key: cord-0856580-b3zfk3ww
authors: Mehta, Rohtesh S.; Saliba, Rima M.; Alsfeld, Leonard C.; Jorgensen, Jeffrey L.; Wang, Sa A.; Anderlini, Paolo; Al-Atrash, Gheath; Bashir, Qaiser; Ciurea, Stefan O.; Hosing, Chitra M.; Im, Jin S.; Kebriaei, Partow; Khouri, Issa; Marin, David; Nieto, Yago; Olson, Amanda; Oran, Betul; Popat, Uday R.; Qazilbash, Muzaffar H.; Ramdial, Jeremy; Rondon, Gabriela; Saini, Neeraj; Srour, Samer A.; Rezvani, Katayoun; Shpall, Elizabeth J.; Champlin, Richard E.; Alousi, Amin M.
title: Bone Marrow versus Peripheral Blood Grafts for Haploidentical Hematopoietic Cell Transplantation with Post-Transplantation Cyclophosphamide
date: 2021-09-16
journal: Transplant Cell Ther
DOI: 10.1016/j.jtct.2021.09.003
sha: 03e89cc73bfbd8c0233bf28b5ff92669c5b229cb
doc_id: 856580
cord_uid: b3zfk3ww

In the coronavirus disease 19 (COVID-19) pandemic era, the number of haploidentical hematopoietic cell transplantations (HCTs) with peripheral blood (PB) grafts increased significantly compared with HCTs with bone marrow (BM) grafts, which may be associated with adverse outcomes. We compared outcomes of HCT in BM graft and PB graft recipients age ≥18 years with hematologic malignancies who underwent T cell- replete haploidentical HCT and received graft-versus-host disease (GVHD) prophylaxis with post-transplantation cyclophosphamide, tacrolimus, and mycophenolate mofetil. Among the 264 patients, 180 (68%) received a BM graft and 84 (32%) received a PB graft. The median patient age was 50 years in both groups. The majority (n = 199; 75%) received reduced-intensity conditioning. The rate of acute leukemia or myelodysplastic syndrome was higher in the BM graft recipients compared with the PB graft recipients (85% [n = 152] versus 55% [n = 46]; P < .01). The median times to neutrophil and platelet engraftment and the incidence of grade II-IV and grade III-IV acute GVHD (aGVHD) were comparable in the 2 groups. Among the patients with grade II-IV aGVHD, the rate of steroid-refractory aGVHD was 9% (95% confidence interval [CI], 5% to 18%) in the BM group versus 32% (95% CI, 19% to 54%) in the PB group (hazard ratio [HR], 3.7, 95% CI, 1.5 to 9.3; P = .006). At 1 year post-HCT, the rate of chronic GVHD (cGVHD) was 8% (95% CI, 4% to 13%) in the BM group versus 22% (95% CI, 14% to 36%) in the PB group (HR, 3.0; 95% CI, 1.4-6.6; P = .005), and the rate of systemic therapy-requiring cGVHD was 2.5% (95% CI, 1% to 7%) versus 14% (95% CI, 7% to 27%), respectively (HR, 5.6; 95% CI, 1.7 to 18; P = .004). The PB group had a significantly higher risk of bacterial and viral infections, with no appreciable advantage in the duration of hospitalization, immune reconstitution, relapse, nonrelapse mortality, or survival. Our data suggest a benefit of the use of BM grafts over PB grafts for haplo-HCT.

In the setting of haploidentical hematopoietic stem cell transplantation (HCT) with post-transplantation cyclophosphamide (PTCy) for graft-versus-host disease (GVHD) prophylaxis, the use of granulocyte colony-stimulating factor (G-CSF)-mobilized peripheral blood (PB) grafts has increased dramatically and has now surpassed that of bone marrow (BM) grafts [1] . The utilization of PB graft increased further during the coronavirus disease 19 pandemic era when the National Marrow Donor Program mandated cryopreservation of unrelated donor (URD) grafts before initiating conditioning [2] . This trend also translated to an increased use of PB grafts in the haploidentical (haplo) HCT setting. With this practice change, we anecdotally noted higher risks of acute GVHD (aGVHD) and chronic GVHD (cGVHD) that led us to conduct this study.

In the absence of a randomized prospective trial, registry studies have assessed the differences in outcomes of haplo-HCT with either BM or PB grafts using PTCy prophylaxis and have yielded controversial results [3À6] . One reason for these conflicting outcomes may be related to the inclusion of patients across centers, which introduces biases that are difficult to control in a retrospective analysis. For instance, practice disparities exist among centers in terms of immunosuppression taper, GVHD management, and post-HCT complications. Moreover, data are lacking on other significant outcomes, such as steroid-refractory (SR) aGVHD, systemic therapy-requiring cGVHD, differences in immune reconstitution, and quality of life (QoL) after haploidentical BM grafts versus PB grafts.

Here we report the outcomes of patients treated at the MD Anderson Cancer Center who underwent T cell-replete haploidentical HCT with either BM or G-CSF-mobilized cryopreserved PB grafts and a uniform GVHD prophylaxis regimen with PTCy, tacrolimus, and mycophenolate mofetil (MMF) (PTCy/Tac/MMF). We hypothesized that PB grafts would be associated with higher risks of aGVHD and cGVHD compared with BM grafts.

We included adult patients age 18 years with any hematologic malignancy who underwent haplo-HCT between January 2015 and July 2020 with any conditioning regimen and PTCy/Tac/MMF GVHD prophylaxis. Patients who received a manipulated graft (T cell depletion or ex vivo engineered T cells) were excluded. The primary objective was to compare the rates of aGVHD (grade II-IV, grade III-IV, SR) and cGVHD (overall, extensive, systemic therapy-requiring) between the BM and PB groups. Secondary outcomes were to compare neutrophil and platelet engraftment, chimerism, nonrelapse mortality (NRM), relapse, progression-free survival (PFS), overall survival (OS), GVHD-free relapse-free survival (GRFS), immune reconstitution, rates of infections, and differences in quality of life (QoL) in the 2 groups. Data on all important outcomes post-HCT, including infections, are routinely prospectively collected and entered into a database by a dedicated team at our institution. We routinely monitor for bacterial infections by blood cultures once weekly and for viral infections, such as adenovirus, cytomegalovirus (CMV), and Epstein-Barr virus (EBV), by polymerase chain reaction (PCR) assay weekly at least until day +100 post-HCT, or longer depending on the clinical situation.

For immune reconstitution, cell subsets in PB samples were analyzed by multicolor flow cytometry for CD3, CD4, CD8, CD19, CD25, CD27, CD45, CD45RO, CD56, CD62L, CD127, and IgM. The absolute numbers of T cells, B cells, regulatory T cells (Tregs), and natural killer (NK) cells, as well as their subsets, were calculated as cells/mL based on their frequency in a lymphoidmononuclear cell-gated population. Chimerism analysis was performed at day +30, day +100, 6 months, and 12 months post-HCT using a PCR assay with primer sets flanking microsatellite repeats.

QoL assessment was done using the Functional Assessment of Cancer Thera-pyÀBone Marrow Transplant (FACT-BMT) scale, version 4 [7] . This scale includes 5 subscales that capture physical, social/family, emotional, and functional wellbeing, along with a bone marrow transplantation (BMT) subscale. An aggregate global score and total scores for each subscale were calculated as suggested previously [7] . All patients who were alive in remission and had at least 1-year of follow-up were selected for QoL assessment and contacted by phone to confirm email addresses, and the survey was then distributed electronically.

Acute GVHD was staged and graded according to the consensus criteria [8] , and cGVHD was staged and graded according to standard criteria [9, 10] . SR-aGVHD was defined as (1) failure to respond after 7 days of treatment with corticosteroids (generally prednisone 2 mg/kg/day or an equivalent dose of methylprednisolone) or (2) progression after 72 hours. Neutrophil engraftment was defined as an absolute neutrophil count of >500 £ 10 6 /L for 3 consecutive days. Platelet engraftment (20K) was defined as a platelet count >20 £ 10 9 /L for 7 days without transfusion. Platelet engraftment (50K) was defined as a platelet count >50 £ 10 9 /L for 7 days without transfusion. Relapse or progression was defined as the time to recurrence or progression of the underlying malignancy, with death without relapse or progression (NRM) treated as a competing risk. PFS was defined as the time from HCT to relapse or progression or death. OS was defined as the time from HCT to death from any cause. GRFS was defined as the absence of grade III-IV aGVHD, systemic therapy-requiring cGVHD, relapse, or death. The study was conducted in accordance with the Declaration of Helsinki and was approved by the institutional Investigational Review Board (MDACC 2021-0103). , and multiple myeloma (1 each) in the BM versus PB groups, respectively, and chronic myeloid malignancies include chronic myelogenous leukemia/myeloproliferative disorders. * Among those who engrafted. Seven patients had graft failure (6 BM and 1 PB); 6 patients received Flu/Mel RIC and 1 received Bu/Flu MAC, with either a child (n = 6) or a sibling (n = 1) donor. Among 6 BM graft failures, 2 patients had donor-specific antibodies (DSAs): 1 with anti-HLA class I antibody and 1 with anti-HLA class I and class II antibodies. Of the remaining 4 patients without DSAs, the median TNC dose in the graft was 1.8 £ 10 8 /kg (range, 1.53 to 2.1 £ 10 8 /kg), which was lower than the median TNC dose (3 £ 10 8 /kg) in the overall BM group. One patient in the PB group with graft failure had anti-HLA class I and class II DSAs. 

Descriptive analyses were performed to summarize clinical and demographic characteristics. Categorical variables were compared between treatment arms using Fisher's exact test, and continuous variables were compared using the Wilcoxon rank-sum test. Engraftment data were analyzed considering death before engraftment as a competing risk. The rate of NRM was estimated in a competingrisks framework with relapse as the competing risk. aGVHD and cGVHD were assessed with competing risks of relapse and death. aGVHD events included both classical (before day +100) and late aGVHD (aGVHD occurring beyond day +100). Kaplan-Meier curves were used to estimate OS, PFS, and GRFS, and the log-rank test was used to test differences by graft source. The Mann-Whitney U test was used to compare QoL data. Statistical analyses were performed primarily with STATA 14 (StataCorp, College Station, TX).

Of the 264 patients, 180 (68%) received BM grafts and 84 (32%) received PB grafts. Recipient age at HCT (median, 50 years) and donor age (median 34 years for BM and 31 years for PB) were similar in the 2 groups. More patients in the BM group than in the PB group had acute myelogenous leukemia/myelodysplastic syndrome (64% versus 43%) and acute lymphoblastic leukemia (21% versus 12%) (P < .001). Approximately three-quarters of the patients in both groups received reduced-intensity conditioning (RIC) with a fludarabine and melphalan-based regimen (74% versus 76%). Busulfan (with pharmacokinetic dose monitoring) and fludarabine-based myeloablative conditioning (MAC) was used in 22% and 23%. Approximately one-half of the patients had an HCT-Comorbidity Index (HCT-CI) of 3 (53% versus 49%; P = .5). Haploidentical children were the most common donors (52% versus 55%), followed by siblings (40% versus 36%). More male patients in the PB group than in the BM group received grafts from female donors (30% versus 18%; P = .04). As expected, there were significant differences in the graft composition, with a significantly lower total nucleated cell dose (median, 3 £ 10 8 /kg versus 8 £ 10 8 /kg), CD34 dose (median, 2.8 £ 10 6 /kg versus 5.8 £ 10 6 /kg), and CD3 dose (median, 0.3 £ 10 8 /kg versus 2.62 £ 10 8 /kg) in the BM group compared with the PB group (Table 1) .

Primary graft failure occurred in 6 of 180 patients (3.33%) in the BM group and in 1 of 84 (1.19%) in the PB group. The median time to neutrophil engraftment was 19 days (interquartile range [IQR], 17 to 22 days) in the BM group and 18 days (IQR, 16 to 21 days) in the PB group (P = .07). Among recipients of an RIC regimen, the median time to neutrophil engraftment was 20 days (IQR, 17 to 22 days) in the BM group and 18 days (IQR, 16 to 20 days) in the PB group (P = .02), and among MAC recipients, it was 19 days (IQR, 17 to 21 days) in the BM group and 18 days (IQR, 16 to 22 days) in the PB group (P = .8) ( Table 2 ). The median duration of hospitalization for HCT was 33 days (IQR, 28 to 39 days) in the BM group versus 31 days (IQR, 26 to 36 days) in the PB group (P = .05).

The median time to platelet engraftment (20K) was 28 days (IQR, 23 to 36 days) in the BM group and 26 days (IQR, 20 to 35 days) in the PB group (P = .3). The median time to platelet engraftment (50K) was 35 days (IQR, 28 to 45 days) in the BM group and 30 days (IQR, 26 to 41 days) in the PB group (P = .06). There was no difference in the median time to platelet engraftment between the groups when analyzed by conditioning intensity ( Table 2) .

Chimerism analysis revealed a median of 100% donor cells in both myeloid and T cell compartments in the BM and PB groups at all time points tested: day +30, day +100, 6 months, and 12 months (Supplementary Table S1 ).

The cumulative incidence of grade II-IV aGVHD was 49% (95% CI confidence interval [], 42% to 57%) in the BM group versus 44% Table 3 Other Outcomes Outcome * Among patients with grade II-IV acute GVHD (95% CI, 34% to 57%) in the PB group (hazard ratio [HR], 0.8; 95% CI, 0.6 to 1.2; P = .3). The cumulative incidence of grade III-IV aGVHD was 7% (95% CI, 4% to 13%) in the BM group versus 12% (95% CI, 6% to 23%) in the PB group (HR, 1.5; 95% CI, 0.7 to 3.6; P = .3). In the patients who developed grade II-IV aGVHD, the incidence of SR-aGVHD in the 2 groups was 9% (95% CI, 5% to 18%) versus 32% (95% CI, 19% to 54%), respectively (HR, 3.7; 95% CI, 1.5% to 9.3%; P = .006) (Figure 1A-C; Table 3 ). When analyzed by conditioning intensity, PB was associated with significantly higher risks of grade III-IV aGVHD (HR, 10.9; 95% CI, 1.3 to 89; P = .03) and SR-aGVHD (HR, 14; 95% CI, 1.8 to 110; P = .01) compared with the BM group in the MAC setting but not in the RIC setting (Table 3) . Patients with SR-aGVHD required second-line treatment with therapies including extracorporeal photopheresis, ruxolitinib, tacrolimus, sirolimus, vedolizumab, and infliximab. Four patients required reinitiation of high-dose corticosteroids on multiple occasions due to recurrence of GVHD after tapering off steroids.

The cumulative incidence of cGVHD at 1 year was 8% (95% CI, 4% to 13%) in the BM group versus 22% (95% CI, 14% to 36%) in the PB group (HR, 3.0; 95% CI, 1.4 to 6.6; P = .005). The cumulative incidence of extensive cGVHD at 1 year in the 2 groups was 3% (95% CI, 1% to 7%) and 18% (95% CI, 10% to 31%), respectively (HR, 6.4; 95% CI, 2.2 to 17; P = .001) and the cumulative incidence of systemic therapy-requiring cGVHD was 2.5% (95% CI, 1% to 7%) and 14% (95% CI, 7% to 27%), respectively (HR, 5.6; 95% CI, 1.7 to 18; P = .004) (Figure 2 and B; Table 3 ). Similar findings were seen when analyzed by conditioning intensity, especially in the RIC setting (Table 3) . Among patients with therapy-requiring cGVHD, 66.7% (n = 14) had moderate cGVHD, 23.8% (n = 5) had severe cGVHD, and 9.5% (n = 2) had mild cGVHD as defined by the 2015 National Institutes of Health consensus criteria [11] . A majority of these patients (57.1%; n = 12) had skin involvement; 38.1% (n = 8) had ocular involvement, 28.5% (n = 6) had oral involvement, 33.3% (n = 7) had gastrointestinal tract involvement, and 2 patients had bronchiolitis obliterans with or without other organ involvement (Supplementary Table S2 ).

The median time to the development of grade II-IV aGVHD was 44 days (IQR, 32 to 73 days) in the BM group versus 55 days (IQR, 35 to 75 days) in the PB group, and that for grade III-IV aGVHD was 44 days (IQR, 32 to 97 days) versus 55 days (IQR, 40 to 119 days), respectively. The median time to the development of cGVHD in the 2 groups was 295 days (IQR, 245 to 344 days) and 240 days (IQR, 183 to 389 days), respectively, and that for systemic therapy-requiring cGVHD was 374 days (IQR, 253 to 524 days) and 244 days (IQR, 200 to 392 days), respectively.

In univariate analysis, receipt of a PB graft was associated with a significantly higher risk of SR-aGVHD (HR, 2.9; 95% CI, 1.1 to 7.3; P = .03) and systemic therapy-requiring cGVHD (HR, 5.6; 95% CI, 1.7 to 18; P = .004). CMV-seropositive patients and those with lymphoid (versus myeloid) malignancies had a higher risk of systemic therapy-requiring cGVHD. No other factor, including conditioning intensity, donor/recipient age, donor relationship, donor/recipient sex, ABO matching, or performance status, was a predictor of either SR-aGVHD or systemic therapy-requiring cGVHD. There were no significant predictors of grade III-IV aGVHD and overall cGVHD (Supplementary Table S3 ). In multivariate analysis, PB graft was the sole factor associated with a significantly higher risk of SR-aGVHD (HR, 2.9; 95% CI, 1.1 to 7.3; P = .03). Predictors of systemic therapy-requiring cGVHD included receipt of a PB graft (HR, 5.4; 95% CI, 1.7 to 17; P = .005) and lymphoid malignancies (HR, 5.4; 95% CI, 1.4 to 20; P = .01) ( Table 4) . No other variable, including female donor to male recipient, was a predictor of either aGVHD or cGVHD. Further subgroup analysis of patients with AML/MDS (n = 151) showed that receipt of a PB graft was associated with substantially greater risks of grade III-IV aGVHD (HR, 3.4; P = .02), SR-aGVHD (HR, 4.6; P = .02), cGVHD (HR, 9.6; P = .006), and systemic therapy-requiring cGVHD (HR N.E Q5 X X.; P = .001).

The rate of NRM at 1 year was 27% (95% CI, 22% to 35%) in the BM group and 28% (95% CI, 19% to 41%) in the PB group (HR, 1.1; 95% CI, 0.6 to 1.8; P = .8) ( Figure 3A ; Table 3 ). On multivariate analysis, age 50 years (HR, 2.8; 95% CI, 1.7 to 4.7; P < .001) and HCT-CI 3 (HR, 2.1; 95% CI, 1.2 to 3.5; P = .005) were predictors of NRM, where there was no effect of graft source (PB: HR, 1.1; 95% CI, 0.7 to 1.9; P = .6) ( Table 4 ). The rate of relapse at 1 year was 21% (95% CI, 16% to 28%) in the BM group and 20% (95% CI, 12% to 32%) in the PB group (HR, 0.8; 95% CI, 0.4 to 1.6; P = .6) ( Figure 3B ; Table 3 ). On multivariate analysis, there was no effect of graft source (PB: HR, 0.97; 95% CI, 0.5 to 1.9; P = .9). High/very high Disease Risk Index (DRI) (HR, 2.2; 95% CI, 1.2 to 3.8; P = .007) was the sole factor associated with a significantly higher risk of relapse (Table 4) . No differences in relapse or NRM between the BM and PB groups were noted when analyzed by conditioning intensity (Table 3) .

PFS at 1 year was 50% (95% CI, 43% to 58%) in the BM group and 52% (95% CI, 40% to 65%) in the PB group (HR, 0.98; 95% CI, 0.7 to 1.5; P = .9). The rate of OS at 1 year in the 2 groups was 58% (95% CI, 51% to 66%) and 61% (95% CI, 49% to 73%), respectively (HR, 0.9; 95% CI, 0.6 to 1.5; P = .8) (Figure 3C and D; Table 3 ). On multivariate analysis, there was no effect of graft source on OS (PB: HR, 1.1; 95% CI, 0.7 to 1.7; P = .7). Age 50 years (HR, 2.4; 95% CI, 1.5 to 3.6; P < .001), HCT-CI 3 (HR, 2.3; 95% CI, 1.5 to 3.6; P < .001), and high/very high DRI (HR, 1.7; 95% CI, 1.1 to 2.5; P = .01) were associated with worse OS (Table 4 ). There were no differences in survival between the BM and PB groups when analyzed by conditioning intensity (Table 3) . Seventy-two patients in the BM group and 27 in the PB group died by 1 year. Infection was the most common cause of death in the BM group (n = 25; 35%), followed by relapse (n = 24; 33%), and GVHD (n = 10; 14%). Despite the higher risk of infections and GVHD in the PB group, organ toxicity (cardiac, pulmonary, liver, or multiorgan failure) was the most common cause of death (n = 10; 37%), followed by relapse (n = 6; 22%), infection (n = 5; 18.5%), and GVHD (n = 3; »11%) (Supplementary Table S4 ).

At 1 year, GRFS was 48% (95% CI, 41% to 56%) in the BM group and 36% (95% CI, 25% to 48%) in the PB group ( Figure 3E ; Table 3 ). On multivariate analysis, there was a significant interaction between graft source and DRI. Receipt of a PB graft was associated with significantly worse GFRS (HR, 2.2; 95% CI, 1.4 to 3.5; P = .001) among patients with low/intermediate DRI. Age 50 years (HR, 1.9; 95% CI, 1.3 to 2.7; P < .001) and HCT-CI 3 (HR, 1.7; 95% CI, 1.2 to 2.4; P = .003) were significant predictors of worse GRFS (Table 4) . 

Day 100 (range, 70 to 147 days) comprehensive immune reconstitution data were available for 56 patients. Except for a greater number of class-switched memory B cells in the PB group compared with the BM group, there were no significant differences in any cell subset analyzed ( Table 5 ). The median absolute lymphocyte count (ALC) was 410 cells/mL (range, 0 to 1900 cells/mL) in the BM group and 530 cells/mL (range, 160 to 2100 cells/mL) in the PB group (P = .4). The absolute CD4 cell count (99 cells/mL versus 100 cells/mL), CD8 cell count (40 cells/mL versus 59 cells/mL), CD4/CD8 ratio (2.2 versus 1.6), absolute Treg count (9 cells/mL versus 10 cells/mL), absolute CD19 B cell count (38 cells/mL versus 51 cells/mL), and absolute NK cell count (165 cells/mL versus 226 cells/mL) were similar in in the 2 groups. Among B cell subsets, the numbers of naïve B cells and IgM memory B cells were similar in the 2 groups, but there were more class-switched memory B cells in the PB group compared with the BM group (median, 2 cells/mL [range, 0 to 25 cells/mL] versus 1 cell/mL [range, 0 to 1193 cells/mL]; P = .02). Among CD4 and CD8 cell subsets, the numbers of naïve, central memory, effector memory, and terminally differentiated effector cells were similar in the 2 groups. Similarly, no between-group differences were noted in Treg subsets, including central memory and naïve Tregs, or in NK cell subsets, including CD56(bright) and CD56(dim) NK cells. Similar results were noted at the day +180 evaluation in a subset of patients (n = 24) in whom comprehensive data were available, which showed no statistically significant between-group differences in any of the cell subsets analyzed (Supplementary  Table S5 ).

The cumulative incidence of any viral infection by day +180 was 7% (95% CI, 4% to 12%) in the BM group and 17% (95% CI, 10% to 27%) in the PB group (HR, 2.4; 95% CI, 1.1 to 5; P = .02). The cumulative incidence of bacterial infections by day +180 in the 2 groups was 4% (95% CI, 2% to 9%) versus 13% (95% CI, 8% to 23%), respectively (HR, 3.1; 95% CI, 1.3 to 7.7; P = .01).

Viral infections constituted the majority of infection events: 32 in the BM group and 26 in the PB group within 6 months of transplantation. These included BK virus (7 in the BM group versus 6 in the PB group), CMV (7 versus 6), adenovirus (0 versus 1), EBV (2 versus 1), human herpesvirus 6 (7 versus 5), herpes simplex virus (1 versus 1), respiratory syncytial virus (0 versus 2), parainfluenza (3 versus 1), rhinovirus (3 versus 1) and others (2 versus 2). Eight bacterial infection events occurred in the BM group, compared with 10 in the PB group. There was only 1 case of fungal infection. Despite the higher rate of infections in the PB group, infection was a less common cause of death in this group.

Ninety-seven patients (75 in the BM group and 22 in the PB group) who were alive and in remission and had at least 1 year of follow-up post-HCT were selected. Of these, 28 patients could not be reached after 2 attempts, 9 patients answered but did not return the survey, 11 patients were either international or did not speak English, and 1 patient refused to participate. The remaining 48 patients (33 in the BM group and 15 in the PB group) comprised the study cohort. The median patient age was 44 years (range, 19 to 72 years) in the BM group versus 45 years (range, 21 to 68 years) in the PB group (P = .71), and the median follow-up in the 2 groups was 1079 days (range, 449 to 2205 days) versus 960 days (range, 394 to 1605 days) (P = . 32) . No between-group differences were noted in the global FACT-BMT score or any subdomains (Supplementary Table S6 ). The most troubling symptoms reported were satisfaction Q7 X Xwith sex life, sleep, and perception of body image. When patients were asked whether they regretted undergoing HCT, the median response was "not at all."

We show that in the setting of haploidentical HCT with PTCy/Tac/MMF GVHD prophylaxis, the use of PB grafts was associated with a significantly higher risk of SR-aGVHD, overall cGVHD, systemic therapy-requiring cGVHD, and bacterial and viral infections compared with the use of BM grafts. Moreover, the PB group had no advantage in terms of engraftment, duration of hospitalization, immune reconstitution, relapse, NRM, or OS. Furthermore, PB was associated with significantly worse GFRS among patients with low/intermediate DRI.

An appreciably greater number of patients in the PB group developed SR-aGVHD or therapy-requiring cGVHD, which likely contributed to their increased risk of both bacterial and viral infections. Although these patients were effectively "salvaged" with treatment, and thus neither NRM nor survival differed between the groups, the clinical burden and morbidity associated with GVHD and its treatment cannot be captured by these statistics. Because the use of PTCy has reduced the risk of GVHD, a much larger study population will be needed to demonstrate any statistical differences in survival. The rate of grade III-IV aGVHD was also higher in the PB group than in the BM group (12% versus 7%), which, although clinically meaningful, did not reach statistical significance owing to the small number of events. The higher graft cell dose in the PB group might have contributed to the greater risk of GVHD [12] .

The rate of graft failure was generally low, and there were no noticeable clinically significant differences in either neutrophil or platelet engraftment between the BM and PB groups in the entire cohort. Among the patients who received RIC, the time to neutrophil engraftment, but not that of platelet engraftment, was 2 days faster in the PB group. This is likely a reflection of the lower BM harvest cell dose achieved than those reported in previous studies [6, 13, 14] . Although there is no specific optimal cell dose for transplantation, generally superior outcomes are seen with higher total nucleated cell and CD34 cell doses in both related and URD HCT [15À18], as well as improved survival in haploidentical HCT [19] .

Encouragingly, the pace of immune reconstitution was similar in the BM and PB groups. This is in contrast to the URD setting with conventional GVHD prophylaxis, in which receipt of a PB graft is associated with faster T cell immune reconstitution [20] , although this is expected to be different with PTCy prophylaxis [21] . Previous studies have assessed immune reconstitution after haploidentical HCT with PTCy and compared it with HCT with other donor types or with antithymocyte globulin [22À28]. However, a direct comparison of immune reconstitution by graft source in haploidentical setting has been lacking, which is provided by our present study. Given that GVHD can negatively affect thymic function and immune reconstitution [29] , the lower incidence of GVHD in the BM group might have facilitated immune recovery.

Several retrospective studies have assessed the differences in outcomes of haploidentical HCT with BM or PB grafts using PTCy prophylaxis and have yielded controversial results. Almost all the studies showed higher risks of aGVHD and/or cGVHD with PB grafts; one study showed an increased risk of both aGVHD and cGVHD [3] , one showed an increased risk of aGVHD but not of cGVHD [4] , and one showed an increased risk of cGVHD but not of aGVHD [5] , Only one study [3] found a greater risk of relapse with BM in patients with acute leukemia. On the other hand, a study in patients with acute lymphoblastic leukemia [6] showed significantly inferior PFS, OS, and GRFS and trends toward higher rates of grade II-IV aGVHD (HR, 1.52; P = .07), cGVHD (HR, 1.58; P = .05), and NRM (HR, 1.66; P = .06) with PB grafts. One reason for these variable outcomes seen in different studies may be related to the inclusion of patients across centers who are treated differently. Our study minimizes these biases and adds to the literature by providing crucial data on SR-aGVHD, therapy-requiring cGVHD, differences in immune reconstitution, and patient-reported QoL by graft source, which were lacking previously. Nevertheless, our outcomes should be validated in future studies involving larger numbers of PB recipients.

We acknowledge the limitations of our study, including a lack of data on the morbidity of GVHD as assessed by longterm complications, including the risk of avascular necrosis, and endocrine and cardiovascular complications, to name a few. Moreover, the QoL assessment was restricted to only those who had at least 1 year of follow-up and was done at the time of this study rather than at a fixed time point post-HCT, and thus it might not be representative of the entire cohort. Moreover, although the survey response rates can vary from 30% to 85% [30, 31] , the completion rate of approximately 50% noted in our study is consistent with previous studies [32] . Thus, the possibility of participation bias should be considered when assessing the QoL data. Future studies should consider the prospective collection of QoL data and a cost-effectiveness analysis and also consider assessing the T cell receptor excision circle and T cell receptor repertoire, data that were lacking in our study. Finally, although data on all post-HCT complications, including infections, are captured prospectively by a dedicated team at our institution, certain infection events might have been missed, especially in the early COVID-19 era for patients who were discharged to home after day +100 to follow-up with their local physicians. Most of those patients are still followed closely at our institution for the first 1 to 2 years post HCT. Because in-person follow-up visits for many patients were limited during the COVID-19 pandemic, most of the patients were still followed-up via virtual televisits.

Our data show a compelling benefit of using BM over PB grafts for haploidentical HCT with PTCy, tacrolimus, and MMF prophylaxis, which was associated with a significantly lower risk of severe aGVHD and cGVHD, fewer bacterial and viral infections, and comparable pace of recovery of neutrophils, platelets, and immune reconstitution, relapse, NRM, and survival.

BM: 58% (95% CI 51-66) PB: 61% (95% CI 49-73) HR: 0.9, 95% CI 0.6-1.5, p=0.8 Naïve Tregs CD3 + CD4 + CD25 + CD127 -CD45RO -CD62L + 1 (0-49) 0.5 (0-4) .9

Central memory Tregs CD3 + CD4 + CD25 + CD127 -CD45RO + CD62L + 8 (1-197) 8 (1-51) .9

Current use of and trends in hematopoietic cell transplantation in the United States

Graft cryopreservation does not impact overall survival after allogeneic hematopoietic cell transplantation using post-transplantation cyclophosphamide for graft-versus-host disease prophylaxis

Mobilized peripheral blood stem cells versus unstimulated bone marrow as a graft source for T-cell-replete haploidentical donor transplantation using post-transplant cyclophosphamide

Bone marrow versus mobilized peripheral blood stem cells in haploidentical transplants using posttransplantation cyclophosphamide

Risk factors for graft-versus-host disease in haploidentical hematopoietic cell transplantation using post-transplant cyclophosphamide

Bone marrow versus mobilized peripheral blood stem cell graft in T-cell-replete haploidentical transplantation in acute lymphoblastic leukemia

Quality of life measurement in bone marrow transplantation: development of the Functional Assessment of Cancer Therapy-Bone Marrow Transplant (FACT-BMT) scale

Consensus Conference on Acute GVHD Grading

Severity of chronic graft-versus-host disease: association with treatment-related mortality and relapse

Chronic graft-versus-host syndrome in man. A long-term clinicopathologic study of 20 Seattle patients

National Institutes of Health Consensus Development Project on Criteria for Clinical Trials in Chronic Graftversus-Host Disease: I. The 2014 Diagnosis and Staging Working Group report

Graft-versus-host disease in unmanipulated haploidentical marrow transplantation with a modified post-transplant cyclophosphamide (PT-CY) regimen: an update on 425 patients

The concentration of total nucleated cells in harvested bone marrow for transplantation has decreased over time

Bone marrow harvest from unrelated donors: up-to-date methodology

Quality of harvest and role of cell dose in unrelated bone marrow transplantation: an Italian Bone Marrow Donor Registry-Gruppo Italiano Trapianto di Midollo Osseo Study

Relevance of bone marrow cell dose on allogeneic transplantation outcomes for patients with acute myeloid leukemia in first complete remission: results of a European survey

Association of CD34 cell dose with hematopoietic recovery, infections, and other outcomes after HLA-identical sibling bone marrow transplantation

CD34 + cell dose predicts survival, posttransplant morbidity, and rate of hematologic recovery after allogeneic marrow transplants for hematologic malignancies

Grade II acute graft-versus-host disease and higher nucleated cell graft dose improve progression-free survival after HLA-haploidentical transplant with post-transplant cyclophosphamide

Kinetics of immune cell reconstitution predict survival in allogeneic bone marrow and G-CSF-mobilized stem cell transplantation

Delayed CD4 + T-cell but faster B-cell immune reconstitution after PTCY-based compared to conventional GVHD prophylaxis after allogeneic transplantation

Impaired T-and NK-cell reconstitution after haploidentical HCT with posttransplant cyclophosphamide

A prospective observational study of immune reconstitution following transplantation with post-transplant reduced-dose cyclophosphamide from HLA-haploidentical donors

Immune reconstitution after T-cell replete HLAhaploidentical transplantation

Impact of KIR/HLA incompatibilities on NK cell reconstitution and clinical outcome after T cell-replete haploidentical hematopoietic stem cell transplantation with posttransplant cyclophosphamide

Impact on early outcomes and immune reconstitution of high-dose post-transplant cyclophosphamide vs anti-thymocyte globulin after reduced-intensity conditioning peripheral blood stem cell allogeneic transplantation

NK cell recovery after haploidentical HSCT with posttransplant cyclophosphamide: dynamics and clinical implications

Unmanipulated haploidentical transplants compared with other alternative donors and matched sibling grafts

Acute graft-versus-host disease transiently impairs thymic output in young patients after allogeneic hematopoietic stem cell transplantation

Nonresponse bias in a survey of patient perceptions of hospital care

Recovery after stem-cell transplantation for hematologic diseases

Patient satisfaction and survey response in 717 hospital surveys in Switzerland: a cross-sectional study

Financial disclosure: Q8 X X Conflict of interest statement: R.S.M. has received research funding from CSL Behring, Kadmon, and Incyte. S.O.C. has served as an advisory board member for Hansa, Spectrum, MolMed, Cytosen, Kiadis, Allogene, CareDx, Acrotech, and Cellularity, and has research funding from Miltenyi Biotec and Kiadis. K.R. has received researching funding from Affimed Therapeutics and Pharmacyclics; has served on advisory boards for Adicet Bio, ViroGen, and GemoAb; holds a patent on the generation of BKV CTLs for the treatment of HC or PML and a patent on the generation of CAR NK cells; and has license and research agreements with Takeda to develop CB-CAR NK cells for the treatment of B cell malignancies and other cancers. The other authors have no conflicts of interest to report. Data availability statement: Deidentified data may be available upon request to the study's principal investigator and will require Institutional Review Board approval.

Supplementary material associated with this article can be found in the online version at doi:10.1016/j.jtct.2021.09.003.