The role of haematopoietic stem cell transplantation (HSCT) in relapsed follicular lymphoma remains controversial. This study analysed 126 patients with relapsed, advanced stage follicular lymphoma who received BEAM (BCNU [carmustine], cytarabine, etoposide, melphalan)-alemtuzumab allogeneic HSCT (BEAM-allo) (n = 44) or BEAM-autologous HSCT (BEAM-auto) (n = 82). The BEAM-allo group had a younger median age (48 years vs. 56 years, P < 0·001) but received a higher median number of therapies pretransplant (P = 0·015) compared with the BEAM-auto group. There was a higher non-relapse mortality (NRM) in the BEAM-allo group compared with the BEAM-auto group at 1 year (20% vs. 2%, P = 0·001). Older age and heavily pretreated patients were associated with a higher NRM and poorer survival in the BEAM-allo group. There was, however, a significantly lower relapse rate (20% vs. 43%, P = 0·01) at 3 years with BEAM-alemtuzumab, with no relapses after 2 years, compared with a continued pattern of relapse in the autologous group. No difference in overall survival (OS) (P = 0·99) or disease-free survival (DFS) (P = 0·90) was identified at 3 years, whereas a plateau in OS and DFS with crossing of the survival curves in favour of BEAM-allo group was observed. Furthermore, the ability to re-induce remissions with donor leucocytes provides additional benefit in favour of allogeneic HSCT.
Follicular lymphoma is typically characterized by an indolent course with recurrent relapses, progressive shortening in the duration of remission and a median survival of 8–9 years (Johnson et al, 1995; Hunault-Berger et al, 2002). Despite modest improvements in progression-free survival (PFS) with monoclonal antibodies and combination chemotherapy, the disease remains incurable with conventional agents. In particular, patients with advanced stage follicular lymphoma or those with high grade transformation show poor survival with standard therapy (Horning & Rosenberg, 1984; Gallagher et al, 1986). Improved PFS was reported with the use of autologous haematopoietic stem cell transplantation (auto-HSCT) in first remission, but no survival advantage over standard chemotherapy has been demonstrated(Lenz et al, 2004; Deconinck et al, 2005). Furthermore, the only randomized study (Chemotherapy, Unpurged-, Purged-stem cell [CUP] trial) performed in relapsed/refractory disease reports both an improved PFS and overall survival (OS) but suffered slow recruitment and contained small patient numbers (Schouten et al, 2003). A major obstacle to the success of auto-HSCT is the high incidence of relapse that becomes apparent on longer follow-up, reported between 43% and 88% (Ratanatharathorn et al, 1994; Verdonck et al, 1997; Mijovic et al, 1998; van Besien et al, 2003). In addition, an excess of secondary malignancies has been reported by several groups following the use of auto-HSCT (Freedman et al, 1999; Micallef et al, 2000; Howe et al, 2003; Deconinck et al, 2005).
BEAM (BCNU [carmustine], cytarabine, etoposide, melphalan) is commonly used as conditioning for auto-HSCT in lymphoproliferative disorders with low transplant-related toxicity (Chopra et al, 1993; Bastion et al, 1995; Mills et al, 1995). The addition of alemtuzumab as part of an allogeneic regimen, with its associated stable donor engraftment, minimal GvHD and modest relapse rates is encouraging (Faulkner et al, 2004). Here, we report the outcome of 126 patients with advanced stage follicular lymphoma, who either received autologous transplantation using BEAM conditioning (BEAM-auto) or allogeneic HSCT incorporating BEAM-alemtuzumab (BEAM-allo).
Design and methods
The outcome of patients with follicular lymphoma who received either BEAM-auto HSCT (n = 82) or BEAM-allo HSCT (n = 44) between 1992 and 2005 was analysed from two UK transplant centres. The diagnosis of follicular lymphoma was made according to the World Health Organization classification (Jaffe et al, 2001). Patients were eligible for BEAM-allo HSCT if they were heavily pretreated at the time of referral, were refractory to salvage chemotherapy, or had failed mobilization of autologous CD34+ cells. Patients were eligible for autologous HSCT if they had disease chemosensitive to salvage chemotherapy or if no donor was available for allo-HSCT. Protocols were approved by the local ethics committee and informed consent was obtained by all patients prior to study entry.
The median age of the BEAM-auto group was 56 years (range 30–74 years) vs. 48 years (range 31–59 years) in the BEAM-allo group. However, BEAM-allo patients were more heavily pretreated, having received more lines of chemotherapy pretransplant at 3 lines (range 1–8) vs. 2 (range 1–6). Of the 44 patients who received BEAM-alemtuzumab conditioning, a sibling donor was used in 29 (66%) and volunteer unrelated donor in 15 (34%) patients. All donor recipient pairs were matched for HLA-A, -B, -C, -DRB1 and -DQB1. Thirty-seven donors were matched at the allelic level for all loci; five patients received cells from a one antigen-mismatched donor and two patients had a >1 antigen-mismatched donor. Unmanipulated peripheral blood progenitor cells (PBSC) were administered to the majority of patients in both the BEAM-auto (93%) and BEAM-allo (86%) groups. The median CD34+ cell dose of PBSC and bone marrow administered to the BEAM-auto group was 3·64 × 106/kg (range 0·8–27·3) and 2·07 × 106/kg (range 0·62–3·24), respectively, and the BEAM-allo group 5·13 × 106/kg (range 2·31–17·28) and 2·89 × 106/kg (range 1·47–3·54) respectively. On comparison of the BEAM-auto and BEAM-allo groups, there was no difference in the median duration from diagnosis to transplantation, in the disease status at the time of transplant, or in the number of patients with prior high grade transformation of disease. Response to chemotherapy was defined according to the National Cancer Institute (NCI) response criteria (Cheson et al, 1999). Table I summarizes the patient and disease characteristics.
Table I. Patient and disease characteristics.
CR, complete remission; PR, partial remission; PBSC, peripheral blood stem cells, BM, bone marrow; N/A, not available.
Number of patients
Median age, years (range)
Time from diagnosis to transplant, months (range)
Number of chemotherapy regimens pretransplant (range)
Disease status at transplant
HLA-matched sibling donor
HLA-matched unrelated donor
HLA-mismatched sibling donor
HLA-mismatched unrelated donor
Stem cell source
Stem cell dose CD34+, 106/kg (range)
High-grade transformation was observed in 11 of the BEAM-auto group, of which two were in complete remission (CR), one had evidence of progressive disease and the remaining eight were in partial remission at the time of transplant. Of the 14 cases of high-grade transformation within the BEAM-allo group, four were in CR, one had evidence of progressive disease and the remaining nine were in partial remission at the time of transplant. Patients in partial remission following chemotherapy were not re-biopsied prior to transplant. We cannot therefore confirm their histological status at the time of transplant and should assume that high grade disease may still be present.
The BEAM protocol consisted of BCNU 300 mg/m2 day-6, cytarabine 200 mg/m2 twice daily day-5 to day-2, etoposide 200 mg/m2 day-5 to day-2 and melphalan 140 mg/m2 day-1. In addition, alemtuzumab (Campath 1-H), either at 10 mg/d or 20 mg/d days-5 to -1, was administered to recipients of allogeneic stem cells. The dose of alemtuzumab administered was dependent on transplant centre. GvHD prophylaxis was achieved with ciclosporin (1·5 mg/kg i.v.) from day-1. Oral ciclosporin was substituted when good oral intake was achieved and tapered from day+56 in the absence of GvHD. Standard nursing care and post-transplant prophylaxis was administered as previously described (Ho et al, 2003). Toxicity was assessed using the common toxicity criteria of the NCI (1999) and GvHD was graded according to the guidelines of the consensus conference on GvHD grading (Przepiorka et al, 1995).
Mobilization of stem cells
Donor choice determined whether peripheral blood stem cells or bone marrow harvest were collected for allogeneic HSCT. Granulocyte colony-stimulating factor (G-CSF; filgrastim) 10 μg/kg/d for 4–5 d was administered prior to PBSC harvest, aiming for a CD34+ cell dose of at least 2 × 106/kg. Cyclophosphamide 1·5 g/m2 or 3 g/m2 i.v. (depending on transplant centre) with G-CSF was used for mobilization of autologous PBSCs. Unselected BM or PBSCs were administered on day 0 of the transplant protocol.
Chimaerism analysis was performed locally following BEAM-allo HSCT at days 28, 56, 100, 6 months, 12 months then yearly or as clinically indicated. A semi-quantitative assessment of donor and recipient chimaerism was performed by polymerase chain reaction and fluorescent analysis of short tandem repeat sequences on unfractionated bone marrow and peripheral blood CD3+ and CD15+ cell fractions. Full donor chimaerism was defined as achievement of >95% donor chimaerism and falling donor chimaerism as >10% decline in donor chimaerism on two consecutive measurements.
Donor leucocyte infusion (DLI)
Incremental doses of DLI were administered after day 100 following BEAM-allo HSCT in the absence of GvHD and complete withdrawal of immunosuppression. Indications for DLI included falling donor chimaerism (defined above) or lymphoma relapse.
Study endpoints and statistical analysis
Patient and transplant characteristics were compared using chi-squared analyses for categorical variables and the Mann–Whitney test for continuous variables. Estimates of non-relapse mortality (NRM) and relapse risk were calculated using cumulative incidence curves, with comparison of curves using Gray’s test. NRM was defined as the date of transplantation to death from any cause other than relapse, with relapse being defined as a competitive risk in the estimate of NRM. NRM was defined as a competitive risk in the estimation of relapse incidence. The probabilities of OS and disease-free survival (DFS) were analysed using Kaplan–Meier estimates. DFS was defined as time to relapse or death because of any cause. Univariate and multivariate analysis used the Cox proportional hazards regression model. Variables analysed included age, sex, history of high-grade transformation, duration from diagnosis to transplant, prior therapy, disease status at time of transplant [CR, partial remission (PR), progressive disease], stem cell dose, stem cell source, donor type (sibling versus unrelated donor) and time to regeneration of both neutrophils and platelets. On multivariate analysis, independent variables with P > 0·1 were excluded from the model. The P-value was set at P < 0·05 for statistical significance. Statistical analyses were performed with the Statistical Package for the Social Sciences (spss) version 14.0 (SPSS Inc., Chicago, IL, USA).
Engraftment and chimaerism
The median time to neutrophil engraftment for the BEAM-auto versus BEAM-allo groups was 12 d (range 9–21 d) vs. 13 d (range 10–31 d), respectively, and to platelet engraftment 12 d (range 7–40 d) vs. 13 d (range 6–38 d) respectively. Two patients in the BEAM-allo group had primary graft failure, defined as failure of neutrophil recovery, with subsequent engraftment in one patient following a second stem cell infusion and death because of adenovirus infection in the second patient. Late graft rejection was observed in two patients, both of whom received a second RIC allo-HSCT. One patient remains alive and in CR at last follow-up, the second died of viral illness.
Donor chimaerism was analysed in 36 evaluable patients. Full donor chimaerism (FDC) (>95% donor) was achieved in 30/36 (83%) patients. On continued follow-up, five of these patients subsequently developed stable mixed donor chimaerism with no evidence of disease recurrence. Of the six patients who never achieved FDC, two patients remain in complete remission with stable mixed donor chimaerism. Four patients sustained a fall in donor chimaerism to <20% CD3+ cells with associated disease relapse.
Donor leucocyte infusion
The median time to first infusion of donor leucocytes was 10 months (range 4–15 months). Cell doses ranging from 5 × 105/kg to 6 × 107/kg were administered. Seven patients in the BEAM-allo group were administered with DLI for falling donor chimaerism; five patients received one dose of DLI and two patients received four incremental doses of DLI to a maximum of 1 × 107/kg. FDC was achieved in three patients and stable mixed chimaerism in a further three patients with no evidence of disease relapse. One patient failed to respond to DLI with subsequent relapse. Six patients received DLI for disease recurrence. All six patients received one dose of DLI, with subsequent complete remission achieved in four of the six patients.
Toxicity and non-relapse mortality
There were three deaths in the BEAM-auto group and 10 deaths in the BEAM-allo group that were identified at latest follow-up and attributed to non-relapse cause. The cumulative incidence of NRM at 1 year was 2% vs. 20% respectively (P = 0·001) (Fig 1A). The cause of death in the BEAM-allo group was identified as bacterial sepsis (n = 2), adenovirus infection (n = 2), cytomegalovirus (CMV) pneumonitis and aspergillosis (n = 1), pseudomonas infection (n = 1), multi-organ failure (n = 2), thrombotic thrombocytopenic purpura (n = 1) and splenic rupture (n = 1). All three patients in the BEAM-auto group died of bacterial sepsis.
Within the BEAM-allo group, 23 patients developed viral complications with four patients developing more than one viral illness. CMV reactivation occurred in 14 patients, only two of which developed CMV disease. Respiratory viruses were detected in five patients: parainfluenza (n = 2) and respiratory syncytial virus (RSV) (n = 3). Other viral complications included adenovirus (n = 3), herpes zoster (n = 4) and BK virus (n = 1).
At the time of censoring of the data, secondary malignancies were identified in five patients within the BEAM-auto group [three myelodysplastic syndrome (MDS), two Hodgkin disease] and one patient in the BEAM-allo group (MDS).
Acute grade I–II GvHD was observed in seven patients who received BEAM-alemtuzumab conditioning. There were no cases of grade III or IV acute GvHD. Of the 36 patients alive at day 100, seven developed chronic GvHD of which three had evidence of chronic extensive GvHD. Two of these patients received unrelated donor stem cells. The cumulative incidence of chronic GvHD of all grades was 20%. A further two patients developed limited chronic GvHD following infusion of donor leucocytes.
The median follow-up of the BEAM-auto and BEAM-allo groups was 7·3 years (range 0·92–13·85 years) and 2·87 years (range 0·55–8·26 years) respectively. Data were censored at 5·5 years follow-up for the purpose of displaying Kaplain–Meier and cumulative incidence curves. A significantly higher relapse rate was observed following BEAM-auto HSCT when compared with BEAM-allo conditioning (P = 0·01). The 3-year cumulative incidence of relapse was 43% vs. 20% respectively (Fig 1B). No significant difference in the 3-year DFS (P = 0·90) or OS (P = 0·99) was observed (Fig 1C and D). The 3-year probability of OS for the BEAM-auto versus BEAM-allo groups was 67% vs. 69%, respectively, and 3-year DFS 56% vs. 58% respectively.
Univariate analysis of factors affecting NRM, OS and DFS for the BEAM-allo group are summarized in Table II. Older age and more lines of chemotherapy pretransplant were the only significant variables which adversely affected NRM, OS and DFS. The use of an unrelated donor or a HLA-mismatched donor did not adversely affect outcome. Furthermore, no significant difference in relapse rate, OS or DFS was observed on comparison of 50 mg vs. 100 mg of alemtuzumab during conditioning. On multivariate analysis, the number of lines of prior chemotherapy was an independent variable adversely affecting OS, whereas age and number of lines of prior chemotherapy adversely affected DFS (Table III). Univariate analysis of the BEAM-auto group failed to identify any significant variables that independently affected NRM, OS or DFS. Age, sex, history of high-grade transformation, time to transplant, number of lines of prior chemotherapy, disease status at the time of transplant, CD34+ cell dose, stem cell source and time to neutrophil and platelet recovery were analysed.
Table II. Univariate analysis of variables affecting NRM, OS and DFS following BEAM-alemtuzumab conditioned HSCT.
Additional variables analysed and found to be non-significant included recipient sex, stem cell dose, alemtuzumab dose and time to platelet and neutrophil recovery.
Bold text denotes P-value which matches statistical significance.
*Denotes factors analysed as continuous variables.
Two patients received a BEAM-allo HSCT at an interval of 8 months and 38 months following a prior autograft using BEAM conditioning. Both patients received stem cells from an unrelated donor. The former patient subsequently relapsed at 14 months after BEAM-allo HSCT, following which DLI was administered and the patient is currently alive and in CR. The latter patient died as a result of multiorgan failure at 30 d postallograft.
The majority of patients with advanced stage follicular lymphoma are currently managed with either combination chemotherapy +/−addition of anti-CD20 monoclonal antibody (rituximab) or high-dose therapy with autologous stem cell rescue. Improved PFS but no survival advantage was reported following the use of high-dose therapy with autologous stem cell rescue when compared with standard chemotherapy alone in previously untreated patients with advanced stage follicular lymphoma (Deconinck et al, 2005). Allogeneic HSCT is considered as a potential curative option for advanced stage lymphoma; however, is only offered to those with relapsed disease as a result of the high regimen-related toxicity. A previous report demonstrates a low incidence of NRM and GvHD with the use of BEAM-alemtuzumab conditioning in a variety of lymphoproliferative disorders (Faulkner et al, 2004). Here, we directly compared the outcome of patients with advanced stage follicular lymphoma who received BEAM conditioning either in an autologous setting or as allogeneic conditioning in combination with alemtuzumab.
We acknowledge that, because of the retrospective nature of this paper, the two groups were not matched for age and prior chemotherapy regimens. Despite a younger median age in the BEAM-allo group, they were more heavily pretreated and were more likely to be resistant to initial salvage chemotherapy than patients who received auto-HSCT. Irrespective of this difference, patients who received BEAM-alemtuzumab conditioning had a higher NRM but significantly lower relapse rate than the BEAM-auto group. Older age and heavily pretreated patients sustained higher regimen-related toxicity, with worse outcome reflected in poorer DFS and OS following BEAM-allo HSCT. The use of volunteer unrelated donors or HLA-mismatched donors did not significantly adversely affect outcome. However, all patients who received HLA-mismatched transplants were heavily pretreated and demonstrated incomplete response to conventional chemotherapy, with, at best, partial remission in six patients, progressive disease in one patient and high-grade transformation in two patients. Details on the International Prognostic Index (IPI) and the follicular lymphoma IPI were unavailable because of the retrospective nature of the study. Current treatment options for patients with relapsed follicular lymphoma are limited, particularly if they have failed mobilization of autologous stem cells. Heavily pretreated patients and older age are both associated with an increased frequency of co-morbidity. Formal assessment of co-morbidity score pretransplant may improve risk stratification in those poor prognostic patients as described recently in the myeloid setting (Sorror et al, 2005; Parimon et al, 2006).
Relapse rates of up to 44% have been reported following the use of T-cell-depleted regimens for lymphoproliferative disorders (Morris et al, 2004). The present study found a relapse rate of 20% at 3 years in the alemtuzumab arm compared with 43% in the autologous group. Indeed the pattern of relapse was quite different between the two groups. A plateau was reached at 2 years in the BEAM-allo group, whereas a continuous pattern of relapse was seen in the autologous group with the 5-year relapse rate reaching 56%. These observations are consistent with a GvLy effect in the BEAM-allo group. No significant difference in the relapse rate was observed within the allo-HSCT group on comparison of 50 mg vs. 100 mg of alemtuzumab. Furthermore, Campath-1H (alemtuzumab) may indeed have a direct anti-lymphoma effect, as low-grade B-cell non-Hodgkin lymphoma (NHL) is known to express CD52 antigen (Salisbury et al, 1994; Rodig et al, 2006). However, in a phase II study of Campath-1H for previously treated low-grade NHL, a partial remission rate of just 14% was achieved in B-cell NHL with a short median time to progression of 4 months (Lundin et al, 1998). These data suggest that the GvLy effect is the most important factor in maintaining remission following BEAM-allo HSCT.
A potential advantage of allogeneic HSCT is the ability to re-induce remissions in the event of relapse or falling donor chimaerism with the use of donor leucocytes, with a low incidence of associated GvHD. Complete remission was achieved in four of six patients who relapsed (three morphological and one molecular) following BEAM-alemtuzumab conditioning, further demonstrating the GvLy effect of allo-HSCT. A state of stable mixed chimaerism with no evidence of relapse was observed in a small proportion of patients (n = 7), suggesting that achievement of full donor chimaerism is not an essential goal. However, rapidly falling donor chimaerism, seen in four patients, was associated with disease relapse and required rapid intervention with DLI.
The comparison of patients who received BEAM-auto with BEAM-alemtuzumab conditioning demonstrated no significant difference in the 3-year PFS or OS. Nonetheless, a plateau was observed following BEAM-allo HSCT with respect to relapse, DFS and OS, with crossing of the survival curves in favour of allo-HSCT despite them being more heavily pretreated. Longer follow-up may enable the identification of a true survival advantage over autologous HSCT. Furthermore, the use of alemtuzumab resulted in a low incidence of GvHD, with no patient in this study developing grade III or IV acute GvHD and only three patients developing chronic extensive GvHD.
A disadvantage of the use of alemtuzumab is the high incidence of viral complications in the post-transplant period as reported by other groups (Chakrabarti et al, 2002a,b,c). Up to 50% of patients in this study developed a viral complication following BEAM-allo conditioning incorporating alemtuzumab. CMV reactivation was the most common finding with a third of patients requiring pre-emptive treatment. Despite this, the mortality from viral illness was low, with only one death as a result of CMV pneumonitis and two resulting from adenovirus infection. Both patients with adenovirus infection also had graft failure at the time of death. Routine monitoring and prompt treatment of patients with anti-viral agents prevent mortality in the majority of cases.
In conclusion, this study demonstrated comparable outcome of patients with relapsed advanced stage follicular lymphoma treated with either BEAM-auto or BEAM-allo HSCT. Longer follow-up will determine whether the reduced relapse risk and plateau in OS observed in the BEAM-alemtuzumab group will translate into a survival benefit over autologous HSCT. Improved patient selection through the use of prognostic scoring systems (Follicular Lymphoma International Prognostic Index) and the use of co-morbidity scores may better predict outcome following HSCT.
Wendy Ingram is supported by the Leukaemia Research Fund.