Research ArticleCancer

The safety and clinical effects of administering a multiantigen-targeted T cell therapy to patients with multiple myeloma

See allHide authors and affiliations

Science Translational Medicine  29 Jul 2020:
Vol. 12, Issue 554, eaaz3339
DOI: 10.1126/scitranslmed.aaz3339

Fighting multiple myeloma with multitargeted T cells

Adoptive cellular therapies are being developed to treat many types of cancer, including multiple myeloma (MM). Lulla et al. performed a clinical study using T cells expanded in response to MM target antigens. The cells were used as adjuvant therapy or in high-risk patients. Therapy was well tolerated and induced clinical responses in a subset of patients. Examination of T cell clonal expansion and antigen loss in MM cells aided understanding of how the multiantigen-targeted T cells were behaving in patients. This approach, which relies on native T cells and does not require expensive or time-consuming gene editing required for some other types of cellular therapy, warrants further investigation for fighting MM.


Multiple myeloma (MM) is an almost always incurable malignancy of plasma cells. Despite the advent of new therapies, most patients eventually relapse or become treatment-refractory. Consequently, therapies with nonoverlapping mechanisms of action that are nontoxic and provide long-term benefit to patients with MM are greatly needed. To this end, we clinically tested an autologous multitumor-associated antigen (mTAA)–specific T cell product for the treatment of patients with high-risk, relapsed or refractory MM. In this study, we expanded polyclonal T cells from 23 patients with MM. T cells whose native T cell receptors were reactive toward five myeloma-expressed target TAAs (PRAME, SSX2, MAGEA4, Survivin, and NY-ESO-1) were enriched ex vivo. To date, we have administered escalating doses of these nonengineered mTAA-specific T cells (0.5 × 107 to 2 × 107 cells/m2) to 21 patients with MM, 9 of whom were at high risk of relapse after a median of 3 lines of prior therapy and 12 with active, relapsed or refractory disease after a median of 3.5 prior lines. The cells were well tolerated, with only two transient, grade III infusion-related adverse events. Furthermore, patients with active relapsed or refractory myeloma enjoyed a longer than expected progression-free survival and responders included three patients who achieved objective responses concomitant with detection of functional TAA-reactive T cell clonotypes derived from the infused mTAA product.


In recent years, several new therapies such as proteasome inhibitors, immunomodulatory drugs, and monoclonal antibodies (mAbs) have prolonged survival of patients with multiple myeloma (MM), but cure remains elusive (14). Most patients eventually relapse or become refractory to treatment, underscoring the need for new therapies.

Adoptive cellular therapy using chimeric antigen receptor (CAR) T cells has demonstrated unprecedented success in the treatment of CD19+ malignancies (58). MM has also proved amenable to CAR T cells (915), and targeting the B cell maturation antigen (BCMA) has produced response rates of >80% in patients with multiply relapsed disease (1216). Unfortunately, despite high initial response rates, disease relapse after CAR T cell therapy is an emerging issue, occurring at a median of ~12 months after treatment (12), thereby highlighting the need for newer therapies with complementary mechanisms of action to sustain or prolong clinical benefit.

We have developed a nonengineered immunotherapeutic approach that relies on the administration of ex vivo expanded autologous T cell lines that simultaneously target a spectrum of MM-expressed TAAs {PRAME, SSX2, MAGEA4, NY-ESO-1, and survivin [multitumor-associated antigen (mTAA) T cells]}. Here, we describe the safety profile of our multiantigen-specific T cell therapy administered to 21 patients with MM. In addition to establishing the feasibility and lack of toxicity of mTAA T cells in patients with MM, we demonstrate a direct correlation between clinical effects and the post-infusion expansion and persistence of functional TAA-reactive T cell clonotypes derived from the infused product.


Patient characteristics

Twenty-three patients with a diagnosis of MM were eligible to participate on the study, and mTAA-specific T cell lines were successfully generated from all individuals with no manufacture failures. Of this group, two patients were not infused—one who withdrew consent and another who developed rapidly progressive disease and chose hospice care. Of the 21 patients who were treated, 9 were infused as adjuvant therapy while in remission after their last line of treatment (median, 3 prior lines of therapy; range, 2 to 7) (group A, Table 1), whereas the remaining 12 received mTAA T cells to treat relapsed or resistant disease after receipt of a median of 3.5 lines (range, 1 to 10) of prior therapies (group B, Table 2). For patients with available pretreatment biopsy samples, all tumors expressed a minimum of two and as many as all five of the target TAAs, as confirmed by immunohistochemistry (IHC) (Tables 1 and 2).

Table 1 Patient characteristics (group A).

ID, patient number; G, gender; M, male; F, female; Std., standard cytogenetic risk [not 17p/tp53 altered or t(4;14), t(14;16)]; hetero, heterozygous p53 mutation/deletion; R-ISS, Revised Myeloma International Staging System; Imid, immunomodulatory class of drugs; PI, proteasome inhibitor class of drugs; MoAb, monoclonal antibody class of drugs; ASCT, number of prior autologous hematopoietic stem cell transplantations; DL, dose level; Y, yes; N, no.

View this table:
Table 2 Patient characteristics (group B).

M-spike, serum or urine myeloma paraprotein quantification by electrophoresis; Std., standard risk [(not 17p/p53 altered or t(4;14), t(14;16)]; quant., quantification.

View this table:

Specificity and clonality of ex vivo expanded mTAA T cells

To manufacture mTAA T cells per our previously published report (17), we exposed peripheral blood mononuclear cells (PBMCs) to pepmix-loaded dendritic cells (DCs) followed by expansion in T cell medium supplemented with T helper 1 (TH1)–polarizing, proproliferative, and prosurvival cytokines. At the time of cryopreservation, mTAA T cells were almost exclusively CD3+ T cells (median, 98.5%; range, 65.3 to 99.7%), with a mixture of CD4+ (26.91%, 2.4 to 93.4%) and CD8+ (56.8%, 3.1 to 94.7%) subsets that expressed both central (CD45RO+/62L+/CCR7+, 0.19%; 0 to 1.4%) and effector (CD45RO+/62L/CCR7, 59%; 4.2 to 88.6%) memory markers (Fig. 1A). We used T cell receptor, complementary determining region 3 (CDR3-TCR) deep sequencing to analyze the clonal diversity of a subset (n = 10 of 23) of the products manufactured for clinical use (limited to residual material available for patients who had completed protocol-specified infusions with over 6 months of follow-up). As shown in Fig. 1B, the mTAA T cell lines were polyclonal with a mean of 4597 individual clones (range, 890 to 13,995), of which 79.5% (median; range, 59 to 95%; fig. S1A) were uniquely identified in the ex vivo expanded product and not in matched patient peripheral blood samples before infusion, thus enabling in vivo tracking of the infused product. To assess the potential for “off-tumor” effects, we measured cytotoxicity against patient-derived phytohemagglutinin (PHA)–stimulated blasts. As shown in Fig. 1C, there was no autoreactivity against nonmalignant cells [mean 2 ± 0.5% specific lysis; effector-to-target (E:T) ratio, 20:1; n = 21], with <10% specific lysis at an E:T ratio of 20:1 set as a study release criterion, which was met by all products.

Fig. 1 Characterization of autologous mTAA-specific T cell products.

Phenotype (A), individual patients assigned a unique symbol (median indicated by solid black line; n = 23). (B) Total number of individual TCR clonotypes (by vβ-TCR DNA deep sequencing, each symbol represents an individual patient’s T cell product, n = 10). (C) Assessment of reactivity to autologous targets at an E:T ratio of 20:1 of 23 manufactured mTAA T cell lines. (D) TAA-directed specificity as measured by ELIspot for all 23 T cell products. Data are means ± SEM, and each color represents each of the five target TAAs. (E) Representative scRNAseq data for patient #8 showing t-SNE plot of TAA-specific IFNG-expressing cells (top) and dot plot (bottom) for expression of IFNG, PRF1, TNF, and GZMB in TAA-stimulated and unstimulated cells. (F) Detection of IFN-γ–producing CD4+ and CD8+ T cells by ICS. (G) Frequency of TAA-reactive clonotypes (confirmed by scRNAseq) per product analyzed. (H) Proportion of TAA-reactive clonotypes that are “unique” to the product (not detectable in the patient before infusion) and proportions of CD4+ and CD8+ TAA-reactive clonotypes per product (n = 8). (I) Cytolytic effects of mTAA T cells against TAA-pulsed autologous cells versus unpulsed controls (n = 7).

The functional specificity of all mTAA T cells was determined by interferon-γ (IFN-γ) enzyme-linked immunospot (ELIspot) assay after stimulation with overlapping peptide libraries or “pepmixes” spanning PRAME, SSX2, MAGEA4, NY-ESO-1, and Survivin (Fig. 1D and fig. S1B). Of the 23 lines generated, 7 had activity [defined as ≥10 spot-forming cells (SFCs)/2 × 105 cells] against all 5 stimulating antigens, 3 lines recognized 4 antigens, 2 lines were trivalent, 1 line was divalent, 4 lines were monovalent, and 6 lines were specific for multiple antigens (1 to 3) below our defined activity threshold. Furthermore, in a subset of lines (n = 8, limited to those with leftover material and with available DNA deep sequencing data), we were able to correlate antigen-specific function with T cell product–derived CDR3-TCR clonotypes using single-cell RNA sequencing (scRNAseq) (Fig. 1E and fig. S1C). Figure 1E shows data for a representative donor, where stimulation with tumor-derived pepmixes identified a population of T cells (top) transcribing genes for multiple effector molecules including IFN-γ, tumor necrosis factor–α (TNF-α), granzyme B, and perforin (bottom). Furthermore, these results were confirmed at the protein level by intracellular cytokine staining (ICS) for IFN-γ, which revealed that reactive T cells were detected in both CD4+ and CD8+ subsets (Fig. 1F). Similar results for seven additional patients are shown in fig. S1C. We determined that up to 36.7% of the sequenced mTAA products were functional based on IFN-γ transcripts (scRNAseq) upon antigen exposure (n = 8, Fig. 1G). Of these, between 12 and 92% (median 73%) represented clones that were unique to the product infused (Fig. 1H). Last, to assess the cytolytic potential of these polyfunctional, TAA-reactive T cells in vitro, we cocultured the T cell products with autologous antigen-loaded targets at a range of E:T ratios. As shown in Fig. 1I, our mTAA-specific T cells were selectively able to kill tumor-expressing targets (mean 38 ± 8.9% specific lysis; E:T, 40:1; n = 7), with minimal activity against autologous controls (mean 9.6 ± 2.8% specific lysis; E:T, 40:1; n = 7).

Safety and clinical responses

Safety. Eighty-five infusions were administered to 21 patients [35 at dose level 1 (DL1), 29 at DL2, and 21 at DL3]. During the dose-limiting toxicity (DLT) observation period, there were 12 adverse events (AEs) (1 at DL1, 11 at DL2, and none at DL3) potentially (defined as “unlikely,” “possibly,” “probably,” or “definitely”) related to the infusions. Of these, only two events (one leukopenia and one thrombocytopenia) seen in the same patient (patient #15, Table 2) were grade III or higher. This patient had baseline leukopenia (at grade II) and thrombocytopenia (at grade III) before infusion that transiently worsened after infusion but subsequently recovered to baseline at 8 and 3 days after infusion, respectively, without the need for growth factor support or transfusions (Table 3). Three patients were hospitalized during the DLT observation period because of a viral bronchitis (n = 1), dehydration from an anxiety attack (n = 1), and transient renal failure (n = 1) due to lisinopril use for persistent hypertension. All three patients were subsequently discharged after a brief hospital stay (1 to 5 days), and the events were deemed unrelated to the study (Table 3). No autoreactivity syndromes, cytokine release syndrome, or immune effector cell–associated neurotoxicities were seen. In summary, infusions of mTAA T cells at all cell doses were deemed safe (<20% DLT rate).

Table 3 Safety profile of mTAA T cells for MM.

View this table:

Clinical responses on study. At the 8-week post-infusion disease assessment, all nine patients in complete remission (CR) at the time of mTAA T cell infusion (group A, adjuvant group) remained in continued complete remission (CCR). On long-term follow-up, only two have relapsed at months 7 and 13 after infusion, respectively, while the remaining patients remain in CCR at a median follow-up of 27.5 months (Fig. 2A and table S1). Five of the nine patients in group A started (n = 3) or continued (n = 2) with lenalidomide maintenance (≤10 mg/day) without any other MM therapies (illustrated in Fig. 2A). Patient #6 subsequently discontinued maintenance therapy (at month 6) because of intolerance.

Fig. 2 Clinical outcomes of infused patients.

Swimmers’ plots depicting outcomes after infusion as well as use and duration of maintenance therapy in patients with MM who were in remission at the time of infusion, group A (A), and those who had active myeloma at the time of infusion, group B (B). (C) Kaplan-Meier estimates of progression-free survival (PFS) in patients (n = 10) who were treated with active myeloma with at least 8-week follow-up (group B) and (D) individual clinical responses as estimated by peak fold change in M protein (involved paraprotein) concentrations (y axis), by the dose of infused T cells (DL1 to DL3, per protocol) the patient was treated on (x axis), within the first 12 months after infusion, before institution of next line of therapy among group B patients.

At the 8-week post-infusion assessment, of the 12 patients infused to treat active disease (group B, persisting or progressive after their last line of therapy; Table 2), 2 were early post-treatment (<8 weeks) at the time of censoring and had not yet undergone disease assessment studies, 9 had stable disease (SD), and 1 achieved a CR by 8 weeks after infusion without other MM-directed therapies including maintenance therapy. Of the 10 patients with longer-term follow-up, 1 was lost to follow-up but was in SD at the time of last assessment (12 months after infusion), 7 experienced disease progression between 3 and 31 months after infusion, whereas the remaining 2 patients continue to have evidence of a response [patient #17, SD at 19 months, and patient #5, ongoing partial response (PR) at 46 months after infusion] for a median progression-free survival (PFS) of 22 months for group B patients who have completed at least 8 weeks on study (Fig. 2, B and C, and table S2). Four of the 10 patients started (3, lenalidomide) or continued (1, pomalidomide) maintenance doses (≤10 mg of lenalidomide or ≤3 mg of pomalidomide without dexamethasone) of immunomodulatory drugs without any other MM therapies (depicted in Fig. 2B). Figure 2D illustrates the best response seen within the first 12 months after infusion and before starting another line of therapy for progressive MM. These data are reported as the fold change in serum M protein concentration from baseline of 10 patients (who completed at least 8 weeks of post-infusion follow-up) treated in the active disease group, 7 of whom had a reduction in serum M protein ranging from 11 to 80%.

Outcomes for patients who progressed or relapsed post-mTAA T cells

Nine of the 21 treated patients have progressed (7 in group B) or relapsed (2 in group A) after treatment, and table S3 shows their outcomes after progression/relapse. None of the patients who have relapsed or progressed after mTAA-specific T cells achieved a durable response to subsequent lines of therapy and most have died of disease progression.

Factors associated with response and immune escape

There was no evidence of a cell dose–dependent effect on either achievement or duration of response in either group. There was no difference in the expansion or persistence of infused T cells as measured by ELIspot and TCR clone tracking (by both scRNA and DNA deep sequencing) across the three DLs or when comparing those who had an objective response to those whose best response was SD in group B. However, when examining the in vivo T cell behavior in patients who received cells to treat active disease (group B) compared with the adjuvant cohort (group A), there were differences noted. Specifically, in patients infused as adjuvant therapy and lacking antigenic (tumor-derived) stimulation, in vivo T cell expansion was modest in comparison to that seen in patients with active disease. This difference was demonstrated by changes in TAA reactivity (1.31- to 9.6-fold mean increase compared with 2.8- to 28.7-fold mean increase in group B based on ELIspot detection of antigen-specific responses; Fig. 3, A and B, and fig. S2, A and B). Tracking scRNAseq-confirmed TAA-reactive clonotypes [median fold change from baseline group A (0.8-fold; range, 0.7- to 1.1-fold) versus group B (1.7-fold; range, 0.99- to 2.5-fold); Fig. 3, C and D] and tracking mTAA product-derived “unique clones” based on deep sequencing [increase from 0% at baseline to group A (0.1%; range, 0.1 to 0.3% total repertoire) versus group B (0.8%; 0.1 to 1.24%); fig. S2C and fig.S2D] confirmed these findings. A similar trend was also noted in the bone marrow (fold change of confirmed TAA-reactive clonotypes by scRNAseq: group A: median, 1.15-fold increase; range, 1.0- to 1.3-fold; and unique clonotypes of median 0.48%, range 0.46 to 0.5% marrow repertoire versus group B: median, 1.8-fold; range, 0.9- to 3-fold; and unique clonotypes of median 0.8%, range 0.2 to 2.2%; Fig. 3, E and F, and fig. S2, E and F).

Fig. 3 In vivo properties of mTAA-specific T cells.

(A and B) Expansion of mTAA-specific T cells after infusion in patients with active disease compared to those without disease at the time of T cell infusion as measured by frequency (±SEM) of functional TAA-specific T cells in peripheral blood based on an IFN-γ ELIspot assay after overnight stimulation with TAA pepmixes. (C to F) In vivo T cell tracking of functional TAA-reactive clonotypes (identified by scRNAseq) in peripheral blood and bone marrow. Reported as fold change in productive frequency (in blood or marrow) from baseline (pre). Each symbol represents an individual patient, and error bars depict standard deviation.

Figure 4 shows the active disease responder (patient #2) who achieved a CR to our therapy. Before mTAA T cells, this patient had evidence of bone marrow disease (20% monoclonal CD138+ plasma cells that costained positive for MAGE expression; Fig. 4, A and B), which resolved after T cell treatment (Fig. 4, C and D). This clinical response correlated with a progressive increase in the circulating frequency of tumor-specific T cells (primarily directed against the tumor-expressed MAGEA4 antigen) in the periphery (Fig. 4E) and in the marrow (Fig. 4F), which were confirmed to be derived from the T cell line infused based on deep sequencing analysis (Fig. 4, G and H). Eventually, patient #2’s disease relapsed 2 years after initial mTAA treatment, with CD138+ tumor cells that were now MAGE negative (Fig. 4, I and J) and thus nonstimulatory to circulating MAGEA4-specific T cells (Fig. 4, K and L). This pattern of elimination of tumor antigen–positive MM cells post-mTAA T cells was observed in multiple patients (fig. S3), demonstrating the selective pressure exerted by the infused cells when administered to treat a heterogeneous disease.

Fig. 4 Elimination of MM cells and induction of antigen loss by TAA-specific T cells.

IHC staining depicting CD138+ (A) and MAGE+ (B) plasma cells in the bone marrow of patient #2 before infusion and at 8 weeks after infusion. (C and D) Frequency of TAA-specific T cells in the peripheral blood (E) and within the bone marrow after infusion (F). Detection of mTAA line-derived unique clones after infusion in the blood (G) and marrow (H). IHC staining depicting CD138+ (I) and MAGE+ (J) plasma cells at the time of disease relapse 2 years later. Frequency of TAA-specific T cells in the peripheral blood (K) and bone marrow (L) at the time of relapse.

Patient #3 (also with active disease) similarly responded initially to mTAA T cells with an increase in circulating and bone marrow–infiltrating (line-derived) T cells directed against a range of tumor-expressed antigens (Fig. 5, A to D) as well as stabilization of their paraprotein concentrations. However, this patient experienced clinical progression 6 months after infusion despite persistent expression of target TAAs that should be T cell stimulatory. To investigate the mechanism of immune escape in this patient, we performed mRNA sequencing of the patient’s PBMCs at the time of relapse (6 months), which we compared to those obtained at three earlier time points: before infusion, week 6 after infusion, and month 3 after infusion (when the patient was responding to mTAA T cells). As shown in Fig. 5E, we found several genes known to be selectively expressed in T cells that were differentially expressed at the time of relapse including the T cell checkpoint antigens CTLA4 and LAG3, which were significantly increased relative to earlier time points (q < 0.01), whereas genes synonymous with T cell stimulation/activation such as MS4A1, IL1B, and CD86 were down-regulated. Last, we examined the most differentially expressed genes [DEGs; gene expression fold change of ≥2 or ≤−0.5 and a false discovery rate (FDR) q ≤ 0.01] within bulk PBMCs (which contained circulating plasma cells at the time of progression) collected at the same time points. We identified 1828 DEGs (q ≤ 0.01), of which 406 met criteria for most differentially expressed. They included genes that have previously been linked with tumor cell proliferation including the p53 signaling pathway, c-myc, miR-23b, and genes clustered at 6p22 (Fig. 5F and data files S1 and S2) (18, 19). Unfortunately, this patient ultimately failed all subsequent lines of therapy and succumbed to disease progression 12 months after the initial mTAA T cell infusion (table S3).

Fig. 5 Other mechanisms of relapse.

(A) Preinfusion bone marrow demonstrating plasma cells in patient #3 that are CD138+ (blue) and costain with Survivin, MAGE, and PRAME (red). (B) Frequency of TAA-specific T cells in the peripheral blood at various time points post-infusion as compared with pre-infusion. TAA-specific T cells (C) and unique clones (D) at various time points post-infusion as compared to pre-infusion where available. (E) Changes in the gene expression profile of immunoregulatory genes within PBMCs at the time of relapse compared with earlier time points. (F) Gene expression profile of 1828 genes that are significantly differentially expressed at the time of progression as compared to any prior time point, including 406 genes that are most differentially expressed (fold change of ≥2 and or ≤−0.5 and an FDR q ≤ 0.01). A list of all DEGs and a full list of fragments per kilobase per million mapped (FPKM) values can be found in the Supplementary Materials (data files S1 and S2).


We report here the safety and clinical effects associated with a multiantigen-targeted adoptive T cell therapy, which relies solely on physiological activation without antecedent cytoreductive conditioning therapy. The therapy, which was administered exclusively in an ambulatory setting, not only was deemed to be safe at the tested DLs but also produced clinical responses and sustained remissions using a mechanism of action that does not overlap with other available MM therapies.

One of the major challenges in the field of adoptive immunotherapy has been balancing potency with safety, given that (i) most tumor-expressed antigens are not exclusively present on malignant cells and (ii) massive in vivo T cell activation and expansion can be associated with serious toxicities. For example, the interaction of transgenic T cells with normal tissues that express the target antigen has led to “on-target, off-tumor” toxicities ranging from mild (e.g., lifelong B cell aplasia in patients treated with CD19-CAR T cells) (58) to severe (e.g., deaths after the infusion of HER2 CAR T cells attributed to the presence of the target antigen on lung epithelial cells and subsequent cytokine release syndrome) (20). Furthermore, in a recent phase 1 clinical trial of 33 patients with MM who received pre-conditioning before BCMA-CAR T cell infusions, all experienced toxic side effects (12), with 32 (97%) having events of grade III or higher, albeit reversible in the majority of cases. Any toxicities must be weighed against the overall potential for benefit. In the case of BCMA-CAR T cells, up to 90% of treated patients (most of whom were heavily pretreated) enjoyed clinical responses that were sustained in at least half the treated cohort for ~12 months (12). Nevertheless, complementary therapies that could prolong responses achieved with conventional and emerging MM therapies without introducing additional toxicities are required.

In the current study, using nonengineered T cells reactive against five TAAs, we achieved clinical benefit without either “off tumor” or inflammatory side effects, establishing the safety of infusing a multiantigen-specific T cell product to patients with MM. Although efficacy was not the focus of this trial, we did observe objective clinical responses in three patients: one patient after administration of two doses of mTAA T cells only and two others who received more than two doses of mTAA T cells and immunomodulatory drug maintenance therapy. We acknowledge that the responses documented in the latter two patients are confounded by the concomitant use of lenalidomide, yet we find it notable that patients in both groups neither progressed nor required additional lines of MM therapy for a longer than expected duration [22 months (group B) to not reached after >2 years (group A) versus 10 to 16 months reported in patients with MM failing two or more lines of therapy (2, 3, 21)]. This highlights that patients can benefit from this therapy even if their best response to mTAA T cells is disease stabilization. The reasons for this likely relate to the natural mechanism of action of these T cells that exist in the circulation of patients at subclinically effective frequencies but that are enriched and reactivated through our ex vivo culture process. It has been hypothesized that failure of endogenous TAA-directed T cells in patients with monoclonal gammopathy of undetermined significance might promote transformation to MM (22), and thus, as observed on this trial, restoration of MM-directed immunity was responsible for inducing clinical responses or prolonging time to progression. Ultimately, efficacy-based clinical trials would be required to confirm the clinical activity seen on this study. A key feature of our approach is that it is nonoverlapping with any currently tested immunotherapies for MM and could be tested in sequence or combined with treatments like BCMA-directed immunotherapies to sustain benefit. Furthermore, the safety profile demonstrated in this study supports the application of this approach even in patients who are not candidates for autologous stem cell transplant (ASCT)– or CAR-based therapies as tested in patients #5, #15, and #20 who were considered too frail (patients #5 and #20) or had BCMA-negative tumors (patient #15).

One of the major objectives of our mTAA approach was to address the emerging issue of immune escape associated with the adoptive transfer of T cell products targeting a single tumor-expressed antigen. For example, numerous groups have reported extremely high initial response rates (80 to 90%) after the adoptive transfer of CD19-CAR T cells to treat acute lymphoblastic leukemia. However, loss or modulation of expression of CD19 has now been recognized as a major mechanism of immune escape (23, 24). In MM, Rapoport et al. (11) have reported loss of the TAA NY-ESO-1 after administering a single NY-ESO-1 epitope–specific TCR-engineered T cell product. Our therapy was expressly designed to target a multiplicity of antigens (PRAME, SSX2, MAGEA4, NY-ESO-1, and Survivin) that are frequently expressed on tumor cells to limit the emergence of antigen-negative clones at relapse. We saw a steady decline in MM cells expressing multiple target antigens, demonstrating the clinical activity associated with a multiantigen-targeted T cell approach. The exception was a single patient in whom we documented immune escape due to target antigen loss upon receipt of a T cell product whose dominant activity was toward MAGEA4. We also uncovered additional immune evasion strategies from in-depth sequencing studies performed on another patient (#3, with active disease) who initially responded to mTAA T cells and subsequently experienced disease progression. In this patient, a range of immunoregulatory genes was altered at the time of progression including the up-regulation of the immune checkpoints CTLA4 and LAG3, down-regulation of immune-activating genes such as MS4A1, and a marked increase in myeloma proliferation genes [such as mir-23b-myc loop (18) and other poor prognosis MM gene sets (19)], which, in combination, overwhelmed both administered and endogenous tumor-reactive effector T cell function. Thus, our analysis highlights mechanisms of failure that may affect other immune-based therapies. As cellular immunotherapy becomes a mainstay treatment for MM, the identification of such common pathways associated with resistance will prove pivotal in advancing the field toward the elusive goal of a cure. Ultimately, future clinical trials may entail combining mTAA T cells with potentially synergistic agents such as checkpoint inhibitors and immunomodulatory agents that can target these pathways (25). Alternatively, the spectrum of target antigens can be further extended to enhance the multiantigen specificity of our tumor-targeted T cells and modified with transgenes to shield them from tumor-mediated inhibitory effects such as the incorporation of additional costimulatory signals (26), secretion of transgenic cytokines [e.g., interleukin-12 (IL-12) or IL-15] (27), expression of inverted cytokine receptors, or a combination of these approaches (28).

Limitations of this study include its small sample size and lack of a placebo control group. As previously indicated, patients were allowed to be on maintenance lenalidomide, which can affect outcome analyses as well. Therefore, although this study can confirm safety, a thorough clinical assessment through a larger multicenter phase 2 or phase 3 clinical trial coupled with correlative end points is necessary to validate these findings.

In summary, our findings demonstrate that autologous mTAA T cells can be manufactured reproducibly from MM patients. These adoptively transferred cells are well tolerated, and we documented in vivo T cell expansion and persistence that was associated with anti-MM effects, resulting in clinical benefit to patients.


Study design

Patients with MM were eligible for infusion on a Baylor College of Medicine and Houston Methodist Hospital (HMH) Institutional Review Board (IRB)–approved protocol (H-35626, NCT02291848) if they had residual disease (defined as persistent or progressive disease following prior lines of treatment) or were considered high risk for relapse (defined as having previously relapsed or achieved less than a complete response after combination therapy with a proteasome inhibitor, immunomodulatory drug, or high-dose chemotherapy and autologous stem cell rescue). The primary objective of this study was to establish the safety (end point: DLT rate at three escalating DLs), whereas the secondary objective was to ascertain the efficacy (end point: response rates and PFS) of mTAA T cells administered to patients with MM. The exploratory objectives were to investigate immunological correlates (e.g., expansion/persistence of infused T cell clones) with clinical outcomes. The sample size was estimated on the basis of the modified continual reassessment method (mCRM) method (see full protocol for details, data file S3), where dose escalations were permitted after two to four patients/arm were treated at a specified DL and the DLT rate was ≤20% at that DL. To be eligible, patients must be ≥18 and have MM that has been treated with at least one line of therapy (that had been assessed before infusion, which included bone marrow disease assessment within 4 weeks of receiving cells, when possible). Patients were required to be off all other investigational therapy for 1 month before T cell infusion, hemoglobin of >8 g/dl (reduced to >7 g/dl in most recent protocol amendment), total bilirubin of ≤2× upper limit of normal (ULN), aspartate aminotransferase (AST) ≤3× ULN, and a serum creatinine of ≤2× ULN and have an available mTAA T cell line. Written informed consent was obtained from all patients before T cell procurement (procurement consent) and again before treatment with T cells (treatment consent). Enrolled patients received at least two infusions of mTAA T cells 2 weeks apart [dose range, 0.5 (DL1), 1 (DL2), or 2 (DL3) × 107 cells/m2] and were eligible to receive up to six additional infusions at the same DL if they remained in CR (high-risk arm) or achieved clinical benefit (defined as SD, PR, or CR) 8 weeks after initial treatment. None of the treated patients received conditioning chemotherapy (i.e., no lymphodepletion). Patients were required to be off conventional therapy for at least 1 week before infusion but were allowed to continue maintenance doses of immunomodulatory medications (lenalidomide, pomalidomide, or other), as summarized in Fig. 2 (A and B). Post-infusion blood was collected for correlative studies on weeks 1, 2, 4, 6, and 8 and months 3, 6, 9, and 12. Marrow was collected at select time points. For the assessment of DLTs, AEs and severe AEs were collected and reported on all patients for the first 8 weeks after infusion. Analysis of disease response to mTAA T cells was performed between weeks 6 and 8 after infusion using the International Myeloma Working Group (IMWG) criteria (29). Per protocol, patients were enrolled on two arms (those who were <90 days from an ASCT and those patients who were >90 days or never had an ASCT). The protocol group assignments are specified in Tables 1 and 2 (“time from last therapy”), but results are grouped into patients treated with active myeloma and those in remission at the time of infusion. The current version of the clinical protocol is provided in the Supplementary Materials (data file S3). Primary data are reported in data file S4.

Generation of mTAA T cells

mTAA T cells were generated as previously described (17). Briefly, monocyte-derived DCs were isolated by plastic adherence from PBMCs isolated from patients, loaded with pepmixes (panels of 15-mer peptides overlapping by 11 amino acids) spanning the TAAs survivin, SSX2, MAGEA4, PRAME, and NY-ESO-1 (JPT Peptide Technologies), and cocultured (1:10) with autologous PBMCs in the presence of a TH1-polarizing cytokine cocktail [IL-7 (10 ng/ml), IL-12 (10 ng/ml), IL-15 (5 ng/ml), and IL-6 (10 ng/ml)]. From day 10, responder T cells were restimulated weekly with irradiated pepmix-pulsed DCs in the presence of IL-2 (50 to 100 U/ml) or IL-15 (5 ng/ml). The manufactured T cell line was then cryopreserved before patient use.

Characterization studies

ELIspot analysis was used to determine the frequency of T cells secreting IFN-γ in response to TAA pepmixes. Aliquots of manufactured T cells were surface-stained with mAbs. We used phycoerythrin-, fluorescein isothiocyanate–, peridinin chlorophyll protein–, allophycocyanin-, Alexa Fluor 700–, phycoerythrin cyanin 7–, Pacific Blue–, or Krome Orange–conjugated anti-CD3 (clone SK7), CD4 (SK3), CD8 (SK1), CD56 (B159), CD19 (SJ25C1), TCR-ab (T10B9), TCR-gd (B1), CD62L (DREG-56), CD27 (M-T271), CD28 (28.2), CD45RA (H100 or 2H4), CD 45RO (UCHL1), CCR7 (3D12), CD86, and PD1 (MIH4) (Becton Dickinson and Beckman Coulter). Control samples labeled with appropriate isotype antibodies were included, and a “fluorescence minus one” strategy was used for multicolor staining. Cells were analyzed using FACScan equipped with a filter set for four fluorescence signals, using CellQuest software, or FACSCanto II, using DIVA software (Becton Dickinson). ICS was performed by exposing mTAA T cells to irrelevant (viral peptides), control (no peptides or dimethyl sulfoxide only), or TAA pepmixes overnight in the presence of CD28 and CD49d (1 μg/ml) (BD Biosciences) followed by the addition of BD GolgiStop and BD GolgiPlug, which contains monensin and brefeldin A, respectively. Before acquisition, T cells are washed, pelleted, and surface-stained with anti-CD8 and anti-CD3, then fixed, and permeabilized with Cytofix/Cytoperm solution (BD Biosciences). After manufacturer-directed incubations and washes, cells were stained with IFN-γ and TNF-α antibodies (BD Biosciences) and were acquired using a Gallios flow cytometer with Kaluza software for analysis. The cytotoxic activity of each mTAA T cell line toward nonmalignant patient–derived PHA blasts (autoreactivity) or mTAA-pulsed patient-derived PHA blasts (tumor reactivity) was measured in a standard 51Cr release assay at varying E:T ratios (80:1, 40:1, 20:1, 10:1, and 5:1), and the percentage of specific lysis was calculated as [(experimental release – spontaneous release)/(maximum release – spontaneous release)] × 100. To assess the clonal diversity and track the expansion of the infused mTAA T cells, we used high-throughput deep sequencing of TCRvβ CDR3 regions (Adaptive Biotechnologies). Deep sequencing, which can detect ~100,000 unique clones, was performed on the infused lines and on peripheral blood and marrow samples collected before and after infusion. Those T cell clones identified within the product but were not detected in patient’s preinfusion repertoire were coded as line-derived unique clones.


For morphologic in situ protein coexpression analysis, IHC stains for CD138 and TAAs of interest were performed with a one- or two-step double-staining technique. Formalin-fixed paraffin-embedded, positively charged, unstained slides (4 to 5 μm thickness) of patient bone marrow core biopsies (lightly decalcified) or nondecalcified clot sections were obtained from the archives at the HMH. Plasma cells were detected with an anti-CD138 (clone MI15, 1:100; Dako) mouse mAb. In addition, the following anti-human primary antibodies were used for detection of TAAs: MAGE mouse mAb (clone 6C1, 1:200), NY-ESO-1 mouse mAb (clone E978, 1:50) (both Santa Cruz Biotechnology), PRAME rabbit polyclonal antibody (1:200, Bioss Antibodies), SSX2 mouse mAb (clone CL3202, 1:500; Atlas Antibodies), and Survivin rabbit mAb (clone 71G4B7, 1:500; Cell Signaling Technology). Standard deparaffinization, rehydration, heat antigen retrieval with citrate buffer (pH 6), dual endogenous peroxidase block (Dako), and nonspecific normal horse serum (2.5%) block were performed with 1× tris-buffered saline and 0.1% Tween 20 washes. Slides were incubated successively with anti-CD138 mAb, horse anti-mouse IgG biotinylated secondary antibody (BA-2000, 1:200; Vector Labs), streptavidin alkaline phosphatase reagent (VECTASTAIN ABC-AP kit; Vector Labs), and alkaline phosphatase substrate mixture (Vector Blue, SK-5300; Vector Labs). The second primary antibody against the TAA of interest was then applied overnight at 4°C. The next day, an anti-mouse secondary antibody (MP-7402, Vector Labs) and then aminoethyl carbazole (AEC; red) or 3,3′-diaminobenzidine (DAB; brown) substrate was added before counterstaining, rinsing in deionized water, and coverslipping in aqueous mounting medium. More recently, stained slides used the ImmPRESS Duet Double Staining Polymer Kit (Vector Labs) as per the kit instructions for one-step double staining of CD138 (DAB, brown) and PRAME or survivin (Vector Red). If cells tested negative on individual TAA stains with appropriate controls, CD138 and TAA costaining was not performed. Testis with intact spermatogenesis served as positive and epididymis as negative control tissue, respectively, for TAAs. These tissues were harvested from autopsies or patients undergoing standard of care surgeries after informed consent (from patient or medical power of attorney) on a HMH IRB–approved repository (IRB #PRO00007175). Immunoreactivity of CD138 and TAA coexpression was scored with membranous staining of CD138 and nuclear/cytoplasmic staining of the TAA of interest as a percentage of CD138-positive plasma cells and graded as follows: 0, <10%; 1+, 10 to 24%; 2+, 25 to 49%; 3+, 50 to 75%; and 4+, >75%. Intensity of IHC staining was also graded as 1+ (dim), 2+ (intermediate), or 3+ (strong). All scoring was performed by a HMH pathologist who was blinded to the clinical outcome but not blinded to the demographics of individual patients. Any level of TAA detection by IHC was considered positive in that given tumor sample.

mRNA sequencing and analysis

RNA sequencing (RNA-seq) was performed on eight peripheral blood specimens collected from patients #2 and #3 before and after T cell infusion. RNA-seq FASTQ files were processed through FastQC (v0.11.5), a quality control tool to evaluate the quality of sequencing reads at both the base and read levels and RNA-SeQC (v1.1.8) (30) to generate a series of RNA-seq–related quality control metrics. All samples passed quality check for this study. STAR 2-pass alignment (v2.5.3) (31) was performed with default parameters to generate RNA-seq BAM files. HTSeq-count (v0.9.1) (32) tool was applied to aligned RNA-seq BAM files to count how many aligned reads overlap with each genes’ exons. The HTSeq raw count data were then processed by DESeq2 (v3.6) (33) software to identify DEGs between two groups. A cutoff of gene expression fold change of ≥2 or ≤−0.5 and an FDR q ≤ 0.01 were applied to select the most DEGs.

Identification of reactive T cell clonotypes by scRNAseq

Preparation of cells for scRNAseq. mTAA T cells were thawed, rested in CTL medium [45% RPMI 1640 (HyClone Laboratories), 45% Click’s medium (Irvine Scientific), 2 mM GlutaMAX TM-I (Life Technologies), and 10% human AB serum (Valley Biomedical)], and resuspended at 2 × 106 cells/ml, and 200 μl was added per well of a 96-well plate. mTAA T cells were incubated overnight with 200 ng of stimulating pepmix (PRAME, SSX2, MAGEA4, NY-ESO-1, and Survivin). The following day, cells were harvested and passed through a 30-μm cell strainer, and dead cells were magnetically removed (Dead Cell Removal kit, Miltenyi Biotec). The resulting single-cell suspensions (1 × 105 cells/100 μl in RPMI + 10% fetal bovine serum) were used for subsequent sequencing studies.

Single-cell TCR and 5′-expression sequencing. Single-cell 5′-end TCR sequencing was performed using single-cell V(D)J and 5′ gene expression analysis platform according to the manufacturer’s protocol (10x Genomics). Cryopreserved mTAA T cells were thawed, prepared as above, then loaded onto a Chromium chip, and captured with barcoded gel beads at a target rate of ~5000 cells per patient sample. Barcoded and amplified complementary DNAs (cDNAs) were divided into two fractions for TCR V(D)J enrichment and 5′ gene expression analysis. V(D)J and 5′ expression libraries were pooled and sequenced by Illumina NovaSeq sequencer [read 1, 26 base pairs (bp); index, 8 bp; read 2, 98 bp].

Single-cell data processing. The analysis pipelines in Cell Ranger were used for sequencing data processing. FASTQ files were generated using cellranger mkfastq with default parameters. Then, cellranger count was run with transcriptome = refdata-cellranger-GRCh38-1.2.0 for each sample, in which reads had been mapped on the human genome (GRCh38/ hg38) using STAR (version 2.5.1b) (30, 34). TCR data were processed by running cellranger vdj with–reference = refdata-cellranger-vdj-GRCh38-alts-ensembl-2.0.0 to assemble TCR α and β chains and determine clonotypes.

Seurat R package (35) was used to run principal components analysis (PCA), t-stochastic neighbor embedding (t-SNE), and k-means clustering algorithms to visualize clustered cells in a two-dimensional space. PCA-reduced data are passed into t-SNE, a nonlinear dimensionality reduction method (36). Individual clonotypes for the TCR β chains were identified for the cytokine-positive barcodes (log2-scaled gene expression, >5%). Identified vβ CDR3 was then manually compared with TCRvβ CDR3 regions (Adaptive Biotechnologies) generated as above for patient product and follow-up blood/marrow samples. Subpopulations of samples were visualized in t-SNE plot using Loupe Cell browser ( DEGs were identified by comparing each cluster to all others using Cell Ranger.

Statistical analysis

Descriptive statistics were calculated to summarize clinical characteristics. Dose escalation was performed per protocol (please see data file S3) using the mCRM with 20% targeted probability of DLT to determine the maximum tolerated doses of mTAA-specific T cells. Treatment-related DLT was defined per protocol based on the National Cancer Institute Common Toxicity Criteria for Adverse Events version 4.X. Group assignments per protocol were based on timing from last ASCT (arm A, those >90 days from their last ASCT or not eligible for an ASCT, and arm B, those who are <90 days from their last ASCT) for safety analysis. However, efficacy analysis as presented here was based on whether the patients were in remission at the time of T cell infusion (group A) or if they had active myeloma at the time of infusion (group B) regardless of timing form their last ASCT. Within group B, patients were further subdivided into having (i) persistent disease (SD) or (ii) progressive disease per IMWG criteria after their last conventional MM therapy. PFS was calculated from the time of the first mTAA T cell infusion to the date of first progression or initiation of another line of MM therapy or was censored at last follow-up. The data cutoff date was 1 November 2019. Survival curves and median survival times were estimated by the Kaplan-Meier method. Graphic works were generated with GraphPad Prism 6.


Fig. S1. TCR profile and functional characteristics of individual patient products.

Fig. S2. In vivo fate of infused mTAA T cell products.

Fig. S3. TAA-directed killing of antigen-positive MM cells.

Table S1. Disease responses and outcomes for individual patients (group A).

Table S2. Disease responses and outcomes for individual patients (group B).

Table S3. After relapse of progression treatments and response.

Data file S1. List of genes differentially expressed in patient #3.

Data file S2. All genes and FPKM values (raw data) for mRNA sequencing performed on patient #3.

Data file S3. Latest version of approved clinical protocol.

Data file S4. Primary data.


Acknowledgments: We thank patients and their families who participated in this trial, their clinicians, staff, and integral personnel in the clinic and the laboratory. We would specifically like to acknowledge D. Lyon and N. Lapteva for quality assurance and quality control and W. Mejia for assisting with formatting figures. We would also like to thank D. Kraushner and R. Chen [Academic Director of the Single Cell Genomics Core (SCGC)] and the GARP core at BCM for assistance with single-cell genomic sequencing. Funding: NIH SPORE in lymphoma 5P50CA126752 (principal investigators: M.K.B. and H.E.H.), Leukemia and Lymphoma Society SCOR award (principal investigator: H.E.H.; project leaders: A.M.L. and P.D.L.), Leukemia Lymphoma Society/Rising Tide Foundation (principal investigator: A.M.L.; co-investigator: P.D.L.), ASH Scholar Award (principal investigator: P.D.L.), Leukemia Texas Research grant (principal investigator: P.D.L.), ASBMT New Investigator Award (principal investigator: P.D.L.), and Edward P. Evans Foundation MDS Discovery Research Grant (principal investigator: P.D.L.). Single-cell experiments were performed at the SCGC at BCM partially supported by NIH shared instrument grants (S10OD018033, S10OD023469, and S10OD025240) and P30EY002520 to R. Chen. Author contributions: Conception and design: A.M.L., H.E.H., M.K.B., and P.D.L.; administrative and regulatory support: B.G. and M.B. (research coordinator); patient care and trial recruitment: P.D.L., G.C., C.A.R., R.K., and H.E.H.; collection and assembly of data: all authors; data analysis and interpretation: P.D.L., A.M.L., H.E.H., M.K.B., J.F.V., M.J., Y.L., S.K. (bioinformatics), S.V., M.W. (biostatistics), T.W. (biostatistics), and B.C. (pathologist); manuscript writing: all authors; final approval of manuscript: all authors; accountable for all aspects of the work: all authors. Competing interests: P.D.L., I.T., G.C., C.A.R., R.K., T.W., M.W., M.B., A.P.G., S.M., B.C., Y.L., M.J., S.K., L.W., A.W., S.V., and M.F.-K. have no conflicts of interest relevant to this work. A.M.L., J.F.V., M.K.B., and H.E.H. are cofounders and equity holders in AlloVir and Marker Therapeutics, companies that aspire to commercialize virus-specific T cells and tumor-specific T cells, respectively. Marker Therapeutics has licensed (from Baylor College of Medicine) patents associated with the manufacture of mTAA T cells. This company could benefit from this research. J.F.V. serves as the chief development officer, and H.E.H. and M.K.B. serve on the advisory board of Marker Therapeutics. B.G. is the owner of QB Regulatory Consulting, LLP, which provides regulatory and project management support to Marker Therapeutics. QB Regulatory Consulting, LLP also provides regulatory support to Lokon Pharma AB and Tessa Therapeutics, which will not benefit from this research. M.K.B. also serves on the advisory board for Tessa Therapeutics, Allogen, and Unum. H.E.H. serves on advisory boards for Gilead Biosciences, Kiadis, Novartis, PACT Pharma, and Tessa Therapeutics and has research support from Tessa Therapeutics and Cell Medica, whereas M.K. serves as a consultant for Allovir, none of which could benefit from this research. A.M.L. and J.F.V. are the inventors on issued patent (US 12/862,409; EP 10814245.6) held by the Baylor College of Medicine that covers the manufacture of mTAA T cells. Data and materials availability: All data associated with this study are in the paper or the Supplementary Materials. scRNAseq datasets are deposited in the Sequence Reads Achieve (SRA) under accession number BioProject ID: PRJNA606116.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article