FLUDARABINE-MELPHALAN AS PREPARATIVE REGIMEN FOR REDUCED-INTENSITY ALLOGENEIC STEM CELL TRANSPLANTATION IN RELAPSED AND REFRACTORY HODGKIN’S LYMPHOMA: THE UPDATED M.D. ANDERSON CANCER CENTER EXPERIENCE

Abstract

Background and Objective

The role of reduced-intensity conditioning (RIC) allogeneic stem cell transplantation (allo-SCT) in relapsed/refractory (R/R) Hodgkin’s lymphoma (HL) remains poorly defined. We hereby present an update of our single-center experience with fludarabine-melphalan (FM) as preparative regimen.

Design and Methods

Fifty-eight patients with R/R HL underwent RIC and allo-SCT from a matched related donor (MRD; n=25) or a matched unrelated donor (MUD; n=33). Forty-eight (83%) had received a prior autologous SCT. Disease status at transplant was refractory relapse (n=28) or sensitive relapse (n=30).

Results

Cumulative day 100 and 2-year transplant-related mortality (TRM) were 7% and 15%, respectively (day 100 TRM MRD vs. MUD 8% vs. 6%, p=ns; 2-year MRD vs. MUD 13% vs. 16%, p=ns). The cumulative incidence of acute (grade II–IV) GVHD (first 100 days) was 28% (MRD vs. MUD 12% vs. 39%, p=0.04). The cumulative incidence of chronic GVHD at any time was 73% (MRD vs. MUD 57% vs. 85%, p=0.006). Projected 2-year overall (OS) and progression-free (PFS) survival are 64% (49–76) and 32% (20–45), with 2-year disease progression/relapse (PD) at 55% (43–70). There was no statistically significant difference between MRD and MUD transplants in OS, PFS and PD. There was a trend for the response status pretransplant to favorably impact PFS (p=0.07) and PD (p=0.049), but not OS (p=0.4).

Interpretation and Conclusions

FM as preparative regimen for RIC allo-SCT in R/R HL is associated with a significant reduction in TRM, with comparable results in MRD and MUD allografts. Optimizing pretransplant response status may improve patient outcome.

INTRODUCTION

Hodgkin’s lymphoma (HL) remains a chemotherapy-sensitive disease with favorable outcomes following combination chemotherapy and radiation therapy. However, the prognosis for many patients with relapsed and refractory disease remains poor ^1–4. Many of them can be successfully salvaged with high-dose chemotherapy and autologous stem cell transplantation (auto-SCT). However, patients who relapse after auto-SCT, have a particularly poor prognosis, especially chemorefractory patients ^4–6.

For these chemoresistant, multiply relapsed patients, allogeneic stem cell transplantation (allo-SCT) employing conventional myeloablative conditioning has generally had poor results ^7–11, with prohibitive transplant-related mortality (TRM) and high relapse rates. Still, a minority (15–20%) of these extensively pretreated patients with advanced, chemoresistant disease have achieved long-term remissions and presumably cure. Data from some of these studies also suggested the presence of a graft-vs.-Hodgkin’s lymphoma (G-vs.-HL) effect ^10–11. This concept has also been supported by case reports of disease response following donor leukocyte infusions (DLIs) ^12–14.

Reduced-intensity conditioning (RIC) with allogeneic stem cell transplantation has been proposed as a means to achieve engraftment and induce graft-vs.-malignancy effects without the morbidity and mortality associated with myeloablative conditioning regimens ¹⁵. This approach has recently been employed in relapsed and refractory HL ^{14, 16–19}. These patients seem well suited for this approach. Despite their young age, they are heavily pretreated and tolerate conventional myeloablative conditioning poorly. In addition, by reducing TRM, RIC could allow a demonstrable and clinically relevant G-vs.-HL effect to emerge, thereby improving outcome.

While published experience in this developing area remains limited, reduced-intensity regimens have been successful in improving TRM ^{14, 16–19}. Data for long-term progression-free survival are still lacking. Outcome data with matched unrelated donors (MUDs) transplants are even more scarce. This has to be viewed as a high-priority area, as most (75–80%) HL patients would otherwise lack an HLA-compatible donor.

We have previously reported preliminary results of a study comparing two fludarabine-based reduced-intensity conditioning regimens. Fludarabine-melphalan (FM) proved more effective than fludarabine-cyclophosphamide (FC)¹⁶. We therefore focused on this RIC approach with the FM regimen, with expanded patient accrual and a special emphasis on matched unrelated donor transplants. We hereby provide an update of our experience with this strategy, including more complete and mature results, as well as an analysis of prognostic factors.

DESIGN AND METHODS

Patient eligibility

All patients with relapsed or refractory HL who underwent allo-SCT at the University of Texas M. D. Anderson Cancer Center (UT-MDACC) during a five-year period (2001–2005) using with fludarabine-melphalan conditioning were analyzed. Study entry criteria were as follows: histologically confirmed HL, chemosensitive or stable disease after salvage treatment, no active or uncontrolled infection as well as adequate cardiac, pulmonary, renal and hepatic function. Patients were required to have either an HLA-identical related donor or an HLA-matched unrelated donor willing and capable of donating filgrastim-mobilized peripheral blood progenitor cells (PBPCs) or bone marrow. Unrelated donors were matched for HLA-A, -B and –C (serologically matched or, more recently, molecularly identical), as well as HLA-DR/DQ compatible (i.e. one single micromismatch allowed) by high resolution molecular typing.

The study was approved by the UT-MDACC Institutional Review Board (IRB). Patients with disease progression and/or lack of insurance coverage for clinical trial participation who were otherwise eligible and transplanted during the same time period following the same treatment plan were also included. They were analyzed as part of a separate IRB-approved protocol. All patients and donors were required to sign written informed consent. All individual patient data presented in this report have been de-identified.

Conditioning regimen

The conditioning regimen consisted of fludarabine, 25 mg/m² daily for 5 days (day −6 to day −2, until 4/04) or 33 mg/m² daily for 4 days (day −5 to day −2; from 4/04 to 8/05) intravenously (IV), melphalan 70 mg/m² IV daily for two days (day −3 and −2). In six patients the melphalan dose was increased to 90 mg/m² daily x2 as part of a dose escalation trial, stopped because of excessive toxicity (i.e. mucositis). Antithymocyte globulin (ATG, thymoglobulin), 2 mg/kg IV daily for 3 days (day −4 to −2), was introduced to ameliorate graft-vs.-host disease (GVHD) ²⁰ in the most recent matched unrelated transplant patients (n=14). Day 0 was the day of the marrow or PBPC infusion.

GVHD prophylaxis and supportive care

All patients received tacrolimus intravenously beginning two days before transplantation, dosed to maintain therapeutic serum levels (4–12 ng/ml) and switched to oral administration as soon as oral intake was feasible. In the absence of persistent or progressive disease, tacrolimus was continued for a minimum of six months and subsequently tapered off. Methotrexate (5/mg m² IV) was administered on day 1, 3, 6. An additional dose of methotrexate was administered on day 11 for matched unrelated donor transplants. Supportive care was administered as outlined previously ²¹.

Criteria for study evaluation

Engraftment was defined as the first of three consecutive days of an absolute neutrophil count (ANC) ≥ 500/μL. Platelet recovery was considered to have occurred on the first of seven consecutive days of an unsupported platelet count ≥ 20,000/μL. Patients were evaluable for engraftment if they survived at least 30 days following transplant and had a chimerism assay performed. Chimerism was determined at day 30–100 post-transplant on bone marrow or peripheral blood samples by restriction fragment length polymorphism (RFLP) and, more recently, with PCR-based microsatellite polymorphism analysis ^22–23.

Acute and chronic graft-vs.-host disease was graded according to established criteria ^24–25. Patients were considered evaluable for acute GVHD if they had achieved engraftment. Transplant-related mortality (TRM) included all causes of death other than disease progression or relapse occurring at any time after transplant. Relapse-related mortality included all deaths in patients with active disease after transplant. However, in patients reinduced into remission (e.g. with salvage chemotherapy and/or donor leukocyte infusions) after disease progression or relapse, deaths were considered as transplant-related.

Response definitions

A complete remission (CR) was defined as disappearance of all clinical and radiological evidence of active disease in all known sites for a minimum of four consecutive weeks. Complete remission, unconfirmed/uncertain (CRU) was defined as the presence of residual radiographic abnormalities of unclear clinical significance, unchanged or decreased in size during an observation period of least four weeks and non-gallium-avid or negative on PET scan (if initially gallium-avid or positive on a PET scan). CRU patients were reclassified as CRs after one full year without disease progression. Partial remission (PR) was defined as an at least 50% decrease in the sum of the products of diameters of any measurable lesion persisting for at least 4 weeks. Stable disease (SD) was defined as any response not meeting PR criteria or lack of evidence of progressive disease. Progressive disease (PD) was defined as at least a 50% increase in measurable disease or the appearance of new disease sites.

A sensitive relapse was defined as the achievement of at least a partial response to salvage treatment, whereas failure to achieve at least a partial response qualified as refractory relapse. Both disease progression and relapse were considered as PD. Patients with evidence of PD and no active GVHD had their immunosuppression tapered or stopped and were eligible, at the discretion of the investigator, to receive donor leukocyte infusions (DLIs), with or without preceding salvage chemotherapy and/or radiation therapy.

Study design

This prospective study was originally conceived as a pilot study evaluating the feasibility of two preparative regimens (FC and FM). Following an initial data analysis ¹⁶, accrual continued only on the FM regimen. Primary study endpoints included engraftment (i.e., neutrophil and platelet recovery) and chimerism, acute GVHD and day 100 TRM. Additional endpoints included two-year TRM, chronic GVHD, disease progression or relapse, overall survival (OS) and progression-free survival (PFS).

Actuarial rates of OS, PFS and time-to-progression (TTP) were estimated by the method of Kaplan-Meier ²⁶. The cumulative incidence method was used to estimate the rates of acute and chronic GVHD, TRM and PD. Acute GVHD (grade II–IV) was estimated prior to donor leukocyte infusions (i.e. first 100 days), as well as before and after DLIs. The 2-year time point was selected as it coincided with the median patient follow-up. Death attributed to disease was considered a competing risk for TRM, and death in remission or without GVHD was considered a competing risk for disease progression and GVHD, respectively.

The Cox proportional hazards model ²⁷ (HR: hazard ratio; CI: confidence interval) was used to evaluate outcomes (OS, PFS, PD) according to response status pre-transplant and donor type, as well as the impact of the CD3+ cell dose infused on acute GVHD among patients who received DLIs. The effect of acute and chronic GVHD on PD was evaluated considering GVHD as a time-dependent variable. Only univariate analysis was possible given the small sample size. OS and PFS were measured in months from the day of transplantation until death or disease progression.

RESULTS

Patient characteristics

Patient characteristics are summarized in Table 1. Median age at transplant was 32 years (range 19–59). The median number of chemotherapy regimens received prior to allo-SCT was five (2–9). Forty-eight (83%) had undergone a prior autologous stem cell transplant (auto-SCT), and the median TTP after auto-SCT was six months (2–38).

Table 1.

Patient characteristics.

Patient number	58
Age (years)	32 (19–59)
Men/women	41/17
Performance status (ECOG)
0	31
1	25
2	1
Unknown	1
Donor
Matched related	25 (43%)
Matched unrelated	33 (57%)
Conditioning regimen*
Fludarabine-melphalan	44 (MRD n=25; MUD n=19)
Fludarabine-melphalan-thymoglobulin	14 (all MUDs)
Patients who had a previous autograft	48 (83%)
No prior autograft	10 (17%)
Time to progression after autograft (months)	6 (2–38)
Response status at transplant	Chemosensitive: n= 30 (52%) CR/CRU: n=14 PR: n=16 Chemorefractory: n= 28 (48%)

ECOG: Eastern Cooperative Oncology Group. For performance status, see reference no. 28

MRD: matched related donor. MUD: matched unrelated donor. CR: complete remission; CRU: complete remission, unconfirmed/uncertain. PR: partial response. See text for response definitions.

^*The first cohort of MUD transplant patients (n=19) did not receive ATG (see text for details). In six patients the melphalan dose was 90 mg/m² daily x2 as part of a dose escalation trial, stopped because of excessive toxicity (i.e. mucositis).

Response status at transplant was almost evenly distributed between sensitive relapse (n=30; MRD n=13, MUD n=17) and refractory relapse (n=28; MRD n=12, MUD n=16), with no significant difference in chemosensitivity between MRD and MUD allografts. Ten patients (17%) underwent upfront allo-SCT without a prior auto-SCT because of refractory disease.

Stem cell source

PBPCs were employed in 28 patients (matched sibling transplants n=24; matched unrelated donor transplants n=4) and bone marrow in 30 patients (all but one unrelated donor transplants). The median CD34+ cell dose infused was 4.7 × 10⁶/kg (0.9–29.1).

Engraftment and chimerism

The median time to neutrophil recovery was 12 days (range 10–24). The median time to platelet recovery was 17 days (range 7–132). Chimerism data at day 30–100 indicated 100% donor-derived engraftment in 57/57 (100%) evaluable patients (one patient expired prior to his chimerism assessment).

GVHD

The actuarial incidence of grade II–IV acute GVHD at day 100 without DLIs was 28% (18–42). The incidence for MRD and MUD transplants was 12% (4–36) and 39% (26–60) respectively (HR 0.3, 95% CI 0.1–0.9; p=0.04). The cumulative incidence of chronic GVHD was 73% (62–87). The incidence for MRD and MUD allo-SCTs was 57% (40–82) and 85% (73–99) respectively (HR 0.4, 95% CI 0.2–0.7; p=0.006). Thymoglobulin administration did not significantly affect the incidence of grade II–IV acute GVHD or chronic GVHD (data not shown).

Patient outcome: transplant-related mortality and causes of death

Early (day 100) and 2-year TRM for the whole group were 7% (2–12) and 15% (8–28), respectively. There was no statistically significant difference between MRD and MUD transplants with regard to day 100 (8% vs. 6%; HR 1.3; 95% CI 0.2–9.4; p=0.8) or 2-year (13% vs. 16%; HR 0.7; 95% CI 0.2–3.1; p=0.7) TRM.

Twenty-two patients expired (38%). The causes of death included PD (n=14), GVHD (n=3), thrombotic thrombocytopenic purpura (n=1), cardiac failure (n=1), pneumonia and/or sepsis (n=3). Of the non-relapse related deaths, four occurred before day 100.

Patient outcome: overall survival, progression-free survival, and disease progression

At the latest follow-up, thirty-six patients are alive (62%), with a median survivor follow-up of 24 months (range 4–78). Twenty-three of these patients are in CR or CRU. OS, PFS (actuarial estimates) and PD (cumulative incidence) at 24 months and at the last follow-up are 64% (95% CI 49–76), 32% (95% CI 20–45) and 55% (95% CI 43–70), and 48% (95% CI 30–64), 26% (95% CI 12–42) and 61% (95% CI 47–80), respectively (Figure 1). The median time to disease progression after allo-SCT was 141 days (29–1047).

Figure 1.

Kaplan-Meier estimates for overall survival (OS) and progression-free survival (PFS) (left) and cumulative incidence of disease progression (right) for the whole group.

Among the forty-eight patients who had failed a prior auto-SCT, thirty-one had achieved a CR/CRU following their auto-SCT. Among these 31 patients, sixteen (51%) achieved a CR/CRU following their allo-SCT. In these 16 patients, the median TTP after auto-SCT was 6 months (2–37). With a median follow-up after achieving CR/CRU following allo-SCT of 16 months (5–52), the median TTP after allo-SCT in the same sixteen patients has not been reached yet, with five patients progressing after a median of 4 months (2–13). The difference in remission duration was statistically significant (p=0.003, log-rank).

Of the ten patients who received an upfront allo-SCT (i.e. no prior auto-SCT), six are alive (five in CR) and four have expired (progressive disease n=3; GVHD after DLI n=1). Among the six patients who received 90 mg/m² daily x2, five patients are alive (three in CR), while one is dead (PD).

Patient outcome according to donor type and response status pretransplant

Donor type and response status prior to allo-SCT were evaluated as prognostic factors (Table 2). When OS and PFS were stratified according to the donor type (MRD vs. MUD), there was no statistically significant difference between MRD and MUD transplants with regard to OS (HR 0.45; 95% CI 0.2–1.3; p=0.1), PFS (HR 0.97; 95% CI 0.5–1.8; p=0.9) (Figure 2) and PD (HR 1.1; 95% CI 0.5–2.3; p=0.8).

Table 2.

Prognostic factors for outcome at two years.

	N	OS (HR, 95% CI, p value)	PFS	PD
Response status at transplant
CR/CRU	14	Reference	Reference	reference
PR	16	2 (0.5–7.8), p= 0.3	2.4 (0.9–6.5), p 0.07	2.7 (0.85–9.0), p= 0.09
Refractory	28	1.5 (0.4–5.7), p= 0.5	2.2 (0.9–5.5), p 0.09	2.9 (0.9–8.8), p= 0.05
CR/CRU vs. all others	14 vs. 44	0.6 (0.2–2.0), p = 0.4	0.4 (0.2–1.1), p 0.07	0.34 (0.1–0.99), p= 0.049
Donor type
MRD	33	0.45 (0.2–1.3), p= 0.1	0.97 (0.5–1.8), p 0.9	1.1 (0.5–2.3), p= 0.8
MUD	25	Reference	Reference	reference
Acute GVHD (grade II–IV)				0.7 (0.3–1.6), p=0.3
Chronic GVHD				2.0 (0.7–5.9), p=0.2

OS: overall survival. PFS: progression-free survival. PD: progressive disease. CR: complete remission; CRU: complete remission, unconfirmed/uncertain. PR: partial response. See text for response definitions. MRD: matched related donor. MUD: matched unrelated donor.

Figure 2.

Kaplan-Meier estimates for progression-free survival (PFS) according to donor type.

OS and PFS were also stratified according to the response status prior to allo-SCT (i.e. CR/CRU vs. all others, including PR, SD and PD). There was a borderline significant trend favoring complete responders with regard to PFS (HR 0.4; 95% CI 0.2–1.1; p=0.07) (Figure 3) and PD (HR 0.34; 95% CI 0.1–0.99; p=0.049), although not OS (HR 0.6; 95% CI 0.2–2.0; p=0.4). When PR patients were analyzed separately, their outcome (i.e. OS, PFS, PD) was not different from chemorefractory patients (data not shown).

Figure 3.

Kaplan-Meier estimates for progression-free survival (PFS) according to response status pretransplant. CR: complete remission; CRU: complete remission, unconfirmed/uncertain (see text for response definitions).

Impact of acute and chronic GVHD on disease progression

The impact of GVHD on PD was evaluated considering GVHD as a time-dependent variable (Table 2). There was no statistically significant impact of acute GVHD (grade II–IV, 100 days) or chronic GVHD (at any time) on PD (HR 0.7; 95% CI 0.3–1.6; p=0.3 and HR 2.0; 95% CI 0.7–5.9; p=0.2, respectively).

Donor leukocyte infusions (DLIs)

Fourteen patients with disease progression received a total of twenty-five DLIs (median 1, range 1–5), with (n= 11) or without (n=3) preceding salvage chemotherapy and/or radiation therapy (Table 3). The overall response rate (PR plus CR/CRU) was 6/14 (43%). The response rate among patients receiving DLIs alone was 1/3 (33%), with two patients achieving SD. Overall cumulative incidence of grade II–IV acute GVHD including before and after DLIs was 32% (22–47).

Table 3.

Donor leukocyte infusions: CD3+ cell dose and patient outcome

UPN	Donor type	Days after allo-SCT and DLI CD3+ cell dose (x 107 CD3+ cells/kg)					Chemotherapy or Radiation Therapy	Acute GvHD Grade II –IV	Best Response	Current Status	Days Post BMT (Death/Follow-Up)
		0.1	0.5	1	5	10
030069	MRD				360	408	Chemotherapy	No	PD	Expired	527
030030	MRD			486			Radiation Therapy	No	CRU	PD	1168
030009	MRD					111	Chemotherapy	No	CR	PD	1147
020161	MRD			600		765, 829	Chemotherapy	Yes	PR	PR	1093
030224	MRD			139	192, 247	304, 354	None	No	PR	SD	716
040362	MRD			124		158	Chemotherapy	Yes	PD	Expired	385
000765	MRD				196	224	Both	Yes	CRU	Expired	355
000741	MUD		215		307		None	Yes	SD	Expired	1005
020181	MUD	739					Chemotherapy	No	PD	Expired	871
040047	MUD			146			None	Yes	SD	Expired	525
030140	MUD	687					Chemotherapy	No	PD	PD	1102
030389	MUD			116	208		Chemotherapy	No	PR	Expired	478
040094	MUD		261				Chemotherapy	No	PD	Expired	732
000674	MUD					157	Chemotherapy	Yes	SD	Expired	267

UPN: unique patient number. MRD: matched related donor. MUD: matched unrelated donor.

PD: progressive disease. CR: complete remission; CRU: complete remission, unconfirmed/uncertain. PR: partial response. SD: stable disease. See text for response definitions.

UPN 020161 received his third DLI (day +829) as previously collected mobilized peripheral blood progenitor cells following melphalan 140 mg/m².

There was a significant correlation between CD3+ cell dose infused and development of GVHD. Grade II–IV acute GVHD developed in 1/19 cases vs. 4/6 cases when the infused CD3+ cell dose was less or more than 0.8 × 10⁷/kg, respectively (HR 16.7; 95% CI 1.8–150.5; p= 0.012). The frequent concomitant administration of salvage chemotherapy (or radiation therapy) and the small sample size precluded any meaningful analysis of prognostic factors for response, such as CD3+ cell dose infused or development of GVHD.

DISCUSSION

RIC allo-SCT is now widely employed for treatment of many hematologic malignancies ¹⁵. Conventional myeloablative allo-SCT has been largely disappointing for HL because of a high rate of toxicity and treatment-related mortality ^7–10. In principle, HL is an attractive malignancy for a RIC approach, as it is a chemotherapy-sensitive malignancy where heavily pretreated patients have a limited ability to tolerate myeloablative therapy. Indeed, preliminary data from this and other studies of reduced-intensity conditioning indicate a substantial reduction in transplant-related mortality ^{14; 16–19}. A retrospective registry analysis of this approach also showed an improvement in survival ²⁹. Still, as with other hematologic malignancies, the comparative roles of the myeloablative and RIC approaches in HL remain, at present, poorly defined.

The sample size of fifty-eight patients is adequate to draw preliminary conclusions. Early TRM was indeed substantially reduced in our patients compared to other recent studies. Patient outcomes reported here are largely consistent with the ones reported recently by Peggs et al ¹⁴, although our approach is substantially different, as it does not include alemtuzumab in the preparative regimen.

Limited data on the use of matched unrelated donor transplants for HL are available in the literature. In this study, unrelated transplants experienced more acute and chronic GVHD, but TRM, OS and PFS were largely comparable to MRD transplants. These findings, along with comparable results reported by others ¹⁴, indicate that matched unrelated transplants are an appropriate option for patients without a matched sibling donor.

While the difference did not quite reach statistical significance, this study supports the prognostic role of the pretransplant response status, particularly a complete response, with regard to patient outcome ^14,17. It seems reasonable to attempt cytoreduction by salvage chemotherapy as effectively as possible prior to allo-SCT. More effective pretransplant salvage strategies and regimens would be desirable, including new effective agents such as gemcitabine ^30–34. In addition, early immunosuppression withdrawal or prophylactic DLIs in high-risk patients should be considered to augment the graft-vs.-malignancy effect.

Whether a graft-vs.-HL effect exists has been the subject of considerable interest and debate ³⁵. This report does not allow any final conclusions because of its small sample size and its limited statistical power to detect such an effect. Acute and chronic GVHD had no measurable impact on PD. However, rapid disease progression post-transplant in many patients (median time to progression was only about 5 months) could conceivably prevent the mounting of an effective graft-vs.-HL reaction, emphasizing once again the key issue of PD prevention. A finding supporting the presence of a graft-vs.-HL effect was a significantly longer median time to progression following allo-SCT in the cohort of complete responders who had previously failed their auto-SCT.

DLI responses are often viewed as the gold standard to establish a graft-vs.-HL effect. There are published and fairly convincing data on this topic, albeit scarce ^{12–14, 35, 36}. In this report, most patients received DLI after salvage chemotherapy, making it difficult to interpret the response data. Whether more frequent or prophylactic DLIs or possibly a higher CD3+ cell dose would be more effective, remains to be determined. The correlation between CD3+ cell dose infused and the development of acute GVHD is noteworthy, although not unexpected ³⁷. Acute GVHD was uncommon with CD3+ cell doses lower than 1 × 10⁷/kg.

In conclusion, RIC allo-SCT using fludarabine and melphalan as the preparative regimen allows a significant reduction in TRM, with comparable results in inducing long-term progression-free survival with matched related and unrelated donor transplants. While these results are encouraging, much work remains to be done. Longer follow-up data are needed to put these results in perspective. Future studies should focus on inducing greater cytoreduction with salvage therapy prior to transplant, as this is likely to improve patient outcome. Prevention of early disease progression and more effective management strategies for GVHD should be high-priority areas as well.