First-in-human phase 1 trial of induced regulatory T cells for graft-versus-host disease prophylaxis in HLA-matched siblings

Adoptive transfer of thymus-derived regulatory T cells (Tregs; tTregs) can ameliorate acute graft-versus-host disease (GVHD; aGVHD) and autoimmunity in mice.^1,2  However, high Treg doses (∼1:1 with donor T cells) are required to reliably prevent disease.^3-5  Historically, clinical application has been substantially hampered by low tTreg frequency and relatively poorer proliferation than T-conventional cells (Tcons) in ex vivo culture.⁶  Although 10-fold more tTregs can be isolated from peripheral blood (PB) than umbilical cord blood (UCB), magnetic cell separation–enriched PB frequently contains CD25⁺ T-effector cells (Teffs) or memory T cells that preferentially expand over tTregs in culture, resulting in a product with reduced or lost suppressor function.⁷ 

PB CD4⁺ Tcons can be driven to a Treg phenotype and function in vitro or in vivo in the context of low-dose or tolerogenic antigen exposure, as occurs after allogeneic hematopoietic cell transplantation (HCT; allo-HCT), leading to the development of induced Tregs (iTregs) that contribute to GVHD amelioration.^8,9  In preclinical murine models, polyclonal or antigen-specific CD4⁺ iTregs suppress murine allogeneic^10,11  and human xenogeneic GVHD.¹²  Previously, we showed that potently suppressive human iTregs could be reliably generated in high numbers from CD4⁺25⁻ T cells by culturing in interleukin 2 (IL-2), rapamycin (rapa), and transforming growth factor β (TGFβ), along with artificial antigen-presenting cells (APCs; aAPCs) expressing the high-affinity Fc receptor, CD64, loaded with a mitogenic anti-CD3 monoclonal antibody. The resulting iTregs provided comparable GVHD suppression to tTregs in a xenogeneic murine model.¹² 

In adult patients, granulocyte colony-stimulating factor–mobilized PB stem cells (PBSCs) are often the preferred graft source. Relative to UCB transplants used in our prior studies,^6,13  mobilized PBSCs contain 10- to 12-fold more CD3⁺ T cells. Therefore, an alternate approach to UCB-derived tTregs was needed. Hence, we developed a manufacturing method that reliably produced large numbers of iTregs with potent suppressive function for testing in recipients of PBSC transplantation (PBSCT). Our aim was to study the safety and efficacy of iTregs in a homogenous population of patients receiving matched sibling PBSCT with the graft and iTregs procured from the same donor using 1 conditioning regimen. We chose a reduced-intensity conditioning (RIC) approach as it is more commonly used at our institution than myeloablative for adult patients with hematological malignancies. Here, we present the results of a first-in-human phase 1 dose-escalation trial testing the safety of iTregs in the context of standard GVHD prophylaxis.

This was a phase 1, single-center, dose-escalation study of iTregs with an extension phase at the maximal tolerable (or achievable) dose (MTD) in adult patients with high-risk malignancy treated with RIC and HLA-identical sibling donor PBSCT. aGVHD immunoprophylaxis consisted of cyclosporine (CSA)/mycophenolate mofetil (MMF) initially with the plan to switch sirolimus for CSA based on preclinical data, suggesting a more permissive effect of sirolimus. The study used a fast-track design with 1 patient per iTreg dose cohort (3.0 × 10⁶/kg, 3.0 × 10⁷/kg, 3.0 × 10⁸/kg, and 10.0 × 10⁸/kg). An extension phase aimed to enroll a total of 10 patients at the MTD.

The primary objective of the phase 1 component was to determine the MTD. The primary objective of the extension phase was to determine the safety profile of iTregs at the MTD or highest achievable dose and sample size estimates for a phase 2 iTreg efficacy trial for prevention of grade 3-4 aGVHD by day 100. Outcomes were compared with a cohort of similar patients treated in the same manner but without iTreg infusion. Clinical and laboratory data were prospectively collected in the University of Minnesota Blood and Marrow Transplant Database and systematically analyzed as of April 2020.

The protocol was approved by the institutional review board at the University of Minnesota. All patients and donors signed institutional review board–approved informed consent in accordance with the Declaration of Helsinki.

Eligible patients were 18 to 75 years of age with high-risk malignancies (acute myelogenous leukemia, acute lymphocytic leukemia, chronic myelogenous leukemia, myelodysplastic syndrome, non-Hodgkin or Hodgkin lymphoma), requiring allo-HCT between June 2014 and September 2016, and deemed eligible to undergo RIC HLA-identical sibling PBSCT with acceptable organ function and Karnofsky ≥60%.¹⁴  Patients were excluded for active uncontrolled infection, pregnancy or breastfeeding, or prior myeloablative HCT within 3 months of study enrollment.

Donors were eligible if they met the following criteria: 12 to 75 years of age, >40-kg body weight and in good health, and HLA-A, -B, -DRB1 identical sibling match to recipient. Donors were required to undergo a nonmobilized apheresis procedure on day −14 for iTreg production and a separate granulocyte colony-stimulating factor–mobilized apheresis collection for PBSCs on day 0 (and on day +1 if needed).

All patients received an RIC regimen consisting of cyclophosphamide 50 mg/kg IV on day −6, fludarabine 30 mg/m² IV on days −6 to −2, and a single fraction of total body irradiation 200 cGy on day −1.¹⁴ 

The iTregs were administered IV over 15 to 60 minutes on day 0 at least 4 hours before the PBSC infusion. MMF was discontinued on day +30 or 7 days after engraftment, whichever was later, if no active GVHD. Targeted CSA levels were 200 to 400 mg/L and target sirolimus levels were 3 to 12 mg/mL. CSA or sirolimus taper began at day 100 if no aGVHD occurred. Patients received standard supportive disease and transplant-related care including antibacterial, antiviral, and antifungal prophylaxis.¹⁴ 

A nonmobilized apheresis product was obtained from an HLA-identical sibling donor on day −14. Enrichment of CD4⁺25⁻ cells was performed by a 2-step method using the CliniMACS device (Miltenyi Biotec, Auburn, CA). First, CD25⁺ cells (including tTregs) were depleted from peripheral blood mononuclear cells (PBMCs) with directly conjugated anti-CD25 magnetic microbeads (Miltenyi Biotec) and the CliniMACS device (depl05). CD4⁺ cells were purified from CD25⁻ PBMCs using directly conjugated anti-CD4 magnetic microbeads (Miltenyi Biotec) and the CliniMACS device (possel). Gammagard S/D (immunoglobulin [human] therapy; Baxter, Deerfield, IL) was used to prevent nonspecific binding.

Purified CD4⁺25⁻ T cells were resuspended at 0.25 × 10⁶ to 0.50 × 10⁶ viable cells per mL in X-VIVO 15 (Lonza, Walkersville, MD) supplemented with 10% human AB serum, heat-inactivated (Valley Biomedical Products and Services, Inc, Winchester, VA), GlutaMAX (2 mM), and N-acetylcysteine (2 mg/mL) in a tissue-culture flask (37°C/5% CO₂). CD4⁺25⁻ T cells were stimulated with anti-CD3 monoclonal antibody (MACS GMP CD3 Pure; Miltenyi Biotec) loaded, irradiated, and K562 transduced with CD64 and CD86 (KT64/86) aAPCs^7,12,13  (produced and tested according to a Biologics master file submitted to the US Food and Drug Administration) at an aAPC-to-Treg ratio of 1:1. Cultures were supplemented with 300 IU/mL IL-2 (Proleukin; Novartis Corporation, East Hanover, NJ) on day 0 and maintained at that concentration (based on total medium) for culture duration. rapa (Rapamune 100 ng/mL; Wyeth Pharmaceuticals, Collegeville, PA) and TGFβ1 (3 ng/mL; R&D Systems, Minneapolis, MN) were added on day 0 and with each culture feed. IL-2 was supplemented for the entire volume whereas sirolimus and TGFβ were supplemented only with new (higher) culture volume. Culture density was adjusted every 2 to 3 days to maintain 0.25 × 10⁶ to 0.50 × 10⁶ viable cells per mL On day 7, rapa/TGFβ iTregs were restimulated with KT64/86 cells at 1:1 (KT64/86 to viable cell). Samples were removed 2 to 3 days prior to final product processing for sterility testing. On day 14 ± 1, cultures were harvested, washed, and resuspended with a 5% human serum albumin solution for patient infusion. Lot-release criteria included: 7-aminoactinomycin D viability ≥70%, CD4⁺CD25⁺ ≥60%, CD4⁻/CD8⁺ cells ≤10%, Gram stain with “no organisms seen,” and endotoxin <5 EU/kg patient weight. These studies, along with our previously published work,¹²  adhere to the minimum information on Treg manufacture as previously described, specifically, expression of CD4, CD25, CD127, Foxp3, Helios, and methylation status of the Foxp3 gene.¹⁵ 

Flow cytometric studies were performed by the use of standard procedures. In brief, human-specific antibodies for CD4 (RPA-T4), CD8 (RPA-T8), CD14 (MSE2), CD19 (HIB19), CD25 (M-A251), CD45RA (HI100), IL-10 (JES3-9D7), and IL-17 (eBio64CAP17) were purchased from eBioscience. Antibodies to interferon γ (4S.B3), tumor necrosis factor α (Mab11), IL-2 (MQ1-17H12), IL-4 (MP425D2), TIGIT (VSTM3), Lag-3 (11C3C65), CTLA4 (BNI3), HLA-DR (L243), programmed cell death 1 (PD-1; EH12.2H7), and inducible costimulator (ICOS; C398.4A) were from BioLegend. Staining for Foxp3 was performed with an anti-human Foxp3 Flow Kit (clone 249D; BioLegend) following the manufacturer’s instructions. Flow data were acquired on an LSRII flow cytometer (BD Biosciences), and the analysis was performed with FlowJo software (TreeStar, Inc). To establish gates for Foxp3⁺ and Foxp3⁺⁺ staining, as well CD39 staining, a single control source of known rapa/TGFβ iTregs, expanded CD4 T cells (non-Tregs), and UCB tTregs was generated prior to analysis of clinical products, which were then thawed and assayed concomitant with each clinical product.

A carboxyfluorescein diacetate succinimidyl ester–based suppression assay was used in which human PBMCs were labeled with carboxyfluorescein diacetate succinimidyl ester according to the manufacturer’s instructions (Invitrogen) and plated at 10⁵ per well in 96-well U-bottom plates with graded titrations of expanded rapa/TGFβ iTregs (1:2 to 1:32 iTreg:PBMCs) in 200 µL of X-VIVO15 culture media (without rapa or TGFβ). Cells were stimulated with anti-CD3–conjugated Dynal magnetic beads (Invitrogen) at a 1:1 bead-to-PBMC ratio. On day 4, cells were harvested and stained with antibodies to CD8. Acquired data were analyzed by the use of the proliferation platform in FlowJo software and the percentage of inhibition of division index (a product of the percentage divided and proliferation index) is presented. As with the flow controls, a single banked sample of research-expanded rapa/TGFβ iTregs were run as a positive control for suppressive function. Suppression is reported for only 6 product samples, the remaining 8 were not adequately cryopreserved, resulting in suppression assays that were unevaluable.

The primary end point was the incidence of grade 3-5 infusional toxicity within 48 hours after iTreg infusions. Adverse events were classified according to the National Cancer Institute’s Common Terminology Criteria for Adverse Events, version 4. A dose-limiting toxicity (DLT) was defined as a grade 3 adverse event of >24-hour duration that occurred within 48 hours of infusion from the following categories: cardiac, immune disorders (including allergic reactions); respiratory, thoracic, and mediastinal disorders (with the exception of mucositis); renal disorders; or central nervous disorders. DLTs also included any grade 4 or 5 adverse event of >24-hour duration with the exception of hematological toxicities or fever. DLT events were captured at 1 to 4 hours after iTreg infusion (and before PBSCT on day 0), and at 24 hours, 48 hours, and 7 days (if DLTs occurred at 48 hours). Secondary end points included the cumulative incidence of grade 2-4 aGVHD at day 100, and the cumulative incidence of chronic GVHD, relapse, and survival at 1 year as previously defined.¹⁴ 

The best available dose or the MTD was determined by design. The experimental design for this study was a fast-track 3+3 trial of a semi-log escalation of iTregs.¹⁶  The trial was conducted in 2 steps with the addition of an extension arm to help confirm safety and provide preliminary estimates of efficacy. The first step enrolled 1 patient per increasing iTreg dose until encountering a DLT. Because there were no DLTs as expected, the second step was initiated in which the cohort size was increased to 3 patients, enrolling 6 patients using the 3+3 design at the highest dose with a regimen of CSA/MMF. After 6 patients were enrolled using CSA/MMF, 4 patients were continuously enrolled at this best available dose to complete the extension arm. A modification of Pocock stopping boundaries was used to continuously monitor for excessive infusional toxicity, primary graft failure, grade 3-4 acute GVHD, transplant-related mortality, and chronic GVHD during study accrual. Given the small numbers, events were primarily summarized using descriptive statistics such as frequencies and proportions. Among the 10 patients at the best available dose, GVHD was estimated using cumulative incidence, treating non-GVHD mortality as a competing risk. Engraftment, relapse, and infection were also estimated with cumulative incidence treating nonevent death as a competing risk. Overall survival was estimated by Kaplan-Meier curves. All analyses were performed using SAS 9.4 (SAS Institute, Inc, Cary, NC).

In total, 55 donor/recipient pairs were screened to enroll 16 donor/recipients. Of the 39 screening failures, 15 were the decision of the treating physician who thought the patient’s disease was too high risk to delay admission for 8 days awaiting iTreg culture. In addition, 11 patients did not meet eligibility criteria, and 13 donors were unable to donate PBSCs twice.

Sixteen patients were enrolled and 16 iTreg products were manufactured. All products met lot-release criteria for CD8 contamination (Figure 1; range, 0.0% to 0.8%), sterility (culture, Gram stain, and mycoplasma), endotoxin, and viability. However, 2 iTreg products (#1 and #4) failed to meet the lot release because of the low percentage of CD4⁺25⁺ and were not infused. The mean theoretical yield of the 14 infused products was 1.6 × 10¹² cells (range, 0.3 × 10¹² to 8.1 × 10¹²). Cell numbers for individual products at different time points in product manufacture are shown (Table 1). For the 14 infused products, the mean total number of nucleated cells (TNCs) at the end of expansion culture was 47 × 10⁹ (range, 15 × 10⁹ to 110 × 10⁹) corresponding to a mean fold-expansion of 624-fold (range,165-fold to 2252-fold) (Table 1). The mean proportion of CD4⁺CD25⁺ cells in products was 95% (range, 86% to 99%). Cell viability of purified CD4⁺25⁻ cells was 97.8% ± 0.6% (range, 92.9% to 99.6%), and at time of infusion was 96.9% ± 0.7% (range, 89.2% to 99.0%).

Approximately 60% of clinical iTreg products expressed Foxp3 with ∼30% Foxp3^bright (Figure 2A-B), with 83% ± 4% (range, 49% to 100%) of CD4⁺ cells expressing the Treg-mediated suppressor antigen, CTLA4 (Figure 2C). The iTregs in the product were potent suppressor cells with a median of 64% ± 5% (range, 48% to 75%) suppression of CD3-stimulated T cells at a 1:2 iTreg-to-PBMC ratio (Figure 2D) in the 6 product samples where there was sufficient iTreg recovery following thaw and wash. Stable Foxp3 expression in tTregs is controlled through the Treg-specific demethylated region in the Foxp3 gene.¹⁷  To date, no method to induce Tregs in murine or human cells has resulted in Treg-specific demethylated region methylation, and we previously found no evidence for rapa/TGFβ iTreg-induced demethylation.¹²  Very few cells in the rapa/TGFβ iTreg products expressed Helios (supplemental Figure 1A), a transcription factor involved in maintaining Foxp3 expression.¹⁸  rapa/TGFβ iTreg products had uniformly high expression of CD45RO, HLA-DR, and ICOS (83% ± 4%, 93% ± 1%, and 96% ± 2%, respectively). Consistent with prior data, rapa/TGFβ iTreg expressed CD39, an enzyme involved in the production of the suppressive nucleoside adenosine, on >60% (Figure 3A), and CD103, a component of the α4β7 receptor required for intestinal homing, on >11% of cells in the clinical iTreg products (Figure 3B).

Upon antigen exposure, CD4⁺ T cells differentiate from naive (CCR7⁺CD27⁺CD45RA⁺, CD95⁻), to central memory (CCR7⁺CD27⁺CD45RA⁻, CD95⁺), and/or effector memory (CCR7⁻CD27^+/−CD45RA⁻, CD95⁺) and to short-lived effector T cells (CCR7⁻CD27⁻CD45RA⁺).^19-21  With continued antigen exposure, T cells can acquire a senescent phenotype, characterized by the graded loss of costimulatory molecules (eg, CD27 and CD28) and increasing expression of inhibitory molecules (eg, PD-1, Lag-3, and TIGIT).¹⁹  Because large-scale in vitro expansion of CD4 and CD8 T cells for adoptive immunotherapy has been linked with senescence and decreased in vivo function,^22-24  we sought to determine the differentiative state and inhibitory receptor expression of expanded rapa/TGFβ iTreg clinical products (supplemental Figure 1). rapa/TGFβ iTreg clinical products were uniformly of effector memory phenotype (CCR7⁻, CD45RA⁻, CD95⁺),²⁵  and maintained high expression of CD27 and CD28, consistent with their potent immune-suppressive function.¹⁹  rapa/TGFβ iTreg clinical products also had low inhibitory receptor expression, suggesting maintenance of function in vivo (supplemental Figure 1G).

The 2 products that failed had post-CD25 depletion recoveries of only 0.17 × 10⁹ and 0.37 × 10⁹ TNCs vs 2.3 × 10⁹ ± 0.25 × 10⁹ TNCs for the other 14 products. One expanded only fourfold, suggesting that the nonmobilized apheresis product was of poor quality as both recovery and expansion were compromised. The other product had a fold expansion on the lower end of the range for the 14 other units (fourfold vs an average of 624- ± 151-fold for the other 14 products), pointing to the CD25 selection step as being suboptimal. Importantly, even though these products failed lot release, they did not secrete effector cytokines when restimulated (supplemental Figure 1H).

Fourteen patients were treated (7 male,7 female; median age, 63 years) with underlying diagnoses of acute myeloid leukemia (n = 8), myelodysplastic syndrome (n = 4), and non-Hodgkin lymphoma (n = 2). As planned, 1 patient each received 3.0 × 10⁶/kg, 3.0 × 10⁷/kg, 3.0 × 10⁸/kg iTregs with a corresponding Tcon-to-iTreg ratio of 86:1, 8:1, and 1:2 (Table 2). As a Tcon-to-tTreg ratio of 1:1 given concurrently was required to optimally suppress GVHD in mice,²⁶  dose escalation was halted at 3.0 × 10⁸/kg, corresponding to a Tcon-to-iTreg of 0.9 ± 0.1 for the patients treated and representing a reproducibly achievable maximum dose after 1 stimulation.

After 3 patients received 3.0 × 10⁸/kg with CSA/MMF immunoprophylaxis with no DLTs, 2 patients received 3.0 × 10⁸/kg iTregs using sirolimus/MMF as planned. However, as both patients developed grade 3 aGVHD, a stopping rule was met for excessive GVHD and the sirolimus/MMF arm was suspended. An additional 7 patients (for a total of 10) received 3.0 × 10⁸/kg iTregs along with CSA/MMF.

Patients receiving iTregs had an increased percentage of circulating Foxp3⁺CD25⁺ among CD4⁺ on days 3 and 7, and the degree of increase was higher in patients receiving 3.0 × 10⁸/kg vs ≤30 × 10⁶/kg iTregs (Figure 4). The percentage of CD4⁺ cells in circulation with a phenotype (Foxp3⁺CD25⁺) consistent with iTregs or activated CD4⁺ T cells returned to basal levels at or before day 15. As predicted, iTreg persistence was increased in patients receiving immunosuppression with sirolimus, compared with CsA, although this increase did not correlate to increased efficacy. To assess CD8 T-cell expansion and effector phenotype in patients receiving CsA vs sirolimus immunosuppression, PBMCs isolated on days 3, 7, 14, and 28 was also stained with antibodies to perforin and granzyme B following phorbol myristate acetate/ionomycin restimulation (Figure 4). Consistent with increased GVHD, patients receiving sirolimus immunosuppression had an increase in percentage of CD8⁺ cells in the periphery on day 28, and an increased percentage of these CD8⁺ cells expressed both perforin and granzyme B.

Infusional toxicities were monitored at baseline before the iTreg infusion, and then at 4 (prior to the PBSC infusion), 24, and 48 hours as well as 7 days. No severe infusional toxicities (grade 4-5) were associated with the administration of iTregs at 1 to 4 hours and none at subsequent time points, as summarized in Table 3.

Transplant outcomes are shown in Table 2. All 14 patients achieved neutrophil engraftment at a median of 7 days (range, 6-13 days) and a platelet engraftment at a median of 16 days (range, 0-34 days). All surviving patients achieved full donor bone marrow and blood CD33⁺ chimerism at 100 days after PBSCT. Full donor blood CD3⁺ chimerism at day 100 was achieved in all surviving patients except for 2 who had 96% and 98% CD3⁺ donor chimerism. By day 180, all patients had full donor blood CD3⁺ chimerism.

Of 10 patients who received 3.0 × 10⁸/kg iTregs using CSA/MMF, 2 developed grade 2 and 1 grade 3 aGVHD by day 100. All 3 had stage III skin involvement with 1 patient also having grade 4 lower gastrointestinal involvement. Five developed chronic GVHD requiring systemic immune suppression. Four of these 5 patients had de novo chronic GVHD. Six patients were alive at 1 year and 3 were alive at 3 years. Cause of death included aGVHD (n = 1), chronic GHVD (n = 1), relapse (n = 3), and infection (n = 2).

Outcomes were compared with those of a contemporary institutional cohort of 203 adult patients with hematological malignancies receiving the same RIC and GVHD prophylaxis, but without iTregs. Neutrophil engraftment was similar between iTreg patients (100% at a median of 7 days) and the contemporary controls (100% at a median of 9 days; P = .22). The cumulative incidence of grade 2-4 aGVHD was 30% (95% confidence interval [CI], 3% to 48%) for iTreg recipients vs 41% (95% CI, 34% to 48%; P = .51) for the contemporary control group (supplemental Figure 2) and 10% (95% CI, 0% to 28%) vs 25% (95% CI, 19% to 31%; P = .29) for grades 3-4 aGVHD.

To determine whether iTreg infusion influenced lymphocyte recovery, we compared lymphocyte recovery in the 10 patients who received the highest dose of iTregs of 3.0 × 10⁸/kg iTregs per kg with CSA/MMF to 69 contemporary controls for whom samples were available. As illustrated in Figure 5, recipients of iTregs had significantly higher absolute lymphocyte counts (P < .01) and higher CD4⁺ cells for the first 6 months after transplant (P < .01). Additionally, absolute CD4⁺Foxp3⁺CD25^hi numbers at day 28 after PBSCT were significantly higher in iTreg recipients than controls (median, 0.0336/μL vs 0.0130/μL; P < .01).

There was no significant difference between the iTreg cohort and the contemporary controls in the incidence of bacteremia at day 100 (60% [95% CI, 26% to 88%] vs 53% [95% CI, 46% to 60%]; P = .75), chronic GVHD at 2 years (50% [95% CI,18% to 82%] vs 45% [95% CI, 37% to 53%]; P = .83), or relapse at 2 years (30% [95% CI, 3% to 57%] vs 36% [95% CI, 29% to 43%]; P = .95). One-year GVHD-free relapse-free survival (grade III/IV GVHD-free, chronic GVHD requiring systemic immunosuppression free, relapse free, survival) was 20% (95% CI, 1% to 45%) for the 10 iTreg patients who received 3.0 × 10⁸/kg iTregs using CSA/MMF and 18% (95% CI, 13% to 23%) for the 203 contemporary control patients (P = .81). Three-year disease-free survival was 30% (95% CI, 7% to 58%) for iTreg patients, and 25% (95% CI, 17% to 32%) for contemporary controls.

This is the first-in-human clinical trial in allo-HCT patients to determine safety and potency of infusing large numbers of iTregs produced in a good manufacturing practice facility. HLA-identical sibling donor iTregs were safely administered at a maximum biological dose of 3 × 10⁸/kg iTregs per kg achieving a Tcon-to-iTreg ratio of ∼1:1, our target ratio to suppress GVHD with CSA/MMF immunoprophylaxis. Although the use of CSA may have negatively affected the iTregs, we observed higher absolute CD3⁺ T cells, CD4⁺ cells, and CD4⁺Foxp3⁺CD25^hi cells at day 28 along with faster lymphocyte recovery in iTreg patients vs contemporary controls. More patients and consideration for further dose escalation will be needed before reaching an efficacy conclusion. Although only 2 patients received sirolimus/MMF, it was ineffective as GVHD prophylaxis in the context of this iTreg clinical trial.

The colon and small intestine are primal GVHD target organs. The α4β7 integrin pair interacts specifically with mucosal addressin cell adhesion molecule-1 (MAdCAM-1) on high endothelial venules in the Peyer patches and intestinal lamina propria and is critical for immune cell homing to these tissues.^27-29  Disruption of α4β7 and MAdCAM-1, either through antibody-mediated blocking of the α4 subunit or gene disruption of the β7 subunit, can reduce graft rejection and GVHD.³⁰  TGFβ induces integrin α4 (CD103) expression and gut homing in T cells. Even though almost no UCB tTreg express CD103,³¹  >11% of PB rapa/TGFβ iTregs were CD103⁺, potentially enhancing the gut-protective effects of iTregs.

Our current good manufacturing practice–compatible preclinical studies indicated that up to 100-fold higher doses of aAPC-expanded iTregs (yield, ≥240 × 10⁹) than bead-expanded UCB tTregs could be achieved even with a shorter culture duration of 12 days vs 18 days. Notably, rapa/TGFβ iTregs and expanded PB tTregs had a similar in vivo potency when directly compared in a xenogenic GVHD model¹²  that served as foundational data for the iTreg trial described here. Murine studies of autoimmunity and GVHD have shown that T-cell receptor–transgenic, antigen-specific Tregs are more effective than polyclonal Tregs at suppressing disease^10,11,32  but this effect is very specific, and HY-Tg iTregs did not suppress GVHD in female mice lacking the HY antigen.¹¹  Although polyclonal iTregs may be less effective on a per-cell basis, GVHD is a multiepitope disease and our production platform generated enough cells to meet the ∼1:1 iTreg-to-Teff ratio that was sufficient to suppress disease in our xenogeneic model of GVHD.

Dose escalation was halted at 3.0 × 10⁸/kg (which equated to ∼1:1 iTregs to Tcons), despite no observed DLTs, as manufacturing costs became prohibitive. As seen with UCB tTreg trials,^6,13  iTregs could be detected for only 14 days. Unexpectedly, significant increases occured in day 28 absolute lymphocyte counts and numbers of CD3⁺ cells, CD4⁺ cells, and CD4⁺25^hiFoxP3⁺ cells. Because neither aGVHD rates nor severity were reduced by iTregs, the mechanism(s) for improved cell recoveries is unknown. Because the major difference between this trial and the historical controls is whether iTregs were infused, it is tempting to speculate that iTregs reduced thymus injury, decreased negative regulatory effects in the host that limited CD4⁺ and CD4⁺25⁺FoxP3⁺ cell recovery, or released soluble molecules that stimulated expansion of the above cell types.

During the dose-escalation phase of the trial, all patients received CSA/MMF as GVHD prophylaxis as standard of care at our institution for over the past decade. Because CSA would not be the optimal GVHD prophylaxis agent as it inhibits IL-2 production and hence may interfere with iTreg survival and function,³³  and rapamycin in the iTreg cultures has been shown to promote selective expansion of tTregs^34-39  and stabilization of Fox3 expression, sirolimus replaced CSA once the maximum dose of 3.0 × 10⁸/kg iTregs was achieved. Although sirolimus/MMF has been shown to be inferior to sirolimus/tacrolimus in preventing aGVHD,⁴⁰  we anticipated it would be sufficient prophylaxis in the setting of matched sibling donor PBSCT with iTreg infusion. However, both patients who received sirolimus/MMF as GVHD prophylaxis developed grade 3 aGVHD within a month of iTreg infusion. Despite the fact that both patients were successfully treated for aGVHD and were alive at a year after transplant, sirolimus/MMF was deemed ineffective as GVHD prophylaxis in this iTreg clinical trial and therefore, GVHD prophylaxis returned to CSA/MMF. Whether the ineffectiveness of sirolimus/MMF as GVHD prophylaxis was due to the regimen per se, independent of iTregs or an adverse effect of sirolimus/MMF on iTregs due to their known antiproliferative effects, is unknown.

A major concern in the field is the potential for instability of iTregs under inflammatory conditions, such as would be present in aGVHD. However, neither rodent¹²  nor these human studies have shown any evidence of increased aGVHD severity or accelerated time to onset when iTregs are infused at the time of allotransplant. In fact, preclinical data have suggested that iTregs should not increase but rather reduce aGVHD risk and potentially promote engraftment; a major concern remained the possibility that iTregs under the highly inflammatory conditions of aGVHD would become Teffs, increasing the incidence and/or severity of aGVHD. Fortunately, this did not occur. In the 10 patients who received 3.0 × 10⁸/kg iTregs using CSA/MMF, the incidence of aGVHD and chronic GVHD, aGVHD organ involvement, and response to therapy did not significantly differ from contemporary controls. Given the timing, logistics of 2 donor aphereses, and costs of manufacturing single-patient Tregs (up to $100 000 per product), the use of third-party Tregs in the future would be a more feasible approach and should be explored in future studies. Further studies also are warranted to evaluate higher Tcon-to-iTreg ratios using a restimulation approach⁷  and testing combined iTreg and tTreg infusion, which synergistically cooperate to treat murine colitis.⁴¹ 

The authors thank the nurses, nurse coordinators, nurse practitioners, physician assistants, social workers, and physicians who cared for these patients and their families. The authors especially thank the patients and their families who have entrusted us with their care.

This work was supported by National Institutes of Health grants R01 HL11879 (B.R.B., K.L.H.) and R01 HL56067 (B.R.B.) from the National Heart, Lung, and Blood Institute, R37 AI34495 (B.R.B.) from the National Institute of Allergy and Infectious Diseases, R01 HL114512 (M.L.M., K.L.H., J.E.W., D.J.W.) from the National Heart, Lung, and Blood Institute, and P01 CA065493 (J.E.W., B.R.B., D.H.M.) from the National Cancer Institute; Leukemia & Lymphoma Translational Research (grant R6029) (B.R.B.); National Institutes of Health National Heart, Lung, and Blood Institute Production Assistance for Cellular Therapies (PACT) 0062 (M.L.M., D.H.M.); and the Children’s Cancer Research Fund (M.L.M.).

Contribution: T.E.D. performed the statistical analysis; M.L.M. had primary responsibility for drafting the manuscript; and all authors contributed equally to the conception, design, and interpretation of data, and reviewed, edited, and approved the manuscript.

Conflict-of-interest disclosure: B.R.B. receives remuneration as an advisor to Magenta Therapeutics and BlueRock Therapeutics; receives research funding from BlueRock Therapeutics, the Children’s Cancer Research Fund, and the KidzFirstFund; and is a cofounder of Tmunity. The remaining authors declare no competing financial interests.

Correspondence: Margaret L. MacMillan, Blood and Marrow Transplant Program, Department of Pediatrics, MMC 484, 420 Delaware St SE, Minneapolis, MN 55455; e-mail: macmi002@umn.edu.