1 Introduction
Hematopoietic stem cell transplantation (HSCT) is the most powerful curative therapy for the majority of hematological malignancies. However, human leukocyte antigen (HLA)-matched HSCT donors have long been scarce in China due to a deficiency of HLA-identical siblings and a relatively small unrelated donor (URD) program [
1,
2]. Haploidentical (haplo) family donors, such as parents or children, offer the benefits of rapid and nearly universal donor availability. However, the HLA barrier of haplo-HSCT has been formidable until recent decades due to the high incidence of rejection and severe graft-versus-host disease (GvHD) [
3–
5]. The three main approaches for haplo-HSCT in the world are as follows [
6–
8]: (1) T cell depletion (TCD)-based regimens, which originated from the Perugia group, Italy [
9–
11]; (2) granulocyte colony-stimulating factor (G-CSF) plus anti-thymocyte globulin (ATG)-based regimens with unmanipulated T cell replete graft, which originated from the Peking group, China [
12,
13]; and (3) post-transplantation cyclophosphamide (PT-CY)-based regimens with unmanipulated T cell replete graft, which originated from the Baltimore group, USA [
14,
15]. The historical aspects of these approaches had been previously reviewed [
7]. The present review will focus on the progress of the Chinese experience with haplo-HSCT and its potential global contributions.
2 Unmanipulated haplo-HSCT with G-CSF and ATG
New approaches to haplo-HSCT have been continually pursued over the past two decades. The Western world first introduced
in vitro TCD in the 1990s. However, this method required expensive laboratory facilities with significant expertise in cell manipulation. Furthermore, this technique was associated with poor T cell function and thus high incidence of mortality because of infections and relapse rates [
7,
9–
11]. Since the 2000s, two haplo-HSCT regimens using unmanipulated T cell replete grafts were developed in China and USA; these regimens have now become popular approaches [
8].
In a series of pilot studies on G-CSF-induced immune tolerance by Huang’s group, bone marrow T cell hyporesponsiveness could be induced by the upregulation of monocytes and plasmacytoid dendritic cells (DC2) and the downregulation of co-stimulatory signals during
in vivo G-CSF administration. The polarization of T cells from Th1 to Th2 could be maintained after
in vitro mixture of G-CSF-mobilized peripheral blood grafts (G-PB) and G-CSF primed bone marrow grafts (G-BM) [
16,
17]. Combinations of G-CSF and ATG were proposed to play a fundamental role in overcoming HLA barriers [
18] through the action of regulatory B cell, regulatory T cell, Th17/Tc 17, and myeloid-derived suppressor cells [
19–
25].
Based on the mechanistic research on cytokine-induced immune tolerance, in 2000 Huang and colleagues at Peking University initiated a pilot study investigating unmanipulated haplo-HSCT without
in vitro TCD for the treatment of acute leukemia. The study mainly included mixed grafts of G-PB and G-BM, modified busulfan/cyclophosphamide (Bu/Cy) plus ATG for myeloablative conditioning, and cyclosporine A+ methotrexate+ mycophenolate mofetil for intensified GvHD prophylaxis. All patients (
n = 58) in the pilot cohort achieved sustained, full donor-type engraftment with an acceptable incidence of grades II–IV acute GvHD (aGvHD, 37.9%) and chronic GvHD (cGvHD, 65.4%). The 2-year disease-free survival (DFS) rates for standard- and high-risk patients were 77.6% and 63.2%, respectively [
26,
27]. The cohort was updated [
13,
28–
30] and expanded to 756 cases in 2010, with 99.5% sustained myeloid engraftment, 43% grades II–IV aGvHD, and 53% cGvHD. The three-year DFS for standard- and high-risk patients was 68% and 49%, respectively [
31]. During the same period, haplo-HSCT following ATG+ G-CSF experience was also developed by the airforce group [
32,
33].
These promising results justified studies on the feasibility of unmanipulated haplo-HSCT with G-CSF and ATG for leukemia. During the last 10 years, several key techniques have been improved by clinical studies from various centers (Table 1), including new conditioning regimens and improved management of complications, such as GvHD, relapse, and infection.
3 Individualized protocols expand the target patient population
ATG is a critical component of the conditioning of T cell replete haplo-HSCT, but its optimal dose remains unknown. In a prospective, randomized trial, the Peking group revealed that 10 mg/kg thymoglobulin (rabbit ATG by Genzyme) reduced grades III and IV severe aGVHD (4.5% vs.16.1%) and moderate-to-severe cGVHD (30.4% vs. 56.3%) compared with 6 mg/kg, with comparable five-year DFS (75.6% vs. 69.6%,
P = 0.283) and improved the 5-year probability of GVHD relapse-free survival (41.0% vs. 26.8%) [
34,
35]. Another haplo-HSCT regimen, which utilized low-dose ATG-F (10 mg/kg rabbit ATG by Fresenius, standard 30–60 mg/kg), introduced by the Zhejiang group also achieved acceptable incidences of severe aGvHD (17.2%) and cGVHD (41.4%) [
36].
In addition to standard Bu/Cy conditioning, a new regimen introduced by the Peking group, which included TBI (700 cGy)/Cy (3.6 g/m
2) plus ATG, was also proven to be feasible for the treatment of unmanipulated haplo-HSCT. Compared with the Bu/Cy regimen, TBI/Cy plus ATG exhibited stable engraftment and a low incidence of liver toxicity (10.5% vs. 37.7%) and hemorrhagic cystitis (23.7% vs. 49.3%) for treating acute lymphoblastic leukemia and provided comparable results to sibling donors in high-risk acute leukemia [
37].
For patients with refractory leukemia, intensified conditioning introduced by the Nanfang group may reduce the high leukemia cell burden and improve outcomes. Using a combination of Flu, cytarabine, TBI, Cy, and etoposide for conditioning in the haplo-setting, Liu
et al. showed that intensified conditioning decreases the five-year relapse rate from 33.9% to 27.3% and might be a good approach for refractory leukemia, as well as acute leukemia of ambiguous lineage [
38–
40]. IDA-intensified haplo-HSCT introduced by the Wuhan Union group improved the dismal prognosis of pre-transplant MRD, yielding 3-year DFS of 47.3% [
41].
Selected older patients (age>50) with low HCT-CI (≤2) and good performance status could tolerate myeloablative haplo-HSCT with similar outcomes compared with younger adults [
42]. For patients above 60 years of age, haplo-HSCT with a reduced intensity regimen by substitution of cyclophosphamide with Flu was proven to be feasible with similar engraftment and relapse rates to myeloablative conditioning [
43]. Utilizing induction chemotherapy with the infusion of HLA-mismatched G-PB, Ai
et al. from the military medical sciences group showed the superiority of “microtransplantation” over chemotherapy for elderly patients suffering from AML (two-year DFS 38.9% vs. 10.0%) [
44]. Further multicenter studies will be needed to compare standard induction followed by microtransplantation with traditional HLA-matched HSCT or haplo-HSCT.
Compared with the mixed grafts of G-PB and G-BM in haplo-HSCT introduced by the Beijing group, the modified G-CSF+ ATG protocols with pure G-BM or G-PB were also feasible as proven by various centers. A propensity score method-based multicenter study demonstrated that haplo-HSCT with mixed grafts achieved better 3-year-DFS compared with G-PB alone (59.9% vs. 44.3%) [
45]. Most unmanipulated haplo-HSCTs with G-CSF and ATG for leukemia are summarized in Table 1 [
46–
55].
Haplo-HSCT has been considered only as a third-line option for patients with severe aplastic anemia (SAA) in the absence of HLA identical sibling donor (ISD) or URD, with a 25% probability of graft failure and five-year OS of less than 30% [
56,
57]. However, a new regimen, including Bu/Cy and ATG combined with unmanipulated grafts, led to the increasing usage of haplo-HSCT in SAA. According to Xu
et al., all patients (
n = 19) in the pilot cohort achieved 100% donor myeloid engraftment, with a 5-year OS of 68.4% [
58]. In a later multicenter study that compared haplo-HSCT and ISD HSCT as salvage treatments for SAA, the rates of grades II–IV aGvHD (33.7% vs. 4.2%) and one-year cGvHD (22.4% vs. 6.6%) were still high in the haplo setting. However, the 3-year OS (89.0% vs. 91.0%), failure-free survival (86.8% vs. 80.3%), and grades III and IV aGvHD (7.9% vs. 2.1%) were similar among the groups [
59]. In a later registry-based analysis of SAA patients without previous failure of immunosuppressive therapy, haplo-HSCT achieved similar results to ISD HSCT as early treatment. Aside from the G-CSF plus ATG-based regimen, the People’s Liberation Army General Hospital introduced a PT-CY regimen in 20 SAA patients, which was associated with 85% engraftment in first transplant and 85% failure-free survival in short follow-up (median 17 months). The efficacy needs to be confirmed using increased sample size and long-term follow-up [
60]. Based on these promising experiences, haplo-HSCT has been promoted to a second-line therapy in the aplastic anemia guidelines of Asia-Pacific Hematology Consortium [
61] and the British guidelines [
62].
4 Donor selection based on a non-HLA system improves outcomes in haplo-HSCT
HLA match plays a predominant role in the selection of the best donor among unrelated transplants but does not influence the outcomes in haplo-HSCT [
63,
64]. A given patient might have multiple choices for a haplo donor, raising the question, “who is the best haplo donor?”
Donor-specific anti-HLA antibodies (DSAs) are associated with primary graft failure and treatment mortality (TRM). Chang
et al. reported that DSAs (MFI≥10 000) were correlated with primary graft rejection and were associated with poor primary graft function (MFI≥2000). This finding supported the rationale for screening DSAs before haplo-HSCT for donor selection [
65].
Basing on a large sample size and relative consistency of transplant variables, young male NIMA-mismatched donors were suggested to reduce the risk of severe GvHD or relapse. As recommended, transplants from older mothers and NIPA-mismatched donors should probably be avoided [
66]. CD4
+CD25
+CD45RA
+ Treg was reported to contribute to low aGvHD in NIMA-mismatched haplo-HSCT. Additionally, other non-HLA systems, such as donor inhibitory killer cell immunoglobulin-like receptors (KIRs), might also facilitate donor selection in haplo-HSCT. Currently, an algorithm based on unmanipulated haplo-HSCT was proposed for donor selection; however, the mechanisms remain to be further elucidated [
67].
5 Risk-adapted intervention to reduce GvHD post-haplo-HSCT
As mentioned above, although rates of grades III and IV aGvHD and extensive cGvHD were comparable between haplo-HSCT and ISD HSCT, total GvHD remained a common problem with high rates observed in haplo-HSCT. As reported in a follow-up of the largest haplo-HSCT cohort, the incidence of grades II–IV aGvHD was 43%, and the two-year cumulative incidence of total cGvHD was 53% [
31]. Meanwhile, rates of grades II–IV aGvHD and cGvHD were generally below 10%–20% in MSD-HSCT [
68–
70].
In a series of studies of biomarkers for GvHD, CD4/CD8 ratios in G-BM allograft≥1.16, CD56
bright NK cells in allograft>1.9×10
6/kg, M-MDSC in allograft<1.22×10
7/kg, and other components of grafts provided predictive markers for the onset of aGvHD after haplo-HSCT [
23,
71,
72]. However, whether these biomarkers would be useful for guiding interventions remained unknown. Chang
et al. reported that a cohort of patients (
n = 228) can be stratified into high-risk and low-risk arms according to CD4/CD8 ratios in allografts. Patients in the high-risk arms (CD4/CD8≥1.16) were randomly assigned at a 1:1 ratio to either the additional low-dose glucocorticoid prophylaxis group or the control group. The cumulative incidence of grades II–IV aGvHD on day 100 was reduced from 48.1% to 20.9% by prophylaxis in the high-risk group, a rate comparable to the low-risk group (25.5%), without an increased rate of infections or delayed immune recovery [
73]. For the first time, intervention based on risk stratification with biomarkers was proven feasible for the management of GvHD. It also established a concrete example of how precision medicine could be incorporated into clinical practice.
Mesenchymal stem cells (MSCs) are multipotent stromal cells with immunomodulatory properties. They have demonstrated promising efficacy for treating steroid-resistant GvHD. Co-transplantation of MSCs and HSCs also reduces aGvHD and cGvHD rates in MSD-HSCT, albeit with a potentially higher likelihood of leukemia recurrence [
74]. The mechanism and usage of MSC infusions for GvHD prophylaxis remain to be evaluated. In a new double-blind trial of haplo-HSCT reported by Zhang
et al., patients without cGvHD at day 100 were randomly selected to receive either umbilical cord-derived MSCs (MSC group: 3×10
7 cells/100 mL per month) or normal saline for more than 4 months after transplantation. The 2-year cumulative incidence of cGVHD was reduced in the MSC group compared with that in the control group (27.4% vs. 49.0%, respectively,
P = 0.021) without increasing the risk of relapse [
75]. This trial might provide hope for preventing cGvHD after haplo-HSCT following an ATG and G-CSF protocol. Further evaluation and adjustment of infusion doses and intervals according to biomarker-directed risk stratification would be desirable.
6 Prophylaxis, intervention, and treatment of relapse post-haplo-HSCT
Relapse remains the most devastating problem after HSCT, accounting for nearly half of deaths [
76]. Recent biomarker advances have facilitated the prediction of relapse risk by dynamic monitoring of MRD, chimerism, and other factors before or after allo-HSCT. Pre- and post-transplant risk-stratification-directed strategies for intervention and treatment might benefit patients by triggering more potent graft-versus-tumor effects without increasing treatment-associated mortality.
Donor lymphocyte infusion (DLI) is one of the most effective strategies for patients with recurrent hematological malignancies after allo-HSCT. However, traditional DLI is more effective for CML than for acute leukemia and carries a relatively high incidence of GvHD. Huang
et al. reported that an infusion of G-CSF-mobilized peripheral blood progenitor cells with a short course of immunosuppression (modified DLI, mDLI) reduces the risk of aGvHD while maintaining the GVL effects after MSD-HSCT and haplo-HSCT [
77,
78]. Chemotherapy followed by mDLI improved the clinical outcomes compared with chemotherapy alone (DFS 36% vs. 0%) [
79]. In a recent trial, MRD and GvHD-guided multiple consolidation chemotherapy and mDLI were proven to prevent a second relapse in patients with acute leukemia relapse post-transplant (CIR 22% vs. 56%). Meanwhile, the cumulative incidences of grades II or higher aGvHD were not associated with chemotherapy or mDLI treatments [
80].
Given its efficient anti-leukemia effects (GVL) and safety by reducing TRM, mDLI could be used for intervention or prophylaxis before hematological relapse. Recent clinical trials and reports regarding the use of MRD or chimerism for guiding pre-emptive intervention post-allo-HSCT have been informative and helpful. In a cohort of 814 patients receiving allo-HSCT, where 105 patients were MRD+, mDLI reduced CIR from 64.4% to 27.8%, with improved OS (28.1% vs. 58.3% previously). However, this result was not significantly different compared with the MRD− patients [
81]. Interferon-α (IFN-α), a well-known antitumor agent for chronic myeloid leukemia, has sparked renewed interest in its use for AML [
82]. Mo
et al. demonstrated that for patients with an unsatisfactory response to MRD-directed mDLI, IFN-α might induce a graft-versus-leukemia effect to improve mDLI efficacy and clear MRD [
83]. Later, among patients who were MRD+, IFN-α was found to be associated with cGvHD more compared with mDLI (90.9% vs. 62.9%,
P<0.001). While NRM and DFS were comparable, IFN-α could be more easily used in an outpatient department [
84].
For advanced leukemia with a high risk of relapse, Huang
et al. showed that prophylactic mDLI can reduce the relapse rate to 51.3% after haplo-HSCT [
78]. Intensified conditioning followed by early immunosuppressant withdrawal and DLI could further reduce the relapse rate of refractory acute leukemia [
40]. In a recent multicenter study, prophylactic DLI at 45–60 days after transplantation followed by MRD and GvHD-guided multiple DLI for patients with refractory/relapsed leukemia further reduced the cumulative relapse rate to 32.4% with improved LFS and OS (50.3% and 51.4%) [
85].
7 Infection management reduces TRM
Invasive fungal disease (IFD) remains a significant threat post-HSCT, especially in the haplo setting. Sun
et al. reported in a single-center analysis that the incidence of IFD after unmanipulated haplo-HSCT is significantly higher than that after MSD-HSCT (7.1% vs. 3.3%), as well as in multivariable analysis (HR= 2.648, 95% CI 1.111–6.310;
P = 0.028). However, the response to antifungal therapy and IFD-attributable mortality were similar between the two types of transplantation. This finding was further validated in the first large-scale observational study of IFD in China, the China Assessment of Antifungal Therapy in Hematological Disease. The incidences of proven IFD following ISD, URD, and haplo-HSCT were 4.47%, 11.64%, and 12.73%, respectively. In total, 83.9% of patients received antifungal prophylaxis. Empirical, pre-emptive, and targeted antifungals were used in 82.3%, 13.6%, and 4.1% of cases, respectively [
86]. This observational study provided abundant information for refining future anti-IFD strategies.
As prolonged severe neutropenia (>14 days) was identified as an independent risk factor for IFD [
86], strategies to address neutropenia are desirable for reducing IFD-associated mortality. In a phase IV trial from the Shanghai group, 206 patients were randomly assigned to receive once-daily subcutaneous GM-CSF, G-CSF, or a combination of both. The authors of the study found that IFD-related mortality was lower in groups that received GM-CSF or G-CSF+ GM-CSF compared with those that received G-CSF (1.47%, 1.45%, and 11.59%, respectively;
P = 0.016). Furthermore, prophylactic GM-CSF was associated with lower 100-day transplantation-related mortality and lower 100-day cumulative mortality, which suggests that incorporation of GM-CSF with G-CSF may promote myeloid engraftment in the haplo setting.
Among post-HSCT viral infections, Epstein–Barr virus (EBV)-related post-transplantation lymphoproliferative disorder (PTLD) after haplo-HSCT can seriously impair patient survival. In a large cohort study, a low absolute count of CD8 T lymphocytes and immunoglobulin M at day 30 and cytomegalovirus DNAemia after HSCT were found to be significantly associated with a high risk of PTLD. Patients who received rituximab-based therapy had significantly better two-year OS (48.2% vs. 13.2%,
P = 0.02). EBV-specific cytotoxic T lymphocyte levels and DLI following rituximab-based therapy were associated with improved outcome after PTLD (five-year OS, 68.9%–70%) [
87].
8 Everyone has a donor with development of the haplo-HSCT system
Numerous advances in stem cell transplantation during the past 16 years have included unmanipulated haplo-HSCT and the integration of G-CSF and ATG into new conditioning regimens; management of complications including GvHD, relapse, and infection; and refinement of donor selection based on a non-HLA system (Table 1). As a result, patient outcomes following this unique protocol have steadily improved, bringing an end to the old era of transplantation plagued by donor shortages.
According to a prospective study by Huang
et al., unmanipulated haplo-HSCT with G-CSF and ATG was proven superior to chemotherapy as a post-remission treatment for intermediate- or high-risk AML or ALL in CR1. The cumulative relapse incidence for haplo-HSCT was 12.0% vs. 57.8% for chemotherapy, and the four-year DFS for haplo-HSCT was 73.1% vs. 44.2% for chemotherapy. Additionally, in multicenter studies, Wang
et al. reported that haplo-HSCT had outcomes comparable to MSD-HSCT for adults with intermediate- or high-risk AML in CR1. Similar results were achieved in Philadelphia-negative high-risk ALL in CR1, MDS, and primary or salvage treatment for SAA [
62,
68–
70,
88]. Haplo-HSCT may also improve outcomes for children compared with umbilical cord blood transplantation [
89]. Accordingly, haplo-HSCT was adopted as the first-choice alternative donor to HLA-ISD. Based on these outcomes, this unique system was named the “Beijing Protocol.” As Kodera
et al. from the Worldwide Network for Blood and Marrow Transplantation (WBMT) commented, “The Beijing Protocol was shown to be a reliable treatment strategy for patients without a suitable HLA-matched donor” [
8].
The “Beijing Protocol” was adopted in the majority of Chinese HSCT centers (
n>90), whereas transplants following PT-CY- or TCD-based protocols were utilized less than 1% in Chinese registries of HSCT. The number of haplo-HSCT cases increased to approximately 2500 per year, making it the largest source of allo-HSCT donors (37.6%–51.5%) in China since 2013 (Fig. 1). The “Beijing Protocol” was also reproduced successfully in Italy, Israel, Korea, and Japan [
4,
47,
90]. By contrast, according to the global survey by WBMT, haplo-HSCT with PT-CY is mainly used in the USA and partially in Europe and Australia [
8]. The European Society for Blood and Marrow Transplantation reported that G-CSF+ ATG-based regimens comprise 43%–45% of haplo-HSCT compared with PTCy, which comprises 27%–57% of haplo-HSCT, in Europe [
91–
93]. The largest cohorts from three different approaches are summarized in Table 2 [
66,
94,
95]. As R. Handgretinger commented, “more than half of the HLA haplotype mismatched transplantations performed worldwide will follow similar protocols (to the Beijing Protocol)”[
96]. Correspondingly, haplo-HSCT has been a global phenomenon as the frequency of haplo-HSCT has grown steadily from 3% to 5% to more than 10% of allo-HSCT in Europe and USA [
97]. The global contribution of the “Beijing Protocol” and PT-CY thus heralds a new era where “everyone has a donor.”
9 Conclusions and perspectives
Substantial progress has been made in the field of haplo-HSCT in recent years in China. From a broader perspective, these innovative efforts from China may have contributed to the worldwide practice of HSCT.
However, several critical questions remain to be addressed. First, with the arrival of a new era where “everyone has a donor” [
98], is there a shift from donor shortage to donor diversification? As most patients have many potential donors, including ISD and alternative donors, the question changes from “who’s the best haplo donor?” to “who’s the best alternative donor?” and, finally, to “who’s the best donor?” Would an ISD donor always remain the first choice regardless of disease status or donor characteristics? Recent advances suggested that ABO incompatibility, age, and sex of donor–patient will have a major impact on outcomes instead of HLA compatibility [
99]. How can we predict the outcomes of different donor sources under a uniform model? Addressing this question through large-scale registry-based studies, especially those involving the cooperation between China and the Western world, is of great importance. In addition, considering the rapid progression of immunotherapies, such as cellular therapy including CAR-T, CAR-NK, leukemia gene-specific CTL, and monoclonal antibodies such as PD-1/PD-L1, how can they be incorporated or bridged with allo-HSCT, especially in the largest pool of haplo-HSCT? Third, though T cell replete haplo-HSCT took place of TCD worldwide [
100], trials of haplo-HSCT following partial depletion of αβ T and B cells continue to be explored [
101]. Can we “design” the graft compositions for haplo-HSCT to improve outcomes following G-CSF+ ATG or PT-CY regimens? Lastly, how can we further improve outcomes for elderly patients or those with comorbidities? Can we make treatment regimen decisions according to a patient’s specific characteristics? All these questions would be critical matters to be addressed at a summit of international scholars.
With the development of international multicenter clinical trials and advances in translational research, the unique Chinese experience can continue to contribute to global practice for haplo-HSCT.
Higher Education Press and Springer-Verlag GmbH Germany, part of Springer Nature