Allogeneic Hematopoietic Stem Cell Transplant Using Mismatched/Haploidentical Donors

Transcription

1 Biology of Blood and Marrow Transplantation 13: (2007) 2007 American Society for Blood and Marrow Transplantation /07/ $32.00/0 doi: /j.bbmt Allogeneic Hematopoietic Stem Cell Transplant Using /Haploidentical Donors Liang-Piu Koh, 1,2 David A. Rizzieri, 1 Nelson J. Chao 1 1 Adult Bone Marrow and Stem Cell Transplantation Program, Duke University Medical Center, Durham, North Carolina; 2 Stem Cell Transplant Program, Department of Hematology-Oncology, National University Hospital, Singapore Correspondence and reprint requests: Nelson J. Chao, M.D., Adult Bone Marrow Transplant Program, Box 3961, Duke University Medical Center, Durham, NC ( Chao0002@mc.duke.edu). Received July 2, 2007; accepted August 13, 2007 ABSTRACT Haploidentical hematopoietic stem cell transplantation (HSCT) provides an opportunity for nearly all patients to benefit from HSCT when a human leukocyte antigen (HLA) genotypically matched sibling is not available. Initial results with the use of mismatched allografts led to limited enthusiasm because of graft-versus-host disease (GVHD) and infectious complications, resulting in an unacceptable treatment-related morbidity and mortality. Recent advances with effective T cell depletion, the use of a megadose of stem cells, earlier detection of severe infections, combined with better antimicrobial therapy and reduced-intensity conditioning (RIC) has significantly decreased the early transplant-related mortality and GVHD, whereas enabling prompt engraftment, hence advancing the therapeutic benefit of haploidentical transplantation. However, the cardinal problems related to delayed immune reconstitution allowing posttransplant infectious complications and relapse remain, limiting the efficacy of haploidentical HSCT. Preliminary data has demonstrated the potential for use of adoptive cellular immunity and selective allodepletion in rapidly reconstituting immunity without GVHD. The encouraging reports from haploidentical transplant using noninherited maternal antigen (NIMA)-mismatched or natural killer (NK) alloreactive donors may greatly increase the donor availability and open the way to more appropriate donor selection in HLA-haploidentical HSCT. Future challenges remain in determining the safest approach for haploidentical transplant to be performed with minimal risk of GVHD, whereas preserving effective graft-versus-leukemia activity and promoting prompt immune reconstitution American Society for Blood and Marrow Transplantation KEY WORDS Immune reconstruction T cell depletion GVHD INTRODUCTION Allogeneic hematopoietic stem cell transplantation (HSCT) has been successfully used to treat many highrisk hematologic malignancies and marrow failure syndromes. The best results with allogeneic HSCT have been obtained in patients receiving an allograft from a human leukocyte antigen (HLA) matched sibling. As the chance of finding an HLA genotypically identical sibling donor is only 25%, much attention has been focused on the use of alternative donors, either from unrelated volunteer adult donors, umbilical cord blood (UCB), or partially matched related donors. Despite the expansion of worldwide unrelated donor registries that have markedly improved the chances of finding a donor for many patients [1], the application of transplantation using unrelated adult volunteer donors remains limited by some major obstacles, including: (1) the variable chance of finding a suitably genotypically matched unrelated donor, from 60%-70% for Caucasians to under 10% for ethnic minorities [2,3]; the cumbersome process of identifying, typing, and harvesting an unrelated donor translating to the median time interval between initiation of a search and the donation of marrow of about 4 months [4], rendering this option less viable for patients who urgently need transplantation. Many such patients do not maintain a remission or survive the long waiting period until a donation is available. Moreover, ablative allogeneic transplant using a matched unrelated donor is still associated with a high transplantrelated mortality (TRM) (30%-40%) and high 1249

2 1250 L.-P. KOH et al. long-term morbidity [5-9]. UCB donations, on the other hand, overcome some of these limitations because of easy procurement, the absence of risk for donors, potential reduced risk of graft-versus-host disease (GVHD) [10], and less stringent criteria for HLA matching for donor-recipient selection. However, engraftment remains a significant concern, in part from the low number of progenitor cells contained in a single UCB unit. Delayed neutrophil recovery and TRM remain the main obstacles for successful UCB transplantation, particularly in patients receiving a myeloablative preparative regimen [11,12]. The use of hematopoietic stem cells (HSC) from relatives who are only partially HLA matched provides some advantages for patients lacking fully HLAmatched sibling or unrelated donors. Virtually all patients have at least 1 HLA-partially matched parent, sibling, or child, who is immediately available to serve as a donor. Further, the immediate availability of this mismatched family member could have important treatment implications as patients will not be lost to early relapse, and financial implications as the considerable expenditure of additional typing and procurement of unrelated donor graft can be avoided. APPROACHES USING ABLATIVE THERAPIES (TABLE 1) Early Studies Using T Cell Replete Marrow Grafts The results of the haploidentical transplant in early series reported in the early 1990s were largely disappointing, mainly because of the high incidence of mortality from the T cell-mediated alloreactions in the graft-versus-host (GVH) directions, which caused lethal GVHD and in the host-versus-graft (HVG) direction, which caused graft rejection [13,14]. In contrast, the results in patients receiving transplant from haploidentical family members for severe combined immunodeficiency (SCID) were more encouraging. Clinical studies have demonstrated that extensive ex vivo T cell depletion (TCD) of bone marrow to a maximum residual of T cells/kg body weight infused to the patient prevents acute and chronic GVHD (agvhd, cgvhd) without any other posttransplant immunosuppressive prophylaxis [15]. Unfortunately, when tested in leukemic patients, haploidentical T cell-depleted bone marrow transplantation was associated with a high incidence of graft rejection because the balance between recipient and donor T cells shifted in favor of the unopposed HVG reaction. In 1 of the earliest and largest reports on the use of mismatched related donor transplant, investigators from Fred Hutchinson Cancer Research Center clearly illustrated that haploidentical transplant is associated with a higher incidence of GVHD, delayed engraftment, and graft failure [16,17]. In patients whose posttransplant immunosuppression consisted of methotrexate (MTX) alone, the risk of agvhd was significantly increased in patients receiving marrow grafts from donors incompatible from 1, 2, or 3 HLA loci. Patients receiving marrow grafts from HLA-incompatible marrow donors had a relative risk for GVHD of 3.23 compared with controls. Notably, the study showed that the use related marrow with no more than a single HLA-A, B, or -DR mismatch provided clinical results comparable to 6 of 6 HLA genotypically identical sibling HSCT. However, the outcome of 2 or 3 loci-mismatched transplantation without graft manipulation remains poor [16]. Subsequent studies of patients whose posttransplant immunosuppression consisted of cyclosporine (CSP) and MTX also showed that the degree of overall HLA incompatibility is inversely correlated with the probability of survival, suggesting that the deleterious effect of GVHD remains prohibitive for these groups of patients. Powles et al. [18] reported the outcome of 35 patients with acute leukemia receiving a 1-3 HLA mismatched marrow graft from parent, child, or sibling, following cyclophosphamide/total body irradiation (TBI) or cyclophosphamide/melphalan conditioning. Graft failure was seen in 29% of patients, and GVHD occurred in 80% of patients, with only 1 death as a direct consequence of GVHD. Fatal complications, likely immune-mediated, consisted of acute pulmonary edema, convulsions, intravascular hemolysis, and renal failure, and were noted in 12 of the 35 patients. The high mortality rate associated with this approach, with nonrelapse death occurring in more than half of the patients, further highlighted the inherent difficulties of haploidentical transplantation, thus limiting its applicability to many patients requiring the transplant. Another analysis reviewed 2000 patients reported to the International Bone Marrow Transplant Registry (IBMTR) [19] showed that TRM was significantly higher after alternative donor transplant compared to 6 of 6 HLA-matched sibling transplants. Patients with early leukemia receiving an HLA matched sibling transplant had the lowest risk of treatment failure and TRM compared with those receiving a 1-2 HLA mismatched related or unrelated transplant. However, the difference in outcome became less striking in patients with advanced leukemia. In this group, treatment failure after 1 HLA-antigen mismatched related donor transplant was similar to 6 of 6 HLA matched sibling transplants, and the risk of other alternative donors was increased approximately 50%. This is primarily explained by the higher treatment relapse rate in HLA matched sibling transplants for advanced disease. Similar results have been demonstrated by another database analysis reported by the Japanese Society for Hematopoietic Cell Transplantation, in which the impact of HLA mismatch on survival was smaller in

3 Haploidentical or Transplant 1251 patients with high-risk disease compared to standardrisk disease [20]. Recent Studies Using T Cell Replete Hematopoietic Cell Grafts Lu et al. [21] from Beijing, China, retrospectively compared a large cohort of patients who received myeloablative transplant from either matched sibling donors or mismatched/haploidentical family donors. Although the transplant procedure involved the use of unmanipulated hematopoietic donor grafts, the inclusion of ATG in the conditioning for the mismatched/ haploidentical transplants, which is known to persist in the plasma for several days, effectively provides in vivo donor T cell depletion. All patients were given CSP, mycophenolate mofetil (MMF), and methotrexate (MTX) as GVHD prophylaxis. There were more patients in the mismatched/haploidentical transplant cohort who received combined granulocyte-colony stimulating factor (G-CSF)-mobilized marrow (G-BM) and peripheral blood stem cell (PBSC) (G-PB) graft. Notably, this comparison study showed that every endpoint in terms of relapse (13% versus 18%), TRM (14% versus 22%), overall (OS 72% versus 71%), and leukemia-free survival (71% versus 64%) between the HLA-matched and mismatched HSCT did not statistically differ, although trended in favor of the matched group. The relatively low incidence of grade III-IV agvhd (11%) among the mismatched cohort in this study was impressive. These findings may be related to (1) the use of ATG providing in vivo depletion of recipient T cell favoring engraftment, but also depletion of donor T cells reducing agvhd and cgvhd; (2) possible effect of combination of CSP, MTX, and MMF as postgrafting immunosuppression; (3) the immunomodulatory effect of T-polarized cells (Th2) and mesenchymal stem cells (MSCs)/mesenchymal (stroma) progenitor cells (MPCs) from the G-CSF mobilized marrow graft and PBSC, respectively. It becomes clear from these early clinical reports that haploidentical or partially matched HSCT with T cell-replete marrow grafts following myeloablative conditioning was associated with high TRM from GVHD, graft failure, delayed immune reconstitution, vulnerability to life-threatening infections, and relapse. Thus, haploidentical transplant could not be widely adopted as a routine procedure, leading to subsequent efforts focusing on strategies to overcome these barriers, which include: (1) reducing the intensity of conditioning (RIC) and hence ameliorating the regimen-related toxicity, (2) promoting engraftment capacity of the graft by using G-CSF-mobilized PBSC and the use of megadose of stem cells, (3) effective T cell depletion methods to decrease both graft rejection and GVHD by different in vivo ex vivo T cell depletion procedures, (4) exploiting the concept of alloreactive NK cells, which may play a vital role in facilitating engraftment and in preventing relapse [22]. The initial attempts in overcoming HLA barrier focused mainly on strategies for effective host and graft T cell depletion. However, the benefit of a decrease in GVHD from donor TCD was offset by a higher incidence of graft rejection, relapse, and infection [14,23]. Another major step toward induction of tolerance was achieved following the pioneering work by Reisner [24] and the clinical results by Aversa et al. [25] using megadoses of hematopoietic stem cells mobilized into the peripheral blood by growth-factor use as a supplement to the heavily T cell-depleted mismatched bone marrow stem cells. A remarkably high 95% engraftment rate was seen in patients receiving an allograft from haploidentical 3 of 6 HLAmatched family members in the absence of severe GVHD despite no postgrafting immunosuppression [25]. The encouraging results of their subsequent report using strategies involving large doses of TCD blood-derived stem cells spurred further interest in exploring the option of using haploidentical/mismatched related donors for patients who may benefit from a transplant but do not have a readily available matched donor. Since the 1990s, several investigators have approached haploidentical transplantation by using partial TCD combined with intensive immunosuppression. In single institution studies, grade II-IV GVHD incidence has ranged from 18% to 40% in recipients of HLA-mismatched marrow after TCD using anti-cd6 or T 10 B 9 monoclonal antibodies (mab) for purging [26-28]. Henslee-Downey et al. [28] were among the early pioneers in exploiting a novel sequential immunomodulation pre- and posttransplant using ex vivo TCD with the T 10 B 9 monoclonal antibody and in vivo T cell lysis with immunotoxin H65-RTA, following intensive total body irradiation (TBI)-based myeloablative conditioning. Seventy-two patients received allografts from haploidentical family members using this treatment protocol. The 88% engraftment rate, 16% probability of grade II-IV agvhd, and 51% probability of extensive cgvhd were encouraging. At a median follow-up of 21.5 months, the 2-year disease free survival (DFS) was 31%, with 53% probability of DFS seen among standard risk patients. The study highlighted the efficacy of partial TCD in preventing GVHD in mismatched transplantation and the potential of postgrafting immunomodulation in lowering the risk of graft failure. In a recent report from the same institution, Mehta et al. [29] reported the outcome of 201 patients with acute leukemia who underwent transplantation from partially mismatched related donors between 1993 to 1999, using bone marrow grafts that were ex vivo TCD with T 10 B 9 ( ) or OKT3 (1995-

4 1252 L.-P. KOH et al. Table I. Myeloablative /Haploidentical Stem Cell Transplantation Institutions (Authors, year) Diagnosis No of Patients Royal Marsden (Powles et al, 1983)[18] IBMTR (Szydlo et al, 1997) [19] Adults/ Children Median Cell dose ( 10 6 /kg) CD34/CD3 Preparative Regimens GVHD Prophylaxis AML, ALL 35 Both N.S. Cy/TBI or Cy/Mel CSP / MTX None CML, Acute leukemia related (N 330) unrelated (N 108) Both N.S. TBI/Cy TBI/Cy/Others TBI/Others CSP / MTX/ Others *TCD Method of T cell depletion N.S. Milwaukee (Drobyski et al, 2002)[30] Leukemia, lymphoma, MDS unrelated (N 58) Hapolidentical (N 48) Both 1.6/N.S. TBI/Cy/AraC/MP TCD Ex vivo TCD with T 10 B 9 or OKT3 JSHCT (Kanda et al, 2003) [20] Leukemia, MDS 142 Adults N.S. TBI/Cy Bu/Cy CSP/MTX (87%) Others (13%) N.S. USC (Mehta et al, 2004) [29] Beijing, China (Lu et al, 2006)[21] Perugia (Aversa et al, 2005)[43] Tuebingen (Lang et al, 2004)[44] Japan Multicenter Study (Kato et al, 2000) [45] Canadian Multicenter Study (Walker et al, 2004)[46] AML, ALL 201 Both 1.9/5 TBI/VP16/AraC/ Cy/ATG Leukemia, MDS 135 Both 2.3/177 Bu/Cy/AraC/ MeCCNU/ATG CSP, partial TCD, MP, ATG CSP/MTX/MMF Ex vivo TCD with T 10 B 9 or OKT3 In vitro TCD with ATG ALL, AML 104 Both 13.8/0.01 TBI/TT/Flu/ATG TCD Ex vivo CD 34 selection negative selection with Anti-CD2 mab MHD, NMD 63 Children 19.5/0.011 TBI or Bu-based with Flu/TT/ ATG/Cy added MHD, NMD 135 Both 3.2/0.06 (BM) 5.5/0.09 (PBSC) 4.9/0.01 (BM PBSC) TBI based regimen in majority of patients TCD CSP/FK506/ MTX/Steroids TCD[*] Ex vivo CD 34 or CD133 selection Ex vivo CD 34 selection AML 11 Adults 13.72/ Mel/TT/Flu/ATG TCD Ex vivo CD 34 selection GVHD, graft-versus-host disease; NRM, non-relapse mortality; USC, University of South Carolina; IBMTR, International Bone Marrow Transplant Registry; JSHCT, Japan Society for Hematopoietic Cell Transplantation; CML, chronic myelogenous leukemia; AML, acute myelogenous leukemia; ALL, acute lymphoblastic leukemia; MDS, myelodysplastic syndrome; AraC, cytarabine; MP, methylprednisolone; TBI, total body irradiation; Cy, cyclophosphamide; Bu, busulphan; TT, thiotepa; Flu, fludarabine; Mel, Melphalan; LFS, leukemia-free survival; OS, overall survival; VP16, etoposide; MeCCNU, methyl-ccnu;atg, antithymocyte globulin; DFS, disease free survival; EFS, event-free survival; TCD, T cell deletion; CSP, cyclosporin; N.S., not stated; MTX, methotrexate; MMF, mycophenolate mofetil; MHD, malignant hematological disease; NMD, non-malignant disease.; BM, bone marrow; PBSC, peripheral blood stem cell. *T cell depletion in some patients. Refers to patients with primary graft rejection after first transplantation.

5 Haploidentical or Transplant 1253 Table I. (Continued) Primary Graft Failure Acute GVHD NRM Outcome/Survival Remarks 29% 80% 57% 31% at 6 months-3 years related: 9%-16% unrelated: 9% unrelated (N 3) Hapoidentical (N 2) Gd II-IV: related: 44%-56% unrelated: 63% Gd III-IV: related 27-36% unrelated: 47% unrelated: Gd II-IV: 33% Hapoldentical: Gd II-IV: 46%> 6.3% Gd III-IV: 30% in 1 locus mismatch 2% Gd II-IV: 13% Gd III-IV: 15% 0% Gd II-IV: 40% Gd III-IV: 16% related: 53%-67% unrelated: 69-79% unrelated: 45% Hapoidentical: 42% related: 15%-36% LFS unrelated: 17%-26% LFS 2 year OS: unrelated: 34% Hapolidentical: 21% 2 year DFS: unrelated: 29% Hapolidentical: 17% 1) Higher treatment failure in alternate donor transplant. 2) Better outcome in matched sibling transplant than 1-2 Ag mismatched related/unrelated transplant for early leukemia 3) Similar outcome in advance leukemia. Overall survival significantly higher after HLA-matched unrelated donor transplant compared with the other two groups. NA N.S. Impact of HLA mismatch on survival was smaller in patients with high-risk disease as compared to standard-risk disease. 51% 5 year OS: 19% 5 year DFS: 18% 22% 2 year OS: 71% 2 year LFS: 64% 7% Gd II-IV: 8% 40% 5 year EFS: 47% (for patients in remission at transplant) Disease status had strong impact on survival 17% Gd II: 7% 29% 3 year DFS: 48% (for ALL/ NHL in remission) 13% Gd II-IV: 21% 47% 5 year DFS: 39% (standard risk patients) 5% (high risk patients) 0% 0% 55% 9 months DFS: 9%, OS: 9%

6 1254 L.-P. KOH et al. 1999) mab. GVHD prophylaxis consisted of CSP, methylprednisolone, and ATG. The median T cell dose in the marrow was /kg. Using an intensive ablative TBI-based conditioning regimen, the transplant resulted in an overall engraftment rate of 98%. The cumulative incidence of grade II-IV agvhd and cgvhd were 13% and 15%, respectively. The 5-year cumulative incidence of relapse and TRM were 31% and 51%, respectively. The 5-year OS and DFS were 19% and 18%, respectively. In the multivariate analysis, younger patients with disease in remission receiving allograft from a less HLA-disparate donor had the best outcome, emphasizing the importance of patient selection and pretransplant disease status in determining the outcome of this highrisk procedure. Drobyski et al. [31], in a single institution analysis, compared the outcome of patients who received transplant from a matched unrelated donor, mismatched unrelated donor, or haploidentical donor. All patients had received a TCD marrow graft, using either T 10 B 9 or OKT3. There was a decrease in relapse and increase in survival in matched unrelated donor recipients compared with the other 2 groups, but the degree of TCD might not have been sufficient for the haploidentical setting. The higher TRM after transplantation from a mismatched unrelated or haploidenticalrelated donor transplant compared to matched unrelated donor transplant demonstrated a clear effect of HLA disparity. Early Studies Using TCD Hematopoietic Cell Grafts via CD34 Selection An important landmark in the development of haploidentical transplant came with the concept of using mega doses of stem cells, which was demonstrated to overcome the major histocompatibility complex (MHC) barrier, presumably by generation of veto effect by the stem cells [24,31]. With these advances also came the realization that natural (NK) cells from HLA-mismatched donors can exert an antileukemia effect against a patient s blasts in the absence of HLA-antigenic engaging killer-inhibitory receptors (KIR). Aversa et al. [32], from the University of Perugia, Italy, pioneered the megadose approach by infusing G-CSF mobilized peripheral blood and bone marrow stem cells, both ex vivo depleted of T cells by soybean agglutination and E-rosetting, following an intensive TBI-based conditioning regimen. Although engraftment was prompt with low occurrence of GVHD, subsequent follow-up showed difficulty with both late rejection and/or graft failure. These events resulted in increased TRM in excess of 60%, leading to subsequent modification of treatment protocols (as discussed below) [33]. This approach was combined with another important breakthrough in the field of haploidentical transplantation: the development of a powerful method to eliminate T cells via selection columns. First introduced by investigators from Tuebingen, the availability of CD34 -enrichment technique using the CliniMACS system (Miltenyi Biotec Gmbh, Bergish Gladbach, Germany) has provided a reproducible TCD of 4.5 logs in several studies [34]. This technology, initially intended to prepare stem cell grafts for autologous transplantation [31,35,36], was later extended to allogeneic transplantation settings [34,37,38,39]. Recent Studies Using T Cell-Depleted Hematopoietic Cell Grafts via CD34 Selection Results from Perugia, Italy. Following the initial success with high incidence of engraftment and low incidence of grade II-IV agvhd, the Perugia group has made modifications that have resulted in marked improvement in clinical outcome: (1) The transplant protocol was modified by substituting cyclophosphamide with fludarabine in an attempt to reduce nonhematologic toxicity. This was based on the observation in the murine model that fludarabine TBI provided equivalent immunosuppressive effects and obviates other toxicities [40]. In addition, the total lung dose of radiation was decreased from 6 to 4 Gy. (2) Further depletion of T cells, hence further diminishing GVHD, was achieved by modification in graft processing. Positive immunoselection of peripheral blood CD34 cells was used instead of the soybean agglutination and E rosetting. The initial technique of 1-round E-rosetting followed by positive immunoselection of the CD34 cells using the Ceprate-SC system [33] was subsequently substituted by the 1-step CD34 cell selection using the Clinimacs device (Miltenyi Biotec) or 2-step (positive/negative selection-anticd34 / anticd2 ) procedure using the Isolex instrument. (3) The use of G-CSF-mobilized peripheral blood progenitor cells (PBPC) instead of bone marrow grafts, which facilitates the transplantation of megadoses of haploidentical cells. (4) The use of G-CSF to promote engraftment was eliminated from the protocol since 1999, following the observation that it impaired immune reconstitution [41,42]. These modifications have resulted in a favorable outcome in a large series of high-risk acute leukemia, with sustained full donor type engraftment in over 95%, rapid hematopoietic recovery, and a very low incidence of agvhd grade II-IV without the need for any posttransplant immune suppression as prophylaxis [9,33]. Importantly, the HSC harvesting or mobilization procedure was well tolerated by the donors. In another recent report, [43] 104 patients with high-risk leukemia [67 acute myelogenous leukemia (AML) and 37 acute lymphoblastic leukemia (ALL)] were transplanted between 1999 and 2004 following the modified transplant protocol of the Perugia group. Engraftment was achieved in

7 Haploidentical or Transplant of 101 evaluable patients. The unfortunate high TRM of 40% (mostly from infection) was anticipated, given that many of these patients were heavily pretreated and had advanced disease status at transplantation (36.5% were in relapse; 10.1% in 3rd or 4th remission). agvhd and cgvhd developed in 8% and 7% of evaluable patients, respectively. Patients in any remission had a remarkable event-free survival of 47% 6%. The cumulative incidence of relapse in patients transplanted in remission from NK alloreactive donors was 14%, whereas it was 28% in the nonalloreactive donors. For patients transplanted in relapse, the corresponding relapse rates were 33% and 45%, respectively, emphasizing the importance of pretransplant disease control and presence of NK alloreactivity. Results from Tuebingen, Germany. Investigators from Tuebingen adopted a similar approach as reported above for haploidentical HSCT using myeloablative conditioning and a high-dose CD34 cell-selected graft. In a recent report, Lang et al. [44] updated their 9-year experience in 63 pediatric patients receiving CD34 or CD133 selected stem cell transplant from haploidentical family donors. G-CSF-mobilized stem cell were selected by either anti-cd34- or anti- CD133-coated microbeads. Using a myeloablative conditioning regimen with no postgrafting immunosuppression given, primary sustained engraftment occurred in 83% of patients (98% after reconditioning) and agvhd occurred in only 7% of evaluable patients. Viral infection was a significant cause of mortality within the first 6 months posttransplant, suggesting that impaired immune recovery remains an important barrier to success using this TCD myeloablative approach. Notably, this study resulted in a long-term survival rate of 48% for children with ALL in remission, which compared favorably with the historical control group, as well as with other pediatric studies using unmanipulated bone marrow from matched unrelated donors. Results from Multicenter Studies. A nationwide survey in Japan of the outcome of haploidentical transplants following negative TCD using the Isolex CD34 selection device was reported by Kato et al. [45]. One-hundred thirty-five patients with various hematologic diseases were transplanted using myeloablative therapy followed by HLA haploidentical related donors, of whom 64 patients received grafts with 2 HLA locus mismatches and 43 patients had 3 loci mismatched. A majority of the patients were given CSP, tacrolimus, MTX, or corticosteroids, either alone or in combination as GVHD prophylaxis. The median CD34 cell dose was lower than that reported by the Perugia group: /kg for patients receiving bone marrow only, /kg for patients receiving PBSC only, and /kg for patients receiving both marrow and PBSC. The median CD3 T cell doses were 6.0, 9.4, and /kg for the 3 groups, respectively. The incidence of grade III-IV agvhd was 8.4%, close to that observed for transplants from 6 of 6 HLA genotypically matched siblings or from 6 of 6 HLA-matched unrelated donors without TCD. Graft failure occurred in 13% of patients, with a higher rate seen in nonmalignant disease (40%) than in hematologic malignancies (13%). DFS at 5 years was 39% in standard-risk patients and 5% in high-risk patients. Results of this study support the hypothesis that standard doses of CD34 cells are inadequate for consistent engraftment of HLA haplotype mismatched donor cells, and that a moderate degree of TCD is inadequate to prevent GVHD without postgrafting immunomodulation. A similar approach based on a modified regimen developed by the Perugia group, using CD34 selection as TCD, was investigated in a Canadian multicenter study [46]. Eleven patients with AML in various stages were accrued. The authors observed engraftment in all 11 patients without occurrence of GVHD. However, 10 of the 11 patients died from relapse or infection. The disappointing results highlighted the problems of graft rejection and GVHD as barriers to haploidentical transplantation can be overcome with this approach, but the slow immune reconstitution limits its general use. Ex Vivo Induction of T Cell Anergy A competing approach not based on megadose stem cells was evaluated by Guinan et al. [47], in which anergy but not TCD was the basis for mismatched transplantation. To induce anergy, donor marrow was harvested and cocultured with irradiated mononuclear cells from the recipient in the presence of soluble CTLA-4-Ig. After myeloablative conditioning, 12 patients, almost all in advanced status of the hematologic malignancies, received marrow containing a median of /kg of CD3 T cells from the mismatched/haploidentical family that had been treated to induce anergy. Evaluable patients engrafted at a median of 20 days (range: 14-23), with full donor chimerism observed in 8 of 10 patients in whom engraftment occurred. Of the 12 patients, 7 died of infections, multiorgan failure, hemorrhage, or relapse and 5 were alive and in remission 132 to 863 days after transplantation. Only 3 patients developed agvhd confined to the gastrointestinal tract, and no deaths were attributable to GVHD. APPROACHES USING NON/MYELOABLATIVE THERAPIES (TABLE 2) Although the highly immunosuppressive and myeloablative conditioning regimen and a megadose of extensively TCD G-CSF mobilized PBSC cells has

8 1256 Table II. Nonmyeloablative /Haploidentical Stem Cell Transplantation Institutions (Authors, years) Diagnosis No of Patients Adults/ Children Preparative Regimens GVHD Prophylaxis Median Cell dose ( 10 6 / kg) CD34/ CD3 Method of T cell Depletion Primary Graft Failure GVHD NRM Outcome/ Survival Remarks Duke University (Rizzieri, 2007)[65] Tuebingen/Dresden (Bethge, 2006)[55] Tuebingen (Handgretinger, 2007)[56] MGH (Spitzer, 2003)[58-60] John Hopkins (O Donnell, 2002)[61] Leukemia, Myeloma Lymphoma, MPD Leukemia/ Lymphoma 49 Adults Flu/Cy/Campath MMF CSP 13.5/460 In vivo ex vivo TCD with Campath 10 Adults Flu/TT/ Melphalan/OKT3 TCD 7.8/0.02 Ex vivo CD3/CD19 negative depletion with anti- CD3 and CD19 mab HM/ AA 38 Both Flu/TT/Melphalan TCD 16/0.049 Ex vivo CD3/CD19 negative depletion with anti- CD3 and CD19 mab Leukemia/ Lymphoma Leukemia/ MDS 12 Adults Cy/Anti-CD2 Mab/Thymic RT 13 Adults TBI/Cy/Flu/Post BMT Cy CsP ex vivo TCD PBSC 10.6*/8.9[* Ex vivo CD 34 selection in vivo TCD using anti-cd 2 mab 6% Gd III-IV: 16% Gd III-IV: 8% 0% Gd II: 60 % Gd IV: 10% 10.2% OS 1 year 30% OS: >1 year 17% Gd II-IV: 27% 2.6% EFS: 70% in good risk patients EFS: 20% in poor risk patients 0% Gd II-IV: 17% 25% 17% DFS, 25% months CsP/MMF 5.3/3.2 Not done 31% Gd II-IV: 54% 8% 38% DFS, 46% >6 months Standard risk patients: OS 1 year Good risk: Patients with non- HM and in CR Poor risk: Patients with chemorefractory disease L.-P. KOH et al.

9 Haploidentical or Transplant 1257 Table II. (Continued) Primary Graft Failure GVHD NRM Outcome/Survival Remarks Method of T cell depletion Median Cell dose ( 10 6 / kg) CD34/ CD3 GVHD Prophylaxis Preparative Regimens Adults/ Children No of Patients Institutions (Authors, years) Diagnosis 4% Gd II: 20% 15% EFS 3 years OS 58% at median FU 664 days 26 Adults Flu/Bu/ATG FK506/MP 6.55/254 In vitro TCD with ATG Leukemia/ Lymphoma Osaka University (Ogawa et al, 2006)[62] 0% Gd III-IV: 9% 17% 35% 1 year CSP/MTX 5.1/260 In vivo TCD with Campath 12 Adults Myeloablative (N 6): TBI/Cy/ Campath Reduced intensity (N 6): Flu/Bu/ Campath TBI4Gy Leukemia/ Lymphoma/ MDS Tokyo University (Kanda et al, 2005)[64] GVHD, graft-versus-host disease; NRM, non-relapse mortality; MPD, myeloproliferative disease; Flu, fludarabine; Cy, cyclophosphamide; MMF, mycophenolate mofetil; CSA, cyclosporin A; OS, overall survival; TT, Thiotepa; HM, hematological malignancies; AA, Aplastic anemia; CR, complete remission; MGH, Massachusetts General Hospital; RT, irradiation; FK506, tacrolimus; mab, monoclonal antibody; MP, methylprednisolone; TCD, T cell depletion; DFS, disease free survival; EFS, event free survival TBI, total body irradiation; Bu, busulphan; FU, follow-up; ATG, antithymocyte globulin; Campath, alemtuzumab. *Cell dose for a subgroup of patients. Cell dose before T cell depletion. demonstrated encouraging survival results, it is not without limitations. First, the procedure is associated with significant regimen-related toxicity and high TRM (between 35% and 40%) [43,44] primarily from infections. Second, a megadose of purified CD34 cells is crucial in overcoming the barrier of residual antidonor cytotoxic T-lymphocyte precursors in TCD mismatched transplant. There is continuing concern with regard to the slow engraftment or graft failure in patients receiving a lower cell dose. Previous studies from Tuebingen have shown delayed engraftment at CD34 doses less than /kg body weight [48]. As such, most physicians would usually target for megadose of stem cell ( CD34 cells/kg body weight) from the donor, whereas planning for haploidentical transplants. This can place considerable demand on both the donors and the pheresis service for the following reasons: (1) The high graft content is an obstacle in large adults. (2) The long hours of multiple days of pheresis can be exhausting, with a slight increase in pheresis-related adverse effects to donors. (3) For the pheresis and stem cell processing laboratory staff, the procedures involved can be time-consuming and labor-intensive. Even with high cell doses, graft failure in the range of 5%-14% has been reported by some [49-51]. Communication from investigators and reports given at conferences on haploidentical transplantation have indicated that both graft failure and GVHD remained a problem, and there were few survivors [52]. Developing new strategies of TCD or graft manipulation in mismatched HSCT, with an aim to improve engraftment with better tolerated, less toxic conditioning, has become an important area of research. Although the number of mismatched allogeneic HSCT has increased steadily over the past few decades, this high-risk ablative procedure can only be offered only to a minority of patients, because most subjects are beyond the age where myeloablative preparative regimens can be delivered with a reasonable degree of safety. GVHD, TRM, and other toxicities remain significant deterrents and have limited use in otherwise healthy, younger patients as well. To extend allogeneic HSCT to older patients with comorbidities, RIC or nonmyeloablative conditioning lacking significant regimen-related toxicities have been developed. Results from Tuebingen/Dresden. Based on the promising experiences gained at St. Jude Children s Research Hospital (SJCRH), Memphis, in the pediatric population [53,54], investigators from Tuebingen explored a new TCD strategy in adult patients following dose-reduced conditioning [55]. Using this new approach, T and B cells (CD3/CD19) are negatively depleted from PBSC with log TCD using anti- CD3- and anti-cd19-coated microbeads on a Clini- MACS device. In contrast to the CD34 selection

10 1258 L.-P. KOH et al. strategy pioneered by the Perugia group, CD3/CD19- depleted grafts harvested using this strategy not only contain CD34 stem cells but also CD34 progenitors and NK, dendritic, and graft-facilitating cells. Dosereduced conditioning consisting of fludarabine ( mg/m 2 ), thiotepa (10 mg/kg), melphalan (120 mg/m 2 ), and OKT-3 (5 mg/day, day 5 to 14) was used. Ten adult patients with a median age of 43, and advanced hematologic malignancies received mismatched transplant using this approach. Rapid engraftment with full donor chimerism was seen after 2 weeks in all patients. Six patients developed grade II GVHD and 1 developed lethal grade IV GVHD. TRM was 30% and OS was 50%, with 4 patients in complete remission with a median follow of 1 year. The fast engraftment seen in this CD3/CD19 group with CD34 doses as low as CD34 cells/kg demonstrates that successful haploidentical transplant may be feasible even without megadoses of CD34 stem cells. Importantly, the favorable immune reconstitution with fast reconstitution of NK cells was noted with this approach, resulting in few infectious complications. In another recent study from Tuebingen, Handgretinger et al. [56] reported the outcome of 38 pediatric patients with high-risk hematologic malignancies and severe aplastic anemia (AA) receiving haploidentical transplant using this approach. The dose-reduced conditioning was modified to a lower dose of fludarabine (to reduce neurotoxicity) and OKT3 was omitted. Primary sustained engraftment occurred in 83% of patients and final engraftment was 98% when the remaining patients with graft failure had a repeat transplant. Grade II-IV agvhd occurred in only 27% of patients. Overall TRM was low at 2.6%. The favorable event free survival of 70% seen only in patients with nonmalignant disease and those in remission at time of transplant, suggests that disease relapse is a major obstacle among patients with refractory malignancies undergoing haploidentical transplant. Results from Massachusetts General Hospital. Based on murine models established by Sykes and colleagues [57], a series of haploidentical stem cell transplantation have been conducted at Massachusetts General Hospital. To address the problems of graft failure and GVHD, the initial regimen has been modified to its current form which includes cyclophosphamide, fludarabine, MEDI-507 (a monoclonal anti-cd2 antibody) and thymic irradiation. Mixed split lineage lymphohaematopoietic chimerism has been achieved in most cases with this strategy, with a predominance of donor myeloid chimerism and a much lower percentage of donor T cell chimerism. In addition, mixed chimerism, including the low percentage of donor T cell chimerism, can be successfully converted to full or nearly full donor chimerism with either no GVHD or manageable, primarily cutaneous GVHD. Recurrent malignancies and late infections have been the chief reasons for treatment failure with this approach. Efforts are underway to optimize the ex vivo TCD of the product and to explore different doses of delayed DLI [58-60]. Results from John Hopkins University. O Donnell et al. [63], from John Hopkins University, have performed nonmyeloablative haploidentical transplant on 13 patients with hematologic malignancies using lowdose TBI 2 Gy and fludarabine (with or without cyclophosphamide) as conditioning. High-dose posttransplant cyclophosphamide, given at 50 mg/kg on day 3, was added onto tacrolimus/mycophenolate mofetil (MMF) to improve GVHD prophylaxis. The median time to absolute neutrophil count 500/ Lin 8 patients with engraftment was 15 days (range, days). agvhd developed in 6 of the 13 patients. Six of the 13 patients were alive, 5 who were in a complete remission at a median of 191 days posttransplant, including 2 patients with graft rejection. The results suggest possible benefits of pre- and posttransplantation cyclophosphamide in promoting engraftment and prevention of GVHD. Results from Osaka University Hospital, Japan. Ogawa et al. [62], from Osaka University Hospital in Japan, investigated the use of ATG-based nonmyeloablative conditioning regimen as previously reported by Slavin et al. [63], in the haploidentical transplant of 26 patients who had hematologic malignancies in an advanced stage or with a poor prognosis. Using a conditioning consisting of fludarabine, busulfan, and anti-t-lymphocyte globulin and GVHD prophylaxis consisting of tacrolimus and methylprednisolone (1 mg/kg/day), 26 patients underwent transplantation using PBSC from an 2-3 antigen HLA mismatched donors. All patients except for 1 achieved donor-type engraftment. Full donor chimerism was achieved by day 14. Only 5 (25%) of 20 evaluable patients developed grade II GVHD. Sixteen of the 26 patients are alive in complete remission. Four died of transplantation-related causes, and 6 died of progressive disease. The event-free survival at 3 years was 55%. Results from Tokyo University, Japan. Kanda et al. [64] evaluated the feasibility of haploidentical unmanipulated PBSC transplantation from 2 or 3 locimismatched family member using in vivo alemtuzumab in 12 patients (median age 49.5 years) with high-risk hematologic malignancies. Six patients received a TBI-based myeloablative regimen, whereas the remaining 6 patients older than 50 years received less intensive or nonmyeloablative fludarabine-based conditioning. Alemtuzumab was added on days 8 to 3 and CSP MTX were used as GVHD prophylaxis. There was no graft rejection, and the incidence of grade III-IV agvhd was only 9%. The nonrelapse mortality (NRM) was observed in only 2 of 12 pa-

11 Haploidentical or Transplant 1259 tients. None of the patients died of infectious causes despite impaired T cells immune reconstitution during the first 2 months after transplantation. Results from Duke University. Rizzieri et al. [65] from Duke University recently reported one of the largest series of adult patients with nonmyeloablative transplant using 3-5 of 6 HLA-matched family donors. Forty-nine patients with hematologic malignancies or marrow failure were accrued. The patients in this group were, in average, older (median age of 48) than most other reported series haploidentical transplantation. Using a nonmyeloablative preparative regimen consisting of fludarabine and cyclophosphamide in combination with alemtuzumab for in vivo and in vitro TCD, the group reported successful engraftment in 94% of patients, low TRM rates of 10.2% and severe GVHD of 8%. With more than half of patients not in first CR at transplantation, the high CR rate of 75% was encouraging. With 4.25 years of median follow-up, 1-year OS in this high-risk group was 31%. Subgroup analysis of 19 standard risk patients showed 63% 1-year OS and 3-year median survival, which compared favorably to reports using alternative matched unrelated donors or cord blood. Despite the use of TCD regimen, immune reconstitution analysis demonstrated encouraging evidence of quantitative lymphocyte recovery through expansion of transplanted T cells by 3 to 6 months. The heterogeneous results reported in the literature is likely a result of a composite of diverse regimens and protocols been employed by different treatment centers. The difference in TRM among various treatment centers is also likely a result of patient selection, TCD methodology, conditioning regimen used, supportive care guidelines, and experience in transplanting physicians. BENEFITS IN IMMUNE RECOVERY FROM USAGE OF HLA MISMATCHED DONORS Effects of NK Cell/KIR Ligand on Outcome of Haploidentical Transplant The translation of NK cell recognition of missingself into clinical practice of haploidentical transplant has opened innovative perspectives in the cure of leukemia. Donor-derived NK cells have the potential to promote engraftment, suppress GVHD, and promote GVT, whereas host-derived NK cells can mediate graft rejection and affect GVHD by eliminating donor HSCs or activated T cells, respectively. NK cells are negatively regulated by MHC class I-specific alleles [66]. Lack of expression of self-mhc molecules on mismatched allogeneic targets results in susceptibility to NK cell-mediated lysis ( missing-self recognition). In humans, inhibitory cell killer immunoglobulin (Ig) receptors (KIRs) recognize groups of HLA-C and HLA-B molecules (KIR ligands). Consequently, when faced with KIR ligand-mismatched allogeneic targets, KIR-bearing NK cells sense the missing expression of self-class I alleles and mediate cell killing. The important role of alloreactive NK cells in the setting of haploidentical transplant has been previously demonstrated. Extensive TCD to prevent GVHD in the setting of haploidentical transplant allows rapid regeneration of NK cells in the graft. Ruggeri et al. [67] have shown that NK alloreactivity reduced the risk of leukemia relapse in 57 AML patients receiving haploidentical transplant, whereas improving engraftment and protecting against GVHD. In a recent updated analysis of 112 adult high-risk AML patients who have received haploidentical transplant from 1993 to 2006 [68], the Perugia investigators demonstrated that transplantation from NK alloreactive donors does not cause GVHD and helps to control leukemia relapse in patients who are transplanted in remission. The marked graft versus-leukemia (GVL) effect has translated into a marked survival advantage [65% event-free survival in patients in any complete remission (CR)]. Although such positive effects of a KIR ligand-mismatched haploidentical transplant is only seen in AML for adults in the Perugia study, similar benefits with lower risk of relapse was also observed in a study in St. Jude s Children Research Hospital among pediatric patients with acute lymphocytic leukemia (ALL) who had received transplant from haploidentical NK-alloreactive donor [69,70]. Several groups of investigators subsequently tested the KIR ligand incompatibility model in patients given grafts from HLA-mismatched unrelated donors [71-74]. Two studies found lower risks of relapse in patients with KIR ligand incompatibility in the GVH direction [71,74], although 2 others did not find such association [72,73]. The heterogeneous results are likely attributable to other factors, such as the extent of donor TCD [75], the speed at which NK cells recover, and/or the use of posttransplant immune modulation. These data point to the need for further study under different transplant procedures and conditions. Nevertheless, the recently demonstrated benefits of NK alloreactivity are expected to encourage greater use of haploidentical transplantation in the future. In patients with advanced or refractory malignancies, the alloreactivity of NK cells has been exploited as a form of adoptive immunotherapy, providing a potential role as an adjunct to HSC transplant. Miller et al. [76] recently demonstrated the safety and potential benefit of adoptive haploidentical-related NK cell therapy without HSCT following high-dose intensity conditioning. All NK cell donors were haploidentical family members; few were KIR-ligand-mismatched in the GVH direction. Twenty-six percent of a small cohort of poor-prognosis patients with AML achieved

12 1260 L.-P. KOH et al. complete hematologic remission of their leukemia. Intriguingly, a significantly higher complete remission rate was observed when KIR-mismatched donors were used. The study also demonstrated in vivo expansion of donor-derived NK cells in the majority of the treated patients, in association with increased levels of endogenous IL-15. More importantly, donor NK-cell infusions were well tolerated without evidence for induction of GVHD. These findings suggest that haploidentical NK cells can persist and expand in vivo, and may have a role in the treatment of selected malignancies when used alone or in association with HSCT. Additional studies are needed to determine how best to exploit the potential benefit of NK cells in allogeneic HSCT by promoting their recovery with cytokines such as IL-15 or by selection of specific subsets. The positive impact of donor versus recipient NK alloreactivity in the haploidentical transplant studies has important implications for donor selection. Ruggeri et al. [68,77] proposed that when it is incorporated as a criteria for donor selection, the random 33% chance of finding an NK-alloreactive donor increases to 49%, which approaches the maximum, because 1/3 of the population expresses all three class I alleles and are resistant to alloreactive NK killing. The near 50% chance of finding NK-alloreactive donors compares favorably with the odds of finding an unrelated donor and has the advantage of no delay between decision making and transplant as haploidentical donors are immediately available. Effects of Nonfetal Maternal Antigen on Outcome of Haploidentical Transplant The potential benefit of feto-maternal immunologic tolerance in allogeneic HSCT was recently demonstrated [78-81], and may serve as a new parameter in selection of donors. Based on the results of a nationwide HSCT survey conducted in Japan [80] and a large IBMTR analysis, maternal stem cell donation was found to be better for HSCT than paternal donation in mismatched transplantation. van Rood et al. [79], in their large IBMTR analysis, have shown that the recipients of non-tcd maternal transplants had a significantly lower incidence of cgvhd than the recipients of paternal transplants in haploidentical 1- or 2-antigen-mismatched transplantations. They have also demonstrated a lower rate of agvhd and TRM in sibling transplantations mismatched for noninherited maternal antigens (NIMAs) compared with those mismatched for noninherited paternal antigens (NIPAs). Separate studies from Japan have confirmed the tolerizing effect of NIMAs after myeloablative [78,82] and RIC [83,84]. Although no differences in risk of clinically significant agvhd were noted in one study, 5-year overall survival was significantly higher and TRM was lower among recipients of maternal grafts compared to paternal grafts. In other studies, significantly lower risks of GVHD were observed among NIMA-mismatched transplant recipients [84]. NIMA-mismatched sibling donor and recipient share the inherited paternal antigens (IPAs) and are mismatched at the maternal antigens, but there are microchimeric cells expressing the NIMAs. These observations support the hypothesis that offspring may be tolerant to haploidentical relatives expressing NIMAs (mother or NIMA-mismatched siblings), and the microchimeric mother may be hyporesponsive to IPAs of the offspring. These encouraging results reported so far provide rationale to assess the feasibility of haploidentical stem cell transplant (SCT) using either myeloablative or nonmyeloablative conditioning regimen, from mother to offspring and vice versa, or from NIMA-mismatched siblings. The approach may provide a more appropriate donor selection in HLA-haploidentical HSCT resulting in both less toxicity and better antitumor effect. Effects of Donor Lymphocyte Infusion on Outcome of Haploidentical Transplant Although there has been a modest trend to reduce TRM in the past few years, there has been no clear reduction in disease relapse, which still surpasses 50% in high-risk patients [28,43]. The risk of day-100 TRM has been reduced to below 20% in several recent series of haploidentical transplant using reduced-intensity conditioning protocols [67,58,64]. The high relapse rate observed in most series, apart from the inclusion of high proportion of high-risk patients or patients with refractory diseases at the time of transplant, is attributable to the delayed immune recovery and abrogated GVL effect with the use of TCD graft. Intensification of conditioning regimens is unlikely to compensate for the loss of T cell-related GVL effect, as the benefit usually is offset with the increase the regimen-related toxicity. Additional posttransplant strategies such as donor lymphocyte infusion are potential therapeutic option for relapse prevention. Donor lymphocyte infusions (DLI) provides direct and potent GVL activity to treat relapse in patients who have undergone HLA-matched, related, or unrelated HSCT [85,86]. Reports on the use of unmanipulated DLI in haploidentical transplant, both prophylactically and therapeutically, remain scanty [29,38,39,87-91]. The diverse results reported in terms of efficacy, adverse events, and survival outcome is more a reflection of the heterogeneity of patients being treated using this therapeutic strategy. In addition, given the limited number of patients in most of these reports, it is difficult to draw definite conclusions about the relationship between cell dose given, GVHD, and GVL effects. Nevertheless, several important observations were made: (1) GVHD remains