Optimizing autologous hematopoietic stem cell transplantation for acute leukemia

Abstract Autologous hematopoietic stem cell transplantation (ASCT) remains an important postremission treatment for acute leukemia (AL). It is known that some prognostic factors, such as age, cytogenetic and molecular risk stratification, and minimal residual disease (MRD) status, are closely related to clinical outcomes following ASCT. Moreover, there are multiple measurements, including pretransplant treatment, stem cell mobilization and collection, conditioning regimens, and maintenance treatment after transplantation, that can affect prognosis after ASCT. Our clinical practice of ASCT should be better standardized to further improve patient outcomes. This review outlines optimization and quality control measures for ASCT developed at the Institute of Hematology and Blood Diseases Hospital of the Chinese Academy of Medical Sciences, the first established and largest autologous stem cell transplant center in China. These measures will enhance the development of best practices and strategies for AL ASCT therapies, thereby improving patient outcomes.

differentiate to common lymphoid progenitors (CLP) and common myeloid progenitors (CMP), thus giving rise to all blood cell types. 3 Stem cells exhibit five minimal functional states, namely self-renewal, multilineage differentiation, apoptosis, resting mode, and trafficking, which constitute the so-called "SMART" model for maintaining stem cell homeostasis in vivo. 4,5 The mechanisms underlying HSC homeostasis involve endogenous and exogenous regulations. Endogenous regulation represents several classical signal pathways (such as the Notch and Wnt signaling pathways) and key transcriptional regulators, 6,7 whereas exogenous regulation is more inclined to refer to the regulatory unit, defined as the bone marrow (BM) niche. The components of the BM niche include cell constituents, cell factors, and metabolites such as reactive oxygen species. 8

| Acute myeloid leukemia
In an early prospective randomized study, AML patients who achieved CR after induction treatment with daunorubicin and cytarabine received a first course of intensive consolidation chemotherapy, combining intermediate-dose cytarabine and amsacrine. Then, these patients were randomly assigned to undergo ASCT or a second course of intensive chemotherapy, and ASCT resulted in better LFS than intensive consolidation. 12 Subsequently, another prospective randomized phase 3 trial evaluated the outcomes of 258 ASCT vs 259 chemotherapy patients who were in CR1 and received 2 cycles of intensive chemotherapy. 13 The results further demonstrated that ASCT, compared with chemotherapy, remarkably reduced the relapse rate (58% vs 70% at 5 years, P = .02) and improved LFS (38% vs 29% at 5 years, P = .065). 13 OS was similar between the two treatment arms (44% vs 41% at 5 years, P = .860), which might be attributable to salvage chemotherapy and HSCT for relapse patients on the chemotherapy arm. 13 Similar results were reported in other studies. 14,15 For example, a total of 465 patients aged 40 to 60 years in CR1 were given chemotherapy or ASCT and the results revealed that ASCT was associated with a reduced risk of relapse (hazard ratio (HR) 0.66, 95% confidence interval [CI] 0.50-0.87, P = .003) and improved LFS (HR 0.69, 95% CI 0.53-0.90, P = .006). 14 In addition, nonrelapse mortality (NRM) was not significantly different when comparing ASCT and chemotherapy (HR 2.13, 95% CI 0.69-6.56, P = .180), supporting the acceptable safety of ASCT. 14 Many studies have compared the clinical outcomes of ASCT with those of allo-HSCT to determine the superior treatment option. 12,[14][15][16][17][18][19] MRD status before ASCT is an independent prognostic factor for both OS and RFS after ASCT. 20 Our data show that MRD detection after one course of consolidation chemotherapy was an independent prognostic factor for 3-year OS (83.1% vs 19.0%; P = .006) and disease-free survival (DFS; 73.9% vs 14.2%; P = .049) in patients with AML who underwent ASCT in CR1. 21 Furthermore, we identified cytogenetic risk as an independent prognostic factor for survival outcomes, with decreasing OS and DFS as risk increased. 21 It is also dem-  20 However, the superiority of ASCT was impaired by a higher incidence of relapse, which resulted in a similar 3-year OS and RFS. 20 In the subgroup analysis of intermediate-risk patients, haplo-HSCT yielded better OS due to higher incidence of relapse in ASCT patients. 20 Collectively, ASCT and allo-HSCT yielded similar outcomes regarding OS and RFS. However, the potential impact of acute and chronic graft vs host disease caused by allo-HSCT on the quality of life of patients was not considered in these studies. Furthermore, regarding the burden of hospitalization and challenging supportive treatment for MUD-HSCT and haplo-HSCT, ASCT is recommended for favorable-risk AML patients in CR1 and is an attractive option for intermediate-risk patients in CR1. In addition, as a result of induction chemotherapy utilizing all-trans retinoic acid and arsenic trioxide, acute promyelocytic leukemia (APL) has become a highly curable disease. Besides, there is a role for ASCT in CR2 patients with APL to achieve a cure. 23  in allo-HSCT patients, respectively (P = 0.001 and P = 0.001). 24

| Acute lymphocytic leukemia
There were studies on ASCT in adult ALL patients including B-cell acute lymphoblastic leukemia (B-ALL) and T-cell acute lymphoblastic leukemia (T-ALL).
A recent study on MRD after first induction treatment (MRD1) in adult ALL patients treated with ASCT showed that the proportion of high-risk immunophenotype (pro-B, pro-T, pre-T, mature T) was significantly higher in MRD1-positive patients than that in MRD1-negative patients (34.6% vs 14.5%, P = .038). Positive MRD1 and high-risk immunophenotype were risk factors for LFS (HR = 3.986, 95% CI 1.813-8.764, P = .001; HR = 2.981, 95% CI 1.373-6.473, P = .006). ASCT could not reverse the poor prognosis of MRD1-positive patients. ASCT treatment is optional for MRD1-negative patients who maintained MRD1 negative during intensive therapy. 25 In ALL patients who achieved CR1 with induction therapy and received at least 4 cycles of early-stage intensive consolidation chemotherapy, ASCT was demonstrated to be a better postremission option compared with chemotherapy. 26 The ASCT group had higher LFS and OS after 3 and 5 years. However, the difference was not statistically significant at 1 year, which reminds us of the importance of long-term follow-ups. 26 To compare the outcomes of ASCT and allo-HSCT in ALL patients, 50 patients with negative MRD receiving ASCT and F I G U R E 1 Prognosis of acute myeloid leukemia (AML) patients after auto-and allo-stem cell transplantation (SCT). 17 A, The overall survival (OS); B, diseasefree survival (DFS); C, nonrelapse mortality (NRM); D, leukemia relapse, in auto group (n = 46, solid line) and allo group (n = 126, broken line) 56 patients receiving allo-HSCT were retrospectively analyzed. 27 There was no significant difference between the two groups in the probability of 3-year OS (74.1% vs 55.1%) and LFS (63.7% vs 53.7%), indicating acceptable efficacy of ASCT in ALL patients with negative MRD. 27 Another study reported a significantly lower NRM and a better OS of ASCT compared with those of haplo-HSCT in 159 ALL patients with CR (2-year NRM, 3% vs 30%, P < 10 À5 ; 2-year OS, 66% vs 40%, P = .010). There was no significant difference in the 2-year LFS, OS and relapse rate between the two groups. 28 Further subgroup analysis found that Ph + ALL patients in CR1 benefited more from ASCT compared with haplo-HSCT. The 2-year LFS, OS, and NRM were 60% vs 26% (P = .005), 76% vs 26% (P = .001) and 4% vs 43% (P < .001) in the ASCT vs haplo-HSCT groups, respectively. 28 These findings indicated that ASCT could be considered as an effective consolidation therapy in CR1 ALL patients at standard risk or with Philadelphia chromosome.
High-risk patients are not suitable for ASCT. An observational study at IH showed that patients with Ph-negative high-risk B ALL had significantly lower OS and LFS in comparison to standard-risk B-ALL patients (3-year OS: 46.1% vs 77.6%, P = .007; 3-year LFS: 39.1% vs 76.3%, P = .001). 29 Positive MRD before ASCT (MRD ≥0.01%) also indicated worse outcomes and the further demand for allo-HSCT. 29 However, it is noteworthy that high risk was not an absolute contraindication. The development of pre-HSCT purification and post-HSCT maintenance therapy can improve the prognosis of ASCT patients. One study reported a 73% 3-year LFS compared to 42.2% for nontreated ASCT, and 50.9% for allo-HSCT in adult patients with T-cell ALL (P < .05). 29 Thus, for those CR patients with no available donor, ASCT combined with graft purification and maintenance therapy can be effectively used as an alternative treatment.
Lyu et al compared the efficacy of ASCT and MSD-HSCT in Ph + ALL patients and found that OS, LFS, and NRM at 3 years were not significantly different between the two groups (P > .05). 30 They also reported that there is no significant difference in terms of OS, LFS, and NRM between the two groups in patients who achieved and remained complete molecular response within 3 months (s3CMR). In patients who did not reach s3CMR, the 3-year cumulative relapse rate in the ASCT group was significantly higher than that of the MSD-HSCT group. Therefore, ASCT is an attractive option for patients with Ph + ALL, especially for those with s3CMR maintained up to transplantation 30 (Figure 3).

Our recommendations
• ASCT is recommended for favorable-risk AML patients in CR1 with negative MRD and an attractive option for intermediate-risk patients in CR1. In addition, there is a role for ASCT in CR2 patients with APL to achieve a cure.
• ASCT is an attractive option for patients with Ph + ALL, especially for those sustaining s3CMR, maintained up to transplantation.

| PRETRANSPLANT TREATMENT
Pretransplant treatment includes induction and consolidation chemotherapy, which are administered according to the National Comprehensive Cancer Network (NCCN) guidelines. 31,32 Although prospective randomized trials are missing, available data suggested that patients with AML in CR1 who were going to be treated with ASCT probably F I G U R E 2 Prognosis of acute myeloid leukemia (AML) patients after auto-and unrelated donor (URD)-SCT. 22  • All patients newly diagnosed with ALL at IH were given a standard induction regimen VDCLP (vincristine + daunorubicin + cyclophosphamide + L-asparaginase + prednisone). After achieving CR, consolidation chemotherapy was administered, which included several regimens such as high-dose MTX, CAM, DOAME, and so on.
• A total of three to four cycles of consolidation chemotherapy should be applied before ASCT.
• Prophylaxis of CNSL is necessary in ALL and intrathecal chemotherapy is typically given throughout the entire course of therapy.
• At least six courses of intrathecal chemotherapy should be given before ASCT for ALL patients.

| STEM CELL MOBILIZATION AND COLLECTION
At IH, most stem cells are harvested from peripheral blood after intensive chemotherapy followed by granulocyte-colony stimulating factor (G-CSF). 29 G-CSF is given at a dose of 10 μg/kg per day for 7 to 10 days after chemotherapy. Peripheral blood stem cells are collected when the white blood cell count rises to 5 to 10 Â 10 9 /L, which is usually achieved at 5 or 6 days after the administration of G-CSF. 35 For patients who failed in the first mobilization (CD34 + cells <1 Â 10 6 /kg in 2 consecutive collection days), another chemotherapy regimen is used to remobilize or collect stem cells from BM as a supplement. There were reports which showed that, Plerixafor, a CXCR4 antagonist, enhanced the engraftment of healthy donor stem cells. Furthermore, plerixafor mobilized and sensitized leukemia cells, which was not suitable for stem cell mobilization in AL patients undergoing ASCT. 36 The ideal collection time is closely related to the patient's age, previous chemotherapy scheme, mobilization plan, and so on. Therefore, each patient needs to be tested to determine the best collection time.
Usually, three factors are considered to determine whether the time is suitable for collection, namely the white blood cell count, the peripheral blood CD34 + cell count and the peripheral blood hematopoietic progenitor cell number. However, with the progress of technology and the establishment of detection standardization, the number of CD34 + cells in peripheral blood before collection proved to be the most accurate predictor of successful collection.

Our recommendations
• G-CSF should be given at a dose of 10 μg/kg per day for 7 to 10 days after chemotherapy. Peripheral blood stem cells should be collected when the white blood cell count rises to 5 to 10 Â 10 9 /L.

| CONDITIONING REGIMENS
At present, total body irradiation (TBI)/cyclophosphamide (Cy) or busulfan (Bu)/Cy are still the two most widely used classic conditioning regimens for ASCT in AL patients. These regimens were originally developed for ASCT and have high immunosuppressive activity, which may not be necessary in the autologous setting. Therefore, researchers have recently attempted to employ novel agents in the conditioning regimens, for example melphalan (Mel), which exhibited stronger antileukemia effects with acceptable or lower hematologic toxicity. Sakaguchi et al retrospectively analyzed the clinical outcomes of 220 AML children who received ASCT after various conditioning regimens. Bu/Cy ± etoposide or Bu/Mel regimens were significantly superior over other Bu-based and TBI-based regimens. 37 Gorin et al also reported 853 patients with AML who underwent ASCT in CR1.
Those patients who received a conditioning regimen with Bu/Mel exhibited a lower relapse risk (RR), better LFS and OS than those who were conditioned with BuCy but NRM was similar in the two groups. 38 In a subsequent study, they further analyzed 1649 adult patients with primary AML and available cytogenetics, autografted from CR1. In the poor risk group, 176 patients received Bu/Cy and 62 patients received Bu/Mel. Bu/Mel was associated with a lower RR at 5 years (53% vs 69%, P = .002), a better LFS (42% vs 25%, P = .002) and a better OS (54% vs 36%, P = .02). However, in the nonpoor risk group, in which 961 patients received Bu/Cy and 450 patients received Bu/Mel, the 5-year RR (50% vs 47%), LFS (45% vs 48%), and OS (56% vs 60%) was not significantly different. The authors concluded that Bu/Mel was the preferable conditioning regimen for poor-risk leukemic patients. 39 In summary, Bu/Mel may be a feasible conditioning regimen for ASCT in patients with AML.
In addition, other conditioning regimens have also been tried. In ASCT, TBI/Cy as pretransplant regimens is associated with a lower relapse incidence and a higher LFS in ALL than Bu/Cy, which demonstrates that TBI is particularly necessary for ASCT in patients with ALL. 41 In a study at IH, fludarabine (Flu 30 mg/m 2 /d Â 3 days) and cytarabine (Ara-c 2 g/m 2 /d Â 3 days) were added to two classical conditioning regimens to treat 27 patients with ALL receiving ASCT to reduce RR. Hematopoietic reconstitution was achieved in all patients except for one patient who died early. The median time of neutrophil and platelet reconstitution was 11 days 9-19 and 16 days 10-53 days, respectively. No serious adverse events occurred during the conditioning, and the 4-year overall RR, transplant-related mortality (TRM) and DFS were 38.8%, 9.3%, and 60.8%, respectively. 42 Thus, the modified standard regimen of patients pretreated by Flu/Ara-c is an effective and safe approach for ALL patients undergoing ASCT.

Our recommendations
• TBI/cyclophosphamide (Cy) or Bu/Cy are still the two most widely used classic conditioning regimens. TBI is particularly necessary for ALL ASCT. The modified standard regimen in which Flu/Ara-c was added to two classical conditioning regimens is an effective and safe approach.  43 Grubovic et al also demonstrated that neutrophil recovery in AML patients was significantly influenced by transfusion support, according to S80 PANG ET AL.
the number of transplanted CD34 + cells and MNC. 44 However, higher CD34 + cell counts (>7 Â 10 6 /kg) are related to higher RR, which might be caused by leukemia residual in the autograft. 45 45 At IH, the median number of infused CD34 + cells was approximately 2 to 3 Â 10 6 / kg, and we also found that CD34 + cells in graft >3.8 Â 10 6 /kg were a poor prognostic factor of DFS for patients with B-ALL in ASCT (P = .021). 25,27,29 Our recommendations • The median number of infused CD34 + cells should be approximately 2 $ 3 Â 10 6 /kg.

| MAINTENANCE THERAPY AFTER ASCT
Relapse is still the main reason for the failure of ASCT in AL patients, especially ALL patients, so it is necessary to explore the protocols of maintenance treatment after ASCT to reduce the relapse rate and improve the curative effect. At present, the application of maintenance therapy after ASCT is mainly given to patients with ALL, and maintenance therapy at IH for ALL after ASCT usually includes maintenance chemotherapy and immunotherapy (including interleukin-2 [IL-2] and interferon-α). 26,29,[48][49][50][51][52] In addition, for Ph + ALL patients, tyrosine kinase inhibitor (TKI) is also in the list of maintenance regimens. 52,53 In one study, posttransplant maintenance was applied in 29 cases with the use of either imatinib (n = 25) or dasatinib (n = 4). 52 In another study, posttransplant maintenance was applied after ASCT with the use of imatinib, nilotinib, or dasatinib. 30   therapies could combine with ASCT to achieve a satisfying prognosis. 58