Outcomes of Related Donor HLA-Identical or HLA-Haploidentical Allogeneic Blood or Marrow Transplantation for Peripheral T Cell Lymphoma

Abstract

The role of allogeneic blood or marrow transplantation (alloBMT) for peripheral T cell lymphoma (PTCL) remains to be defined. There is growing interest in reduced-intensity conditioning (RIC) regimens and/or utilization of human leukocyte antigen haploidentical (haplo) grafts given concerns about treatment-associated toxicities and donor availability. We reviewed the outcomes of 44 consecutive, related donor alloBMTs for PTCL performed at Johns Hopkins Hospital from 1994 to 2011, including 18 RIC/haplo alloBMTs. Patients receiving RIC (n = 24) were older, with median age of 59 years (range, 24 to 70), than patients receiving myeloablative conditioning (MAC, n = 20), with median age of 46 years (range, 18 to 64), P =.01. The median age at RIC/haplo alloBMT was 60 years. The estimated 2-year progression-free survival (PFS) was 40% (95% confidence interval [CI], 26% to 55%) and overall survival (OS) was 43% (95% CI, 28% to 59%). In older patients (≥60, n = 14), the estimated 2-year PFS and OS were 38% (95% CI, 18% to 79%) and 45% (95% CI, 24% to 86%), respectively. On unadjusted analysis, there was a tendency toward superior outcomes for alloBMT in first remission versus beyond first remission, with an estimated 2-year PFS of 53% (95% CI, 33% to 77%) versus 29% (95% CI, 9% to 45%), P = .08. On competing risk analysis, the 1-year cumulative incidence of relapse was 38% for MAC/HLA-identical alloBMTs and 34% for RIC/haplo alloBMTs. Estimated 1-year nonrelapse mortality was 10% for MAC and 8% for RIC (11% for RIC/haplo alloBMT). On unadjusted landmark analysis, patients with acute grade II-IV or chronic graft-versus-host disease (GVHD) had a 17% probability of relapse (95% CI, 0% to 39%), compared with 66% (95% CI, 48% to 84%) in patients without GVHD, P = .04. Utilization of RIC and alternative donors expands treatment options in PTCL to those who are older and unable to tolerate high-dose conditioning, with outcomes comparable with approaches using myeloablative regimens and HLA-matched donors. AlloBMT may be appropriate in first remission in select high-risk cases.

INTRODUCTION

Peripheral T cell lymphomas (PTCLs) have variable responses to chemotherapy and often relapse, even after high-dose therapy with autologous stem cell rescue [1–4]. Thus, there is growing interest in allogeneic blood or marrow transplantation (alloBMT) as an alternative given its potential for a graft-versus-lymphoma (GVL) effect [5–9]. Older patient age, difficulty identifying human leukocyte antigen (HLA)-identical donors, and concerns about transplant-related toxicities have limited the exploration of this approach. Yet, recent advances in alloBMT platforms, including high-dose post-transplantation cyclophosphamide for graft-versus-host disease (GVHD) prophylaxis, have significantly reduced the morbidity of alloBMT, including HLA-haploidentical (haplo) alloBMT [10,11]. Herein, we report the Johns Hopkins Hospital outcomes of related donor alloBMT for PTCL, half of which used haplo grafts.

METHODS

Patients

After institutional review board approval, the Johns Hopkins BMT registry was screened for diagnoses of PTCL or natural killer (NK) cell lymphoma. Forty-four consecutive alloBMTs for PTCL in adults were identified from January 1994 to June 2011 and selected for analysis. Diagnostic pathology was centrally reviewed before alloBMT. For the purposes of this article, histologic diagnoses were designated using the 2008 World Health Organization classification of mature T cell and NK cell neoplasms [12]. International Prognostic Index or Prognostic Index for T cell lymphoma scores could not be tabulated for many of the patients due to missing data from the time of PTCL diagnosis and are thus not available for this alloBMT cohort.

The presence of high-risk disease features as well as a strong institutional research focus on alternative donor transplantation factored into decisions regarding autologous BMT versus alloBMT, with 80% of patients in this cohort enrolled in an alloBMT clinical trial. The decision to take a patient to alloBMT in first remission was dependent on the treating physician and was based in part on the patient’s candidacy for myeloablative conditioning (MAC), with in reduced-intensity conditioning (RIC)/haplo alloBMT prioritized over autologous BMT for those unable to tolerate MAC.

Transplantation

MAC regimens included Bu/Cy (busulfan pharmacokinetically dosed in 16 doses over 4 days to achieve an area under the curve of 800 to 1400 mmol × min/L and cyclophosphamide 50 mg/kg i.v. daily for 4 days), Cy/TBI (cyclophosphamide 50 mg/kg i.v. daily for 4 days and 1200 cGy total body irradiation given as 300 cGy per day for 4 days), or Bu/Flu (busulfan pharmacokinetically dosed as above and fludarabine 40 mg/mm i.v. daily for 4 days). RIC regimens were fludarabine-based, either Flu/Cy/TBI (fludarabine 30 mg/mm i.v. daily for 5 days, cyclophosphamide 14.5 mg/kg i.v. daily for 2 days, and 200 cGy TBI) or Flu/TBI. Supportive care measures were according to Johns Hopkins Hospital institutional standards.

Through 2006, cyclosporine was used for GVHD prophylaxis for all MAC alloBMTs. After 2006, PT/Cy (post-transplantation cyclophosphamide 50 mg/kg i.v. on days 3 and 4 after alloBMT) was used as sole GVHD prophylaxis for MAC/HLA-identical alloBMTs. For RIC/HLA-identical alloBMTs, the GVHD prophylaxis regimen was PT/Cy (days 3 and 4) with mycophenolate mofetil 15 mg/kg orally three times daily up to 1000 mg/dose given days 4 to 33 after transplantation. For RIC/haplo alloBMTs, the GVHD prophylaxis regimen was PT/Cy (days 3 and 4), tacrolimus, and mycophenolate mofetil (days 5 to 35). In the absence of GVHD, tacrolimus was continued with goal trough of 5 to 15 ng/mL until day 180 after transplantation.

Definitions of Disease Status and Clinical Outcomes

Primary induction failure (PIF) was defined as progressive disease during first-line treatment or disease relapse within 2 months of treatment completion. Chemoresistant disease was designated in patients with primary induction failure or relapse refractory to salvage chemotherapy. Disease status at the time of alloBMT was determined in accord with standard response criteria for non-Hodgkin lymphoma [13].

Neutrophil recovery time was the number of days from alloBMT to the first of 3 consecutive days with an absolute neutrophil count above 500/μL. Platelet recovery time was the number of days from alloBMT to the first day with a platelet count of ≥20,000 without platelet transfusion in the preceding week. Donor chimerism was determined by restriction fragment length polymorphisms or polymerase chain reaction of variable nucleotide tandem repeats. Full donor chimerism was defined as achievement of ≥95% donor T cells in the blood at post-transplant day 30 or in the blood or bone marrow at post-transplant day 60. Mixed chimerism was defined as 5% to 94% donor cells. Primary graft failure was defined as <5% donor cells by posttransplant day 60. Day 60 engraftment was determined by donor chimerism, if available, or count recovery criteria. Acute GVHD was graded per the Keystone criteria, and chronic GVHD was graded per the 2005 National Institutes of Health Working Group Report and Seattle standard guidelines [14–16].

Progression-free survival (PFS) was the time from alloBMT to disease relapse, progression, or death from any cause. Overall survival (OS) was the time from alloBMT to death from any cause. Nonrelapse mortality (NRM) was defined as death without disease relapse.

Statistics

The dataset was locked for analysis on December 20, 2011. Descriptive statistics were used to summarize baseline patient and alloBMT characteristics and compared using the unpaired t test. The probabilities of PFS and OS were estimated using the Kaplan-Meier method with 95% confidence intervals (CIs) and compared using the two-tailed Gehan-Wilcoxon test [17]. Cumulative incidences of relapse, NRM, and GVHD were estimated by competing-risk analysis using Gray’s method [18]. Relapse and NRM were competing risks for each other. Relapse, death, and graft failure were competing risks for GVHD. An unadjusted landmark analysis setting day 18 (the median time to engraftment) as the landmark time was used to estimate the cumulative incidence of relapse with respect to GVHD (acute grades II-IV or chronic) as a time-dependent variable, treating NRM as a competing risk for relapse [19]. Data were analyzed with the R program, version 2.12 (R Core Development Team, Vienna, Austria).

RESULTS

Table 1 presents patient characteristics. This was a poor-risk cohort, with 60% receiving at least two prior chemotherapy regimens, 50% having history of chemorefractory disease, and 25% having active disease at alloBMT. Table 2 presents transplantation characteristics. Donors were haplo (n = 22) or HLA-identical (n = 22) first-degree relatives. Grafts in all but one case were marrow derived. Over two thirds of alloBMTs (68%) incorporated high-dose PT/Cy for GVHD prophylaxis, including all RIC alloBMTs (n = 24). The median follow-up for surviving patients was 3.9 years (range, .5 to 13.7 years). At last follow-up, 16 of 44 patients were alive, 15 without relapse. Overall, 32 of 36 assessable patients (92%) engrafted by day 60; of the 4 that did not engraft, 3 had autologous hematopoietic recovery and 1 received a second marrow infusion with subsequent engraftment.

Table 1.

Patient Characteristics

Characteristic	Result
Age at alloBMT, yr, median (range)	51 (18–70)
MAC	46 (18–64)
RIC	59 (24–70)
Male sex, n (%)	24 (55)
Histology, n (%)
Nodal	23 (52)
PTCL, not otherwise specified	7
Angioimmunoblastic T cell lymphoma	6
Anaplastic large cell lymphoma, ALK negative	5
Anaplastic large cell lymphoma, ALK positive	2
Anaplastic large cell lymphoma, ALK unknown	3
Extranodal	14 (32)
Enteropathy-associated T cell lymphoma	6
Hepatosplenic gamma-delta T cell lymphoma	6
Extranodal NK/T cell lymphoma, nasal type	1
Subcutaneous panniculitis-like gamma-delta T-cell	1
Other	7 (16)
Adult T cell leukemia/lymphoma	3
Mycosis fungoides	2
Cutaneous NK cell lymphoma	1
Blastic NK cell lymphoma	1
Remission status at alloBMT, n (%)
First partial or complete remission, without PIF	14 (32)
Second or greater remission, without PIF	8 (18)
Remission with history of PIF	11 (25)
Active disease at alloBMT	11 (25)
Prior chemotherapy resistance, n (%)	22 (50)
Number of prior therapies, n (%)
1	18 (40)
2	13 (30)
3 or more	13 (30)
Prior autologous BMT, n (%)	4 (9)
Years from diagnosis to alloBMT, median (range)	.7 (.2–11)
Follow-up of survivors, yr, median (range)	3.9 (.5–13.7)
MAC	4.8 (.8–13.7)
RIC	1.9 (.5–5.1)

PIF indicates primary induction failure; ALK, anaplastic lymphoma kinase.

Table 2.

Transplantation Characteristics

Characteristic	Result
Number of transplants	44
MAC	20
RIC	24
Transplant type, n (%)
RIC/HLA haploidentical	18 (41)
RIC/HLA identical	6 (14)
MAC/HLA identical	16 (36)
MAC/HLA haploidentical	4 (9)
HLA-haploidentical regimens, n (% of 22)
Flu/Cy/TBI with PT/Cy, tacrolimus, and MMF	18 (82)
Bu/Cy with PT/Cy, tacrolimus, and MMF	3 (14)
Cy/TBI with cyclosporine	1 (4)
HLA-matched regimens, n (% of 22)
Bu/Cy with cyclosporine	8 (36)
Cy/TBI with cyclosporine	5 (23)
Flu/TBI with PT/Cy and MMF	5 (23)
Bu/CY with PT/Cy alone	2 (9)
Bu/Flu with PT/Cy alone	1 (4)
Flu/Cy/TBI with PT/Cy, tacrolimus, and MMF	1 (4)
Donor relationship, n (%)
Sibling	35 (80)
Child	7 (16)
Parent	2 (4)
Unrelated	0
Donor age, yr, median (range)	47 (13–74)
Female donor and male recipient	10 (23)
Stem cell source, n (%)
Bone marrow, T cell replete	31 (70)
Bone marrow, T cell depleted	12 (27)
Peripheral blood, unprocessed	1 (2)
Graft dose, median (25th, 75th percentile)
Total nucleated cells × 108/kg	3.25 (.6, 4.2)
CD34+ cells × 106/kg	3.32 (2.5, 4.8)
CD3+ cells × 107/kg	3.30 (.04, 3.9)
Days to count recovery, median (range)
Neutrophils (≥500/mL)
MAC (n = 19)	18 (11–25)
RIC (n = 20)	18 (13–28)
Platelets (≥20,000/mL)
MAC (n = 16)	25 (18–43)
RIC (n = 20)	24 (18–65)
Donor chimerism after RIC, n (%)
Day 30 (n = 24)
Full (≥95% donor)	16 (67)
Mixed (5%–94% donor)	6 (25)
<5% donor	2 (8)
Day 60 (n = 20)
Full	14 (70)
Mixed	4 (20)
<5% donor	2 (10)

MMF indicates mycophenolate mofetil.

Donor lymphocyte infusion (DLI) was used sparingly in this cohort and was not given for mixed chimerism alone. Decisions to pursue DLI in the setting of relapse were based on the timing of relapse, GVHD history, and the use of haplo versus HLA-matched grafts. Three patients did receive DLI. In two cases, DLI was pre-emptively given within the 6 months after myeloablative, HLA-matched alloBMT in patients without GVHD and with very high-risk disease (hepatosplenic gamma-delta T cell lymphoma and blastic NK lymphoma). Nonetheless, both patients died of relapsed disease shortly after DLI. In the third case, DLI along with involved field radiation therapy was given to a patient with disease relapse 4 years after RIC/haplo alloBMT, resulting in ongoing disease-free survival at the time of data analysis.

The estimated 2-year PFS for all patients was 40% (95% CI, 26% to 55%) and OS was 43% (95% CI, 28% to 59%) (Figure 1A). On unadjusted analysis, there was a tendency toward superior PFS for alloBMT in first remission versus beyond first remission, with an estimated 2-year PFS of 53% (95% CI, 33% to 77%) versus 29% (95% CI, 9% to 45%), P =.08 (Figure 1B). Of the 15 patients who remain alive in sustained remission, 10 received alloBMT in first remission; represented among these 10 are especially poor-risk histologies, including adult T cell leukemia/lymphomas, hepatosplenic gamma-delta T cell lymphoma, and subcutaneous panniculitis-like gamma-delta T cell lymphoma.

Figure 1.

Kaplan-Meier survival estimates. (A) PFS (dashed line) and OS (solid line) for all patients (n = 44). (B) PFS, stratified by remission. Patients receiving alloBMT in first remission (n = 21, solid line) compared with patients receiving alloBMT beyond first remission (n = 23, dashed line), P =.08. (C) PFS (dashed line) and OS (solid line) for patients receiving MAC (n = 20). (D) PFS (dashed line) and OS (solid line) for patients receiving RIC (n = 24).

To investigate other potential prognostic variables with respect to survival outcomes, Kaplan-Meier curves were constructed to compare outcomes for patients with nodal histologies versus other histologies (including extranodal, leukemic, and cutaneous), patients aged ≥60 versus aged <60, and patients in remission at alloBMT versus those with stable or progressive disease at transplant. No differences in PFS or OS estimates were found between groups for any of these three variables. For exploratory purposes, an unadjusted landmark analysis was performed to investigate the relationship between GVHD and relapse. Overall, patients with GVHD (acute grades II-IV or chronic) had a lower probability of relapse of 17% (95% CI, 0% to 39%) than patients without GVHD of 66% (95% CI, 48% to 84%), P = .04.

Patients receiving RIC were older, with a median age of 59 years (range, 24 to 70), than patients receiving MAC, with median age of 46 years (range, 18 to 64), P =.01. The median age at RIC/haplo alloBMT was 60 years. In patients aged 60 or older (n = 14), the estimated 2-year PFS and OS were 38% (95% CI, 18% to 79%) and 45% (95% CI, 24% to 86%), respectively. This was similar to survival outcomes for patients younger than age 60 (n = 30), with 2-year PFS and OS estimates of 41% (95% CI, 27% to 64%) and 43% (95% CI, 28% to 66%), respectively. For MAC alloBMTs (n = 20, 16 HLA-identical), the estimated 2-year PFS and OS were 44% (95% CI, 26% to 73%) and 42% (95% CI, 25% to 72%), respectively (Figure 1C). For RIC alloBMTs (n = 24, 18 haplo), 2-year estimates of PFS and OS were 37% (95% CI, 21% to 65%) and 44% (95% CI, 26% to 72%), respectively (Figure 1D).

The estimated 1-year cumulative incidences of NRM following alloBMT were 10% for patients receiving MAC (6% for MAC/HLA-identical transplants) and 8% for patients receiving RIC (11% for RIC/haplo transplants) (Figure 2). Causes of NRM after MAC were GVHD (n = 1), cardiac arrest (n = 1), multisystem failure (n = 1), and donor-derived lymphoma (n = 1). Causes of NRM after RIC were GVHD (n = 1), pulmonary embolism (n = 1), and stroke (n = 1). The median time to relapse was 3.2 months (interquartile range, 1.8 to 8.8 months). The estimated 1-year cumulative incidence of relapse after alloBMT was 40% for patients receiving MAC (38% for MAC/HLA-identical transplants) and 38% for patients receiving RIC (34% for RIC/haplo transplants) (Figure 2). Only five relapses occurred beyond 1 year post-alloBMT, with only two relapses occurring beyond 2 years post-alloBMT and only one occurring beyond 3 years post-alloBMT. Of note, three of the five late relapses were in patients with enteropathy-associated T cell lymphoma histology. Cumulative incidences of acute grade II-IV GVHD and chronic GVHD were 16% and 19%, with no differences between HLA-identical and haplo grafts.

Figure 2.

Cumulative incidence of relapse and nonrelapse mortality for RIC (n = 24, solid lines) and MAC (n = 20, dashed lines) alloBMT.

DISCUSSION

We present nearly two decades of experience with alloBMT for PTCL, in which two thirds of the patients received RIC regimens, haplo grafts, or both. Our results confirm and extend the results of previous reports showing long-term disease-free survival in a subset of patients with PTCL after alloBMT [7–9,20–24].

If the efficacy of alloBMT is due largely to both a GVL effect and avoidance of graft contamination by tumor, then MAC may not be a requisite in all cases. Although the median age at PTCL diagnosis is 60 [25,26], the subset able to receive MAC is most often two to three decades younger [20,24,27]. Utilization of RIC regimens has permitted alloBMT in older and more heavily pretreated patients; indeed, the median age at RIC/haplo alloBMT in this study was 60 years.

Although the potential for a GVL effect motivated the use of alloBMT, GVHD and transplant-related mortality have been major concerns. As in the treatment of other hematologic malignancies [10,11,28,29], we report here that high-dose PT/Cy for GVHD prophylaxis is associated with modest and comparatively favorable rates of GVHD, even in recipients of haplo grafts [11]. This is the first PTCL-restricted series to include a substantial fraction of patients who received RIC/ haplo alloBMT, all of whom received high-dose PT/Cy as part of their GVHD prophylaxis. The incidences of GVHD, relapse, and NRM for RIC/haplo alloBMT in this series were comparable with previous reports of outcomes after RIC/haplo alloBMT with PT/Cy and furthermore closely approximate the outcomes reported with RIC/HLA-identical platforms [11,28,30–32]. Such alloBMT platforms may be ideally suited for patients with PTCL, addressing the older age of this patient population and concerns about treatment-related toxicities.

There has been controversy with regard to the role and optimal timing of alloBMT. Although reserving alloBMT for relapsed disease spares the overtreatment of some, those who relapse are often resistant to salvage therapy. Both alloBMT in first remission and at relapse were represented in this series, with remarkably favorable outcomes for alloBMT in first remission and very poor outcomes for those with relapsed or refractory disease. The favorable outcomes for patients transplanted in first remission are notable considering the decision to proceed directly to alloBMT at first remission was typically reserved for patients with highly unfavorable features, such as a history of primary induction failure, central nervous system involvement, or very high-risk histology, such as gamma-delta subtypes or adult T cell leukemia/lymphoma. In a recently published prospective study of upfront autologous BMT for PTCL patients in pure first remission, the estimated 5-year OS was 61% for those receiving upfront autologous BMT, with PFS not reported for this subgroup, but with an estimated 5-year PFS of 44% for the intention-to-treat population [33]. In our series, patients receiving alloBMT in first remission, including those with primary induction failure, had an estimated 5-year PFS of 54%, with an estimated 5-year OS also of 54%, comparing favorably to the upfront autologous BMT outcomes, particularly considering that some of the alloBMT patients were spared MAC and that the rates of GVHD and NRM were low in this study. Given the retrospective nature of this study, it remains uncertain if these first remission BMT patients simply had favorable disease biology and would have had similar outcomes regardless of alloBMT, although this first remission cohort included aggressive subtypes with historically dismal outcomes, supporting a benefit of early alloBMT in select high-risk cases.

The RIC/haplo platform presented here offers a feasible and comparatively safe alloBMT option for the great majority of PTCL patients, including older patients and those with only partially HLA-matched donors. Given the poor outcomes for relapsed PTCL patients both in this series and historically, consolidation of first remission with RIC/haplo alloBMT may ultimately improve outcomes. Despite the increasing feasibility and safety of alloBMT, disease relapse after transplant continues to pose some limitations in the efficacy of this treatment modality for patients with PTCL. We found, however, that most relapses occurred early in the post-transplant course, adding further weight to the suggestion of previous studies that the subset of PTCL patients with long-term disease-free survival after alloBMT are likely cured [7–9,20–24]. Approaches to alloBMT that are able to maximize the GVL effect while minimizing transplant-related toxicities and GVHD are critical to improving outcomes in these diseases. Although the retrospective nature of this study and the heterogeneity of the lymphomas and their treatment limit definitive conclusions, this study provides evidence that there can be long-term disease free survival after alloBMT for PTCL, even in older patients with partially matched related donors. We hope that the ability to transplant a much larger subset of PTCL patients will encourage a cooperative group-based trial to prospectively evaluate the role of alloBMT in the management of these patients.