD45⁺ cells (median CD45⁺/CD3⁺: 63.8% vs. 78.5% for Tisa-cel and Axi-cel, respectively; P < 0.005) and of CAR⁺ among CD3⁺ cells as compared with Axi-cel (median CD3⁺/CD45⁺: 27.3% vs. 72.9% for Tisa-cel and Axi-cel, respectively; P < 0.001). On the contrary, Tisa-cel displayed a higher CD4⁺/CD8⁺ ratio among CAR⁺ cells as compared with Axi-cel (median ratio CD4⁺/CD8⁺: 2.86 vs. 1.37 for Tisa-cel and Axi-cel, respectively; P < 0.005; Fig. 2A–C). The presence of different CAR⁺ cell differentiation subsets in infusion products was then assessed (25–27). We analyzed the frequency of T central memory [T_CM (CD45RO⁺/CD197⁺/CD62L⁺)], T naïve-like [T_N-like (CD45RO⁻/CD197⁺/CD62L⁺)], T effector [T_E (CD45RO⁻/CD197⁻/CD62L⁻)] and T effector memory [T_EM (CD45RO⁺/CD197⁻/CD62L⁻)] subsets among CD4⁺ and CD8⁺ CAR⁺ cells within Tisa-cel and Axi-cel infusion products (n = 46).

Collectively CAR T-cell infusion products were largely composed of differentiated T-cell subsets such as CD4⁺ T_EM (21.9%), CD8⁺ T_EM (12.4%) and of CD4⁺ T_CM (18.4%), CD8⁺ T_CM (11.4%), whereas T_N-like and T_E cells were less represented (CD4⁺ T_E 0.01%, CD8⁺ T_E 0.03%, CD4⁺ T_N-like 0.04%, CD8⁺ T_N-like 0.09%).

In Tisa-cel (n = 19), CAR⁺ cells were enriched in CD4⁺ cells with central memory and effector phenotypes (median CAR⁺/CD4⁺ T_CM: 37.7% in Tisa-cel vs. 14.1% in Axi-cel; P < 0.005 and median CAR⁺/CD4⁺ T_E: 0.056% in Tisa-cel vs. 0.005% in Axi-cel; P < 0.005) while Axi-cel (n = 27) displayed an enrichment in CD8⁺ cells with effector memory and naïve-like phenotypes (median CAR⁺/CD8⁺ T_EM: 6.07% in Tisa-cel vs. 23% in Axi-cel; P < 0.005 and median CAR⁺/CD8⁺ T_N-like: 0.04% vs. 0.15%; P < 0.05). Of note, no significant differences were found when comparing the frequencies of the other T-cell populations (CAR⁺/CD4⁺ T_EM, CAR⁺/CD4⁺ T_N-like, CAR⁺/CD8⁺ T_CM and CAR⁺/CD8⁺ T_E; Fig. 2D–K). All data are reported in Supplementary Table S2.

To evaluate whether quantitative or qualitative features of CAR T-cell infusion products could affect in vivo CAR T-cell fate, expansion kinetics data of 53 patients with available longitudinal in vivo CAR T-cell records were analyzed. Cellular parameters, including concentration of CAR⁺ cells at day 7 (C₇) and at day 10 after infusion (C₁₀), time to maximal expansion (T_max), concentration of CAR+ cells at T_max (C_max) and the magnitude of expansion up to 30 days (calculated as the area under the curve, extended, AUC_0–30) were evaluated. Overall, the kinetics curves displayed an initial increase in CAR T-cell numbers within the first 2 weeks of infusion, followed by a progressive decrease in all treated patients (Fig. 3A). Median T_max was 10 days for both products. Of note, both C_max and C₁₀ were found to be strongly correlated with the AUC_0–30 (Pearson correlation 95% CI; r = 0.9588; P < 0.0001 and r = 0.8774; P < 0.0001, respectively).

On the basis of this observation, the C₁₀ was selected as the surrogate measure to quantify CAR T-cell expansion and the median C₁₀ (24.5 CAR T cells/μL) was used as a cutoff to dichotomize the cohort of patients into “expanders” (patients with a C₁₀ ≥ 24.5 CAR T cells/μL) and “poor-expanders” (patients with a C₁₀ < 24.5 CAR T cells/μL). When looking for possible differences in the infusion products administered to these two groups, we established that expanders had received infusion products significantly enriched in CAR⁺CD8⁺ cells with a central memory phenotype as compared with poor-expanders (median CAR⁺/CD8⁺ T_CM: 13.8% in expanders vs. 4.5% in poor-expanders; P < 0.005) thus suggesting that CAR⁺/CD8⁺ T_CM have the highest ability to expand (Fig. 3B–I; Supplementary Table S3). Of note, in our cohort total metabolic tumor volume (TMTV) values did not differ between expanders and poor-expanders (Supplementary Fig. S6). Similarly, the use of tocilizumab and steroids did not affect CAR T-cell expansion (OR, 2.301; χ² test 95% CI, 0.804–6.561; P = ns; OR, 2.929; χ² test 95% CI, 0.670–7.400; P = ns, respectively).

The fact that the CAR⁺/CD8⁺ T_CM population was the only population among all the tested ones positively associated with CAR T-cell expansion, but not differentially present in Tisa-cel and Axi-cel infusion products, is consistent with Tisa-cel– and Axi-cel–treated patients displaying similar expansion kinetics (Fig. 3J and K). In fact, no discrepancy in the number of CAR T cells/μL either at day 10 or at the T_max was observed (median C₁₀: 34 CAR T cells/μL for Tisa-cel vs. 22.9 CAR T cells/μL for Axi-cel; P = ns and median C_max: 59.6 CAR T cells/μL for Tisa-cel vs. 54 CAR T cells/μL for Axi-cel; P = ns); the magnitude of expansion was also similar in patients receiving the two products (median AUC_0–30: 128.8 for Tisa-cel and 113.3 for Axi-cel; P = ns; Fig. 3L). Notably, CAR T-cell persistence was similar in Tisa-cel- and Axi-cel–treated patients (Supplementary Fig. S7A) and was not associated with a prolonged DoR (Supplementary Fig. S7B).

As CAR T-cell expansion after infusion has been correlated with clinical response in other studies (4, 14), we set to analyze whether this was true also in our real-life setting. Patients who achieved complete response (CR) or partial response (PR) by day 90 after CAR T-cell infusion were defined as responders (RE); patients with stable, progressive disease (SD, PD) and patients who died for progression before day 90 were defined as nonresponders (NR). Accordingly, among the 53 patients with available longitudinal expansion assessment and clinical evaluation, 31 (59.1%) were RE (28 CR, 15 Tisa-cel and 13 Axi-cel; 3 PR, all Axi-cel) and 22 (40.9%) were NR (all PD, 11 Tisa-cel and 11 Axi-cel).

Consistent with previous reports, responders showed a superior expansion of CAR T cells as compared with NR, who exhibited reduced peak levels and a more rapid contraction (Fig. 4A and B). In detail, C₇, C₁₀, C_max, and AUC_0–30 were detected in RE when compared with NR. In particular median C₇ was 22.8 CAR T cells/μL in RE versus 6.6 CAR T cells/μL in NR (P ns), C₁₀ was 67.3 CAR T cells/μL in RE versus 13.7 CAR T cells/μL in NR (P < 0.01), median C_max was 89.9 CAR T cells/μL in RE versus 25.3 CAR T cells/μL in NR (P < 0.005) and median AUC_0–30 was 202.4 in RE versus 61.1 in NR (P < 0.001; Fig. 4C). Of note, we observed that among expanders, 21 (77.8%, 95% CI, 59.2%–89.4%) patients were RE, in contrast to 10 (38.5%; 95% CI, 22.4%–57.5%) in the poor-expander group. On the basis of these data, we found that expansion was significantly positively associated with response (OR, 5.600; χ² test 95% CI, 1.681–18.65; P < 0.005; Fig. 4D). This association remained significant even after correcting for other parameters linked with outcome in a multiple logistic regression analysis (C₁₀ OR, 1.268; 95% CI, 1,062–1,676; P < 0.05; Supplementary Table S4; Supplementary Fig. S8). Consistently, expanders had significantly longer survival rates when compared with poor-expanders (PFS median, not reached vs. 3.7 months, respectively; P < 0.05; Fig. 4E).

To the best of our knowledge, this is the first study performing correlative analysis of the T-cell immunophenotypic features of Tisa-cel and Axi-cel infusion products and in vivo CAR T-cell expansion and efficacy. CAR T-cell expansion dynamics have been indicated as surrogate markers for tumor response and disease control in exploratory analysis of clinical trials involving patients with R/R lymphoma but properties of CAR T cells that enable their proliferation in vivo have still to be consistently defined in real-life settings. In the ZUMA-1 trial, peak CAR T-cell levels correlated with outcome but expansion of CAR⁺ cells was not relevant (4). These findings were further confirmed after normalizing peak CAR⁺ cells to tumor burden (28). In this study, poorly differentiated CAR T cells with a T naïve-like phenotype were enriched in expanders and early CAR T-cell expansion was associated with efficacy. In another retrospective analysis of biomarker data from Axi-cel–treated patients, cells from the infusion products of patients with ongoing CR at 3 months had an enrichment of CD8 central memory T-cell phenotypes compared with patients with PR/PD (15). This is in line with a previous study that evaluated CAR T cells with CD3ζ and 4–1BB signaling domains in the setting of CLL, in which a memory CD8 phenotype was associated with superior responses (16). In the Juliet trial, Tisa-cel expansion was positively correlated with the duration of response but no correlative analysis on infusion product CAR T-cell characteristics and proliferation was reported (3).

In our study, we highlighted that an enrichment of CD8⁺ T_CM phenotypes in CAR T-cell products is associated with improved in vivo CAR T-cell expansion. This is true for both Tisa-cel and Axi-cel and is consistent with what observed by Deng and Fraietta, but in contrast to Locke and colleagues who defined the CAR⁺ T stem cell memory (T_SCM) population in infusion products as the one with the highest ability to expand (15, 16, 28). Of note, even in our cohort expanders have higher frequencies of CD8⁺CAR⁺ naïve-like memory cells expressing CD45RA, CD197, and CD62L when compared with poor-expanders although the difference did not reach statistical significance. Interestingly, transcriptomic analyses revealed that T_SCM cells are similar to T_CM and CD45RO⁺ CCR7⁺ T_CM cells are known to play a key role in antitumor immunity. The frequency of T_CM cells in blood was in fact a positive predictor for both response to anti–PD-1 immunotherapy and survival.

Collectively previous data and our current analyses point to the essential role of CAR⁺ cells with a central memory phenotype in mediating in vivo CAR T-cell expansion which is required for their clinical efficacy. In fact, in our study cohort, the improved proliferation of circulating CAR⁺ cells in vivo correlated with clinical outcome and progression free survival. Similarly Ayuk and colleagues have used the median CAR T-cell peak concentration to define “weak expanders” and “strong expanders” and showed a significantly increased survival of “strong expanders” among 21 patients treated with Axi-cel (10).

While estimating the AUC could be considered a valid tool to describe and quantify aspects of the concentration–time profile of CAR T cells in PB, for AUC_0–30 to be calculated but also for the peak concentration to be defined, a number of longitudinal assessments within the first month postinfusion are needed. On the other hand, the C₁₀ can be more conveniently used as a surrogate measure of CAR T-cell expansion. Given the high correlation we established among all the estimates of expansion analyzed (namely the AUC_0–30, C_max and C₁₀), in our study expanders and poor-expanders have been dichotomized based on the CAR T-cell concentration at day 10 post infusion which represents an early parameter of expansion that can be easily monitored in routine clinical practice.

Although target cell manipulation could provide better quality CAR T cells, to date, autologous cells from leukapheresis remain the most common starting material for CAR T-cell production as manufacturing defined composition of CAR T cells is difficult, costly and challenging (29). This is probably the reason why very limited information is provided when commercial products are released. These include the safety requirements and cell viability and cell dose for Tisa-cel but only the range of CAR⁺ viable cells for Axi-cel (0.4–2 × 10⁸ cells). Nevertheless the use of out-of-specification products, designated as such because of viability <80%, were reported to be similarly effective and safe when compared with products with viabilities ≥80% (30). The dose range of Tisa-cel and Axi-cel was approved based on the pivotal trials, in the absence of formal dose finding first-in-human studies and there is no clear understanding either on the minimum effective CAR T-cell dose required or on the subtypes of infused CAR T cells mediating effective in vivo proliferation and efficacy.

The significantly different composition of Tisa-cel and Axi-cel infusion products we describe, with Tisa-cel being characterized by fewer CAR⁺ cells among CD3⁺ lymphocytes and less CD8⁺CAR⁺ cells, is surprising given their similar expansion kinetics in our cohort of patients. Notably, CD8⁺CAR⁺ central memory cells are among the few T-cell populations that were equally represented in Tisa-cel and Axi-cel infusion products, consistent with their role in mediating CAR T-cell expansion. In fact, another interesting finding of our study is that CAR⁺ cells dynamics were similar in patients treated with Tisa-cel and Axi-cel, in line with their similar efficacy, despite the different manufacturing processes and costimulatory domains that are thought to induce CAR T-cell proliferation with distinctive kinetics (31, 32).

In conclusion, our study describes for the first time the clinical utility of characterizing the T-cell composition of commercial infusion products and of real-time enumeration of expanding CAR T cells in patients with lymphoma. Our results suggest that despite the fact that Tisa-cel and Axi-cel infusion products are significantly different, Tisa-cel and Axi-cel expansion kinetics are similar and are mediated by the presence of CAR T cells with central memory phenotypes in infusion products. Consistently, for both Tisa-cel and Axi-cel, the number of circulating CAR T cells, a surrogate of expansion, can be used as predicting biomarker of response. Despite requiring validation in a larger series of patients, these observations indicate that Tisa-cel and Axi-cel infusion product characteristics can be interrogated early to gain insight into the expansion capabilities of CAR T cells in vivo that are necessary for ultimate CAR T-cell efficacy.

C. Monfrini reports grants from Alleanza Contro il Cancro (ACC), Associazione Italiana contro le Leucemie-Linfomi e mieloma (AIL), the Italian Ministry of Health, and Associazione Italiana Ricerca sul Cancro (AIRC) during the conduct of the study. F. Stella reports grants from ACC, AIL, AIRC and the Italian Ministry of Health during the conduct of the study. V. Aragona reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study. M. Magni reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study. S. Ljevar reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study. C. Vella reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study. E. Fardella reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study. A. Chiappella reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study and personal fees from Celgene-BMS, Clinigen, Gilead Sciences, Janssen, Roche, Takeda, AstraZeneca, Incyte, and Novartis outside the submitted work. F. Nanetti reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study. M. Pennisi reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study. A. Guidetti reports personal fees from Gilead during the conduct of the study; personal fees from GSK, nonfinancial support from Janssen, and nonfinancial support from Roche outside the submitted work. P. Corradini reports personal fees from Gilead, Novartis, Celgene BMS, Sanofi, Incyte, and Takeda outside the submitted work. C. Carniti reports grants from ACC, AIL, AIRC, and the Italian Ministry of Health during the conduct of the study. No disclosures were reported by the other author.

C. Monfrini: Data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. F. Stella: Data curation, formal analysis, validation, investigation, writing–original draft, writing–review and editing. V. Aragona: Formal analysis, writing–original draft. M. Magni: Data curation, validation, investigation, visualization, writing–original draft, writing–review and editing. S. Ljevar: Data curation, formal analysis. C. Vella: Methodology. E. Fardella: Writing–review and editing. A. Chiappella: Clinical data collection. F. Nanetti: Writing–review and editing. M. Pennisi: Writing–review and editing. A. Dodero: Clinical data collection and analysis. A. Guidetti: Clinical data collection and analysis. P. Corradini: Conceptualization, resources, supervision, funding acquisition, writing–review and editing. C. Carniti: Conceptualization, supervision, validation, writing–original draft, project administration, writing–review and editing.

This work has been supported by Alleanza Contro il Cancro, Associazione Italiana contro le Leucemie-Linfomi e mieloma, Associazione Italiana Ricerca sul Cancro, and the Italian Ministry of Health (Ricerca Finalizzata). The authors would like to thank the patients who made this research possible; Filippo Bagnoli and Vincenzo Marasco for critical revision of the manuscript; Paolo Longoni for technical assistance; Ilaria Lo Russo and Chiara Ghidoli for the collection of blood samples; and the study coordinators Debora Degl'Innocenti and Anisa Bermema.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.