Single-center randomized trial of T-reg graft alone vs T-reg graft plus tacrolimus for the prevention of acute GVHD

Allogeneic hematopoietic cell transplantation (HCT) has proven to be a curative therapy for patients with a broad range of hematological malignancies, genetic disorders, and bone marrow failure states. Acute and chronic graft-versus-host diseases (GVHDs) still cause significant morbidity and mortality after myeloablative, HLA-matched allogeneic HCT.¹ Standard immunosuppressive GVHD prophylaxis (PPX) with a calcineurin inhibitor and methotrexate features an incidence of grade 3 to 4 acute GVHD of 10% to 20% by 100 days and moderate-to-severe chronic GVHD of 40% to 50% by 1 year.^2-4 Both calcineurin inhibitor-based and calcineurin inhibitor-free approaches have failed to show improvement in the composite end point of chronic GVHD-free or relapse-free survival, which remain significant problems.⁵ 

One way to prevent GVHD is through the use of donor T-regulatory cells (Tregs). These cells are a rare immune population that limit immune responses.⁶ Tregs in the donor graft have been shown in preclinical studies to reduce acute GVHD while retaining graft-versus-tumor effects.^7-10 Clinical studies have suggested that enrichment of Tregs in the donor graft or the addition of expanded third-party cord blood Treg added to the donor graft may reduce acute GVHD in haploidentical or cord blood HCT, respectively.^11-15 In haploidentical transplantation, this strategy has resulted in low relapse rates suggesting retention of graft-versus-tumor effects when donor grafts are enriched for Tregs.¹⁶ 

We previously reported the results of a phase 1 dose escalation study in HLA-matched HCT of an engineered donor graft in which highly purified CD4⁺CD25⁺CD127^lo Tregs were administered with hematopoietic stem and progenitor cells (HSPCs) followed 2 days later by conventional CD3⁺ T cells (termed T-reg graft).¹⁷ In this phase 1 study, a dose of 2 × 10⁶ to 3 × 10⁶ Tregs and 3 × 10⁶ CD3⁺ T cells per kg administered with single-agent GVHD prophylaxis appeared safe with little GVHD seen in the small number of patients treated. Prior clinical studies have suggested that as few as 5 × 10⁴ CD3⁺ T cells per kg without GVHD prophylaxis and as few as 1 × 10⁶ CD3⁺ T cells per kg with single-agent GVHD prophylaxis and without the addition of Tregs result in significant GVHD.^18,19 

Here, we report the results from stage 1 of our completed single-center phase 2 trial evaluating T-reg graft. In the first stage of this trial, we investigated the incidence of both acute and chronic GVHD among patients randomized to receive T-reg graft single-agent prophylaxis or no prophylaxis. This study found that rates of acute and chronic GVHD were very low in patients treated with T-reg graft and single-agent prophylaxis and that prophylaxis was preferred over T-reg graft alone for the prevention of GVHD.

The trial was conducted according to the principles of the Declaration of Helsinki and Good Clinical Practice. The overall trial was an investigator-initiated, open-label 2-stage phase 2 trial (NCT01660607) evaluating the feasibility, safety, and efficacy of the T-reg graft engineered product to prevent grade 3 to 4 acute and moderate-to-severe chronic GVHD after HCT from matched, related or unrelated donors. The first stage was designed to test if the incidence of GVHD was different between subjects who received Tregs alone (n = 12) and those who received Tregs and single-agent prophylaxis (n = 12). If the incidence of acute GVHD was noninferior between the no prophylaxis and single-agent prophylaxis arms, a predetermined phase 2 stage 2 extension cohort would be enrolled without prophylaxis, otherwise prophylaxis would be used. In the second stage, GVHD-free and relapse-free survival (RFS) at 1 year would be assessed. The results from stage 1 of the trial are presented here. For purposes of general comparison, a contemporaneous propensity score–matched standard-of-care (SOC) cohort of patients with HLA-matched HCT was identified, who consented to an institutional biobanking and clinical monitoring protocol. The trial protocol was approved by the institutional review board at Stanford University (study 21275). The senior author served as the Food and Drug Administration sponsor and holds the Investigational New Drug (IND) at Stanford University (IND no. 014686). All patients gave written informed consent. Data were collected and trial procedures were overseen by the trial investigators and the clinical research staff of the Cancer Clinical Trials Office of the Stanford University Comprehensive Cancer Institute. The Cancer Clinical Trials Office managed an independent data and safety monitoring committee which reviewed phase 2 trial conduct.

All authors had access to the data, vouched for the completeness and accuracy of the data and for the fidelity of the trial to the protocol, contributed to the writing and reviewing of the manuscript, had final responsibility for the manuscript content, and made the decision to submit the manuscript for publication. Public and sponsored research commercial support was obtained for trial support.

Eligible patients were aged between 13 and 72 years. Patients had acute leukemia (primary refractory, first complete remission, beyond first complete remission, or any high-risk features), chronic myelogenous leukemia, myelodysplastic syndrome (intermediate-2 or high-risk, Revised International Prognostic Scoring System [IPSS-R]), myeloproliferative disorders, or non-Hodgkin lymphoma with poor risk features not suitable for autologous HCT. Patients were eligible if they had a Karnofsky performance-status score ≥ 70%, a left ventricular ejection fraction of at least 45%, a diffusing capacity of the lung for carbon monoxide of at least 50%, a creatinine clearance of at least 60 mL per minute, and serum levels of alanine aminotransferase and aspartate aminotransferase that were no more than 7.5 times the upper limit of the normal range. Patients were required to be candidates for myeloablative conditioning with suitable 10/10 HLA-matched, related and unrelated donors or 9/10 HLA-matched, unrelated donors (based on high-resolution typing). Only HLA-matched or 9/10-permissive mismatches (at HLA-DP) were enrolled in stage 1.

Patients received different myeloablative conditioning regimens depending on the disease characteristics and included fractionated total body irradiation (fTBI; 1320 cGy, fractionated over 4 days), VP16 (60 mg/kg, as a single infusion), and cyclophosphamide (60 mg/kg, as a single infusion); or busulfan (3.6 mg/kg q24 initially, infused over 4 days, with targeting to busulfan level of 800-900 nM) and cyclophosphamide (60 mg/kg, per infusion over 2 infusions); or carmustine (300 mg/m²), VP16 (60 mg/kg), and cyclophosphamide (100 mg/kg). Patients received fresh HSPCs and Tregs on day 0, followed by thawed cryopreserved T conventional cells (Tcons) from the original apheresis material on day +2, dosed according to prefreeze CD3⁺ T-cell content. Initially in the trial, there was the option for either tacrolimus or sirolimus prophylaxis, but because of an observed incidence of sinusoidal obstructive syndrome (SOS) with busulfan and sirolimus and a reluctance by treating physicians to use sirolimus, tacrolimus became standard prophylaxis, unless there was a concern of intolerance. Tacrolimus was initiated at 0.015 mg/kg per day on day +3 (at least 12 hours after Tcon infusion). Doses of tacrolimus were targeted to serum trough levels of 6 to 8 ng/mL (Stanford phase 2) and tapered with a goal of discontinuation by day +180. Supportive care was provided according to institutional guidelines.¹⁷ 

Donor cells were apheresed from related and unrelated volunteer donors after 5 daily doses of 10 to 16 mcg/kg recombinant human granulocyte colony stimulating factor (rhG-CSF, Neupogen, Amgen) using institutional continuous-flow cell separator (Spectra Optia, Terumo BCT). Collections occurred as a single-day, large-volume collection (target of 24-30 L) or 2 consecutive apheresis collections on days 4 and 5 of G-CSF and were pooled or as a single large volume collection on day 5. Production of each batch of cellular product proceeded under good manufacturing practice conditions within the Stanford Health Care Cellular Therapy Facility (Palo Alto, CA) or at the Orca Biosystems, Inc Cellular Therapy Facility (Sacramento, CA), as has been previously described with a targeted Tcon-to-Treg ratio of 1.¹⁷ 

This trial used a Simon optimal 2-stage design with the primary phase 2, stage 1 end point to determine whether single-agent GVHD prophylaxis or no prophylaxis should be used based on noninferiority assessment of the randomized 2-arm comparison.²⁰ Adverse events were evaluated according to the revised National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.5, and their potential relationship to T-reg graft was assessed by the treating physician and the team of investigators.

We performed the noninferiority hypothesis test for the first segment using the method of Farrington and Manning implemented in the R package ratesci with the skewness corrections developed by Laud, with a noninferiority margin of 10%.^21,22 For this 1-sided test, we used the left tail of the distribution for the probability of rejecting the null hypothesis. With 24 evaluable subjects, this method has 80% power to test for a 10% margin of noninferiority at α = 0.05, assuming a variance in the incidence of acute GVHD of 0.35. For GVHD, RFS, and GVHD-free/RFS (GRFS),²³ prophylaxis and no prophylaxis groups were compared using log-rank tests. Multiple comparisons were performed using Kruskal-Wallis tests with Dunn correction for multiple comparisons, and pairwise comparisons were performed using Mann-Whitney U test. Linear regression was used for comparisons of fit curves and to evaluate associations among variables. P ≤ .05 was determined as the threshold for statistical significance for all tests. GraphPad Prism 9.5 or R 4.2 was used for all statistical analyses.

Peripheral blood samples were collected at baseline and at defined times after HCT. Platelet and peripheral blood immune cell populations were measured using clinical hemocytometer or flow cytometry. Chimerism was measured by Clinical Laboratory Improvement Amendments–certified short tandem repeat and sequence-based methodology by the participating sites’ clinical laboratories. Plasma concentrations of regenerating islet–derived protein 3α and suppression of tumorigenicity 2 were assessed through Viracor (Eurofins) clinical test at day 7 after HCT. For DNA methylation studies, bulk DNA was isolated from PAXgene DNA tubes containing whole patient blood collected on day 14 after transplantation (Qiagen), and bisulfite modification and sequencing was performed by EpigenDx. For bulk RNA sequencing (RNA-seq) analysis, CD4⁺ Tcons or CD4⁺ Tregs were sorted from thawed day 14 peripheral blood mononuclear cell samples using fluorescence-activated cell sorting and prepared for RNA-seq. Raw sequences were pseudoaligned to the GRCh38 Homo sapiens genome using Kallisto.²⁴ Differential expression analysis was performed in R 4.2 with RStudio using the DEseq2 package²⁵ and log₂ fold-change values were shrunk using the shrinkage estimator ashr.²⁶ Gene-set enrichment analysis was performed using the fgsea package with log₂ fold change set as ranks. Cell-cycle analysis was performed using the Seurat package,²⁷ treating each bulk RNA-seq sample as a single cell and assigning a cell-cycle score based on expression of G2M- and S-phase–associated genes.²⁸ 

A total of 24 patients were enrolled between April 2019 and February 2020 for stage 1 of the trial, and the last follow-up was June 2022. All patients were randomized to T-reg graft with either single-agent prophylaxis or no prophylaxis; patient characteristics are summarized in Table 1.

Cell-products processing occurred without manufacturing failure for all products with infusion within 72 hours for all but 1 product, which was received within this vein-to-vein window but administered at 76 hours. Table 2 shows the cell-dose yields of subjects, with a median of 2.6 × 10⁶ Tregs per kg administered (range, 0.7 × 10⁶-3.1 × 10⁶ Tregs per kg) and median Treg purity was 93% (range, 85%-97%) based on Foxp3 positivity by intracellular staining. All patients achieved a CD34⁺ dose > 2.5 × 10⁶ cells per kg, with a median dose of 6.3 × 10⁶ cells per kg (range, 2.9 × 10⁶-12.2 × 10⁶).

No infusion reactions were attributed to T-reg graft by trial investigators. The cumulative number of nonhematologic grade ≥3 severe adverse events occurring in the first 100 days after T-reg graft infusion totaled 7 events across 5 different patients amongst all 24 patients (21%). No exceptional infections or cytomegalovirus (CMV) reactivation with organ involvement were reported on study with the cumulative incidence of infectious complications approximating that seen in SOC at Stanford. Of the recipients who were CMV seropositive before receiving transplant (n = 7 with PPX and n = 11 without PPX), 8 (n = 2 with PPX and n = 6 without PPX) required antiviral therapy for CMV viremia. The incidence of SOS cases and treatment response and other complications are provided in Table 2.

Among the 24 patients who were randomized, the incidence of grade 2 to 4 acute GVHD was 7 of 12 patients (58%) who received T-reg graft alone vs 1 of 12 patients (8%) who received T-reg graft plus single-agent prophylaxis. All acute GVHD manifestations were steroid responsive (Table 3). The noninferiority hypothesis test with a 10% margin for the difference in acute GVHD incidence between the arms did not reject the null hypothesis that T-reg graft alone results in worse GVHD outcomes than T-reg graft plus single-agent prophylaxis (P = .982). The estimated difference in grade 2 to 4 acute GVHD incidence was 50% (95% confidence interval [CI], 13%-78%) higher for T-reg graft alone vs T-reg graft plus single-agent prophylaxis (Figure 1A). Thus, T-reg graft without prophylaxis was determined to be inferior to T-reg graft plus single-agent prophylaxis. The incidence of acute GVHD grade 3 to 4 was of 17% for T-reg graft without prophylaxis vs 0% with single-agent prophylaxis (P = .149). Acute GVHD characteristics are shown in supplemental Table 1. T-reg graft plus single-agent prophylaxis also appeared to be superior to a selected SOC plus transplant (SOC cohort at Stanford, n = 102), which exhibited an incidence of grade 2 to 4 acute GVHD of 34% (95% CI, 25%-44%; P = .069).

Moderate-to-severe chronic GVHD manifestations occurred within 1 year in 3 of 12 patients (28%; 95% CI, 10%-65%) who received T-reg graft alone with a median follow-up of 870 days (76-1091) and in 0 of 12 trial patients who received single-agent prophylaxis, with a median follow-up of 738 days (range, 261-906; Figure 1B). In concordance with typical results from other centers, 44 of 102 patients (43%; 95% CI, 33%-54%, at 1 year) had extensive chronic GVHD within the comparator SOC cohort, with a median follow-up of 473 days (range, 42-1286).

The nonrelapse mortality for patients without GVHD prophylaxis was 8% (1 of 12 patients) because of the onset of a new malignancy, and there was no nonrelapse mortality observed in the arm with single-agent prophylaxis. A total of 1 patient of 12 relapsed on the T-reg graft alone arm 217 days after HCT, whereas 4 of 12 patients relapsed on the T-reg graft with prophylaxis arm at a median of 241 days (range, 82-375) after transplantation with no significant differences between each arm (P = .42). RFS at 1 year after HCT was therefore 83% (95% CI, 41%-91%) with T-reg graft alone and 75% (95% CI, 48%-96%) when single-agent prophylaxis was added. The unadjusted 1-year GRFS estimates include 37% (95% CI, 11%-63%) for T-reg graft alone and 75% (95% CI, 41%-91%) for the cohort that received T-reg graft plus prophylaxis. In comparison, the 1-year GRFS estimate of the contemporaneous SOC cohort was 33% (95% CI, 24%-43%).

All patients who received graft with median times to neutrophil and platelet engraftment of 12 days (range, 9-19) and 15 days (range, 12-19; Figure 2A), respectively. No primary graft failures were observed, and 1 patient had graft loss associated with later relapse. Median time from transplant to discharge was 17 days (range, 13-53) without prophylaxis and 14 days (range, 12-17) with single-agent prophylaxis.

Full donor (as defined by ≥95% donor type) whole-blood and CD15⁺ myeloid chimerism was achieved in all patients in the trial by day 90 after transplantation. Full CD3⁺ T-cell chimerism was observed in all patients who did not receive prophylaxis and in 5 of 8 measured patients (63%) with single-agent prophylaxis by day 90 after transplantation (Figure 2B). There was no relationship between split T-cell chimerism and relapse, although this study was not powered for this evaluation. Immune reconstitution of total white blood cells and CD3⁺ T cells in evaluable patients did not significantly differ between patients on either arm, though CD19⁺ B-cell absolute counts appear higher over time for patients receiving single-agent prophylaxis, possibly because of reduced incidence of and treatment for acute GVHD in this arm (Figure 2C). A notable confounding factor is the number of patients who received steroids to treat GVHD in the patients who received T-reg graft alone. Furthermore, in patients who received tacrolimus prophylaxis, lower average tacrolimus blood concentrations may be associated with higher donor T-cell chimerism 90 days after HCT (supplemental Figure 1).

Although total CD3⁺ T-cell, CD4⁺ T-cell, and Treg reconstitution was not significantly different between arms, there was a modest nonsignificant trend for patients who received T-reg graft alone to have higher absolute T-cell counts than those who also received single-agent prophylaxis. This also was true for the ratio of CD4⁺ T cells to Tregs in these patients. Notably, patients receiving T-reg graft alone exhibited a higher CD4⁺ T-cell–to–Treg ratio than T-reg graft plus GVHD prophylaxis on day 14 after HCT (P = .056), an important timepoint because it precedes the onset of acute GVHD (Figure 2D). Importantly, although the effect of steroids may be a confounding factor, the earliest day +7 and +14 time points were before any steroid treatment.

In prior studies, FOXP3 promoter methylation has been shown to be linked with decreased Treg functionality.²⁹ Therefore, in order to understand how single-agent prophylaxis might relate to T-cell and Treg function, we first evaluated FOXP3 promoter methylation in peripheral blood cells and found that it was generally similar between both arms of the trial and the selected SOC cohort 14 and 30 days after transplantation (Figure 3A). Bulk RNA-seq analysis of sorted CD4⁺ Tcons and CD4⁺ Tregs collected 14 days after transplantation suggests that T-cell populations from patients who did not receive GVHD prophylaxis more frequently expressed genes closely associated with G2M or S-phase compared with the single-agent prophylaxis or SOC cohorts (Figure 3B). Consistent with this, gene-set enrichment analysis of CD4⁺ Tcons and CD4⁺ Tregs from patients who did not receive GVHD prophylaxis revealed an upregulation of gene sets associated with several proinflammatory and proliferative pathways relative to single-agent prophylaxis, and this was also observed when comparing CD4⁺ Tregs from patients receiving T-reg graft alone to SOC (supplemental Figure 2). Serum regenerating islet–derived protein 3α levels were generally low (median, 25 ng/mL; range, 16-271), whereas suppression of tumorigenicity 2 levels (median, 40.9 ng/mL; range, 11.1-160) were often elevated without clear association of either biomarker with acute GVHD (Figure 3C).

This single-center stage 1 of a concluded phase 2 trial reinforces the notion that concomitant GVHD prophylaxis is more effective for the prevention of GVHD in patients receiving T-reg graft products than in those receiving T-reg graft without GVHD prophylaxis. T-reg graft had no associated infusion reactions or exceptional adverse events.

Importantly, neutrophil engraftment and full-donor CD15 myeloid chimerism were achieved early in all patients, independent of the presence of single-agent prophylaxis. This finding could be in part due to the removal of methotrexate, an immunosuppressive agent often avoided as it may exacerbate mucositis and delay neutrophil engraftment. However, it is also possible that donor Tregs may facilitate engraftment as well, which has been reported preclinically.^30,31 In general, patients were taken off prophylaxis by 4 to 8 months after HCT and showed excellent overall recovery and few late-term complications, especially from chronic GVHD. We did not see any exceptional infectious disease risks or complications, which may compare favorably to studies involving T-cell depletion.^32-34 

To our knowledge, this is the first phase 2 study application of high-throughput flow cytometric purification of primary Tregs to create a precision-engineered donor cell-therapy product. Preclinical studies demonstrate that high-purity Treg can counterbalance an equivalent dose of T cells when administered in a timed, sequential manner for effective GVHD prevention, thereby providing a rationale to maintain this ratio in human studies.^35,36 Thus, this trial marks the novel implementation of a new category of T-cell–reduced grafts (3.0 × 10⁶ cells per kg) as compared with T-cell depleted grafts (<10⁵ cells per kg), standard peripheral blood grafts (1 × 10⁸-8 × 10⁸ cells per kg), or prior unsuccessful clinical trial evaluations of T-cell–reduced dose (1× 10⁸-3 × 10⁶ cells per kg) with single-agent prophylaxis.¹⁸ 

Although Tregs at this dose and with a 1:1 Treg-to-Tcon ratio do not sufficiently control donor T-cell alloreactivity and do not prevent GVHD without prophylaxis, the addition of relatively low-dose tacrolimus (6-8 ng/mL target) achieves the prevention of both acute and chronic GVHD. It has been reported that Tcons may be more sensitive to tacrolimus than Tregs.^37,38 Therefore, calcineurin inhibitors could be more effective at controlling donor T-cell engraftment and alloreactivity in the T-reg graft setting compared with standard peripheral blood grafts in which Tregs are rare. Indeed, we unexpectedly found that donor T-cell chimerism was achieved earlier and more completely without prophylaxis and in patients with lower tacrolimus serum concentrations. In addition, the absolute number of CD4⁺ T cells and CD4⁺ T-cell–to–Treg ratio was greater without prophylaxis, suggesting that the addition of single-agent prophylaxis may favor better control of alloreactivity by Tregs. Of note, although sirolimus has been suggested to be more synergistic with Tregs, we included the use of tacrolimus because of the concern for increased risk of SOS in patients receiving sirolimus and busulfan.^39,40 In the phase 2 extension of this trial, we mandated the use of only tacrolimus to remove the use of 2 different prophylactic agents as a confounding factor in our analyses of trial end points.

A major effect of calcineurin inhibition is the reduction of interleukin-2 production and the limitation of T-cell proliferation.^41,42 RNA-seq analysis of sorted Tregs and Tcons 14 days after transplantation revealed a more proliferative and perhaps more inflammatory expression profile in both cell types from patients who received T-reg graft alone than those from patients who received single-agent prophylaxis. Furthermore, Treg and Tcon messenger RNA from both T-reg graft arms indicated greater expression of proliferative genes than SOC comparison samples, consistent with the removal of methotrexate from T-reg graft GVHD prophylaxis or possibly as a result of the graft’s T-cell composition. Although this study was not powered to identify significant differences among immune cell, epigenetic, and transcriptional measurements, the results from these analyses represent interesting hypotheses to be tested in future, larger studies.

Although patients without prophylaxis had a relatively high rate of acute GVHD comparable to SOC, all cases were steroid responsive and did not evolve into chronic GVHD and only 1 of 12 relapsed. A remarkable feature of our trial is the 90% (9/10) response rate of acute GVHD manifestations across all patients to primary steroid therapy and appears much higher than that seen in SOC. One patient with steroid-refractory disease responded readily to secondary ruxolitinib therapy. Successful responses to GVHD therapy were also reported in earlier trials involving haploidentical transplantation with Treg-enriched donor grafts, suggesting an underlying role for Tregs in the responsiveness of GVHD to steroid therapy.¹⁴ 

Another noteworthy feature of these trial results is the low incidence of chronic GVHD manifestations, which affect the quality of life and drive long-term morbidity and mortality. The exact mechanism by which Treg cells improve immune reconstitution such that the pathological immune states that underly chronic GVHD are blunted or subverted is not clear; however, it is possible that donor Tregs could interact with B cells or pro–B cells affecting long-term reconstitution.

Although the patients on trial had heterogenous primary diseases requiring transplantation, we were unable to find evidence of impaired graft-versus-tumor effects in T-reg graft recipients. We observed a number of patients with very high risk, minimal residual disease positivity, or active disease including myelofibrosis exhibited good engraftment, achieved minimal residual disease-negative status, and remain in durable remissions. We did not observe a relationship between mixed T-cell donor chimerism within or beyond 90 days and relapse. Although the distribution of conditioning regimens did not differ significantly by treatment arm, it remains a possibility that the multiple conditioning regimens used in this trial could have affected the subsequent incidence of acute GVHD. We did not observe a correlation between the incidence of acute GVHD and conditioning regimen, but this is a topic that could be revisited in the future given that in this cohort conditioning was dictated primarily by disease diagnosis.

One limitation of this study is the small sample size, which was not designed to assess outcomes beyond the primary end point of acute GVHD. We were, therefore, unable to test the important question of whether the addition of posttransplant prophylaxis affects subsequent graft-versus-tumor effects and patient relapse. It remains possible that certain patients, such as those with high-risk or active disease, may benefit from the absence of posttransplant immunosuppression. Additional studies are needed to test the comparative effects of GVHD prophylaxis and GVHD therapy on graft-versus-tumor effects to improve the strategy we propose here. Furthermore, although we have explored a 1:1 Treg-to-Tcon ratio with Tcon administration occurring on day +2 after HCT, different ratios and timings, such as those implemented by the University of Perugia group,¹⁴ are important to examine to identify an optimal strategy. Additional limitations of this study include the use of multiple conditioning regimens and variable single-agent prophylaxis. When combined with our initially small sample size and patient and disease heterogeneity, we are unable to completely remove these variables as potential confounders in our analysis of acute GVHD across treatment arms.

Although T-reg graft without prophylaxis performed as well as or better than SOC with respect to incidence of grade 3 to 4 acute GVHD and moderate-to-severe chronic GVHD, this study demonstrates that T-reg graft with single-agent GVHD prophylaxis produced superior results and is, therefore, the preferred approach. Future larger studies will help to further elucidate the impact of T-reg graft on GVHD, relapse risk, and overall survival.

The study was supported by National Institutes of Health National Heart, Lung, and Blood Institute grants RO1 HL114591 (R.S.N.), K08HL119590 (E.H.M.), an Orca Sponsored Research Agreement (E.H.M.), the Amy Manesevit award (E.H.M.), and National Cancer Institute grant K00CA245728 (C.S.B.).

Contribution: E.H.M. was responsible for overall trial conduct, analysis, and writing of the manuscript; A. Pavlova, R.L., R.S.N., and S.K. contributed to trial conduct, analysis, and writing of the manuscript; R.L., R.S.N., L.S.M., V.A., P.S., S.A., L.J.J., A.R.R., W.-K.W., D.B.M., M.J.F., S.B., S.S., and J.A.S. contributed to patient care and accrual; B.J.X. contributed to correlative science studies; C.S.B. contributed to correlative science studies and writing of the manuscript; N.B.F. and A. Putnam contributed to trial conduct and cell-product manufacturing; and J.S.T. contributed to statistical evaluations.

Conflict-of-interest disclosure: A. Pavlova, N.B.F., A. Putnam, and S.K. are employed at and E.H.M. has received research funding from Orca Biosystems, Inc. B.J.X. is employed by 3T Biosciences. The remaining authors declare no competing financial interests.

Correspondence: Everett H. Meyer, Stanford Blood and Marrow Transplantation and Cellular Therapy Division, Stanford School of Medicine, Stanford University, 269 Campus Dr, CCSR 2250, Stanford, CA 94305; email: evmeyer@stanford.edu.