Haploidentical Stem Cell Transplantation With Posttransplant Cyclophosphamide Therapy vs Other Donor Transplantations in Adults With Hematologic Cancers

Key Points

Question

What are the outcomes of haploidentical stem cell transplantation with posttransplant cyclophosphamide therapy compared with other transplant types in adults with hematologic cancers?

Findings

In this meta-analysis of 30 studies including 22 974 patients, the use of haploidentical stem cell transplantation with posttransplant cyclophosphamide therapy appeared to show increased all-cause mortality compared with matched related donors, similar all-cause mortality compared with matched unrelated donors, and reduced all-cause mortality compared with mismatched unrelated donors. Relapse appeared to be increased in lymphoma and similar in acute leukemia compared with matched unrelated donors.

Meaning

Haploidentical stem cell transplantation with posttransplant cyclophosphamide therapy may be preferred to mismatched unrelated donors, and prospective comparisons with matched donors are needed.

Abstract

Importance

Use of haploidentical (HAPLO) stem cell transplantation with posttransplant cyclophosphamide is rapidly increasing in adults with hematologic cancers. However, its specific role compared with other transplant strategies has yet to be identified.

Objective

To synthesize the existing evidence regarding outcomes of stem cell transplantations comparing HAPLO stem cell transplantation and posttransplant cyclophosphamide therapy with transplantations from matched related donors (MRDs), matched unrelated donors (MUDs), or mismatched unrelated donors (MMUDs).

Data Sources

PubMed, Cochrane Library, ClinicalTrials.gov, and meeting abstracts were searched for the key words haploidentical and cyclophosphamide from inception through March 1, 2019.

Study Selection

Studies comparing HAPLO stem cell transplantation and posttransplant cyclophosphamide therapy with transplantations from other donors in adults with hematologic cancers were eligible for meta-analysis.

Data Extraction and Synthesis

Pooled odds ratios (ORs) and 95% CIs were calculated using a random-effects model.

Main Outcomes and Measures

Main outcomes were all-cause mortality, nonrelapse mortality, and relapse.

Results

A total of 30 studies including 22 974 participants were analyzed. HAPLO stem cell transplantation with posttransplant cyclophosphamide therapy was associated with increased all-cause mortality compared with MRDs (OR, 1.17; 95% CI, 1.05-1.30), similar all-cause mortality compared with MUDs (OR, 1.06; 95% CI, 0.96-1.18), and reduced all-cause mortality compared with MMUDs (OR, 0.75; 95% CI, 0.61-0.92). Regarding nonrelapse mortality, HAPLO stem cell transplantation with posttransplant cyclophosphamide was associated with worse outcomes compared with MRDs (OR, 1.20; 95% CI, 1.04-1.40) but better outcomes compared with MUDs (OR, 0.75; 95% CI, 0.61-0.92) and MMUDs (OR, 0.51; 95% CI, 0.25-1.02). In terms of relapse, HAPLO stem cell transplantation with posttransplant cyclophosphamide was associated with similar outcome compared with MRDs (OR, 1.01; 95% CI, 0.86-1.17) and MMUDs (OR, 1.06; 95% CI, 0.77-1.47) but showed increased relapse compared with MUDs (OR, 1.20; 95% CI, 1.03-1.40).

Conclusions and Relevance

Results of this meta-analysis suggest that MRDs, if available, remain the optimal donors regarding mortality and HAPLO stem cell transplantation with posttransplant cyclophosphamide may be preferred over MMUDs. Prospective comparisons with MUDs are needed.

This systematic review and meta-analysis assesses outcomes of stem cell transplantations using the haploidentical stem cell transplantation with posttransplant cyclophosphamide therapy compared with other mismatched or matched donor transplants in adult patients with hematologic cancers.

Introduction

Allogeneic hematopoietic stem cell transplantation with human leukocyte antigen (HLA)–identical siblings or unrelated donors is a curative treatment option for several hematologic cancers.¹ Over the past 15 years, HLA-haploidentical donors (HAPLOs) were increasingly used for stem cell transplantation for patients who lack an HLA-matched donor or those who need an urgent allograft.¹ These donors are siblings or other family members of the patient who are matched at all HLA loci on 1 chromosome. Considering HAPLOs may provide the possibility of finding a donor more quickly and may therefore potentially reduce overall costs.² Historically, however, early HAPLO attempts with or without in vivo or ex vivo T-cell depletion have been limited by higher rates of graft-vs-host disease (GVHD), nonrelapse mortality, and graft rejection.^3,4

Currently, rates of stem cell transplantation using HAPLOs continue to expand around the world, owing primarily to the application of posttransplant cyclophosphamide therapy, a low-cost and easy-to-use alkylating agent, which has been found to successfully control the intense alloreactive reactions seen in this setting.^5,6,7 Using the novel posttransplant cyclophosphamide approach, pioneered at The Johns Hopkins University,⁸ such transplantations have been effective at deleting alloreactive donor T cells, limiting GVHD as well as early treatment-related mortality, although questions remain about effective prevention of relapse using HAPLO grafts.⁹ To our knowledge, only 4 reports synthesized findings from studies on HAPLOs, but these reports either summarized different HAPLO strategies using various GVHD prophylaxis, included pediatric patients, or only investigated the role of graft type.^10,11,12,13

Herein, we aimed to synthesize the recent evidence regarding outcomes of stem cell transplantations specifically using the HAPLO stem cell transplantation with posttransplant cyclophosphamide strategy compared with other mismatched or matched donor transplants in adult patients with hematologic cancers.

Methods

Review and meta-analysis methods followed the Cochrane Handbook,¹⁴ and reporting was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline¹⁵ and the Meta-analysis of Observational Studies in Epidemiology (MOOSE) Checklist.¹⁶

Search Strategy

Medline and the Cochrane Library were searched (from inception until March 1, 2019). In addition, meeting abstracts archived between 2017 and 2019 from the American Society of Hematology, American Society for Transplantation and Cellular Therapy, Center for International Blood and Marrow Transplant Research, and European Society for Blood and Marrow Transplantation were searched. Review of ClinicalTrials.gov was performed until March 1, 2019. The search strategy consisted of key words specific to each database, considered all trial designs of humans, and was not restricted by language. Search terms included all subject headings and associated key words for haploidentical and cyclophosphamide. In addition, the reference lists of relevant studies and reviews were reviewed.

Study Selection and Data Extraction

Two reviewers (N.G. and N.K.) independently screened titles, abstracts, and the full text of relevant articles based on prespecified inclusion and exclusion criteria. Disagreements were resolved by consensus and by a third reviewer (A.B.) or contacting the study authors when needed. Studies were included if they fulfilled the following criteria: adult patients with hematologic cancers; prospective or retrospective studies reporting on more than 10 patients undergoing HAPLO stem cell transplantation with posttransplant cyclophosphamide therapy; comparison with transplants from matched related donors (MRDs), matched unrelated donors (MUDs), mismatched unrelated donors (MMUDs), or peripheral blood or bone marrow as graft type; and full data on conditioning intensity.

The following information was extracted from the included studies: the name of the first author, year of publication, study design, GVHD prophylaxis, type of disease, disease status at transplantation, disease risk at transplantation, type of transplantation, number of participants, type of conditioning regimen, and length of follow-up. Definitions from the primary studies for disease status at transplantation were used. High-risk groups in the studies were mostly defined according to the disease risk index or the comorbidity index score 3 or greater as previously reported.^17,18 Main end points for data synthesis were all-cause mortality, nonrelapse mortality, relapse, overall chronic GVHD, acute GVHD grades 2 to 4 and 3 plus 4 at day 100 or 180 post transplant, and GVHD-free and graft relapse-free survival (GRFS). The composite end point of GRFS was defined as being alive without evidence of relapse, and without acute GVHD grades 3 plus 4 or chronic GVHD requiring immunosuppressive treatment or severe chronic GVHD.^19,20

Risk of Bias and Quality Assessment

We addressed the risk of bias using the tool for assessing risk of bias in nonrandomized studies of interventions developed by the Cochrane Collaboration.²¹ Response options for each domain are yes, probably yes (low risk of bias), probably no, and no (high risk of bias). Domains for potential bias are the following: confounding, selection of participants, classification of intervention, deviations from intended interventions, missing data, measurement of outcome, and selection of the reported result. Overall judgment for risk of bias was categorized as low, moderate, serious, or critical.

We evaluated the certainty of the evidence for each outcome using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach²² including considerations of risk of bias, inconsistency, indirectness, imprecision, and publication bias. The resulting overall certainty of the evidence was assessed as high, moderate, low, or very low. We then created a GRADE evidence profile using GRADEpro software.²³

Statistical Analysis

Odds ratios (ORs) and 95% CIs were calculated by pooling the results from studies using the Mantel-Haenszel method and the random-effects model. Heterogeneity was assessed using the I² index and was deemed as moderate to high with an I² value greater than 50%.¹⁴ Prespecified subgroups for analyses were donor relation and disease (ie, studies evaluating HAPLO stem cell transplantation with posttransplant cyclophosphamide only in patients with lymphoma or acute leukemia). To examine the association of prespecified continuous moderator variables with study effect size, a mixed-effects model was used for meta-regression.²⁴ With 2-tailed, unpaired testing, all values with P < .05 were considered statistically significant. Analyses were performed with R statistical software, version 3.4.3, using the meta²⁵ and metafor²⁶ packages.²⁷

Results

A total of 2315 unique citations were identified from the electronic database search and other sources, including meeting abstracts. Based on title and abstract screening, 2197 citations were excluded and 88 citations were excluded based on full-text screening. Reasons for exclusion were pediatric population; lack of direct comparison results; HAPLO group not restricted to patients receiving posttransplant cyclophosphamide; no explicit distinction of control groups into MRD, MUD, or MMUD cohorts; umbilical cord as the only comparator; and nonhematologic cancers. Thirty studies^{28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57} were included in quantitative analyses (Figure 1).

Figure 1. Study Retrieval and Identification for Meta-analysis.

Study inclusion and exclusion. MMUD indicates mismatched unrelated donor; MRD, matched related donor; and MUD, matched unrelated donor.

Study characteristic details are reported in eTable 1 in the Supplement. A total of 22 974 participants were included in the 30 studies.^{28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57} One study was a prospective phase 2 trial,²⁸ while the remaining 29 studies were retrospective multicenter, registry, or single-center studies. The reference groups for HAPLO stem cell transplantation with posttransplant cyclophosphamide were the following: MRD only (11 studies), MUD only (7 studies), MRD and MUD (8 studies), MMUD only (1 study), MUD and MMUD (2 studies), and MRD, MUD, and MMUD (1 study). In the HAPLO stem cell transplantation with posttransplant cyclophosphamide group, 5 studies only used peripheral blood as graft type and 3 studies only evaluated bone marrow transplants. Patients received either reduced-intensity only (4 studies), myeloablative only (4 studies), nonmyeloablative only (1 study), or various regimens (21 studies). Follow-up ranged from 18 to 49 months.

Risk of Bias and Publication Bias

Low risk of bias was assessed in 26 studies. Assessment yielded moderate risk of bias in 3 studies and critical risk of bias in 1 study. Overall, the risk of bias of the included studies was low and is summarized in eTable 2 in the Supplement. The main dimension for possible risk of bias was missing data owing to the retrospective nature of most studies.

Visual inspection of funnel plots and the Begg test for each outcome revealed no significant asymmetry, arguing against publication bias (eFigure 1 in the Supplement). Review of ClinicalTrials.gov revealed no additional studies with results not already included in the literature. In accordance with the GRADE approach, quality assessment started at moderate quality of evidence after downgrading owing to the retrospective design of most studies. Overall quality of evidence ranged from very low to moderate. A summary of the quality of evidence and assessment of each dimension according to outcome is depicted in eTable 3 in the Supplement.

Mortality and Relapse

All-cause mortality was assessed in all 30 studies at 18 to 49 months’ follow-up.^{28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57} The overall pooled OR was 1.06 (95% CI, 0.99-1.15; P = .10) with no relevant heterogeneity (I² = 11%) (eFigure 2 in the Supplement). Subgroup analyses showed difference in outcomes according to the comparator (P = .002). The pooled ORs were 1.17 (95% CI, 1.05-1.30; I² = 1%) compared with MRDs and 1.06 (95% CI, 0.96-1.18; I² = 0%) compared with MUDs. HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with better outcomes compared with MMUDs, with a pooled OR of 0.79 (95% CI, 0.65-0.97; I² = 0%). The estimated certainty of the evidence for the overall comparison was assessed as moderate.

Nonrelapse mortality was assessed in 28 studies (20 862 patients).^{28,29,30,32,33,34,35,36,37,38,39,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57} The pooled OR was 0.88 (95% CI, 0.75-1.03; P = .11) (Figure 2) with observed heterogeneity (I² = 60%). Subgroup analyses identified possible differences in outcome according to the comparator (P < .001). HAPLO stem cell transplantation with posttransplant cyclophosphamide was associated with a possibly increased nonrelapse mortality compared with MRDs, with an OR of 1.20 (95% CI, 1.04-1.40; I² = 5%). In contrast, in pooled ORs, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with better outcomes compared with MUDs (OR, 0.75; 95% CI, 0.61-0.92; I² = 46%) and MMUDs (0.51; 95% CI, 0.25-1.02; I² = 86%). The estimated certainty of the evidence was assessed as low because of inconsistency owing to relevant heterogeneity.

Figure 2. Nonrelapse Posttransplantation Mortality.

Comparison of transplantations from haploidentical donors and posttransplant cyclophosphamide (HAPLO-PTC) with matched related donors (MRDs), matched unrelated donors (MUDs), and mismatched unrelated donors (MMUDs) regarding outcome in nonrelapse mortality. HAPLO indicates haploidentical; OR, odds ratio.

Relapse was assessed in 29 studies comprising 22 681 patients.^{28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,54,55,56,57} The overall pooled OR was 1.09 (95% CI, 0.98-1.21; P = .11) (Figure 3). No relevant heterogeneity (I² = 47%) and no between-group differences were identified (P = .27). With pooled OR analysis, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with similar outcomes in relapse compared with MRDs (OR, 1.01; 95% CI, 0.86-1.17; I² = 43%) and MMUDs (OR, 1.06; 95% CI, 0.77-1.47; I² = 55%). HAPLO stem cell transplantation with posttransplant cyclophosphamide was associated with increased relapse compared with MUDs, showing a pooled OR of 1.20 (95% CI, 1.04-1.40; I² = 42%). The estimated certainty of the evidence for the overall comparison was moderate.

Figure 3. Posttransplantation Relapse.

Comparison of haploidentical donors and posttransplant cyclophosphamide (HAPLO-PTC) with matched related donors (MRDs), matched unrelated donors (MUDs), and mismatched unrelated donors (MMUDs) regarding outcome in relapse. HAPLO indicates haploidentical; OR, odds ratio.

Chronic GVHD

Twenty-seven studies (17 115 patients) assessed chronic GVHD.^{28,29,30,31,32,33,34,35,37,38,39,41,42,43,44,45,46,47,48,49,50,51,52,53,54,56,57} Overall, HAPLO stem cell transplantation with posttransplant cyclophosphamide was associated with better outcomes, with a pooled OR of 0.49 (95% CI, 0.39-0.61; P < .001). Heterogeneity was identified (I² = 80%) (eFigure 3 in the Supplement). The pooled ORs were 0.46 (95% CI, 0.33-0.62; I² = 81%) compared with MRDs, 0.49 (95% CI, 0.34-0.71; I² = 80%) compared with MUDs, and 0.74 (95% CI, 0.54-1.03; I² = 21%) compared with MMUDs. The overall estimated certainty of the evidence was judged to be moderate after upgrading for strong pooled effect estimate.

Acute GVHD

Twenty-three studies (13 795 patients) assessed acute GVHD grades 3 plus 4.^{28,29,30,32,33,34,35,38,39,41,42,43,44,45,46,47,49,50,51,52,53,56,57} HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with reduced acute GVHD grades 3 plus 4, showing a pooled OR of 0.72 (95% CI, 0.60-0.88; P < .001), and heterogeneity was not observed (I² = 34%) (eFigure 4 in the Supplement). A between-group difference was detected (P = .008). The pooled OR was 0.89 (95% CI, 0.70-1.13; I² = 18%) compared with MRDs. HAPLO stem cell transplantation with posttransplant cyclophosphamide was associated with reduced acute GVHD grades 3 plus 4 compared with MUDs (OR, 0.63; 95% CI, 0.48-0.83; I² = 23%) and MMUDs (OR, 0.38; 95% CI, 0.23-0.64; I² = 0%). The overall estimated certainty of the evidence was assessed as moderate.

Twenty-five studies (16 389 patients) assessed acute GVHD grades 2-4.^{28,29,30,32,33,34,35,38,39,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56} The overall pooled OR was 0.95 (95% CI, 0.80-1.13; P = .56). Heterogeneity was relevant (I² = 72%) (eFigure 5 in the Supplement) and between-group difference was identified (P < .001). HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with increased acute GVHD grades 2 to 4 compared with MRDs (OR, 1.32; 95% CI, 1.07-1.62; I² = 59%). In contrast, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with better outcome compared with MUDs (OR, 0.76; 95% CI, 0.62-0.93; I² = 42%) and MMUDs (OR, 0.51; 95% CI, 0.32-0.81; I² = 64%). The estimated certainty of the evidence for the overall comparison was assessed as very low owing to serious inconsistency and imprecision in the overall pooled estimate.

Graft Relapse-Free Survival

The composite end point of GRFS was assessed in 14 studies (9305 patients).^{31,34,38,39,40,41,42,44,45,47,52,53,54,56} Overall, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with increased GRFS, with a pooled OR of 0.80 (95% CI, 0.66-0.97; P = .02). Relevant heterogeneity (I² = 60%) (eFigure 6 in the Supplement), but no between-group difference, was identified (P = .36). HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with similar GRFS compared with MRDs (OR, 0.87; 95% CI, 0.66-1.15; I² = 60%) and better GRFS compared with MUDs (OR, 0.69; 95% CI, 0.52-0.93; I² = 60%). Only one study, by Brissot et al,⁴² evaluated the comparison with MMUDs and showed an OR of 0.99 (95% CI, 0.62-1.58). The estimated overall certainty of the evidence was assessed as low because of inconsistency owing to observed heterogeneity.

Outcomes in Lymphoma and Acute Leukemia

Subgroup analyses according to disease were performed for HAPLO stem cell transplantation with posttransplant cyclophosphamide compared with matched donors because only 1 study compared HAPLO stem cell transplantation with posttransplant cyclophosphamide and MMUDs.⁴² Seven studies (4984 patients) reported on HAPLO stem cell transplantation with posttransplant cyclophosphamide outcomes in patients with lymphoma, including 3 on Hodgkin lymphoma,^38,51,56 1 on diffuse large B-cell lymphoma,⁴⁷ and 3 on various lymphomas.^32,43,53 HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with an increase in nonrelapse mortality compared with MRDs (OR, 1.35; 95% CI, 1.02-1.79) and better outcomes compared with MUDs (OR, 0.77; 95% CI, 0.59-1.00). HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with reduced relapse compared with MRDs (OR, 0.78; 95% CI, 0.63-0.96) and increased relapse compared with MUDs (OR, 1.25; 95% CI, 1.00-1.56). Overall, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with reduced chronic GVHD and improve GRFS. The remaining pooled ORs are listed in the Table.

Table. Outcomes of HAPLO Stem Cell Transplantation With Posttransplant Cyclophosphamide Compared With Matched Donors According to Disease.

Donor	Lymphoma		Acute Leukemia
	Random-Effects Model OR (95% CI)	I2 (P Value)	Random-Effects Model OR (95% CI)	I2 (P Value)
All-Cause Mortality
Overall	1.01 (0.87-1.16)	0 (.84)	1.13 (1.01-1.25)	0 (.83)
Matched related	1.07 (0.89-1.29)	0 (.84)	1.11 (0.96-1.30)	0 (.54)
Matched unrelated	0.92 (0.75-1.14)	0 (.89)	1.14 (0.98-1.32)	0 (.78)
Nonrelapse Mortality
Overall	1.07 (0.83-1.37)	43 (.07)	1.02 (0.84-1.24)	43 (.06)
Matched related	1.35 (1.02-1.79)	17 (.30)	1.20 (0.99-1.45)	0 (.49)
Matched unrelated	0.77 (0.59-1.00)	0 (.94)	0.89 (0.67-1.20)	48 (.09)
Relapse
Overall	0.93 (0.76-1.15)	49 (.04)	1.05 (0.91-1.20)	23 (.21)
Matched related	0.78 (0.63-0.96)	19 (.29)	0.98 (0.84-1.14)	0 (.86)
Matched unrelated	1.25 (1.00-1.56)	0 (.78)	1.06 (0.80-1.40)	55 (.05)
Chronic GVHD
Overall	0.30 (0.20-0.44)	79 (<.001)	0.64 (0.43-0.96)	82 (<.001)
Matched related	0.31 (0.17-0.55)	84 (<.001)	0.55 (0.33-0.92)	85 (<.001)
Matched unrelated	0.28 (0.16-0.51)	73 (.01)	0.81 (0.45-1.47)	67 (.02)
Acute GVHD Grades 3-4
Overall	0.57 (0.43-0.75)	12 (.33)	0.83 (0.60-1.05)	7 (.38)
Matched related	0.65 (0.47-0.90)	0 (.44)	1.00 (0.74-1.34)	0 (.60)
Matched unrelated	0.45 (0.29-0.70)	13 (.33)	0.65 (0.47-0.91)	0 (.62)
Acute GVHD Grades 2-4
Overall	1.20 (0.96-1.51)	56 (.01)	1.03 (0.80-1.32)	61 (.005)
Matched related	1.47 (1.15-1.88)	34 (.18)	1.16 (0.97-1.39)	9 (.36)
Matched unrelated	0.89 (0.71-1.10)	0 (.73)	0.82 (0.53-1.27)	62 (.03)
GRFS
Overall	0.74 (0.51-1.09)	70 (.002)	1.00 (0.81-1.25)	0 (.50)
Matched related	0.72 (0.36-1.42)	82 (<.001)	1.02 (0.78-1.34)	0 (.45)
Matched unrelated	0.76 (0.51-1.14)	42 (.18)	0.87 (0.46-1.64)	40 (.20)

Abbreviations: GRFS, graft-vs-host disease-free and relapse-free survival; GVHD, graft-vs-host disease; HAPLO, haploidentical; OR, odds ratio.

In further analyses of lymphoma studies, a between-group difference according to subtype was identified for chronic GVHD (P = .003) and acute GVHD grades 3 plus 4 (P = .04). The association of HAPLO stem cell transplantation with posttransplant cyclophosphamide and decreased chronic GVHD appeared to be stronger in diffuse large B-cell lymphoma (OR, 0.28; 95% CI, 0.20-0.38) compared with Hodgkin lymphoma (OR, 0.47; 95% CI, 0.26-0.85). Regarding acute GVHD grades 3 plus 4, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with a reduced incidence in diffuse large B-cell lymphoma (OR, 0.47; 95% CI, 0.29-0.79) and similar incidence in Hodgkin lymphoma (OR, 0.89; 95% CI, 0.58-1.39). No difference according to subtype was apparent for the remaining outcomes.

Eleven studies (13 240 patients)^{30,35,36,39,41,42,46,49,50,52,54} included only patients with acute leukemia, of which 8 studies analyzed acute myeloid leukemia,^{30,35,39,41,42,49,50,54} 1 study analyzed acute lymphoblastic leukemia,³⁶ and 2 studies included both.^46,52 Overall, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with increased all-cause mortality (pooled OR, 1.13; 95% CI, 1.01-1.25). HAPLO stem cell transplantation with posttransplant cyclophosphamide seemed to be associated with increased nonrelapse mortality compared with MRDs (OR, 1.20; 95% CI, 0.99-1.45) and decreased nonrelapse mortality compared with MUDs (OR; 0.89; 95% CI, 0.67-1.20). HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with similar outcomes in relapse overall (OR, 1.05; 95% CI, 0.91-1.20), with pooled ORs of 0.98 (95% CI, 0.84-1.14) compared with MRDs and 1.06 (95% CI, 0.80-1.40) compared with MUDs. The remaining pooled ORs for GVHD and GRFS are listed in the Table.

Discussion

This meta-analysis of 30 studies including 22 974 patients compared outcomes of allogeneic hematopoietic stem cell transplantation using HAPLO stem cell transplantation with posttransplant cyclophosphamide with those from other matched or mismatched donors for adult patients with hematologic cancers. Matched related donors, if available, appear to remain the preferred option regarding mortality. However, HAPLO stem cell transplantation with posttransplant cyclophosphamide seemed to be associated with reduced chronic GVHD and better GRFS. Furthermore, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with reduced nonrelapse mortality but increased relapse compared with MUDs. In comparison with MMUDs, HAPLO stem cell transplantation with posttransplant cyclophosphamide may be the preferred treatment option in terms of mortality, relapse, and GVHD.

Subgroup analyses identified possible differences in outcomes according to underlying diseases for the comparison of HAPLO stem cell transplantation with posttransplant cyclophosphamide and matched donors. In studies including only patients with lymphoma, outcomes in all-cause mortality appeared to be similar. In terms of relapse, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with better outcomes compared with MRDs but worse outcomes compared with MUDs. The association of HAPLO stem cell transplantation with posttransplant cyclophosphamide with reduced chronic and acute GVHD seemed to be even more pronounced. In contrast, in studies evaluating only patients with acute leukemia, all-cause mortality apparently was increased after transplants using HAPLO stem cell transplantation with posttransplant cyclophosphamide compared with matched donors. Nevertheless, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with similar outcomes in relapse compared with matched donors.

Our results may reflect the trend toward an increased use of HAPLO stem cell transplantation with posttransplant cyclophosphamide over the past years.⁷ Half of the 30 studies included in this meta-analysis were published between 2018 and early 2019. A current position statement suggests that HAPLO stem cell transplantation with posttransplant cyclophosphamide may be an option for patients lacking a matched sibling or unrelated donor.⁵⁸ Moreover, insufficient data were available to prefer a certain HAPLO strategy over MMUDs.^58,59 Hence, recommendations may benefit from evidence specific to the HAPLO stem cell transplantation with posttransplant cyclophosphamide strategy. Our results, in association with previous data syntheses of various HAPLO strategies,¹⁰ suggest what may be a stronger association with GVHD and nonrelapse mortality favoring HAPLO stem cell transplantation with posttransplant cyclophosphamide compared with MUDs, suggesting that the existing association with HAPLO transplantation itself may even be intensified in conjunction with posttransplant cyclophosphamide. However, the role of relapse and the utility of the composite end point of GRFS remains to be addressed in a prospective manner.

In standard practice, posttransplant cyclophosphamide is given at 50 mg/kg/d on posttransplantation days 3 and 4 and is used alone or in combination with other GVHD prophylactic agents depending on the characteristics, including donor type.⁶⁰ Regardless of the strategy, posttransplant cyclophosphamide is historically known to prevent both severe acute and chronic GVHD,^3,4 minimize the need for other immunosuppression,^61,62 and preserve infectious immunity.^63,64,65 Ultimately, use of this therapy may result in low rates of nonrelapse mortality, which could be confirmed in our analysis with significantly lower rates after HAPLO stem cell transplantation with posttransplant cyclophosphamide in comparison with unrelated donors, while nonrelapse mortality was higher compared with MRDs.

Recent findings suggest that posttransplant cyclophosphamide may not fully eliminate alloreactive T cells.⁶⁶ Rather, posttransplant cyclophosphamide may induce functional impairment of CD4⁺ and CD8⁺ alloreactive T cells, thereby preventing new donor T cells from causing GVHD.^66,67 In our analysis, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be strongly associated with reduced chronic and acute GVHD grades 3 plus 4, while the association with acute GVHD grades 2 to 4 was less pronounced. This finding is in line with previous reports identifying frequent occurrence of grade 2 acute GVHD after posttransplant cyclophosphamide, which may be even related to better cancer control.^68,69,70

The results of our analysis may strengthen the evidence of HAPLO stem cell transplantation with posttransplant cyclophosphamide as a potential alternative to MUD transplants. Furthermore, HAPLO stem cell transplantation with posttransplant cyclophosphamide provides higher donor availability (>95%), faster graft acquisition, and shorter time between collection and infusion, and may ease the procedure of a second transplantation owing to easier repeated donations.⁷¹ In terms of posttransplant outcome, we found that HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with reduced nonrelapse mortality, chronic GVHD, acute GVHD grades 3 plus 4, and better GRFS. Regarding relapse, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to be associated with increased relapse in patients with lymphoma but similar outcome in acute leukemia. These results need to be validated in prospective controlled trials.

Limitations

Our study has several limitations. Concurrent GVHD prophylactic agents differed between studies, which could not be controlled for, limiting the interpretation of specific comparisons. In studies comparing HAPLO stem cell transplantation with posttransplant cyclophosphamide with other donors receiving antithymocyte globulin for GVHD prophylaxis, the association between HAPLO stem cell transplantation with posttransplant cyclophosphamide and GVHD and nonrelapse mortality was smaller, while all-cause mortality and relapse outcomes were even similar compared with MUDs.^38,41,47 Furthermore, we conducted our meta-analysis at the study level; variables at the patient level were not available. Our meta-analysis is limited by the heterogeneous population included in the studies, most of which were retrospective.

Sources of heterogeneity may be different graft types used for transplantation, conditioning, diseases, or disease status before transplantation. First, no confounding association for graft type or conditioning intensity was identified in meta-regression, which is supported by previous data synthesis, prospective, and retrospective studies.^{13,72,73,74,75,76} Second, we aimed to address the association between different diseases in subgroup analyses of studies including only patients with lymphoma or acute leukemia. In lymphoma, comparing HAPLO stem cell transplantation with posttransplant cyclophosphamide showed possibly similar all-cause mortality and relapse compared with MRDs but increased relapse compared with MUDs, which is in line with one of the first studies on HAPLO stem cell transplantation with posttransplant cyclophosphamide in adults and children with Hodgkin lymphoma over a decade ago.⁷⁷

In contrast, HAPLO stem cell transplantation with posttransplant cyclophosphamide appeared to show increased all-cause mortality but similar relapse in studies of patients with acute leukemia. Third, different disease status may affect posttransplant outcome, which we aimed to account for using meta-regression. Therefore, we included the number of patients who were in complete remission before HAPLO stem cell transplantation with posttransplant cyclophosphamide therapy and found no confounding factors for the comparison with other donor transplants overall and in studies including only lymphoma or acute leukemia. One recent study⁵⁰ in acute myeloid leukemia suggested that associations between donor type and outcome did not depend even on status of minimal residual disease before transplantation. However, evidence remains limited and studies are needed specifically evaluating the role of transplant- and disease-related factors on outcome of HAPLO stem cell transplantation with posttransplant cyclophosphamide compared with other donors.

Conclusions

Matched related donors, if available, appear to remain the optimal donor in terms of mortality after stem cell transplantation. Nevertheless, the results presented herein indicate that HAPLO stem cell transplantation with posttransplant cyclophosphamide therapy may be preferred in terms of chronic GVHD and GRFS. Furthermore, HAPLO stem cell transplantation with posttransplant cyclophosphamide seems to be preferred in comparison with MMUDs. Prospective controlled trials are warranted to further investigate the role of HAPLO stem cell transplantation with posttransplant cyclophosphamide with respect to relapse, especially compared with MUDs. These findings could facilitate clinical decision making and assist in the design and interpretation of future trials.

Supplement.

eTable 1. Study Characteristics

eTable 2. Risk of Bias

eTable 3. Quality of Evidence

eFigure 1. Funnel Plots

eFigure 2. All-Cause Mortality

eFigure 3. Chronic GVHD

eFigure 4. Acute GVHD Grades 3 + 4

eFigure 5. Acute GVHD Grades 2-4

eFigure 6. GRFS