T-cell Exhaustion in Multiple Myeloma Relapse after Autotransplant: Optimal Timing of Immunotherapy

High-dose chemotherapy followed by autologous stem cell transplantation (ASCT) can produce complete responses (CR) in up to one third of patients with multiple myeloma (1). Nevertheless, most patients achieving a CR after ASCT eventually relapse, and in patients who fail to attain a CR, progression of disease is inevitable (2). This has remained true even with lenalidomide maintenance therapy, which is now standard after ASCT and extends progression-free survival from approximately 2 to 4 years (3–5).

The immune system participates in the control of multiple myeloma, whereas compromised immunity contributes to its evolution. Myeloma-reactive T cells are present in active disease (6–9) and correlate with disease burden (9). The expansion of T-cell clones after ASCT (10) and the emergence of antigen-specific T cells after allogeneic stem cell transplantation (11, 12) are associated with improved clinical outcomes. The loss of tumor-specific T cells characterizes progression from the benign precursor condition, monoclonal gammopathy of undetermined significance (MGUS), to multiple myeloma (13, 14). Malignant plasma cells themselves also evade immune surveillance by various mechanisms (15–19), including the upregulation of PD-L1 (17–19).

Incorporating immune-based therapies into post-ASCT treatment regimens to induce or restore antitumor immunity offers a promising approach to target residual multiple myeloma. The minimal residual disease state and lymphopenia after ASCT afford a unique platform for promoting antitumor immune responses by limiting tumor-driven immunosuppression (20), eliminating cytokine sinks (21), and transiently depleting regulatory T cells (Treg; ref. 22). Posttransplant reconstitution of immune cell subsets, however, occurs with disparate kinetics that can affect the outcome of immunotherapy. The immunomodulatory effect of lenalidomide maintenance therapy on the dynamics of immune recovery also remains undefined.

The rational development of immunotherapeutic interventions after ASCT, in which relapse remains the primary cause of treatment failure, requires a comprehensive understanding of the immunologic milieu. We therefore performed a prospective analysis of immune reconstitution in 55 multiple myeloma patients undergoing ASCT and lenalidomide maintenance therapy to define patterns of lymphocyte recovery and to identify immunologic markers of relapse, which could elucidate potential targets for converting durable responses to long-term cures.

Fifty-five patients were evaluated (Table 1). All patients received pre-ASCT induction regimens that included bortezomib and/or lenalidomide, with 40 (72%) patients receiving both agents. All patients were conditioned for ASCT with high-dose melphalan (140 mg/m² or 200 mg/m², depending on age and comorbidity risk) and received post-ASCT lenalidomide maintenance therapy. Disease status was assessed for response to induction therapy within 30 days before ASCT, at 3 and 12 months after ASCT, and at the time of relapse where applicable. Myeloma response was assessed using International Myeloma Working Group (IMWG) criteria (23).

Peripheral blood mononuclear cells (PBMC) were obtained by centrifugation over Ficoll-Paque PLUS (GE Healthcare) from peripheral blood or leukocyte concentrates from patients and healthy volunteers. Biospecimen sample collection and use adhered to protocols approved by the Institutional Review and Privacy Board of Memorial Hospital, Memorial Sloan Kettering Cancer Center (MSKCC). Leukocyte concentrates (buffy coats) purchased from the Greater New York Blood Center, American Red Cross, were also used as a source of healthy donor cells.

PBMCs were incubated with fluorochrome-conjugated mAbs and analyzed on either an FC 500 (Beckman Coulter) or an LSRFortessa (Becton Dickinson) flow cytometer. FITC-, PE-, PE-Texas Red–, ECD-, APC-, PE-Cy5–, PE-Cy7–, PerCP-Cy5.5–, Pacific Blue–, and AF700-conjugated mouse anti-human mAbs included anti-CD3, anti-CD4, anti-CD8, anti-CD11c, anti-CD14, anti-CD16 (clone 3G8), anti-CD19, anti-CD25, ant-CD28, anti-CD45RA, anti-CD45RO, anti-CD80, anti-CD86, anti-CD123, anti–CTLA-4, anti–HLA-DR, anti-IL2, anti–Ki-67 (BD Pharmingen), anti-CD56, anti-CD83 (Beckman Coulter), anti-CD127, anti-IFNγ, anti–LAG-3, anti–PD-1, anti-TNFα (eBioscience), anti-CCR7, anti–TIM-3 (R&D Systems), and anti-CD57 (BioLegend). Nonreactive isotype-matched antibodies (Becton Dickinson; eBioscience; R&D Systems) were used as controls. The LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Life Technologies) facilitated exclusion of dead cells. Gates were set for collection and analysis of at least 20,000 live events. Data were analyzed with FlowJo 9.5 software (TreeStar).

Monocyte-derived dendritic cells (MoDC) were generated from PBMCs using media, media supplements, and cytokines, exactly as published (24). Xenogeneic plasma or serum (e.g., FCS) was never used in any cultures. T cells were obtained from tissue culture plastic-nonadherent PBMCs, further purified by elution from nylon wool columns (Polysciences), achieving >90% purity.

MoDCs were added in serial doses (1:30 to 1:3,000, moDC:T) to triplicate wells of 1 × 10⁵ allogeneic T cells in 96 round-bottomed well plates (Corning Life Sciences). Final volume was 100 μL/well in RPMI with 10% heat-inactivated, pooled normal human serum. For PD-1 blockade, cells were cultured at a 1:30 moDC:T-cell ratio with nivolumab (10 μg/mL final; Bristol-Myers Squibb) or IgG4 isotype control added at culture initiation. Responder allogeneic T-cell proliferation was measured after 5 days by a colorimetric assay according to the manufacturer's instructions (CellTiter96 Aqueous One Solution Cell Proliferation Assay MTS; Promega) or by intracellular staining with Ki-67 (BD Pharmingen).

WT1 mRNA transcription was performed exactly as published (25, 26). Immature moDCs were electroporated with WT1 mRNA on days 5 to 6. After electroporation, cells were immediately transferred to culture and terminally matured by exposure to inflammatory cytokines for 48 hours (24).

Mature WT1 mRNA–electroporated autologous moDCs were added in serial doses to triplicate wells containing 1 × 10⁵ T cells in a 96 round-bottomed well plate (Corning Life Sciences). Final volume was 100 μL/well of RPMI-10% heat-inactivated, autologous serum, supplemented with recombinant human IL15 (10 ng/mL; R&D Systems). After 7 days of moDC-T cell culture, 5 × 10³ target cells were added directly to each well, and cytolytic activity exerted by responder T lymphocytes was assessed after 4 to 6 hours with a colorimetric cytolytic T lymphocyte (CTL) assay (27). These data represented the total cytolytic activity generated in each culture according to the primary stimulation conditions, rather than per number of effector T cells irrespective of their frequency in the primary cultures. Target cells were 697 cells (HLA-A*0201⁺, WT1⁺ cell line). SKLY-16 cells (HLA-A*0201⁺, WT1^neg cell line) served as a negative control.

Descriptive and graphical measures were used to characterize leukocyte subpopulation patterns both longitudinally and by disease status following ASCT. Unpaired t tests were used to explore mean differences in expression across disease categories, whereas paired t tests were used to compare across time points and treatment conditions. A P value less than 0.05 was considered statistically significant. All statistical analyses were calculated using Prism 6 software (GraphPad).

We evaluated absolute lymphocyte count (ALC) after ASCT to determine the kinetics of lymphocyte reconstitution. ALC nadir occurred at day +5, followed by early recovery at day +12 (Fig. 1A) and complete recovery by day +30 (Fig. 1B). Reconstitution of CD8⁺ T cells, however, outpaced that of CD4⁺ T cells, most likely due to the homeostatic proliferation of peripheral T cells that phenotypically resemble memory cells after chemotherapy-induced lymphopenia (28). This resulted in an inverted CD4/CD8 ratio lasting up to 1 year (Fig. 1B). CD4⁺CD45RO⁺ memory T cells represented the majority of CD4⁺ T cells at day +12 (Fig. 1C; 61.11% ± 3.27%), whereas CD4⁺CD45RA⁺ naïve T cells remained low at 1 year (Fig. 1C; 10.13% ± 1.5%). CD8⁺CCR7^negCD45RO⁺ effector memory and CD8⁺CCR7⁺CD45RO⁺ central memory cells comprised the majority of CD8⁺ T cells at day +12 (Fig. 1D; 39.26% ± 2.8% and 35.75% ± 3.15%, respectively), with low levels of CCR7⁺CD45RO^neg naïve CD8⁺ T cells present at 1 year (Fig. 1D; 8.81% ± 1.79%). Natural killer (NK) cells (CD3^negCD56⁺CD16^neg and CD3^negCD56^dimCD16⁺) exhibited rapid and sustained recovery after ASCT (Fig. 1E). The recovery of CD19⁺ B cells lagged in comparison with the other lymphocyte subsets but recovered by 3 months (Fig. 1E). Plasmacytoid dendritic cells (CD123⁺DR⁺CD11c^neg) were present at similar levels before and after ASCT (Fig. 1E). Subgroup analysis based on 3-month post-ASCT disease response (i.e., partial response vs. very good partial response vs. CR) revealed no statistically significant differences in the pattern of lymphocyte reconstitution between groups (data not shown).

The balance between Tregs and effector T cells shapes antitumor immune responses and the efficacy of immune-based interventions (29). We compared CD3⁺CD4⁺CD25^brightCD127^neg Tregs with CD3⁺CD8⁺CD25⁺ effector T cells after ASCT. As shown in Fig. 1F, the Treg:CD8⁺ effector T-cell ratio at day +12 (0.59 ± 0.21) was significantly lower than before transplant (1.04 ± 0.23; P < 0.05) or day +30 after transplant (1.51 ± 0.27; P < 0.001). Tregs therefore decline early after nadir as CD8⁺ T-cell recovery occurs, resulting in a markedly lower Treg:CD8⁺ effector T-cell ratio and providing a critical early window for the introduction of immune-based posttransplant consolidation therapies.

In the nontransplant setting, there are reports of defective dendritic cell (DC) function in multiple myeloma (30, 31). To evaluate the integrity of DCs from patients after transplant, moDCs were generated from PBMCs (24) from patients in CR 3 months after transplant, patients who relapsed after transplant, and healthy donors. DCs were assessed by flow cytometry for the upregulation of the prototypical DC maturation marker, CD83 (32). Neither posttransplant status nor disease relapse impaired terminal, inflammatory cytokine–induced maturation (Fig. 2A). We then evaluated the functional capacity of these moDCs in the allogeneic mixed leukocyte reaction (alloMLR), which is a standard assay of overall DC immunogenicity (24). MoDCs from each group stimulated comparably vigorous proliferation of allogeneic T cells (Fig. 2B).

We also assessed the generation of antigen-specific CTLs by mRNA-electroporated moDCs from patients who had undergone ASCT for multiple myeloma. MoDCs were electroporated with WT1 mRNA and then matured and activated with inflammatory cytokines for 48 hours (25, 26). The resulting mature moDCs were added in serial 3-fold dilutions to a fixed number of purified autologous T cells obtained before and after ASCT (days +12, 30, and 90). Cultures were supplemented with recombinant human IL15, because unlike CD34⁺ HPC-derived Langerhans-type DCs, moDCs do not provide sufficient endogenous IL15 for CTL stimulation (25). Autologous WT1 mRNA–electroporated moDCs plus IL15 stimulated potent antigen-specific CTLs after only 7 days of stimulation, demonstrating that active cellular immune responses can be elicited in vitro as early as day +12 from multiple myeloma patients after ASCT (Fig. 2C). We therefore conclude that the DC compartment causes no alteration in immune responsiveness in multiple myeloma patients after ASCT.

Immune checkpoint pathways can impede antitumor immune responses and contribute to persistent and/or relapsed malignancy (33). To assess this key regulatory axis in the setting of ASCT, we compared T cells before and after ASCT for expression of the inhibitory receptors CTLA-4 (Fig. 3A), PD-1 (Fig. 3B), LAG-3 (Fig. 3C), and TIM-3 (Fig. 3D). CD4⁺ and CD8⁺ T cells maintained the expression of each inhibitory receptor at 3 and 12 months after transplant, with CTLA-4 and PD-1 showing higher overall expression than LAG-3 and TIM-3. These findings provide a rationale for the inclusion of checkpoint inhibition to augment T-cell responses after ASCT.

Treg expansion has been implicated in multiple myeloma pathogenesis (34, 35) and is associated with inferior survival (36). Early NK-cell reconstitution correlates with improved progression-free survival after ASCT (37), and impaired NK-cell function correlates with multiple myeloma progression (38). We compared patterns of Treg and NK-cell content in patients who remained in a continuous CR 1 year after ASCT with a subset of patients who initially achieved a CR at 3 months but subsequently relapsed after ASCT (mean time to relapse: 12.3 months; range: 5–21 months). The fold change in Tregs was significantly greater in relapsed compared with nonrelapsed patients, 1.65 ± 0.25 and 0.94 ± 0.13, respectively (Fig. 4A). Treg:CD8⁺ effector T-cell ratio was also increased in relapsed compared with nonrelapsed patients, but did not reach statistical significance (data not shown). The fold increase in NK cells was greater in patients remaining in a continuous CR compared with relapsed patients, 2.98 ± 0.63 and 1.38 ± 0.42, respectively (Fig. 4B).

T-cell exhaustion and senescence occur under conditions of chronic antigen stimulation and contribute to impaired immune responses (39, 40). Although investigators have not fully distinguished the molecular characteristics of exhaustion and senescence (41), cell surface markers can be used to segregate these T-cell states phenotypically. Downregulation of the costimulatory marker, CD28, is a feature of both exhaustion and senescence (41). Upregulation of the glycoepitope CD57 can be used to identify T cells with low proliferative capacity (42). Senescent cells can also express PD-1, a prototypical marker of exhaustion (39, 40). Using these markers, we assessed the pattern of T-cell exhaustion/senescence in the posttransplant setting. As shown in Fig. 5A, patients with relapsed disease after ASCT had a higher percentage of CD8⁺CD28^neg cells (62.94% ± 4.86%), compared with age-matched (mean age, 53.3; range 47–66) healthy donors (23.69% ± 2.03%; P < 0.0001) or patients in a continuous CR after ASCT (36.83% ± 3.95%; P < 0.001). Patients with relapsed disease after ASCT also had a similar increase in the fraction of CD8⁺CD28^negCD57⁺ (Fig. 5B) and CD8⁺CD28^negPD-1⁺ (Fig. 5C) T cells. CD4⁺ T cells showed the same trends in the frequencies of CD28^neg, CD28^negCD57⁺, and CD28^negPD-1⁺ cells, albeit at lower levels of expression (Fig. 5D). The expression of the inhibitory receptors, CTLA-4, LAG-3 and TIM-3, was not significantly different between the groups examined (data not shown).

We also compared interval changes in T-cell exhaustion/senescence profiles among patients who initially achieved CRs at 3 months after ASCT, but then either relapsed thereafter or remained in a continuous CR at 12 months. As shown in Fig. 5E, patients with relapsed disease had a higher percentage of exhausted/senescent CD8⁺CD28^neg and CD8⁺CD28^negPD-1⁺ T cells at the time of relapse than did patients remaining in a continuous CR at 12 months. Of note, patients in the relapsed group had higher baseline levels of CD8⁺CD28^neg, CD8⁺CD28^negCD57⁺ and CD8⁺CD28^negPD-1⁺ T cells at the 3-month mark, preceding the onset of clinical relapse. Together, these findings indicate that impaired immune function due to exhaustion and/or senescence may contribute to relapse after ASCT. The senescence/exhaustion T-cell phenotype could also provide a predictive marker supporting earlier intervention to prevent relapse.

Using cells from patients in continuous CR 1 year after ASCT and from patients who initially achieved CR but relapsed beyond 3 months after ASCT, we tested the capacity of CD8⁺CD28^negPD-1⁺ T cells to respond to stimulation by moDCs derived from healthy donors in alloMLRs as a generic response assay, with and without the PD-1 inhibitor, nivolumab. In all cases, PD-1 inhibition increased CD8⁺CD28^negPD-1⁺ T-cell proliferation (Fig. 6A; 61.33% ± 10.58% vs. 27.82% ± 6.78% for isotype control; P < 0.01). Nivolumab treatment also induced a concomitant increase in the secretion of the cytokines IFNγ, IL2, and TNFα. These findings demonstrate augmented responsiveness of a significant fraction of cells in the setting of PD-1 inhibition, thus underscoring the potential benefit of checkpoint blockade to enhance/restore T-cell responses in this patient population.

This study identifies several key immunologic parameters of immune reconstitution and relapse in multiple myeloma after ASCT. Lymphocyte recovery from high-dose chemotherapy-induced lymphopenia relies primarily on the peripheral expansion of memory cells with only a minor contribution from their thymic-derived naïve counterparts. The early posttransplant period is marked by a transient decline in the Treg-to-effector T-cell ratio, which reverts to pretransplant levels by 1 month. DCs are functionally intact, comparable with those from healthy donors, and induce autologous antigen-specific T cells with robust lytic activity in vitro. CD4⁺ and CD8⁺ T cells retain immune-inhibitory receptor expression after transplant, however. A significantly increased subset of these T cells with an exhausted/senescent phenotype and reversible hyporesponsiveness to PD-1 inhibition characterizes patients who relapse after transplant. Increased Tregs and decreased NK cells further characterize immune dysfunction in these patients.

Consistent with previous reports (28), disruption of the normal balance of immune cells with a bias toward T-cell memory phenotypes, protracted diminution of naïve T-cell output from the thymus, and inversion of CD4⁺/CD8⁺ T-cell ratios characterize the immune reconstitution in our study population. Other groups have shown that T-cell repertoire diversity of expanding lymphocytes after lymphopenia is limited in the absence of adequate thymopoiesis (43, 44). These changes contribute to a state of prolonged posttransplant immune deficiency and predispose patients to infections and compromised antitumor immune surveillance (28), but the specific dynamics of T-cell recovery in the context of our current understanding of exhausted/senescent T cells and checkpoint inhibitors have not been defined.

During initial lymphocyte recovery, the Treg:CD8⁺ effector T-cell ratio is significantly reduced. This shift provides an opportunity to stimulate maximal antitumor responses in the absence of Treg-mediated suppression. DC-based vaccination in this setting offers one approach to redirect recovering T cells toward specific MHC-restricted antigen. Our data show the induction of potent autologous antigen-specific CTLs by moDCs supplemented with IL15 at day +12 after transplant, thus confirming preservation of DC and T-cell function in vitro. Non–DC-based pre-ASCT vaccination supplemented with the adoptive transfer of ex vivo–expanded T cells promotes accelerated reconstitution of immune responses against both microbial and multiple myeloma tumor antigens in vivo (45, 46), supporting the concept of immunotherapy in the setting of posttransplant lymphopenia. We are testing the feasibility of early posttransplant vaccination using autologous Langerhans-type DCs (LC) electroporated with mRNA encoding three multiple myeloma–associated antigens in a phase I clinical trial (NCT01995708). LCs provide sufficient endogenous IL15, obviating the need for IL15 supplementation that moDCs require to induce CTLs (25). Adoptive T-cell infusions should not be necessary in the setting of DC-based vaccines, unless perhaps used in a prime-boost sequence.

Inhibitory receptor blockade augments vaccine-induced T-cell responses (19, 47), reviving antigen-reactive T cells from their exhausted state (39). Our data demonstrate the persistent expression of CTLA-4, PD-1, LAG-3, and TIM-3 by T cells after ASCT, providing a rationale for early inhibition of checkpoint blockade in posttransplant treatment strategies. Such intervention should promote the recognition of multiple myeloma–associated antigens, boost antigen-specific T-cell activation and proliferation, enhance immune reconstitution, expand anti–multiple myeloma immune responses, and counteract relapse from minimal residual disease. Using the alloMLR as a proxy assay for robust immune reactivity, our data in fact demonstrate a role for PD-1 blockade in reversing the hyporesponsiveness of exhausted/senescent CD8⁺ T cells in vitro.

Patients who initially achieved a CR at 3 months but subsequently relapsed after ASCT were assessed for immune markers of relapse. Analysis of Tregs showed a relative increase among the relapsed cohort, a finding that is consistent with recent reports showing greater risk for disease progression with increased Tregs (35) and an association of long-term disease control with decreased Tregs (36). NK cells exhibit rapid recovery after ASCT but are less abundant in patients who relapse, thus supporting their role in conferring an anti–multiple myeloma benefit after ASCT as well (37, 38).

T-cell dysfunction, including exhaustion and/or senescence, contributes to tumor persistence and progression (39, 40). Our studies define a T-cell exhaustion/senescence phenotype associated with post-ASCT relapse, which is most pronounced in the CD8⁺ T-cell compartment. These cells downregulate CD28, an essential costimulatory receptor involved in the activation and modulation of key cellular functions (48), upregulate CD57, a marker of low proliferative capacity (42), and display increased PD-1 expression. Patients with relapsed multiple myeloma had the highest levels of CD8⁺CD28^neg, CD8⁺CD28^negCD57⁺ and CD8⁺CD28^negPD-1⁺ T cells. Patients in the continuous CR group, however, also had higher levels of exhausted/senescent T cells compared with healthy controls, possibly due to chronic malignancy and related treatments.

An intriguing finding is that patients in the relapsed group had higher baseline levels of exhausted/senescent T cells at the 3-month mark before the detection of clinical relapse, suggesting the utility of these immune biomarkers for identifying patients at higher risk of relapse who are candidates for early immunotherapy. Because T cells become exhausted/senescent from chronic antigen stimulation, one might ask whether patients who eventually relapse have enough minimal residual disease persistence, despite high-dose chemotherapy, to provide sufficient chronic antigen stimulation after ASCT to cause T-cell exhaustion/senescence. Alternatively, although it is not a standard practice to characterize the cell composition of autografts, apart from their CD34⁺ content, our findings lend sufficient rationale to evaluating whether exhausted/senescent T cells, as well as Tregs, are over-represented in the autografts of patients more likely to relapse.

Results from many studies suggest that exhaustion and senescence are mechanistically distinct, but the underlying molecular pathways differentiating the two have yet to be fully elucidated (41). Distinguishing T-cell exhaustion from senescence and delineating the plasticity and function of these cells remain important unknowns. Exhausted T cells can be reactivated by blocking the PD-1 pathway (49). Senescent T cells can transiently upregulate telomerase activity and proliferate (40), and blockade of p38 MAPK reverses senescence via an mTOR-independent mechanism (50). In this study, PD-1 inhibition significantly enhanced the proliferation and cytokine secretion of CD8⁺CD28^negPD-1⁺ T cells, demonstrating the potential to revive these hyporesponsive cells with checkpoint blockade. Additional studies of this T-cell subpopulation, including the contribution of other inhibitory receptors and immunosuppressive factors, are warranted to gain a more complete understanding of the biology of these cells with regard to post-ASCT disease status and immunotherapy.

These results provide a rationale for the early introduction of immunotherapeutic modalities like vaccines and checkpoint blockade agents to induce antitumor immunity after ASCT. Our findings also underscore the contribution of immune dysregulation to disease relapse, with T-cell exhaustion/senescence as an immune biomarker that deserves validation by prospective testing because of its potential value in identifying candidates for early therapeutic intervention. The reservoir of DC immunostimulatory function and the potential reactivity of T cells and NK cells comprise an untapped resource to alter the natural history of multiple myeloma in the setting of ASCT.

A.M Lesokhin has received speakers bureau honoraria from and is a consultant/advisory board member for Bristol Myers-Squibb and Janssen. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D.J. Chung, A.M. Lesokhin, S.A. Giralt, J.W. Young

Development of methodology: D.J. Chung, S. Sharma, S.A. Curran, J.W. Young

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.J. Chung, J.A. Shyer, S. Sharma, A.M. Lesokhin, J.W. Young

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.J. Chung, K.B. Pronschinske, J.A. Shyer, S. Sharma, S.M. Devlin, S.A. Giralt, J.W. Young

Writing, review, and/or revision of the manuscript: D.J. Chung, S.A. Curran, A.M. Lesokhin, S.M. Devlin, S.A. Giralt, J.W. Young

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.J. Chung, K.B. Pronschinske, J.A. Shyer, S. Sharma, J.W. Young

Study supervision: D.J. Chung, J.W. Young

The authors gratefully acknowledge the various contributions of members of the Laboratory of Cellular Immunobiology to the development of this work. They thank the nurses, advanced practice providers, and physicians of the Adult Bone Marrow Transplant and Myeloma Services at MSKCC for assistance with sample procurement. They also thank the patients and healthy volunteers who provided samples for research.

This work was supported by Mr. William H. Goodwin and Mrs. Alice Goodwin and the Commonwealth Foundation for Cancer Research and The Experimental Therapeutics Center of Memorial Sloan Kettering Cancer Center (to D.J. Chung), The Society of Memorial Sloan Kettering (to D.J. Chung), Cycle for Survival (to D.J. Chung and A.M. Lesokhin), P30 CA008748 from the NCI, NIH (to S.M. Devlin), Thomas Israel Myeloma Research Fund (to S.A. Giralt), U10HL069294 from the National Heart, Lung, and Blood Institute and the NCI, NIH (to S.A. Giralt), Swim Across America (to J.W. Young), and P01 CA23766 from the National Cancer Institute, NIH (to J.W. Young). This research was also funded in part through the NIH/NCI Cancer Center Support Grant P30 CA008748.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.