Comparison of Mobilized Peripheral Blood Stem Cells Versus Unmanipulated Bone Marrow Haploidentical Transplantation Using Post-Transplant Cyclophosphamide: A Retrospective Study of SFGM-TC in 176 Patients

Introduction

The administration of high dose cyclophosphamide post-transplant (HD-Cy) has improved the outcome of patients who receive an unmanipulated haploidentical graft. The modalities described by the group of Baltimore originally associated a non myeloablative conditioning regimen and the use of a T cell replete bone marrow (BM) graft. Nevertheless, mobilized peripheral blood (PB) is increasingly used as it is more convenient. A very recent report from the United States showed a lower rate of graft versus host disease (GVHD) with BM but an increased risk of relapse in leukemic patients, as compared to PB, leading to a similar overall survival (OS). We report the French experience of haploidentical transplantation using HD-Cy and unmanipulated BM or PB, with a particular interest on the contents of the grafts.

Patients and methods

One hundred seventy-six patients with hematologic malignancy received a haploidentical transplant in 21 French centers from May 2012 to December 2015. We compared the outcome of patients who received a BM graft (n=89) with those receiving a PB graft (N=87). Median age at transplant was 44 year-old (2.5-72). One hundred thirteen (64%) patients had myeloid disease, 19 (11%) had acute lymphoblastic leukemia and 44 (25%) had lymphoid neoplasm. Prophylaxis of GVHD consisted in anticalcineurin and mycophenolate in 94% of the cases. All patients received HD-CY post-transplant (100 mg per kg total dose).

Results

The 2 groups were comparable for age, disease status at transplant, type of conditioning regimen (RIC 71% and 80% for BM and PB respectively; p=0.16), recipient cytomegalovirus (CMV) seropositive status, ABO mismatch. PB recipients were more likely to receive antithymocyte globulin (ATG) (N=19 vs 0) and to have a higher disease risk index (DRI). The number of CD34+ and CD3+ infused cells was higher in PB grafts (median 2.64x106 CD34/Kg and 26 x106 CD3/Kg for BM and 6.03x106 CD34/Kg and 191 x106 CD3/Kg for PB. Median follow up was 281 days (6-1851).

In univariate analysis, the rate of patients who recovered neutrophils and platelet (>50G/L) were similar in both groups (91% and 75% vs 96% and 75% for BM vs PB respectively). However, median time to neutrophil engraftment was significantly longer in BM than in PB group (20 days (13-59) vs 18 days (13-45) respectively, p=0.004). Median time to platelet recovery was not significantly different (30 days (17-250) vs 24 days (13-52) in BM and PB, p=0.14). Patients who recovered neutrophils received a graft which contained more CD3+ cells than patients who did not (p=0.007).

At one year, OS, progression free survival (PFS) and relapse were similar in the 2 groups (OS = 64% vs 53% (p=0.1), PFS= 53% vs 46 % (p=0.17) and relapse rate 28% vs 34% (p=0.19) in BM and PB respectively). Acute GVHD II-IV or III-IV and chronic GVHD were similar in both groups (aGVHD II-IV= 18% vs 28% (p=0.09) and aGVHD III-IV = 7% vs 9% (p=0.48), cGVHD=13% vs 15% (p=0.6) in BM and PB respectively), resulting in a similar GVHD-free relapse-free survival (GRFS) at 1 year (48% vs 37% (p=0.09)). Interestingly, the number of infused CD3+ cells/Kg impacted the incidence of acute GVHD II-IV (p=0.049).

High DRI and the lack of complete remission at transplant had a significant pejorative effect on OS, PFS, relapse, GRFS and platelet recovery. CMV seropositive recipients had a worse GRFS (p=0.045).

Conclusion

In this retrospective study the use of PB instead of BM does not significantly impact the outcome of patients who received a haploidentical unmanipulated graft for a malignant disease with HD-CY. PB graft contained more CD3+ and CD34+ cells than BM and were associated with a faster neutrophils recovery. A higher number of CD3+ cells increased the risk of acute GVHD II-IV in univariate analysis, without significant effect on GRFS. Prospective randomized studies are needed to evaluate the effect of PB vs BM in haploidentical settings.