ALLOGENEIC TRANSPLANTS IN FOLLICULAR LYMPHOMA: HIGHER RISK OF DISEASE PROGRESSION AFTER REDUCED INTENSITY COMPARED TO MYELOABLATIVE CONDITIONING

Abstract

Reduced intensity conditioning (RIC) regimens have been increasingly used for allogeneic hematopoietic stem cell transplantation (HSCT) in Follicular lymphoma (FL). We compared traditional myeloablative conditioning regimens to RIC in FL. Outcomes of HLA-identical sibling HSCT for follicular lymphoma in 208 recipients reported to the Center for International Blood and Marrow Transplant Research between 1997 and 2002 were studied. Conditioning regimens were categorized as myeloablative (N=120) or reduced-intensity (RIC; N=88). Use of RIC regimens increased from <10% of transplants in 1997 to >80% in 2002 signaling a major shift in practice. Patients receiving RIC were older and had a longer interval from diagnosis to transplant. These differences did not correlate with outcomes. Median follow-up of survivors was 50 mo (4–96 mo) after myeloablative conditioning versus 35 mo (4–82 mo) after RIC (p<0.001). At 3 years, overall survival (OS) for the myeloablative and RIC cohorts were 71 (63–79%) and 62 (51–72%; p=0.15) and progression free survival (PFS), 67 (58–75%) and 55 (44–65%; p=0.07) respectively. Lower Karnofsky performance score (KPS) and resistance to chemotherapy were associated with higher treatment related mortality (TRM), lower OS and PFS. On multivariate analysis, an increased risk of lymphoma progression after RIC was observed (RR=2.97, p=0.04) RIC has become the de facto standard in allogeneic HSCT for FL and appears to result in similar long term outcomes. Although disease free survival is similar compared to myeloablative conditioning, an increased risk of late disease progression after RIC is concerning.

INTRODUCTION

Conventional chemotherapy, for follicular lymphoma, although much improved, is not curative(1)’. Allogeneic hematopoietic stem cell transplantation (HSCT) has a low recurrence rate and can sometimes cure follicular lymphoma, but traditional myeloablative conditioning regimens are associated with substantial transplant-related mortality (TRM), especially in older persons with low performance scores and chemotherapy-resistant lymphoma(2, 3).

Reduced-intensity conditioning (RIC) regimens were designed to decrease TRM after allogeneic HSCT while retaining anti-tumor effects of the allograft(4, 5). Most prior reports of RIC in FL are small retrospective series or subsets of larger series that included different types of NHL(6–8). The current study examined 208 HSCTs for follicular lymphoma between 1997 and 2002 and compared outcomes after myeloablative conditioning and RIC.

PATIENTS AND METHODS

Data Sources

The Center for International Blood and Marrow Transplant Research (CIBMTR) is a voluntary group of more than 500 transplant centers worldwide. Participating centers register basic information on all consecutive HSCTs to a Statistical Center at the Medical College of Wisconsin. Detailed demographic and clinical data are collected on a representative sample of registered patients using a weighted randomization scheme. Compliance is monitored by on-site audits. Patients are followed longitudinally, with yearly follow up. Computerized checks for errors, physician reviews of submitted data and on-site audits of participating centers ensure the quality of data.

Patients

The initial study population included all persons with follicular NHL receiving an HSCT from an HLA-identical sibling reported to the CIBMTR between 1997 and 2002. Recipients of unrelated donor transplants and or in vitro T-cell depleted grafts were excluded.

Definitions of RIC and Myeloablative Conditioning Regimens

The degree of myelosuppression after RIC is not entirely predictable and cannot reliably be used as a feature to discriminate RIC from conventional myeloablative conditioning since rare cases of irreversible myelosuppression have been reported after “non myeloablative regimens”’(9) and spontaneous hematopoietic recovery is sometimes observed after administration of “myeloablative” doses of total body irradiation (TBI)(10, 11). RIC regimens also have varying levels of organ toxicity. For example, the commonly used fludarabine – melphalan RIC regimen has considerable extramedullary toxicity and can cause irreversible myelosuppression(12).

In the absence of a robust biological classification, the Regimen Related Toxicity Working Committee of the CIBMTR developed the following working criteria to define RIC: TBI doses <5 Gy as a single fraction, busulfan doses <9 mg/kg (oral / IV equivalent), melphalan doses ≤150 mg/m² and fludarabine-based regimens without myeloablative doses of TBI, busulfan or melphalan. Myeloablative conditioning regimens were defined as: TBI with 1 fraction of ≥5 Gy or fractionated doses of ≥8 Gy, busulfan doses ≥9 mg/kg or melphalan doses ≥150 mg/m². This consensus definition reflects the practice of a large segment of the transplant community and has also been proposed and used by others(4, 13, 14).

Study Endpoints

Outcomes included transplant-related mortality (TRM), progression, progression-free survival (PFS) and overall survival (OS). TRM was defined as death within 28 days post-transplant or death without lymphoma-progression. Progression was defined as progressive lymphoma post transplant (≥28 d) or lymphoma recurrence. It could follow a period of “stable” disease post transplant, or a partial or complete remission. For PFS, subjects were considered treatment-failures at the time of lymphoma-progression or death from any cause. OS was defined as time from the date of transplant to the date of death or last contact. Other outcomes analyzed included acute- and chronic graft-versus-host disease (aGvHD and cGvHD) and cause of death (COD). Acute GvHD was defined and graded using established criteria. Chronic GvHD was defined as the development of any cGvHD based on clinical criteria. Cause of death was reported by the individual teams involved in the subject’s care. Treatment of relapse or progression of lymphoma including the use of donor lymphocyte infusions was not analyzed.

Statistical Analysis

Patient-, disease-, and transplant-related variables for patients receiving RIC and myeloablative transplants were compared using chi-square statistic for categorical variables and the Kruskal-Wallis test for continuous variables. Univariate probabilities of developing aGvHD and cGvHD, TRM and lymphoma-progression were calculated using cumulative incidence curves to accommodate competing risks(15). Probabilities of overall and progression-free survival were calculated using Kaplan-Meier estimator(16). Confidence intervals (CI) were calculated with a log-transformation.

To compare the outcomes of TRM, lymphoma-progression, PFS and survival after RIC and myeloablative conditioning, Cox proportional hazards models were used to adjust for potential imbalances in baseline characteristics between treatment cohorts. A stepwise forward method was used to identify covariates which influenced outcomes. Any covariate with a p-value ≤0.05 was considered significant and included in the final model. The proportionality assumption for Cox-regression was tested by adding a time-dependent covariate for each risk factor and each outcome. A time-dependent covariate with an appropriate cut point was fit in cases where the proportional-hazards assumption was not met(17). Final results were expressed as relative risks (RR) of the event with 95% confidence intervals.

The following variables were considered in model building: conditioning - RIC vs. myeloablative (main effect), age at transplantation, recipient sex, KPS at transplantation, lymphoma stage at diagnosis, lymphoma histology, B-symptoms, number(s) of pretransplant chemotherapy regimens, bone marrow involvement, lymphoma response to chemotherapy at transplantation, disease state and sensitivity to chemotherapy at transplant, CMV-state of recipient and donor, use of pretransplant rituximab, time from diagnosis to HCT and donor-recipient sex match. Type of conditioning regimen (RIC vs. myeloablative) was retained in all steps of model-building since it was the main effect of interest. No interactions were detected between the main effect and all significant risk factors tested. The final multivariate analyses reported are based on outcomes censored at 3 years to balance the difference in length of follow up between the cohorts. Analyses were performed using SAS software, version 8.2 (SAS Institute). The study had 80% power to detect statistically significant difference of 20% in OS between the cohorts.

RESULTS

Patient, Disease- and Transplant-Related Variables

One hundred and twenty subjects received myeloablative conditioning and 88, RIC (Table 1). Completeness of follow-up (the ratio of the sum of the observed follow up time to the sum of the potential follow-up time for all subjects(18)) was 90% for both cohorts.

Table 1.

Preparative Regimens - Conventional Myeloablative versus Reduced Intensity.

Preparative Regimens	N	N (%)	N	N (%)
Conventional Myeloablative	120
CY+TBI (TBI dose > 500 cGy)		80 (67)		---
BU+CY		30 (25)		---
TBI dose ≥ 500 cGy		10 ( 8)		---
Reduced Intensity Conditioning			88
Fludarabine + MEL ± ATG		---		16 (18)
Fludarabine + BU (dose < 9 mg/kg) ±ATG		---		22 (25)
Fludarabine + TBI		---		9 (10)
Fludarabine + CY		---		37 (42)
Fludarabine ± other*		---		3 ( 4)
TBI = 300 cGy		---		1 ( 1)

Abbreviations: CY = Cyclophoshamide; TBI = Total body irradiation; BU = Busulfan; MEL = Melphalan; VP16 = Etoposide; Ara-C = Cytarabine; ATG = Antithymocyte globulin.

^*

Others include (n=3)

1. Fludarabine + Campath (n=1)
2. Fludarabine + Ara-C (n=2)

Table 2 describes subject-, disease-and transplant-related variables of the cohorts. The proportion of RIC transplants was less than 10% in 1997 and steadily increased reaching >80% by 2002. Median follow-up of survivors of myeloablative transplants was 50 mo (range, 4–96 mo) and 35 mo (4–82 mo;(p=0.001) in the RIC cohort. Median age at transplant was 44 y (range, 27–70 y) in the myeloablative cohort compared to 51 y (27–70 y) in the RIC cohort (p>0.001). Pretransplant KPS did not differ significantly between the cohorts.

Table 2.

Patient, disease and transplant related characteristics.

	Myeloablative		RIC
Variable	N	N (%)	N	N (%)	P-value
Number of patients		120		88
Age, median (range), years	120	44 (27 – 70)	88	51 (27 – 70)	<0.001
Age at transplant, years	120		88		<0.001
21–30		6 ( 5)		3 ( 3)
31–40		36 (30)		11 (13)
41–50		60 (50)		30 (34)
51–60		17 (14)		37 (42)
≥ 61		1 ( 1)		7 ( 8)
Male sex	120	68 (57)	88	46 (52)	0.53
Karnofsky score pre-transplant	119		86		0.49
<90		40 (34)		25 (29)
≥90		79 (66)		61 (71)
Disease stage at diagnosis	115		85		0.10
I		2 ( 2)		7 ( 7)
II		11 ( 9)		9 (11)
III		30 (26)		26 (31)
IV		72 (63)		43 (51)
Disease related
Histology	120		88		0.09
Small cleaved		59 (49)		35 (40)
Mixed Small and Large cell		49 (41)		35 (40)
Predominantly Large cell		12 (10)		18 (20)
B- Symptoms at diagnosis	112	30 (27)	75	23 (31)	0.56
Number of prior chemotherapy regimens	118		85		0.28
1		19 (16)		9 (11)
2		38 (32)		21 (25)
3		23 (20)		18 (21)
4+		38 (32)		37 (43)
Disease status at transplant	79		77		0.01
Primary Induction Failure		28 (36)		20 (26)
CR1		6 ( 8)		2 ( 3)
REL1		28 (35)		18 (23)
CR2+		9 (11)		17 (22)
REL2+		8 (10)		20 (26)
Bone marrow involvement	108		79		0.028
Yes		64 (59)		34 (43)
No		44 (41)		45 (57)
Chemosensitivity disease at transplant	118		86		0.74
Sensitive		85 (72)		58 (68)
Resistant		27 (23)		23 (26)
Untreated/Unknown		6 ( 5)		5 ( 6)
Coexisting disease at transplant	118		86		0.15
Yes		39 (33)		37 (43)
Fungal infection at transplant	118		87		0.11
Yes		2 ( 2)		5 ( 6)
Serum creatinine at transplant median (range), mg/dL	118	0.9 (0 – 1.8)	88	0.9 (0 – 1.9)	0.15
Transplant related
Previous autologous transplant	120	7 ( 6)	88	9 (10)	0.24
Time from autologous to allogeneic transplant, median (range), months	6	16 (10 – 50)	9	26 (12 – 60)	0.29
Donor/Recipient CMV status	117		83		0.16
+/+		39 (33)		41 (49)
+/−		17 (15)		9 (11)
−/+		23 (20)		12 (15)
−/−		38 (32)		21 (25)
Time from diagnosis to transplant, median (range), months	120	25 (3 – 198)	88	36 (6 – 195)	0.002
Time from diagnosis to transplant, months	120		88		0.002
Less than 12 mo		22 (18)		9 (10)
12–24 mo		37 (31)		13 (15)
More than 24 mo		61 (51)		66 (75)
Rituximab pre-transplant	120	31 (26)	88	40 (45)	0.003
Donor-recipient gender match	118		88		0.83
M-M		39 (33)		24 (27)
M-F		27 (23)		21 (24)
F-M		28 (24)		22 (25)
F-F		24 (20)		21 (24)
Source of stem cells	120		88		<0.001
BM		44 (37)		8 ( 9)
PBSC		76 (63)		80 (91)
Year of transplant	120		88		<0.001
1997		35 (29)		2 ( 2)
1998		21 (17)		5 ( 6)
1999		32 (27)		8 ( 9)
2000		14 (12)		17 (19)
2001		11 ( 9)		24 (27)
2002		7 ( 6)		32 (36)
GVHD prophylaxis	116		87		<0.001
FK506/CsA + MTX		99 (85)		43 (49)
FK506/CsA without MTX ± Others		17 (15)		44 (51)
Radiation post transplant	119		87		0.39
Yes		1 ( 1)		3 ( 4)
No		2 ( 2)		1 ( 1)
No post transplant therapy given		116 (97)		83 (95)
DLI post transplant	120		88		0.82
Yes		1 ( 1)		1 ( 1)
No		119 (99)		87 (99)
Median hospital stay, days*	95	28 (6 – 100)	71	26 (2 – 100)	0.17
Median follow-up of survivors, months	82	50 (4 – 96)	54	35 (4 – 82)	<0.001

Abbreviations: CR = complete response; REL = relapse; CMV = cytomegalovirus; BM = bone marrow; PBSC = peripheral blood stem cell; GVHD = graft vs host disease; FK506 = tacrolimus; CsA = cyclosporine; MTX = methotrexate; DLI = donor leukocyte infusion.

^*In patients who survive > 100 days.

Lymphoma-stage at diagnosis was similar between the cohorts; >80% were ≥stage-3. Distribution of subjects was similar across the 3 histological-grades of FL. Although the RIC cohort had a higher proportion of subjects at or beyond 2nd complete remission (CR2 and subsequent relapses or remissions; 21% versus 48%: p=0.01), there was a higher proportion of patients with primary induction failure in the myeloablative group (36% versus 26%; p=0.01). Time from diagnosis to transplant was a median of 25 mo (3–198 mo) in the myeloablative versus 36 mo (6–195 mo) in the RIC cohort (p=0.002). There was a higher incidence of bone marrow involvement in the myeloablative cohort (59% versus 43%, p=0.03). Few subjects in either cohort (6% in myeloablative and 10% in RIC; p=0.24) had prior autotransplants. Subjects receiving an RIC transplant were more likely to have received rituximab pretransplant (26% in myeloablative versus 45%, p=0.003).

Conditioning Regimens

Amongst myeloablative regimen recipients, cyclophosphamide (CY) and total body radiation (CY-TBI) and busulfan (BU) and CY (BU-CY) accounted for 92% of conditioning regimens; others are summarized in Table 1. The most common RIC conditioning regimens were fludarabine (Flu) and CY (42%), Flu and BU (25%), Flu and melphalan (18%) and Flu with TBI (10%). Other RIC regimens are summarized in Table 1. GvHD prophylaxis schemes varied between cohorts. Seventy-nine percent of subjects in the myeloablative cohort received a calcineurin inhibitor with methotrexate, this was used in only 48% of RIC recipients (p=0.001).

Outcomes

Outcomes are summarized in Table 4.

Table 4.

Univariate probabilities of transplant outcomes.

		Myeloablative		RIC
Outcome event	N	Prob (95% CI)	N	Prob (95% CI)	P-valuea
30 day mortality	120	8 (4 – 14)	88	5 (1 – 10)	0.26
100 day mortality	120	18 (12 – 26)	88	14 (7 – 22)	0.36
ANC>0.5 × 109/L	117		83
@ 28 days		97 (94 – 99)		100 (0 – 100)	0.07
@ 100 days		98 (95 – 100)		100 (0 – 100)	0.13
Acute GVHD @ 100 days, grades (2–4)	115	36 (28 – 45)	87	44 (34 – 55)	0.26
Acute GVHD @ 100 days, grades (3–4)	115	13 (7 – 20)	87	21 (13 – 30)	0.13
Chronic GVHD	118		86
@ 1 year		44 (34 – 53)		58 (47 – 68)	0.06
@ 3 years		46 (36 – 56)		62 (51 – 72)	0.03
TRM	120		88
@ 1 year		23 (16 – 31)		23 (15 – 32)	0.97
@ 3 years		25 (17 – 33)		28 (19 – 38)	0.60
Progression/Relapse	120		88
@ 1 year		7 (4 – 13)		15 (8 – 23)	0.11
@ 3 years		8 (4 – 14)		17 (10 – 26)	0.06
PFS	120		88
@ 1 year		70 (61 – 78)		62 (52 – 72)	0.26
@ 3 years		67 (58 – 75)		55 (44 – 65)	0.07
Overall survival	120		88
@ 1 year		74 (66 – 81)		69 (59 – 78)	0.44
@ 3 years		71 (63 – 79)		62 (51 – 72)	0.15

Abbreviations: ANC = neutrophil recovery; TRM = treatment-related mortality; PFS = progression-free survival; PROB = probability; CI = confidence interval.

^a30-day mortality and 100-day mortality were tested using the chi-square test. Probabilities of neutrophil recovery, acute GVHD, chronic GVHD, treatment-related mortality and progression/relapse were calculated using the cumulative incidence estimate. Progression-free survival and overall survival was calculated using the Kaplan-Meier product limit estimate.

Graft Versus Host Disease

Cumulative incidences of aGvHD ≥ grade-2 by day 100 were 36% (28–45) in the myeloablative cohort and 44% (34–55) in the RIC cohort (p= 0.26). Incidence of chronic GvHD at 3 years was 46% (36–52) in the myeloablative cohort and 62% (52–72; p= 0.03) in the RIC cohort (Table 4).

TRM

Day 30, day 100, 1 year and 3 year mortality rates were similar between the cohorts (Table 4). Cumulative incidence estimates of TRM at 1 year were 23% in both cohorts. In multivariate analyses, the type of conditioning did not correlate significantly with TRM (RR for RIC =1.10; 95% CI=0.64–1.91, p=0.72; Table 4, Table 5 and Figure 1). A lower performance score, KPS <90% (RR=2.20, 95% CI=1.27, 3.82, p=0.005) and chemotherapy-resistant lymphoma (RR=1.94, 95% CI=1.12, 3.37, p=0.019) were associated with higher TRM. None of the imbalances between the cohorts (age, disease stage, interval from diagnosis to transplant, bone marrow involvement, or prior rituximab) were associated with TRM.

Table 5.

Multivariate analysis of treatment-related mortality, relapse/progression, progression-free survival and overall survival.

Variables:	N	Relative Risk (95% CI)	P-value
Treatment-related mortality
Main effect:
Myeloablative	120	1.00
RIC	88	1.10 (0.64 – 1.91)	0.72
Other significant covariates:
Karnofsky performance status at transplant
≥90%	140	1.00
<90%	65	2.20 (1.27 – 3.82)	0.005
Chemosensitivity disease at transplant
Sensitive	143	1.00
Others	61	1.94 (1.12 – 3.37)	0.019
Relapse/progression
Main effect:
Myeloablative	120	1.00
RIC	88	2.97 (1.03 – 8.55)	0.044
Progression-free survival
Main effect: (Risk of Treatment failure)
Myeloablative	117	1.00
RIC	84	1.38 (0.85 – 2.21)	0.19
Other significant covariates:
Karnofsky performance status at transplant
≥90%	137	1.00
<90%	64	1.89 (1.16 – 3.06)	0.010
Chemosensitivity disease at transplant
Sensitive	141	1.00
Others	60	1.96 (1.21 – 3.18)	0.006
Overall survival:
Main effect: (Risk of Mortality)
Myeloablative	117	1.00
RIC	84	1.24 (0.76 – 2.02)	0.39
Other significant covariates:
Karnofsky performance status at transplant
≥90%	137	1.00
<90%	84	2.04 (1.25 – 3.35)	0.005
Chemosensitivity disease at transplant
Sensitive	141	1.00
Others	60	1.95 (1.19 – 3.20)	0.008

Abbreviations: CI = confidence interval.

Figure 1.

Cumulative Incidence of TRM after Allogeneic transplants for Follicular Lymphoma by Conditioning Regimen – Conventional Myeloablative vs. RIC

Progression

The 3-year univariate probabilities of progression were 8% (4–14%) in the myeloablative versus 17% (10–26%) in RIC cohort. In multivariate analyses, there was a statistically significant higher risk of progression with RIC (RR=2.97, 95% CI= 1.03–8.55, p=0.044; Table 4, Table 5 and Figure 2). No other variables were associated with TRM.

Figure 2.

Cumulative Incidence of Progression/Relapse after Allogeneic transplants for Follicular Lymphoma by Conditioning Regimen – Conventional Myeloablative vs. RIC

Progression Free Survival and Treatment Failure

The 3-year univariate probabilities of PFS were 67% (58–75%) in the myeloablative and 55% (44–65%) in the RIC cohort (p=0.07; Table 4). This difference was not significant in multivariate-analyses (RR of treatment failure in RIC cohort =1.38, 95% CI= 0.85–2.21, p=0.19; Table 5 and Figure 3). A KPS <90 (RR=1.89, 95% CI=1.16–3.06, p=0.01) and chemotherapy-resistant lymphoma (RR=1.96, 95% CI=1.21, 3.18, p=0.006) were strongly associated with lower PFS. None of the variables that were imbalanced between the cohorts were associated with PFS.

Figure 3.

Adjusted Probability of PFS after Allogeneic transplants for Follicular Lymphoma by Conditioning Regimen – Conventional Myeloablative vs. RIC

Survival and Mortality

The 3-year univariate probability of survival was 71% (63–79%) in the myeloablative versus 62% (51–72%) in the RIC cohort (p=0.15; Table 4). Type of conditioning was not significantly correlated with survival in the multivariate model (RR of mortality for RIC=1.24, 95% CI= 0.76–2.02, p=0.39; Table 5 and Figure 4). In multivariate analyses, a KPS <90 (RR=2.04, 95% CI= 1.25–3.35, p=0.005) and chemotherapy-resistant lymphoma (RR=1.95, 95% CI=1.19–3.20, p=0.008) were associated with more deaths. None of the variables that were imbalanced between the cohorts were associated with survival.

Figure 4.

Adjusted Probability of Overall Survival after Allogeneic transplants for Follicular Lymphoma by Conditioning Regimen – Conventional Myeloablative vs.

Statistical models were constructed to adjust for potential time-varying effect of RIC versus myeloablative transplants. There was no difference in OS and PFS within 9 months, and RIC had worse OS and PFS probabilities after 9 months of transplantation. A Cox model(19) stratified on treatment groups, did not show a significant difference in OS or PFS between the two treatment groups at fixed late time points.

Causes of Death

There were 38 deaths in the myeloablative cohort and 34 in the RIC cohort. Seven deaths in the myeloablative cohort and 9 deaths in the RIC cohort were attributed to progressive lymphoma. Other causes of death are summarized in Table 6.

Table 6.

Causes of death for patients who received allogeneic matched sibling donor transplants for Follicular Lymphoma

	Myeloablative		RIC
Causes of death	N eval	N (%)	N eval	N (%)
Number of patients	38		34
Primary disease		7 (18)		9 (26)
GVHD		4 (11)		6 (18)
Pulmonary		2 ( 5)		3 ( 9)
Infection		8 (21)		7 (20)
Organ Failure		11 (29)		4 (12)
New malignancy		0		1 ( 3)
Other (Hemorrhage/Vascular/Unknown)		6 (16)		4 ( 2)

Abbreviation: GVHD = graft vs host disease.

Hospital Stay of Survivors

The median hospital stay of recipients who survived more than 100 days after myeloablative conditioning and RIC were similar at 28 days (6–100) and 26 (2–100) days respectively.

DISCUSSION

The initial studies of allogeneic transplantation in follicular lymphoma and an analysis of 113 patients reported to the International Bone Marrow Transplant Registry (IBMTR) transplanted between 1984 and 1995 showed low rates of disease recurrence but a TRM up to 40%(2)’(20). Poor performance status at transplantation, chemotherapy refractory disease and age >40 were predictors of worse outcome. The RIC approach was developed with the intent of further reducing the TRM associated with allogeneic transplantation while preserving graft versus lymphoma (GVL) effects and is increasingly used as an alternative to myeloablative transplantation. Excellent results were reported after RIC in low grade lymphoma with chemotherapy sensitive disease(6, 8, 21–23). However high TRM as well as increased relapse risk have been suggested in recent studies(24, 25)

Our analysis shows that the practice of allogeneic transplantation in FL has shifted in favor of RIC conditioning which by 2002 represented 80% of allogeneic transplants for FL. There were intrinsic differences in stem cell source and GVHD prophylaxis regimens between those receiving myeloablative and RIC conditioning. For RIC transplants peripheral blood stem cell grafts were almost exclusively used and GVHD prophylaxis commonly did not include methotrexate. These aspects of treatment are typical of the RIC conditioning approach and their effect on outcome cannot be separated from that of the choice of conditioning method. As in other comparative studies(25) of RIC and myeloablative conditioning, these differences do not preclude a comparison of the conditioning modalities.

There were several differences in the patient characteristics. Those undergoing RIC were on average older, but very few patients in either cohort were below 40, the age cutoff under which we previously have demonstrated a superior outcome(3). RIC transplant recipients also had a longer delay between diagnosis and transplant, more often were in second or subsequent remission, and had more frequently received rituxan prior to transplant. Those undergoing myeloablative conditioning had more commonly primary refractory disease and more frequently bone marrow involvement at the time of transplant. None of these features had prognostic significance in multivariate analysis. The groups were well balanced for performance status and chemotherapy sensitivity of lymphoma, the two most consistent prognostic features in our previous analyses and confirmed again in the current study.

We found that RIC and myeloablative transplantation for FL had very similar TRM. These results are consistent with other recently published comparisons of RIC and myeloablative regimens in FL, NHL and Hodgkin lymphoma (24–26) but are in apparent contrast with data from the Seattle group, who found that TRM was reduced in patients receiving the RIC regimen of fludarabine and low dose TBI(23, 27). The risk of acute GVHD was similar across the two groups but there was an increased risk of chronic GVHD at 3 years in the RIC group. This may be related to the differences in GVHD prophylaxis regimens as well as the greater use of peripheral blood stem cells (PBSC) grafts in patients receiving RIC(28, 29).

In multivariate analysis there was an increased risk of lymphoma progression after RIC compared to myeloablative conditioning. No other patient, disease or transplant related factor that we examined was significantly associated with risk of progression. Despite the increased risk of relapse after RIC, there were no significant differences in overall survival or PFS between the two cohorts.

The equivalence of RIC and myeloablative regimens contrasts with recent single institution reports and has several possible explanations. One is that outcomes after myeloablative transplants have improved over the past decade and comparison of RIC with historical myeloablative transplant data is no longer appropriate. An IBMTR analysis of follicular lymphomas undergoing myeloablative transplants between 1990 and 2000 found that TRM had steadily declined in later years with a survival of 75% at 2 years for patients undergoing myeloablative HSCT for FL between 1997 and 1999(3). It is also possible that the decision to use RIC may have involved pretransplant variables unknown to us for which we could therefore not adjust. However, within the limits of a retrospective study, we could not detect any difference in PS, comorbidities or frequency of prior infections between the two groups. Since the overall annual number of transplants for FL was unchanged over the interval we studied, it is likely that the observed shift to RIC transplants represents a change in practice pattern as opposed to increased access to allogeneic transplantation for patients with lower PS. The lack of difference in survival between myeloablative and RIC cohorts could also arise from the diversity of RIC conditioning regimens. Specific RIC conditioning regimens may have different safety and efficacy profiles but individual regimens could not be further analyzed because there were too few subjects in each subgroup.

Our analysis did not have the ability to evaluate costs or non fatal toxicity such as mucositis, but one important parameter, the median duration of hospitalization of survivors was not different between the groups. We were also unable to analyze the quality of life impact of conditioning regimen intensity. Although such additional benefits may exist, the higher risk of late progression in RIC without a demonstrable benefit in decreasing TRM is concerning, particularly since the median follow up in the RIC cohort was shorter. Higher relapse/progression risk after RIC compared to myeloablative regimens has also recently been reported in other analyses of RIC in NHL(25) (40% with low grade NHL), CLL(30) and multiple myeloma(14).

Importantly, these data demonstrate that regardless of conditioning intensity, allogeneic HSCT remains an effective treatment modality in patients with advanced follicular lymphoma with excellent 3 yr disease free survival rates of 55% to 67%. While it is disappointing that RIC did not lead to significant benefits in terms of reduced TRM or improved survival for follicular NHL, this approach has gained rapid acceptance in practice in recent years. The overall therapeutic equivalence, presumed quality of life benefits and tolerability of RIC in older patients with more advanced disease are driving practice patterns. However, in view of the increased risk of progression that we observed, further validation is recommended in a larger dataset with longer follow-up as well as quantification of potential non lethal toxicity and quality of life benefits. Certain RIC regimens and or GVHD prophylaxis schemes may ultimately prove to be more successful than others and should be tested further in hypothesis-driven prospective trials.

Supplementary Material

01

Supplement Figure 1: Cumulative incidence of TRM by the major conditioning regimens

02

Supplement Figure 2: Probability of PFS by the major conditioning regimens

03

Supplement Figure 3: Probability of OS by the major conditioning regimens

Table 3.

Variables tested in Cox proportional hazards regression models.

Main effect variablea

Conditioning regimen: Myeloablative vs RIC

Patient-related variables:

Age at transplant: 21–30 vs 31–40 vs 41–50 vs 51–60 vs ≥61 years

Gender: female vs male

Karnofsky performance status at transplant: ≥90% vs <90%

Disease stage at diagnosis: I/II vs III/IV

Disease-related:

Histology: small cleaved vs others

B-symptoms at diagnosis: yes vs no

Number of prior chemotherapy regimens: ≤2 vs >2

Bone marrow involvement: yes vs no

Chemosensitivity disease at transplant: sensitive vs others

Disease status and chemosensitivity at transplant:

Sensitive CR vs Sensitive Relapse vs PIF-sensitive vs PIF-resistant vs Resistant

Relapse vs Unknown

Treatment-related:

Donor/Recipient CMV status: +/+ vs +/− vs −/+ vs −/− vs not tested/inconclusive

Time from diagnosis to transplant: <12 vs 12–24 vs >24 months

Rituximab pre-transplant: yes vs no

Donor-recipient gender match: M-M vs M-F vs F-M vs F-F

^aIncluded in all models.

APPENDIX

Other authors for the Writing Committee include: J. Douglas Rizzo, James O. Armitage, Asad Bashey, Willem A. Bujan-Boza, John Gibson, Roger H. Herzig, Armand Keating, James R. Mason, Richard T. Maziarz, Philip L. McCarthy, Alan M. Miller, Alvaro Urbano-Ispizua and Peter H. Wiernik. These authors participated in analysis and discussion of results.