Unrelated donor peripheral blood stem cell transplants incorporating pre-transplant in-vivo Alemtuzumab are not associated with any increased risk of significant acute or chronic graft-versus-host disease
Dr Bronwen Shaw, The Anthony Nolan Research Institute, Royal Free Hospital, Pond Street, Hampstead, London NW3 2QG, UK. E-mail: email@example.com
There is little information published comparing peripheral blood stem cells (PBSC) with bone marrow (BM) as the stem cell source in the long-term outcome in recipients of T-cell depleted (TCD) unrelated donor (UD) transplants. We present retrospective outcome data on 306 recipients of myeloablative, human leucocyte antigen-matched UD allografts using pre-transplant in-vivo Alemtuzumab. Transplants were performed between 2000 and 2007 for chronic myeloid leukaemia in first chronic phase and acute leukaemia in first or second complete remission; 184 patients received BM and 122 PBSC. The median age was 28·9 years (<1–58) and the median follow-up was 48 months. Overall survival at 8 years was 53%. The incidence of acute graft-versus-host disease (GvHD) was significantly higher in PBSC (65%) than BM recipients (49%; P = 0·012). This represented only grade 1 GvHD with no difference in grade II–IV aGvHD (BM 23% PBSC 24%). The incidence of chronic GvHD, either overall (BM 47%, PBSC 49%) or extensive (BM 15%, PBSC 13%) was not increased with PBSC. The incidence of relapse, non-relapse mortality and survival were not significantly different. Whilst accepting the limitations of retrospective analyses, we suggest the increased risk of GvHD in recipients of PBSC in T-replete transplants is offset by in-vivo Alemtuzumab, and that either stem cell source can be used with good outcomes in this setting.
The use of unrelated donors (UDs) for allogeneic transplantation has extended the availability of this potentially life-saving therapy to those who lack a sibling donor. In view of the increased genetic disparity between recipient and donor the incidence of some complications, such as graft-versus-host disease (GvHD), are increased. Recent advances in technology, including those related to human leucocyte antigen (HLA)-typing, have resulted in the outcome of UD transplants reaching that of transplants using sibling donors (Goldstone & Rowe, 2009; Lee et al, 2009), in part due to a reduction in the incidence of severe GvHD. One highly successful strategy to reduce GvHD is the use of T cell depletion (TCD) – either in vivo or ex vivo (Kottaridis et al, 2000; Marks et al, 2000).
A commonly used method of in-vivo TCD is Alemtuzumab, given for several days pre-transplant. Alemtuzumab is an antibody against CD52 and as such depletes not only T cells, but also other cell subsets, such as B cells and dendritic cells (Hale et al, 1998). Alemtuzumab has been shown to persist in the patient’s circulation for up to 30 d post-infusion, allowing for depletion of CD52-positive cells from the patient as well as from the donor cells on infusion (Rebello et al, 2001).
Historically, all donors provided stem cells via a bone marrow (BM) harvest. Since the 1990s we have had the ability to collect granulocyte colony-stimulating factor (GCSF)-mobilized peripheral blood stem cells (PBSC) and their use in UD has increased considerably since 2000 (Ljungman et al, 2009). In view of the low incidence of significant short-term side effects and the ease of donation, c. 70–80% of UDs opt to donate via this method in preference to BM harvesting (Miller et al, 2008; Pulsipher et al, 2009). It was clear from early experience that the number of cells (CD34+ and CD3+ cells) harvested via PBSC was greater than obtained by BM harvesting (Pavletic et al, 1998). While this was associated with significantly shorter engraftment times, there were concerns that the incidence of GvHD would be increased with the use of PBSC. Indeed, a number of studies, predominantly in sibling transplants, showed an increased incidence of both acute and/or chronic GvHD in PBSC recipients (Storek et al, 1997; Schmitz et al, 2005, 2006) without, however, an impact on patient outcome (Ringden et al, 2002; Schmitz et al, 2005; Gallardo et al, 2009; Friedrichs et al, 2010). More recent studies have considered UD transplants (Blau et al, 2001; Remberger et al, 2001, 2005; Garderet et al, 2003; Eapen et al, 2007), again confirming an increase in GvHD with PBSC, but not consistently showing any difference in outcome. However, most of these studies included recipients of both HLA-matched and -mismatched grafts and all report series where transplants were predominantly T cell replete.
A few studies have suggested a differential effect on survival due to stem cell source, both in sibling (Eapen et al, 2004; Stem Cell Trialists’ Collaborative Group., 2005) and UD transplants (Garderet et al, 2003; Eapen et al, 2007). The large meta-analysis including 1111 adult recipients of HLA-identical sibling grafts (Stem Cell Trialists’ Collaborative Group., 2005), showed that PBSC was associated with increased incidences of both acute and chronic GvHD. However, PBSC was also associated with a decrease in relapse in both early and late stage disease, with a significantly improved survival in recipients of PBSC with late stage disease. Conversely, in another study (Eapen et al, 2004) in children receiving HLA-identical sibling allografts, PBSC was associated with a higher incidence of chronic GvHD, as well as a worse overall survival (OS) and a higher transplant-related mortality (TRM). Furthermore in a Center for International Blood and Marrow Transplant Research study of UD transplant outcome, Eapen et al (2007) showed a significantly higher incidence of both grade II–IV acute GvHD and of chronic GVHD. Although there was no detrimental effect on OS in the whole group, the subgroup of patients with chronic myeloid leukaemia (CML) in first chronic phase (CP1) receiving PBSC did have a significantly worse OS, with a trend to an increase in TRM. Garderet et al (2003), reported a trend to an increase in acute (but not chronic) GvHD in patients with acute lymphoblastic leukaemia receiving PBSC, which was associated with an inferior survival compared to BM (without an impact on relapse).
We hypothesized that any negative impact caused by increased GvHD using PBSC compared to BM might be abrogated in recipients of UD transplants incorporating in-vivo Alemtuzumab conditioning for T cell depletion. Here we present data on the outcome in 306 UK transplant recipients with leukaemia regarding the impact of stem cell source on outcome.
Materials and methods
Patients fulfilling the requirements for this study were identified from the database held by the British Society for Blood and Marrow Transplantation (BSBMT) registry in London, UK. Data are reported at fixed time points in the form of the MED A and B forms derived from the European Group for Blood and Marrow Transplant (EBMT)). Data collected by the Anthony Nolan Trust from a previous study were added to the BSBMT database. Three hundred and six patients from 19 centres were identified. Supplementary data requests were sent to all centres.
Patients, definitions and inclusion criteria
Criteria for inclusion were: (i) Transplant at a UK centre between 2000 and 2007; (ii) Myeloablative conditioning; (iii) In-vivo pre-transplant T cell depletion using alemtuzumab; (iv) HLA-matched (10/10) UD; (v) Standard risk leukaemia (acute myeloid or lymphoblastic leukaemia in first or second complete remission (CR), CML in CP1).
Human leucocyte antigen typing was performed in the local Histocompatibility and Immunogenetics laboratories by molecular methods to achieve high-resolution types.
Alemtuzumab schedule and dose were given according to institutional protocols. Both the overall dosing and the days of infusion differed between institutions. However the dose of Alemtuzumab within an institution did not differ depending on whether the stem cell product was BM or PBSC. Cytomegalovirus (CMV) was treated pre-emptively. Donor leucocyte infusions (DLI) were not used routinely, but given in some cases according to institutional protocols for mixed chimaerism, relapse or viral reactivations.
Primary graft failure was defined as a failure to achieve a neutrophil count of 0·5 × 109/l by day 28 and was evaluable only in those surviving at least 28 d after the transplant. Relapses were defined as haematological, cytogenetic or molecular. Acute and chronic GvHD were defined using internationally accepted criteria.
The protocols used were approved by the individual institutional review boards of all of the contributing hospitals. Informed consent for the transfer of data to, and the analysis by, the BSBMT was obtained from all patients.
Relationships between categorical variables were analysed by Fisher’s exact test. Continuous variables such as age were tested using the k-sample equality of medians test. OS was calculated by Kaplan–Meier analysis. Univariate analysis of OS and PFS was performed using the log-rank test, and multivariate analysis by Cox regression. Cumulative incidence of non-relapse mortality (NRM) was calculated by competing risks regression, with death from relapse as the competing risk. Relapse rate was calculated as a cumulative incidence by competing risks regression, with transplant-related death and chronic GvHD as the competing risks. Cumulative incidence of cGvHD was calculated by competing risks regression with death and relapse as competing risks. All multivariate models were designed to include any factor that had a P-value of <0·2 in univariate analysis plus the source of stem cells. In view of the relatively small dataset and number of significant variables, stepwise inclusion/exclusion were not applied.
Statistics tests were performed using Stata (StataCorp, College Station, TX, USA; URL: http://www.stata.com), with competing risks calculated using ‘stcompet’ (Enzo Coviello, Italy; May Boggess, StataCorp); and using R (R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org), with competing risks calculated using the package ‘cmprsk’ (Gray, 1988; Fine & Gray, 1999).
Patient and transplant characteristics
Three hundred and six patients were included in the study. Patient, disease and transplant characteristics are presented in Table I. The majority of conditioning regimens included total body irradiation (92%), usually in combination with cyclophosphamide (84%), and additional GvHD prophylaxis with ciclosporin alone or ciclosporin and short-course methotrexate was used. In all patients T cell depletion was performed using pre-transplant in vivo Alemtuzumab. The two groups were similar, except for an increased use of BM in patients with CML (P = 0·007) and the disease stage differed significantly between the groups (P = 0·001). Twenty-four patients received DLI)post-transplant, 17 for disease relapse, two for viral reactivations and five for mixed chimaerism.
Table I. Patient and donor demographics.
BM (n = 184)
PBSC (n = 122)
Year of transplant (median)
M, male; F, female; TBI, total body irradiation; Cy, cyclophosphamide; Bu, busulphan; CMV, cytomegalovirus; pos, positive; neg, negative; AML, acute myeloid leukaemia; ALL, acute lymphoblastic leukaemia; CML, chronic myeloid leukaemia; AL, acute leukaemia; CR, complete remission; CP, chronic phase.
Patient age, years (median, range)
Donor M, recipient M
Donor M, recipient F
Donor F, recipient M
Donor F, recipient F
Donor/patient CMV status
Donor pos, recipient neg
Donor pos, recipient pos
Donor neg, recipient neg
Donor neg, recipient pos
CD34+ cell dose (median, range)
One percent (4/306) of patients had primary graft failure. Of these, two received BM and two received PBSC (two had a second allograft and are alive, while two succumbed from infection or disease progression, one each in the BM and PBSC groups). A further four patients who received BM had secondary graft failure (two of whom remain alive following further cellular therapy) while none of the PBSC recipients developed this complication. There was no statistically significant difference between stem cell sources in the incidence of graft failure overall (P = 0·306). The median time to neutrophil engraftment was 18 d (range: 9–61 d). This was significantly faster in recipients of PBSC (15 d) than BM (20 d, P < 0·0001). Likewise, platelet engraftment was significantly faster in recipients of PBSC (17 d) than BM (28 d, P < 0·0001), with a median of 24 d (5–469).
All patients who achieved neutrophil engraftment were considered eligible for analysis. The incidence of acute GvHD (aGvHD) of any grade, grade II–IV and grade III–IV were 56%, 23% and 4%, respectively. There was a significantly higher risk of developing aGvHD in the PBSC recipients (65%) compared to BM recipients (49%, P = 0·012), but this risk was only apparent in grade 1 disease. Most importantly, there were no significant differences in severe aGvHD between the groups (grade III/IV: BM 4% PBSC 5%, P = 0·554) (Table II). No other factor had a significant impact on the incidence or severity of aGvHD in either univariate or multivariate analysis.
Table II. Incidence of acute GvHD comparing PBSC to BM.
*Refers to the lack of statistical significance between those with grade III/IV acute GvHD in the BM and PBSC group (i.e grade III/IV vs 0–II).
Overall chronic GvHD (cGvHD) developed in 119/235 patients (48% at 5 years). There was no significant difference in the incidence of cGvHD dependant on stem cell source (5 years: 47% following BMT and 49% following PBSCT; P = 0·48) (Fig 1). Strikingly, extensive cGvHD occurred in only 15% of BM recipients and 13% of PBSC recipients (P = 0·831). Factors associated with cGvHD were CMV seropositivity in the patient [Hazard ratio (HR) =1·56; 95% Confidence interval (95% CI) 1·07–2·27, P = 0·029] and disease in CR1/CP1 (HR 0·66; 95% CI 0·43–1·00, P = 0·054). In multivariate analysis, including these two factors and source of stem cells and age, only disease status at transplant showed a trend towards significance (HR=0·68; 95% CI 0·44–1·05, P = 0·081).
Non-relapse mortality (NRM)
Non-relapse mortality was 9% and 20% at day 100 and 1 year, respectively (Fig 2). There were no significant differences in NRM between recipients of BM or PBSC (day 100: 9% and 11%, 1 year: 20% and 20% and 5 years: 23% and 25%, respectively, P = 0·77). There was no significant impact on NRM due to the presence or absence of aGvHD. Factors associated with NRM in univariate analysis were: patient CMV (HR 2·15; 95% CI 1·34–3·47, P = 0·001) and donor (HR 1·54; 95% CI 0·93–2·53, P = 0·090) CMV seropositivity. In multivariate analysis (including these factors, patient age and stem cell source) the only factor to remain significantly associated with a higher mortality was CMV seropositivity in the patient (HR=1·98; 95% CI 1·16–3·38, P = 0·012). The causes of NRM were: GvHD (22%), infection (62%) and other (16%), which did not differ significantly depending on stem cell source (P = 0·358).
The cumulative incidences of relapse were 36% and 41% at 2 and 5 years respectively (Fig 3). Relapse was predominantly haematological in patients with acute leukaemia (AL) (86%), with 6% defined as cytogenetic and 8% defined as molecular. In contrast, only 41% of relapses in CML were haematological with 26% defined as cytogenetic and 33% defined as molecular. There was a trend to a difference in relapse rate according to whether the patient received BM or PBSC (5 years: 45% and 33% respectively, P = 0·096). The only factor to impact significantly on disease relapse was disease type, with a relapse risk of 58% at 5 years in CML patients compared to 33% in AL patients (P < 0·001). There was a trend to an increase in relapse in CMV seronegative patients (HR=0·69; 95% CI 0·46–1·04, P = 0·075). There were no differences in relapse risk between patients with AL transplanted in CR1 or CR2 (both 33%). In multivariate analysis, including disease, patient and donor CMV and stem cell source, the only significant factor was the underlying disease (HR=2·03; 95% CI 1·38–2·98, P < 0·001). There was no significant impact on relapse due to the presence or absence of aGvHD.
The median follow up was 48 months (3–110 months). This was significantly longer in recipients of BM than PBSC (59 vs. 36 months, P = 0·001) (Fig 4). The OS in the whole group was 53% at 8 years, with progression-free survival of 36% at the same time point. There were no significant differences in survival between patients who received BM or PBSC (8 years: 54% and 52% respectively, P = 0·571). Likewise progression-free survival was similar (8 years: 32% and 42% respectively, P = 0·225). There was no significant impact on OS due to the presence or absence of aGvHD. Two factors were significantly associated with an improved survival in univariate analysis: CML compared to AL (P = 0·005) and CMV seronegativity in patients (P = 0·025). Both of these factors remained significant in multivariate analysis (HR 0·57; 95% CI 0·37–0·86, P = 0·010 and HR 0·70; 95% CI 0·49–0·99, P = 0·046 respectively), while PBSC vs. BM was not significant (HR 1·03; 95% CI 0·73–1·49, P = 0·826).
This study is the first to consider the impact of stem cell source in a cohort of in-vivo TCD HLA-matched UD allograft recipients. In contrast to previous studies of T cell-replete UD transplants, we found no increase in the risk of grade II–IV acute GVHD or of chronic GVHD in recipients of PBSC. The only significant difference between BM and PBSC recipients in this study was a higher incidence of grade I aGvHD in the PBSC group.
In 617 adult recipients of T cell-replete UD transplants, Eapen et al (2007) reported an increase in the incidence of overall, but not grade III/IV, aGvHD with PBSC. In addition, unlike our report, the incidence of cGvHD was also found to be significantly higher in the patients receiving PBSC. Two smaller studies (Garderet et al, 2003; n = 213; Blau et al, 2001; n = 74) reported significant increases in aGvHD (only in HLA-mismatched patients in the latter study), but no increase in cGvHD. Another group, reporting on both short (Remberger et al, 2001) and long term (Remberger et al, 2005) follow-up in 214 adult transplant recipients, found no increase in aGvHD while the incidence of chronic extensive GvHD was significantly higher when PBSC was used compared to BM.
There are a number of clear differences between these studies, the most important being the use or not of T-cell depletion, in particular, Alemtuzumab. A possible additional factor explaining the lack of impact of stem cell source on cGvHD is the differing length of follow-up in each study. The median follow-up reported by Remberger et al (2005) was over 4 years (which is the same as in our current study), while that in the other studies ranges between 1·5 and 3 years and, in some, differed between the PBSC and BM groups. The pattern of GvHD may also be of interest. All of the studies to date reported outcomes using the ‘old’ classification of GvHD i.e. using a simple time cut-off between acute and chronic GvHD. It may be that use of the reclassification of GvHD in future studies, would provide further insights (Filipovich, 2008). In our study we noticed that the time to onset of cGVHD was shorter in the PBSC than the BM group (data not shown), suggesting that a proportion of the ‘chronic’ GvHD would now be reclassified as late onset acute or persistent rather than de novo chronic GvHD. It is well recognized that chronic GvHD is more likely in patients who have suffered acute GvHD (especially of higher grades) which may explain the increase in cGvHD in the study reported by Eapen et al (2007) (where the incidence of acute GvHD was higher than in our study), while this may be less discernable in the studies by Blau et al (2001) and Garderet et al (2003), where the overall and grade II/IV aGvHD incidence was less.
T cell depletion using Alemtuzumab was universal in our study. This method of GvHD prophylaxis was either absent or used in a very small minority of patients in the other studies (Blau et al, 2001; Garderet et al, 2003; Eapen et al, 2007) except for one (Remberger et al, 2005), where one third of patients received antithymocyte globulin (ATG). Direct comparisons between ATG and Alemtuzumab cannot be made because Alemtuzumab depletes all CD52-positive cells and not just T cells. The use of ATG in only a proportion of patients and the likelihood of ‘hidden’ HLA mismatches in that study, especially for HLA-C (low resolution HLA-A and -B and high resolution -DRB1 matching criteria were used) may account for the high incidence of chronic GvHD that they observed. All of the recipients included in our study had high resolution typing performed.
Neither our or other UD studies have reported any significant differences due to stem cell source in TRM, relapse or OS in the overall group studied. In the study reported by Eapen et al (2007) there was a trend to a higher TRM, with a significantly lower OS, in recipients with CML in CP1, although the reason for this is not clear. In the study reported by Garderet et al (2003) patients with ALL receiving PBSC had a significantly lower disease-free survival and OS. Besides the obvious differences with regards to TCD, other differences between the studies are noted, such as the inclusion of both paediatric and adult patients, the degree of HLA matching, the diseases and stage of disease included, medical GvHD prophylaxis and the year of transplantation. None of the studies in UDs are randomized and the results of the Blood and Marrow Transplant Clinical Trials Network 0201 randomized trial (Peripheral Blood versus Bone Marrow Grafts from UDs) (http://www.cibmtr.org/Studies/ClinicalTrials/BMT_CTN/Protocols/0201/index.html) are keenly awaited.
It is recognized that the current study has a number of limitations. In common with most retrospective studies there is some heterogeneity in the study group. Although TCD using Alemtuzumab was universal, the schedule and dosage varied between centres. Importantly, however, the dose of Alemtuzumab within an institution did not differ depending on whether the stem cell product was BM or PBSC. In addition the use of cyclosporin (with or without methotrexate) differed by centre. DLI were given for various reasons, but none as part of a planned immunotherapy program. Any impact of DLI on outcomes should be negligible, as most were given for relapse, and this was taken into account as a competing risk for chronic GvHD. Although the number of CD34+ cells is known, the data on CD3+ and other cell subset content of the graft were not available.
In conclusion, this study confirmed the low incidence of severe GvHD following TCD myeloablative transplantation from UD for patients with standard risk leukaemia, with excellent long-term survival rates. The use of either BM or PBSC was associated with equally good outcomes. The only significant difference between the groups was an increase in grade 1 acute GvHD in the PBSC recipients. The risk of extensive cGVHD was low and did not differ with stem cell source. These data suggest that either stem cell source can be used with equivalent results and that there is no deleterious effect from the use of PBSC in Alemtuzumab-containing UD transplants.
We would like to thank all of the UK transplant teams who have contributed patients to this study.
JLB, NHR and BES contributed to the design and analysis. RP performed the statistical analysis. All authors contributed to the data collection, writing and review of the manuscript.
Conflict of interest
There are no conflicts of interest to disclose for any of the authors.