Graft-Versus-Host Disease and Graft-Versus-Tumor Effects After Allogeneic Hematopoietic Cell Transplantation

Abstract

Purpose

We designed a minimal-intensity conditioning regimen for allogeneic hematopoietic cell transplantation (HCT) in patients with advanced hematologic malignancies unable to tolerate high-intensity regimens because of age, serious comorbidities, or previous high-dose HCT. The regimen allows the purest assessment of graft-versus-tumor (GVT) effects apart from conditioning and graft-versus-host disease (GVHD) not augmented by regimen-related toxicities.

Patients and Methods

Patients received low-dose total-body irradiation ± fludarabine before HCT from HLA-matched related (n = 611) or unrelated (n = 481) donors, followed by mycophenolate mofetil and a calcineurin inhibitor to aid engraftment and control GVHD. Median patient age was 56 years (range, 7 to 75 years). Forty-five percent of patients had comorbidity scores of ≥ 3. Median follow-up time was 5 years (range, 0.6 to 12.7 years).

Results

Depending on disease risk, comorbidities, and GVHD, lasting remissions were seen in 45% to 75% of patients, and 5-year survival ranged from 25% to 60%. At 5 years, the nonrelapse mortality (NRM) rate was 24%, and the relapse mortality rate was 34.5%. Most NRM was a result of GVHD. The most significant factors associated with GVHD-associated NRM were serious comorbidities and grafts from unrelated donors. Most relapses occurred early while the immune system was compromised. GVT effects were comparable after unrelated and related grafts. Chronic GVHD, but not acute GVHD, further increased GVT effects. The potential benefit associated with chronic GVHD was outweighed by increased NRM.

Conclusion

Allogeneic HCT relying on GVT effects is feasible and results in cures of an appreciable number of malignancies. Improved results could come from methods that control progression of malignancy early after HCT and effectively prevent GVHD.

INTRODUCTION

Allogeneic graft-versus-leukemia effects were first suggested by Barnes et al¹ in 1956 after studying a murine model of syngeneic and H-2–incompatible hematopoietic cell transplantation (HCT). Reports linking graft-versus-host disease (GVHD) to antileukemic effects in humans were published many years later.^2–5 Importantly, the International Bone Marrow Transplant Registry study⁴ showed that allogeneic recipients with little or no GVHD experienced less relapse than identical twin recipients, suggesting antileukemic effects even in the absence of clinical GVHD. Subsequently, donor lymphocyte infusions have been used to control leukemic relapse after HCT,⁶ and graft-versus-tumor (GVT) effects have been described for other hematologic malignancies (reviewed in Roddie and Peggs⁷).

We developed a minimal-intensity conditioning regimen for allogeneic HCT that includes low-dose total-body irradiation (TBI) with or without fludarabine.⁸ Postgrafting immunosuppression with mycophenolate mofetil (MMF) and a calcineurin inhibitor serves to aid engraftment and control GVHD. The approach was designed for patients with advanced hematologic malignancies who could not tolerate high-dose conditioning regimens historically used for HCT because of age, comorbidities, or unsuccessful prior HCT with high-dose conditioning. The use of a minimal-intensity conditioning regimen allows for the purest assessment of GVT effects apart from conditioning and the best determination of GVHD not augmented by regimen-related toxicities. The current retrospective analysis of data in 1,092 patients has enabled a comprehensive assessment of the success and limitations of this approach.

PATIENTS AND METHODS

Patients

Between December 16, 1997, and October 30, 2009, 1,092 consecutive patients with hematologic malignancies were entered onto prospective, multicenter trials registered with ClinicalTrials.gov (Appendix Table A1). The institutional review board at each collaborating center approved the trials. Patients signed institutional review board–approved consent forms.

Table 1 lists patient characteristics. Grafts consisted of granulocyte colony-stimulating factor-mobilized blood mononuclear cells containing medians of 7.8 × 10⁶ CD34⁺ cells/kg (range, 0.8 to 42.6 × 10⁶ CD34⁺ cells/kg) and 3.1 × 10⁸ CD3⁺ cells/kg (range, 0.2 to 68.0 × 10⁸ CD3⁺ cells/kg). Patients had received a median of three prior chemotherapy regimens (range, zero to 18 regimens). HCT comorbidity index (HCT-CI) scores were assigned as described.⁹ Forty-five percent of patients had serious comorbidities with scores of 3 and higher.

Table 1.

Patient Demographics and Clinical Characteristics

Characteristic	Patients
	All (N = 1,092)		Related Recipientsa(n = 611)		Unrelated Recipientsb(n = 481)
	No.	%	No.	%	No.	%
Sex
Female	402	37	226	37	176	37
Male	690	63	385	63	305	63
Age, years
0-19	15	1	3	< 1	12	2
20-29	34	3	16	3	18	4
30-39	67	6	41	7	26	5
40-49	167	15	117	19	50	10
50-59	424	39	241	39	183	38
60-69	348	32	170	28	178	37
70-79	37	3	23	4	14	3
Median	56		56		58
Range	7-75		17-74		7-75
Diagnosis
ALL	46	4	11	2	35	7
AML	289	26	127	21	162	34
CLL	122	11	67	11	55	11
CML	47	4	26	4	21	4
HL	61	6	41	7	20	4
MDS	92	8	47	8	45	9
MPN	18	2	7	1	11	2
MM	217	20	170	28	47	10
NHL	190	17	109	18	81	17
Waldenström's syndrome	10	1	6	1	4	1
Regimen
TBI 2 Gy	222	20	222	36	—	—
TBI 2 Gy + fludarabinec	840	77	378	62	462	96
TBId 3 Gy + fludarabine	26	2	10	2	16	3
TBId 4 Gy + fludarabine	4	< 1	1	< 1	3	1
Prior transplantation
None	657	60	345	56	316	66
Planned autologouse	214	20	164	27	50	10
Failed autologous	174	16	87	14	87	18
Other autologousf	22	2	8	1	10	2
Failed allogeneic	25	2	7	1	18	4
Year of transplantation
1997-1999	71	6	71	12	—	—
2000-2004	601	55	344	56	257	53
2005-2009	423	39	196	32	224	47
Follow-up of surviving patients, yearsg
Median	5		5.1		5
Range	0.6 to 12.7		0.6 to 12.7		0.6 to 10.8
HCT-comorbidity index scoresh
0	210	24	124	27	86	20
1	107	12	54	12	53	12
2	168	19	85	19	83	19
3	209	24	104	23	105	24
4	80	9	34	7	46	11
5	55	6	25	6	30	7
6	38	4	18	4	20	5
7+	22	2	10	2	12	3
CMV positivei
Patient	641	60	374	63	267	56
Donor	485	46	339	58	146	31

NOTE. Fifty-five percent of patients were enrolled at Fred Hutchinson Cancer Research Center and 45% at collaborating centers. Donors and recipients were matched for HLA-A, -B, and -C, DRB1, and DQB1 by high-resolution typing except for 54 unrelated donor-recipient pairs.

Abbreviations: ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; CLL, chronic lymphocytic leukemia; CML, chronic myelocytic leukemia; CMV, cytomegalovirus; HCT, hematopoietic cell transplantation; HL, Hodgkin lymphoma; MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasms; MM, multiple myeloma; NHL, non-Hodgkin lymphoma; TBI, total-body irradiation.

^aFive phenotypic HLA matches.

^bFifty-four mismatches for one HLA class I allele. In addition to mycophenolate mofetil/tacrolimus, 61 unrelated recipients received sirolimus as part of a randomized trial.

^cFludarabine was added to TBI to reduce the risk of graft rejection.

^dThirty patients with advanced myeloid malignancies received 3 Gy (n = 26) or 4 Gy (n = 4) TBI; even at these doses, TBI is nonmyeloablative, sublethal, and well tolerated.

^eIncluding one syngeneic transplantation.

^fTransplantation for other malignancy and now referred for secondary MDS/AML.

^gData reflect events as of the date of last contact before the database was locked on March 1, 2011.

^hMissing for 203 patients.

ⁱMissing for 24 patients and 45 donors.

Conditioning and Postgrafting Immunosuppression

Patients received low-dose TBI with (n = 840) or without (n = 222) fludarabine 30 mg/m² per day on days 4, 3, and 2 before HCT (Table 1). Immunosuppression after HCT included MMF (28 days for related⁸ and at least 96 days for unrelated¹⁰ recipients) and a calcineurin inhibitor (180 days), either cyclosporine (CSP; n = 814) or tacrolimus (TAC; n = 278). In our experience, CSP and TAC were equally effective for GVHD prevention.¹¹ Infection prophylaxis and treatment followed institutional standard practice guidelines.

Monitoring After Transplantation

Patients had marrow aspirations on days 28, 84, 180, and 360, and then annually to assess disease status. Donor chimerism was evaluated on days 28, 84, and 365 after HCT. Acute and chronic GVHD were graded as described.^12,13 Relapse was defined as recurrence of malignancy based on one or more of the following: marrow morphology, flow cytometry, cytogenetics including fluorescence in situ hybridization, electrophoresis, immunofixation assays, polymerase chain reaction–based assays for disease markers, or imaging results. Disease progression was based on at least 50% increase in disease burden.¹⁴ Donor lymphocyte infusions were given to 33 patients for relapse or progression and to 19 patients for declining donor chimerism.

Causes of Death

Relapse mortality included deaths after relapse or progression of disease present before transplantation, regardless of other events. Nonrelapse mortality (NRM) included all deaths without relapse or progression. GVHD mortality included all deaths in patients with a history of GVHD who died from infections or organ failure during immunosuppressive therapy. Infection was listed as cause of death in patients without relapse or progression or a history of GVHD.

Statistical Analysis

Survival was estimated using the Kaplan-Meier method.¹⁵ Rates of acute and chronic GVHD, relapse or progression, and NRM were estimated according to standard methods.¹⁶ Death was treated as a competing risk for all other end points. Relapse was treated as a competing risk for NRM. Cox regression analysis was used for risk factor analysis and for comparisons between groups defined by donor relation and other characteristics. Risk factors for NRM occurring with a history of grade 2 to 4 acute GVHD or chronic GVHD were evaluated using a stepwise backward elimination procedure. All candidate risk factors were included in the initial model, and the least significant factors were excluded in sequential fashion until all remaining factors were significant at the P = .05 level. Candidate risk factors included HCT-CI scores of ≥ 3, unrelated transplantation, recipient cytomegalovirus (CMV) seropositivity, prior cytoreductive autologous HCT, transplanted CD34 and CD3 cell doses, allele-level HLA class I mismatch, grafts from female donors into male recipients, the addition of fludarabine to TBI, CSP versus TAC, and recipient age. Missing HCT-CI scores and CMV serostatus were accommodated by including missing categories for these variables. Relapse rates per patient year were calculated by dividing the total number of relapse events by the total years of follow-up in the time period of interest. Follow-up in a time period was truncated by the earliest occurrence of relapse, death, last contact, or the end of the time period.

RESULTS

Engraftment/Chimerism

Sustained engraftment was documented in 96% of patients, whereas 4% experienced graft rejection or failure between days 13 and 1,123 (median, 82 days). Engraftment was prompt, as evidenced by increasing neutrophil and platelet counts following the nadirs after conditioning. On days 28, 84, and 365, the median proportions of donor-derived cells were 95%, 99%, and 100% in the marrow; 98%, 100%, and 100% in neutrophils in blood; and 84%, 90%, and 100% in CD3⁺ T cells in blood, respectively. Forty-four percent of patients were never hospitalized at any time, whereas 56% had a median of 10 hospitalization days (range, 1 to 233 days), including the single overnight hospitalization for infusion of unrelated grafts.

Among the 48 patients with graft rejection (n = 41) or graft failure (n = 7), 36 had myeloid malignancies, and 13 received TBI (2 Gy) alone. Of the 48 patients, 36 died between 49 and 3,936 days (median, 284 days) after HCT, 29 of relapse and seven of nonrelapse causes. Twelve patients are alive and in remission, seven after second HCT, two after treatment with pentostatin immunosuppression and donor lymphocyte infusion, and three, all with chronic myelocytic leukemia, after successful treatment with tyrosine kinase inhibitors.

Relapse or Progression

By 5 years, 43.5% of the 1,092 patients had disease relapse or progression (related recipients, 45%; unrelated recipients, 42%), and 34.5% had died from relapse. Among 98 patients alive at 5 years after relapse or progression, 58% had stable complete remission after discontinuing immunosuppression, after chemotherapy with or without donor lymphocyte infusion, or after another allogeneic HCT; 5% were in very good partial remission; 5% were in molecular relapse (chronic myelocytic leukemia); 12% had stable disease; 5% had progression; and the remaining 15% were unknown. Relapse rates varied depending on underlying diseases and disease stages (Table 2). The designation of low, standard, or high relapse risk for each disease and disease stage followed a previously described algorithm.¹⁴ The lowest risk group included patients with B-cell malignancies, especially non-Hodgkin lymphoma (NHL) at any stage except for aggressive NHL not in remission. The highest risk group included patients with advanced malignancies not in remission. All 18 patients with acute lymphoblastic leukemia (ALL) in ≥ second complete remission experienced relapse. Relapse rates were highest during the first 2 years after HCT, whereas rates during years 3 to 5 were low, except among patients with multiple myeloma (MM) or Hodgkin lymphoma (HL) not in remission at the time of HCT. The cumulative incidence rates of relapse or progression in the three risk groups followed the expected pattern (Fig 1). Median days until relapse in the low-, standard-, and high-risk groups were 182, 174, and 107, respectively.

Table 2.

Relapse Rates Per Patient Year Among 1,092 Patients

Diagnosis*	Stage	No. of Patients	Relapse Rate
			Years 1 and 2	Years 3-5
Low risk
MPN	Any	18	0.10	0.00
CLL	CR	9	0.11	0.14
Waldenström's syndrome	Any	10	0.13	0.06
NHL	Any stage of mantle cell and low-grade; aggressive CR	140	0.16	0.02
ALL	CR1†	28	0.17	0.04
MM	CR	38	0.19	0.06
Standard risk
CLL	No CR	113	0.24	0.05
CML	CP1	24	0.24	0.00
MM	No CR	179	0.32	0.17
AML	CR‡	191	0.33	0.02
MDS	RA/RARS	30	0.35	0.00
High risk
NHL	Aggressive; no CR	50	0.48	0.00
AML	No CR; evolved from MDS	98	0.65	0.04
HL	After failed autologous HCT	61	0.61	0.14
MDS	RAEB; CMML; second	62	0.65	0.04
CML	CP2; AP; BC	23	0.71	0.07
ALL	≥ CR2; no CR	18	1.03	—

Abbreviations: ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; AP, accelerated phase; BC, blast crisis; CLL, chronic lymphocytic leukemia; CML, chronic myelocytic leukemia; CMML, chronic myelomonocytic leukemia; CP, chronic phase; CR, complete remission; HCT, hematopoietic cell transplantation; HL, Hodgkin lymphoma; MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasms; MM, multiple myeloma; NHL, non-Hodgkin lymphoma; RA, refractory anemia; RAEB, refractory anemia with excess blasts; RARS, refractory anemia with ring sideroblasts.

^*There were 243 patients in the low-risk group (53% related and 47% unrelated donors); 537 patients in the standard-risk group (58% related and 42% unrelated donors), and 312 patients in the high-risk group (54% related and 46% unrelated donors).

^†Before HCT, 14% of patients had minimal residual disease.

^‡Before HCT, 13% of patients had minimal residual disease.

Fig 1.

Cumulative incidence rates of (A) relapse or progression, depending on risk group and (B) progression-free survival.

GVHD

Incidence.

The overall cumulative incidence rates of grades 2, 3, and 4 acute GVHD at 120 days were 37%, 9%, and 3%, respectively (Fig 2A); rates were lower for related than for unrelated recipients (32% v 44%, 8% v 11%, and 3% v 4%, respectively; P < .0001, both overall and grades 3 and 4 separately). NRM increased in direct relation to increasing grades of acute GVHD (Fig 2B). The 3-year cumulative incidence rates of chronic GVHD were comparable among related and unrelated recipients (49% and 50%, respectively). The cumulative incidence of any acute or chronic GVHD was 75% (71% for related recipients and 79% for unrelated recipients; P < .0001).

Fig 2.

Cumulative incidence rates of (A) acute graft-versus-host disease (GVHD) among 1,092 patients and (B) nonrelapse mortality in relation to acute GVHD grades.

NRM.

The day-100 cumulative incidence of NRM was 5% overall (4% for related recipients and 5% for unrelated recipients). The 1-year NRM was 15%. The 5-year NRM was 24% overall (20.6% for related recipients and 28.4% for unrelated recipients; Table 3). Most 5-year NRM was associated with GVHD (20.2%), either acute GVHD with (8.6%) or without (7.4%) chronic GVHD or de novo chronic GVHD (4.2%). Overall GVHD-associated mortality was 16.6% among related recipients and 24.9% among unrelated recipients. Forty patients (3.8%) died before 5 years without a history of GVHD, 18 with bacterial, fungal, or viral infections. Causes related to pretransplantation comorbidities predominated among the remaining non-GVHD deaths, including six cardiac and three pulmonary deaths.

Table 3.

5-Year NRM: Effects of Comorbidity^* and GVHD

GVHD	5-Year NRM (%)
	All Patients	Patients With Related Donors		Patients With Unrelated Donors
		HCT-CI 0-2	HCT-CI ≥ 3	HCT-CI 0-2	HCT-CI ≥ 3
Acute grade 0-1	3.8	2.7	6.1	2.5	4.2
Acute grade 2-4
No chronic	8.6†	5.3	7.8	8.1	12.2
With chronic	7.4‡	4.6	6.4	11.6	13.3
De novo chronic	4.2	2.1	4.8	3.8	6.3
Combined acute grade 2-4 and de novo chronic	20.2	12	19	23.5	31.8
Overall	24	14.7	25.1	26	36

Abbreviations: GVHD, graft-versus-host disease; HCT-CI, hematopoietic cell transplantation comorbidity index; NRM, nonrelapse mortality.

^*HCT-CI scores had no association with incidence rates of acute or chronic GVHD.

^†NRM rates for acute grade 2 and grade 3 or 4 GVHD were 3.3% and 5.3%, respectively.

^‡NRM rates for acute grade 2 and grade 3 or 4 GVHD were 5.2% and 2.2%, respectively.

Comorbidities.

Table 3 shows the combined effects of comorbidities and GVHD on cumulative 5-year NRM. Related recipients with HCT-CI scores of 0 to 2 experienced 14.7% NRM, of which 12% was NRM associated with acute or chronic GVHD; these figures increased to 25.1% and 19%, respectively, when scores were ≥ 3. Rates of NRM and NRM associated with acute or chronic GVHD were appreciably higher for unrelated recipients (26% and 23.5%, respectively, for those with scores of 0 to 2, and 36% and 31.8%, respectively, for those with scores of ≥ 3).

GHVD, relapse, NRM, and overall mortality.

Approximately half of the overall relapse incidence occurred in patients without a history of GVHD. During the first 2 years, their relapse rate was 0.50 (0.51 for related recipients and 0.48 for unrelated recipients) per patient year compared with 0.23 in patients with GVHD. Table 4 shows an association of acute GVHD with NRM but no significant GVT effects. Chronic GVHD was associated with significantly lower relapse or progression rates, but high associated NRM offset this benefit. These conclusions held true when patients with low-, standard-, and high-risk disease were analyzed separately (data not shown).

Table 4.

Time-Dependent Analysis of GVHD in Relation to Relapse, NRM, and Overall Mortality at 5 Years

GVHD	Relapse or Progression			NRM			Overall Mortality
	HR	95% CI	P	HR	95% CI	P	HR	95% CI	P
All patients (N = 1,092)
No acute or chronic GVHD	1.0			1.0			1.0
Acute GVHD, grade 2*	0.92	0.7 to 1.2	.48	2.22	1.4 to 3.6	.001	1.09	0.9 to 1.4	.48
Acute GVHD, grade 3 or 4*	0.63	0.4 to 1.0	.06	13.8	9.0 to 21.1	< .0001	3.00	2.3 to 3.9	< .0001
Chronic after acute GVHD	0.55	0.4 to 0.8	.0002	4.83	3.2 to 7.8	< .0001	1.10	0.9 to 1.4	.42
De novo chronic GVHD	0.46	0.3 to 0.7	< .001	3.52	2.2 to 5.8	< .0001	0.88	0.7 to 1.1	.32
Related recipients with HCT-CI scores of 0-2 (n = 263)
No acute or chronic GVHD	1.0			1.0			1.0
Acute GVHD, grade 2*	0.75	0.4 to 1.3	.31	1.62	0.5 to 5.8	.46	0.91	0.5 to 1.7	.78
Acute GVHD, grade 3 or 4*	1.25	0.6 to 2.7	.57	15.4	5.5 to 42.8	< .0001	3.63	2.0 to 6.6	< .0001
Chronic after acute GVHD	0.43	0.2 to 0.8	.008	3.46	1.2 to 9.6	.02	0.95	0.6 to 1.5	.82
De novo chronic GVHD	0.48	0.3 to 0.9	.02	1.58	0.5 to 5.1	.44	0.62	0.4 to 1.1	.08
Unrelated recipients with HCT-CI scores of 0-2 (n = 222)
No acute or chronic GVHD	1.0			1.0			1.0
Acute GVHD, grade 2*	0.90	0.5 to 1.5	.70	1.57	0.5 to 5.1	.46	0.78	0.4 to 1.4	.38
Acute GVHD, grade 3 or 4*	0.57	0.2 to 1.6	.29	13.1	4.4 to 36.7	< .0001	1.93	1.0 to 3.6	.04
Chronic after acute GVHD	0.60	0.3 to 1.2	.14	4.14	1.5 to 12.4	.009	1.09	0.6 to 1.8	.75
De novo chronic GVHD	0.51	0.2 to 1.3	.17	4.10	1.3 to 14.0	.02	1.16	0.6 to 2.2	.64

NOTE. The analysis was adjusted for relapse risk (low, standard, or high), comorbidity scores (0-2, ≥ 3, or missing), and GVHD prophylaxis (cyclosporine or tacrolimus).

Abbreviations: GVHD, graft-versus-host disease; HCT-CI, hematopoietic cell transplantation comorbidity index; HR, hazard ratio; NRM, nonrelapse mortality.

^*Before chronic GVHD.

When data among 263 patients with related grafts and low comorbidity scores were analyzed separately (Table 4), acute grade 3 or 4 GVHD was not associated with reduced relapse or progression but showed an association with high NRM and high overall mortality. Chronic GVHD arising from acute GVHD and de novo chronic GVHD remained associated with significantly reduced risk of relapse or progression, but this was offset by increased NRM, resulting in overall mortality hazard ratios (HRs) of 1.21 and 0.75, respectively. No significant GVT effects were seen with either acute or chronic GVHD among 222 unrelated recipients with low comorbidity scores (Table 4). Acute grade 3 or 4 GVHD was associated with increased risk of NRM. Results among all 481 unrelated recipients were comparable except for decreased relapse or progression with de novo chronic GVHD (HR, 0.50; 95% CI, 0.3 to 1.0; P = .04).

Risk factors for GVHD-related NRM.

A multivariate analysis using backward elimination showed HCT-CI scores of ≥ 3 to be the most significant risk factor for GVHD-related NRM (HR, 1.70; 95% CI, 1.3 to 2.3; P < .001), followed by unrelated transplantation (HR, 1.41; 95% CI, 1.1 to 1.9; P = .002) and CMV-positive status (HR, 1.35; 95% CI, 1.0 to 1.8; P = .04). Patients given cytoreductive autologous HCT for either MM or advanced NHL, 40 to 281 days before the allograft, had a reduced risk of GVHD-related NRM (HR, 0.55; 95% CI, 0.4 to 0.8; P = .003). Transplanted CD34 or CD3 cell doses, allele-level HLA class I mismatch, grafts from female donors into male recipients, the addition of fludarabine to TBI, the nature of the calcineurin inhibitor, and recipient age did not have significant effects on GVHD-related NRM, after adjustment for the four factors noted.

Survival

Five-year survival rates for patients in the low relapse risk group with HCT-CI scores of 0 to 2 and ≥ 3 were 60% and 42%, respectively. The figures were 50% and 35%, respectively, for patients in the standard relapse risk group and 30% and 25%, respectively, for patients in the high relapse risk group.

DISCUSSION

The current regimen avoids serious toxicities and lacks profound cytotoxic antitumor activities. Its main role is to enable sustained allogeneic engraftment, which was accomplished in 96% of patients, thereby setting in motion immunologic GVT effects.

The 5-year survival rate ranged from 25% to 60%, largely influenced by the following three factors: comorbid conditions, disease risk, and GVHD. GVHD in this older population with often considerable comorbidities was not more frequent than reported for younger patients given myeloablative conditioning.^13,17 Powerful GVT effects were seen across all disease stages, except for advanced ALL. Although unrelated recipients experienced more GVHD than related recipients, the magnitude of GVT effects was comparable in the two patient groups. These effects presumably result from graft-versus-host reactions against minor histocompatibility antigens expressed on hematopoietic cells; however, natural killer cells may also be powerful contributors, especially for myeloid malignancies.^18–22 Similar findings were reported in a Center for International Blood and Marrow Transplant Research analysis of patients with acute and chronic leukemias given a variety of myeloablative conditioning and postgrafting immunosuppression regimens.²³ Seventy percent of relapse or progression occurred during year 1, and much of the remainder occurred during year 2. We hypothesize that early disease regrowth reflected a blunting of GVT effects as a result of early post-transplantation immune compromise and immunosuppression. Later, the immune system recovered, and immunosuppressive drugs were tapered or discontinued, permitting the donor immune cells to engage in GVT effects. Consistent with the hypothesis, relapse rates for all diseases, except MM and HL, were markedly reduced during years 3 to 5. The reasons for the partial escape of MM and HL cells from GVT effects are not clear.

Heightened GVT effects among current patients were associated with chronic GVHD but not with acute GVHD. Reducing or eliminating clinically significant acute GVHD would, therefore, not be expected to increase the risk of relapse. We have observed that CSP-treated related recipients whose donors were taking statins had no grade 3 or 4 acute GVHD and no apparent increase in relapse.²⁴ If ongoing prospective trials confirm this retrospective observation, the 7.5% incidence of mortality associated with grade 3 or 4 acute GVHD could theoretically be eliminated, decreasing overall 5-year GVHD-related NRM from 20.2% to 12.7%. Most residual NRM (11.6%) would likely be related to chronic GVHD. In line with previous observations,^4,17,25,26 chronic GVHD was associated with significantly decreased relapse risk compared with no GVHD, but this benefit was offset by an overall significantly higher NRM. As a result, patients with chronic GVHD preceded by acute GVHD experienced significantly increased mortality, whereas survival of patients with de novo chronic GVHD was comparable to that of patients without GVHD. Similarly, in subgroups of related and unrelated recipients with low comorbidity scores, neither de novo chronic GVHD nor chronic GVHD arising from acute GVHD conveyed significant survival advantages. These findings differed from early data published in 2005, which seemed to show a survival advantage for patients with chronic GVHD, attributable to associated GVT effects.²⁵ The main difference is the reference groups. The current study calculated HRs relative to patients who had neither chronic nor acute GVHD, whereas the 2005 analysis was calculated relative to patients who had no chronic GVHD without regard for acute GVHD. Moreover, current patient numbers were much larger, and median follow-up time (5 years in the current study v 1.5 to 2 years in the 2005 analysis) was much longer, which more definitively captured late mortality both from relapse or progression and from chronic GVHD.

Taken together, current data suggest that effective prevention of chronic GVHD would decrease NRM but increase mortality related to relapse or progression. Although the resultant trading of one cause of death for another would not change overall mortality, effective prevention would avert the considerable morbidity associated with chronic GVHD and its treatment. Treatment of recipients with a statin medication offers a possible avenue for reducing the risk of chronic (but not acute) GVHD,²⁷ although this approach might also increase the risk of relapse or progression.

Recent publications have addressed the use of reduced-intensity conditioning regimens for allogeneic HCT.^27–34 As a rule, the regimens were more intense and thus less reliant on GVT effects than the current regimen. Most reports did not provide information on comorbidities. Five of the reports were retrospective National Marrow Donor Program/Center for International Blood and Marrow Transplant Research/European Bone Marrow Transplantation Group analyses,^28–30,35 one was a multicenter French study,³² and three provided data from individual centers.^31,33,34 In these studies, median follow-up after HCT was 3 years (range, 1 to 5 years). Across the studies, median event rates were 34% (range, 6% to 38%) for NRM and 43% (range, 22% to 65%) for relapse, with overall survival at 38% (range, 23% to 60%). One additional study showed that GVT effects were mainly associated with National Institutes of Health–defined chronic GVHD.²⁶

Options for decreasing the risks of relapse and progression are limited. The use of more intense conditioning regimens would increase the risk of toxicity in elderly patients and those with comorbidities and, moreover, offer no benefit to the majority of current patients who did not experience relapse. Given these limitations, we hypothesize that improved cure rates could emerge from approaches that delay disease progression until the grafted immune system has recovered sufficiently to generate GVT effects. Well-tolerated targeted drugs or antibodies that are not curative on their own could set the stage for curative GVT effects. Candidate agents include antibodies to CD20 (NHL) and CD30 (HL),³⁶ phosphoinositide 3′-kinase δ inhibitor CAL101 (CLL),³⁷ proteasome inhibitors (MM), and FLT3 inhibitor AC220 (AML).³⁸ As an example of such an approach, 3-year survival rates were 62% for patients with Ph1-positive ALL in first remission treated with imatinib for 1 year after HCT and 73% for the subgroup without minimal residual disease before HCT, which is impressively better than previous results without imatinib.³⁹

We propose that GVT effects were present even without GVHD, assuming that relapse or progression would have been inevitable without HCT. This hypothesis is consistent with the notion that hematopoietic cells are targets for such reactions and echoes the International Bone Marrow Transplant Registry report describing less relapse in allogeneic recipients without GVHD compared with syngeneic recipients.⁴ Current patients who experienced chronic GVHD but not acute GVHD had significantly less relapse. Overall, the beneficial GVT effects associated with GVHD were offset by increased NRM. Major efforts should be directed toward methods that control progression of malignant disease early after HCT and more effectively prevent clinically significant GVHD.

Appendix

Table A1.

Allogeneic Hematopoietic Cell Transplantation Protocols and Corresponding ClinicalTrials.Gov Registry Identifiers

Protocol No.	ClinicalTrials.gov Identifier
1209.00	NCT00003145
1225.00	NCT00003196
1383.00	NCT00003954
1406.00	NCT00005801
1409.00	NCT00005803
1463.00	NCT00005799
1533.00	NCT00006251
1581.00	NCT00036738
1591.00	NCT00040846
1596.00	NCT00014235
1641.00	NCT00027820
1654.00	NCT00045435
1668.00	NCT00078858
1711.00	NCT00060424
1732.00	NCT00052546
1743.00	NCT00054353
1813.00	NCT00075478
1840.00	NCT00104858
1898.00	NCT00089011
1938.00	NCT00105001
1959.00	NCT00118352
2056.00	NCT00397813

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTSOF INTEREST

The author(s) indicated no potential conflicts of interest.

AUTHOR CONTRIBUTIONS

Conception and design: Rainer Storb, Mohamed L. Sorror, David G. Maloney, Brenda M. Sandmaier

Financial support: Rainer Storb, Mohamed L. Sorror

Administrative support: Rainer Storb, H. Joachim Deeg

Provision of study materials or patients: Rainer Storb, Mohamed L. Sorror, Karl Blume, Dietger Niederwieser, Thomas R. Chauncey, Michael A. Pulsipher, Finn B. Petersen, Firoozeh Sahebi, Edward D. Agura, Parameswaran Hari, Benedetto Bruno, Peter A. McSweeney, Michael B. Maris, Richard T. Maziarz, Amelia A. Langston, Wolfgang Bethge, Lars Vindeløv, Georg-Nikolaus Franke, Ginna G. Laport, Andrew M. Yeager, Kai Hübel, H. Joachim Deeg, George E. Georges, Mary E.D. Flowers, Paul J. Martin, Marco Mielcarek, Ann E. Woolfrey, David G. Maloney, Brenda M. Sandmaier

Collection and assembly of data: Boglarka Gyurkocza, Mohamed L. Sorror, Karl Blume, Dietger Niederwieser, Thomas R. Chauncey, Michael A. Pulsipher, Finn B. Petersen, Firoozeh Sahebi, Edward D. Agura, Parameswaran Hari, Benedetto Bruno, Peter A. McSweeney, Michael B. Maris, Richard T. Maziarz, Amelia A. Langston, Wolfgang Bethge, Lars Vindeløv, Georg-Nikolaus Franke, Ginna G. Laport, Andrew M. Yeager, Kai Hübel, H. Joachim Deeg, George E. Georges, Mary E.D. Flowers, Paul J. Martin, Marco Mielcarek, Ann E. Woolfrey, David G. Maloney, Brenda M. Sandmaier

Data analysis and interpretation: Rainer Storb, Boglarka Gyurkocza, Barry E. Storer, Mohamed L. Sorror, Karl Blume, Ginna G. Laport, Marco Mielcarek, David G. Maloney, Brenda M. Sandmaier

Manuscript writing: All authors

Final approval of manuscript: All authors