Risk score for outcome after allogeneic hematopoietic stem cell transplantation

A Retrospective Analysis

Authors

  • Alois Gratwohl MD,

    Corresponding author
    1. Department of Hematology, University Hospital, University of Basel, Basel, Switzerland
    • Hematology, University Hospital Basel, Petersgraben 4, CH-4031 Basel, Switzerland
    Search for more papers by this author
    • Fax: (011) 41 61 265 42 54

  • Martin Stern MD,

    1. Department of Hematology, University Hospital, University of Basel, Basel, Switzerland
    Search for more papers by this author
  • Ronald Brand,

    1. Department of Medical Statistics and Bioinformatics, Leiden University Medical Center, Leiden, Netherlands
    Search for more papers by this author
  • Jane Apperley MD,

    1. Department of Hematology, Hammersmith Hospital, London, United Kingdom
    Search for more papers by this author
  • Helen Baldomero,

    1. Department of Hematology, University Hospital, University of Basel, Basel, Switzerland
    Search for more papers by this author
  • Theo de Witte MD,

    1. Department of Hematology,University Medical Center St. Radboud, Nijmegen, Netherlands
    Search for more papers by this author
  • Giorgio Dini MD,

    1. Department of Pediatric Hematology and Oncology, Institute G. Gaslini, Genova, Italy
    Search for more papers by this author
  • Vanderson Rocha MD,

    1. Department of Hematology, St. Louis Hospital, Paris, France
    Search for more papers by this author
  • Jakob Passweg MD,

    1. Division of Hematology, Department of Internal Medicine, University Hospital, Geneva, Switzerland
    Search for more papers by this author
  • Anna Sureda MD,

    1. Clinical Hematology unit, Santa Creu i Sant Pau Hospital, Barcelona, Spain
    Search for more papers by this author
  • André Tichelli MD,

    1. Department of Hematology, University Hospital, Basel, Switzerland
    Search for more papers by this author
  • Dietger Niederwieser MD,

    1. Division of Hematology and Oncology, University Hospital, Leipzig, Germany
    Search for more papers by this author
  • for the European Group for Blood and Marrow Transplantation and the European Leukemia Net

    Search for more papers by this author
    • Alois Gratwohl, Martin Stern, Ronald Brand, and Helen Baldomero represent the European Group for Blood and Marrow Transplantation Activity Survey Office; Theo de Witte represents the Chronic Leukemia Working Party; Giorgio Dini represents the Pediatric Working Party; Vanderson Rocha represents the Acute Leukemia Working Party and the Eurocord Office; Jakob Passweg represents the Aplastic Anemia Working Party; Anna Sureda represents the Lymphoma Working Party; André Tichelli represents the Working Party Late Effects; Dietger Niederwieser is the President of the European Group for Blood and Marrow Transplantation.


Abstract

BACKGROUND:

It was investigated whether the European Group for Blood and Marrow Transplantation risk score, previously established for chronic myeloid leukemia, could be used to predict outcome after allogeneic hematopoietic stem cell transplantation (HSCT) for hematological disease in general.

METHODS:

Age of patient, disease stage, time interval from diagnosis to transplant, donor type, and donor-recipient sex combination were used to establish a score from 0 to 7 points. Its validity was tested in 56,505 patients, 33,113 (58%) male, 23,392 female, median age 33 years (range, 0.5-77 years), with an allogeneic HSCT for a hematological disorder between 1980 and 2005.

RESULTS:

Survival probability at 5 years decreased from 71% (95% confidence interval [CI], 69%-73%) for risk score 0 for the whole cohort (75%, 95% CI, 72%-78% for the most recent time cohort) to 24% (95% CI, 21%-27% for risk score 6 and 7; 25%, 95% CI, 22%-29% most recent cohort). Transplant-related mortality increased from 15% (95% CI, 14%-17%) for risk score 0 (11%, 95% CI, 9%-13%, most recent cohort) to 47% with risk score 6 and 7 (95% CI, 44%-50%) for the whole cohort (45%, 95% CI, 42%-48%, most recent cohort). The risk score was predictive in all disease categories, over all time periods, and was not altered by transplant techniques.

CONCLUSIONS:

Five well-defined pretransplant patient and donor characteristics give a reasonable risk estimate of allogeneic HSCT. This risk score can provide a basis for the decision between transplant and nontransplant strategies. Cancer 2009. © 2009 American Cancer Society.

Over the past 50 years, allogeneic hematopoietic stem cell transplantation (HSCT) has evolved into an established therapy for many severe congenital or acquired disorders of the hematopoietic system.1, 2 Stem cells from bone marrow, peripheral blood, or cord blood are used. They can be obtained from human leukocyte antigen (HLA)-identical siblings, from other family donors, from 1 of the >12 million HLA-typed, registered volunteer donors, or from 1 of the >300,000 registered public cord blood products (www.wmda.org). Today, >30,000 such allogeneic transplants are currently performed worldwide per year.3

Allogeneic HSCT can provide a cure in otherwise lethal disorders. It gives proof of principle of an immunological control of malignancy by a graft-versus-tumor effect. Its advantage over nontransplant treatment strategies in well-defined situations is documented.4 Novel approaches have expanded allogeneic HSCT to older patients and to patients with comorbidities.5, 6 Still, allogeneic HSCT remains associated with significant mortality and morbidity. At the same time, nontransplant strategies have improved as well.7, 8 The decision to proceed with an allogeneic HSCT no longer depends on the sole availability of a donor, and allogeneic HSCT is not always considered the first choice in a treatment strategy.4, 9, 10 Moreover, major progress has been made in risk assessment of the disease.11 The lack of an appropriate risk assessment of the transplant procedure becomes apparent.

Ten years ago, the European Group for Blood and Marrow Transplantation defined a risk score for patients with chronic myeloid leukemia, the most frequent indication for an allogeneic HSCT at that time.12 The risk score was based on 5 criteria, disease stage, patient age, donor type, time interval from diagnosis to transplantation, and donor-recipient sex combination. The score was validated in several independent patient cohorts and confirmed over time.13 In view of the need for a pretransplant risk assessment in general, we tested the applicability of this European Group for Blood and Marrow Transplantation risk score in a broad range of hematological disorders.

MATERIALS AND METHODS

Study Design and European Group for Blood and Marrow Transplantation Data Collection System

This is a retrospective observational study from the mandatory Med-A data set of the European Group for Blood and Marrow Transplantation and the annual report of all transplants in the preceding year through the European Group for Blood and Marrow Transplantation activity survey program (www.ebmt.org).14

Database Population

The European Group for Blood and Marrow Transplantation megafile database holds information on 77,568 patients with an allogeneic HSCT performed between 1980 and 2005. This time frame was selected to exclude the very early experimental transplants and to have sufficient follow-up. We concentrated on patients with an HLA-identical sibling or an unrelated donor transplant. Syngeneic twin donor and nonidentical family donor transplants were excluded. We restricted the analysis to disease categories of at least 1000 patients with acquired hematological disease. A total of 65,193 patients (57,220 [88%] of them transplanted between 1990 and 2005) fulfilled these criteria. Of these, 56,505 (87%) had complete information; 49,372 (88 %) were transplanted between 1990 and 2005. There was no difference in survival, transplant-related mortality, or relapse incidence between patients with full or partially missing data (data not shown). According to the activity survey (www.ebmt.org), the 57,220 patients in the Med-A database between 1990 and 2005 represent 87% and the 44,721 patients with full data 68% of all HSCTs in Europe performed during this time spam.

Patient Population

The final analysis includes 56,505 patients, 33,113 male (58%), 23,392 female (41%), median age 33 years (range, birth to 77 years). The transplants were performed at 384 institutions in 47 countries. Their disease characteristics are summarized in Table 1. The patient population represents patients with leukemia (n = 46,557; 83%), lymphoproliferative disorders (n = 5396; 10%), and aplastic anemia (n = 4252; 8%). It describes transplants from HLA-identical siblings (n = 41,545; 74%) or unrelated donors (n = 14,960; 26%) with bone marrow (n = 35,871; 63%), peripheral blood (n = 19,782; 35%), or cord blood (n = 852; 2%) as stem cell source. Conditioning intensity was classified, according to the teams definition and European Group for Blood and Marrow Transplantation guidelines (www.ebmt.org) in myeloablative (n = 43,530; 86%) or reduced intensity conditioning (n = 6946; 14%).

Table 1. Characteristics of 56,605 Patients With an Allogeneic HSCT for Hematological Disorders in Europe
Risk CategoryAMLALLCMLMDSMMNHLAACombinedP
  1. HSCT indicates hematopoietic stem cell transplantation; AML, acute myeloid leukemia; ALL, acute lymphoblastic leukemia; CML, chronic myeloid leukemia; MDS, myelodysplastic syndrome; MM, multiple myeloma; NHL, non-Hodgkin lymphoma; AA, aplastic anemia; HLA-id.sib., human leukocyte antigen-identical siblings; RMDF, recipient male, donor female; CMV, cytomegalovirus; neg, negative; pos, positive.

Patient sex, No. (%)        <.001
 Male8515 (53)7435 (64)7667 (59)3254 (56)819 (61)2915 (67)2508 (59)33,113 (59) 
 Female7598 (47)4243 (36)5314 (41)2531 (44)532 (39)1430 (33)1744 (41)23,392 (41) 
Stem cell source        <.001
 Bone marrow9409 (58)8234 (71)9555 (74)2739 (47)730 (54)1830 (42)3374 (79)35,871 (64) 
 Peripheral blood6502 (40)3103 (27)3359 (26)2944 (51)621 (46)2478 (57)775 (18)19,782 (35) 
 Cord blood202 (1)341 (3)67 (1)102 (2)037 (1)103 (2)852 (2) 
Regimen intensity        <.001
 Myeloablative12,554 (87)9715 (96)10,773 (93)3764 (72)964 (78)2451 (62)3309 (85)43,530 (86) 
 Reduced1845 (13)393 (4)858 (7)1445 (28)265 (22)1537 (38)603 (15)6946 (14) 
Patient age, y        <.001
 Median3520374446421833 
 Range0-770-730-700-7520-770-750-680-77 
Age classes, y, No. (%)        <.001
 <203298 (21)5779 (50)1206 (9)962 (17)0518 (12)2445 (58)14,208 (25) 
 20-407154 (44)4391 (38)7134 (55)1609 (28)327 (24)1536 (35)1537 (36)23,688 (42) 
 >405661 (35)1508 (13)4641 (36)3214 (56)1023 (76)2288 (53)270 (6)18,605 (33) 
Disease stage, No. (%)        <.001
 Early9847 (61)5573 (48)9685 (74)1588 (28)38 (3)642 (15)425231,565 (56) 
 Intermediate2744 (17)4112 (35)2711 (21)2158 (37)921 (68)1826 (42) 14,472 (26) 
 Advanced3522 (22)1993 (17)645 (5)2039 (35)392 (29)187 (43) 10,468 (19) 
Time interval, mo, No. (%)        <.001
 <1215,151 (94)10,744 (92)6271 (48)3873 (67)717 (53)1344 (31)2682 (63)40,782 (72) 
 >12962 (6)934 (8)6710 (52)1192 (33)634 (47)3001 (69)1570 (37)15,723 (28) 
Histocompatibility, No. (%)        <.001
 HLA-id.sib.12,442 (77)8139 (70)9146 (71)3616 (63)1232 (91)3574 (82)3378 (79)41,545 (74) 
 Unrelated donor3671 (23)3539 (30)3817 (29)2169 (38)119 (9)771 (18)874 (21)14,960 (26) 
Sex combination, No. (%)        <.001
 Other12,647 (79)8549 (73)9908 (76)4524 (78)1020 (76)3127 (72)3130 (74)42,905 (76) 
 RMDF3466 (22)3129 (27)3073 (24)1261 (22)331 (25)1218 (22)1122 (26)13,600 (24) 
Risk score, No. (%)        <.001
 01243 (8)893 (8)305 (2)87 (2)049 (1)923 (22)3500 (6) 
 13677 (23)2512 (22)2172 (17)460 (8)3 (0)241 (6)1613 (38)10,678 (19) 
 24259 (36)3235 (28)3830 (30)1080 (19)118 (9)472 (11)446 (11)14,190 (25) 
 32997 (19)2909 (25)3662 (28)1409 (24)431 (32)831 (19)70 (2)12,685 (23) 
 42287 (14)1505 (13)2040 (16)1385 (24)455 (34)1227 (28)4 (0)8969 (16) 
 51301 (8)520 (5)800 (6)985 (17)263 (20)1072 (25)04945 (9) 
 6-7349 (2)104 (1)172 (1)379 (7)80 (6)450 (10)01534 (3) 
Year of transplant, No. (%)        <.001
 1980-1985814 (5)719 (6)721 (6)55 (1)17 (1)54 (1)461 (11)2481 (5) 
 1986-19901577 (10)1292 (11)1683 (13)263 (5)144 (11)259 (6)485 (11)5703 (10) 
 1991-19952722 (17)2183 (19)2893 (22)736 (13)277 (21)542 (13)895 (21)10,248 (18) 
 1996-20004580 (28)3510 (30)4798 (37)1681 (29)522 (39)1250 (29)1103 (26)17,444 (31) 
 >20006420 (40)3914 (34)2886 (22)3050 (53)391 (29)2240 (52)1308 (31)20,269 (36) 
CMV status, No. (%)        <.001
 Recipient neg2111 (38)2062 (50)1895 (42)900 (43)205 (40)709 (45)523 (41)8405 (43) 
 Recipient pos3410 (62)2057 (50)2580 (58)1204 (57)309 (60)886 (56)761 (59)11,207 (57) 
Total16,11311,67812,981578513514345425256,505 

There were significant differences between the disease categories, among the risk groups, and over time (Table 1). Patients with aplastic anemia and acute leukemia were generally younger. Age and advanced disease stage increased over time. Patients with leukemia more frequently had early disease and were more frequently transplanted within 1 year of diagnosis. More patients with leukemia or myelodysplastic syndrome had unrelated transplants. The population of female unrelated donors for male recipients decreased over time. Bone marrow was the sole source of stem cells in 1990; it gradually decreased over time since 1993 to 31% in 2005. Reduced intensity conditioning was more frequently used in recent years and for patients with lymphoma or myelodysplastic syndromes.

Definitions of Risk Score

The risk score for this analysis used the same 5 pretransplant risk factors as initially defined12: age of the patient, disease stage, time from diagnosis to transplant, donor type, and donor-recipient sex combination, with 0 to 1 or 2 points for each factor as outlined in Table 2. Age was categorized as <20 years (0), 20 to 40 years (1), and >40 years (2). Disease stage was classified for each main disease category, based on previous analyses, as follows. Early disease stage (0) included: acute leukemia transplanted in first complete remission, myelodysplastic syndrome transplanted either untreated or in first complete remission, chronic myeloid leukemia in first chronic phase, and non-Hodgkin lymphoma and multiple myeloma transplanted untreated or in first complete remission. Intermediate disease stage (1) included: acute leukemia in second complete remission; chronic myeloid leukemia in all other stages than chronic phase or blast crisis; myelodysplastic syndrome in second complete remission or in partial remission; and non-Hodgkin lymphoma and multiple myeloma in second complete remission, in partial remission, or stable disease. Late stage disease (2) included: acute leukemia in all other disease stages, chronic myeloid leukemia in blast crisis, myelodysplastic syndromes in all other disease stages, and multiple myeloma and lymphoma in all other disease stages than those defined as early or intermediate. Stage was not applicable for patients with aplastic anemia. Time from diagnosis to transplant was categorized into <12 months (0) and >12 months (1). Donor type separated HLA-identical sibling transplants (0) from unrelated donor transplants (1). Donor-recipient sex combination separated all others (0) from the male recipient with a female donor (1). Hence, the score ranged from 0 to a maximum of 7 risk points.

Table 2. European Group for Blood and Marrow Transplantation Risk Score Definition
Risk FactorScore Point
  • HLA indicates human leukocyte antigen.

  • *

    See text for the definitions according to main disease category; does not apply for patients with severe aplastic anemia (score 0).

  • Does not apply for patients transplanted in first complete remission (score 0).

Age of the patient, y 
 <200
 20-401
 >402
Disease stage* 
 Early0
 Intermediate1
 Late2
Time interval from diagnosis to transplant, mo 
 <120
 >121
Donor type 
 HLA-identical sibling donor0
 Unrelated donor1
Donor-recipient sex combination 
 All other0
 Donor female, male recipient1

Furthermore, we tested the possibility of integrating 2 other, previously described factors. We looked for impact of cytomegalovirus (CMV)15 serostatus in 19,612 pairs for whom information was available. We also looked for the impact of Karnofsky score16 in 8668 pairs with this information.

Statistical Analysis

Endpoints analyzed were overall survival, transplant-related mortality (TRM), and relapse. Univariate survival rates were estimated by the Kaplan-Meier method. Univariate TRM rates were calculated as cumulative incidences, treating development of disease relapse as a competing risk. All survival and TRM rates are 5-year estimates. Multivariable Cox models were used to compare the impact of pretransplant risk factors. All risk factors contained in the European Group for Blood and Marrow Transplantation risk score were forced into the Cox models, along with year of transplant. To adjust for possible center effect, transplant center was added to all multivariate models as a frailty variable. Models including all patients with their different diseases were adjusted for disease type. For both outcome measures, the impact of the covariates was quantified by hazard ratios. Models were stratified for intensity of conditioning regimen (reduced vs myeloablative). The proportional hazards assumption was validated by using a time-dependent covariate method.17

The predictive power of the models was compared by computing the receiver operating characteristics area under the curve (ROC AUC) of each model using Harrell's C-Index statistics.18 The same model was used to compare the predictive power of the European Group for Blood and Marrow Transplantation risk score to the original Cox model incorporating the variable that defines the risk score.

The analysis followed the principles of the Strengthening the Reporting of Observational Studies in Epidemiology initiative.19

RESULTS

Survival

At the time of the analysis, 30,485 of the 56,505 patients were alive (54%), and 26,020 had died (46%), 16,116 from transplant-related causes (28%) and 9904 (17%) of their disease. Median follow-up was 37 months for the surviving patients. For the whole group, probability of survival at 5 years was 48% (95% confidence interval [CI], 48%-49%) with an estimated TRM rate of 31% (95% CI, 30%-31%). At 10 years, survival was 44% (95% CI, 44%-45%), and TRM was 32% (95% CI, 32%-33%).

Impact of the Risk Score on Survival and TRM

Survival and TRM were significantly influenced by the risk score in a systematic and monotonic way (dose-response relationship). Five-year survival rates decreased from 71% (95% CI, 69%-75%; risk score, 0) to 24% (95% CI, 21%-27%; risk score, 6 or 7), whereas 5-year TRM rates increased from 15% (95% CI, 14-17; risk score, 0) to 47% (95 CI, 44%-50%; risk score, 6 or 7; Fig. 1). In the cohort treated with HSCT since 2000, survival decreased from 75% (95% CI, 72%-78%) to 25% (95% CI, 22%-29%) with increasing risk score, and TRM increased from 11% (95% CI, 9%-13%) to 45% (95% CI, 42%-48%).

Figure 1.

Survival (Top) and transplant-related mortality (TRM) (Bottom) of 56,605 patients with an allogeneic hematopoietic stem cell transplantation (HSCT) for an acquired hematological disorder is shown by risk score. Graphs reflect probability of survival (Top) and transplant-related mortality (Bottom) over the first 5 years after HSCT.

All risk factors had a significant impact on the probability of survival and TRM in all disease categories (Table 3). The relative weight of the individual risk factors varied. Impact of age of the patient was significantly greater in aplastic anemia or acute lymphoblastic leukemia. Impact of disease stage was very similar in all disease categories (aplastic anemia excluded). Impact of the unrelated donor was most marked in aplastic anemia. Time interval had no impact on patients with acute leukemia transplanted in first complete remission, and an interval between diagnosis and transplant of >12 months had no significant negative impact on TRM in patients with lymphoma. The female donor/male recipient sex combination increased risk of TRM and decreased survival in all disease categories. This combined effect of risk score points led to a perfectly systematic decrease of survival and increase of TRM in all disease groups.

Table 3. Relative Risk Impact of Pretransplant Risk Factors on Survival and Impact of Cox or Score Analysis on ROC AUC
Risk FactorsAML (95% CI)ALL (95% CI)CML (95% CI)MDS (95% CI)MM (95% CI)NHL (95% CI)AA (95% CI)Combined (95% CI)
  1. ROC AUC indicates receives operating characteristics area under the curve; AML, acute myeloid leukemia; CI, confidence interval; ALL, acute lymphoblastic leukemia; CML, chronic myeloid leukemia; MDS, myelodysplastic syndrome; MM, multiple myeloma; NHL, non-Hodgkin lymphoma; AA, aplastic anemia; HLA-id.sib., human leukocyte antigen-identical siblings; RMDF, recipient male, donor female.

Age class, y        
 <201.001.001.001.001.001.001.001.00
 20-401.20 (1.12-1.28)1.44 (1.36-1.53)1.18 (1.07-1.31)1.43 (1.25-1.64)1.14 (0.96-1.34)1.02 (0.88-1.18)1.45 (1.29-1.63)1.34 (1.30-1.39)
 >401.45 (1.36-1.56)1.94 (1.78-2.11)1.58 (1.42-1.75)1.80 (1.58-2.06) 1.03 (0.89-1.20)2.90 (2.40-3.52)1.67 (1.60-1.73)
Disease stage        
 Early1.001.001.001.001.001.001.001.00
 Intermediate1.47 (1.39-1.58)1.48 (1.39-1.57)1.75 (1.65-1.86)1.22 (1.09-1.36)1.59 (0.95-2.67)1.64 (1.40-1.93) 1.52 (1.48-1.57)
 Advanced2.92 (2.76-3.09)2.58 (2.37-2.80)3.31 (3.01-3.64)1.71 (1.54-1.90)2.34 (1.38-3.96)2.58 (2.20-3.03) 2.52 (2.43-2.65)
Time interval, mo        
 <121.001.001.001.001.001.001.001.00
 >120.90 (0.82-0.98)1.26 (1.14-1.40)1.24 (1.17-1.31)1.03 (1.94-1.13)1.17 (1.01-1.36)0.75 (0.68-0.83)1.58 (1.22-1.55)1.12 (1.09-1.16)
Histocompatibility        
 HLA-id.sib.1.001.001.001.001.001.001.001.00
 Other1.30 (1.25-1.36)1.27 (1.22-1.33)1.39 (1.32-1.46)1.32 (1.24-1.42)1.46 (1.20-1.76)1.28 (1.17-1.40)2.09 (1.90-2.31)1.35 (1.32-1.38)
Sex combination        
 Other1.001.001.001.001.001.001.001.00
 RMDF1.07 (1.01-1.13)1.04 (0.98-1.10)1.22 (1.15-1.29)1.05 (0.95-1.16)1.15 (0.98-1.35)1.16 (1.05-1.28)1.10 (0.98-1.25)1.10 (1.07-2.13)
Explanatory content        
 ROC AUC        
  Cox0.6480.6340.6420.6030.5970.6160.6560.634
  Score0.6300.6160.6340.6000.5910.5770.6330.621

The predictive power and the explanatory content of the score as measured by ROC AUC varied between the individual disease groups and ranged for survival from 0.577 (non-Hodgkin lymphoma) to 0.634 in chronic myeloid leukemia (Table 3). Explanatory content was uniformly but only slightly better by the more precise Cox model than with the simplified score (Table 3) and slightly more so by the Cox model when year of transplant, center, and conditioning were integrated as interaction factors (data not shown). The benefit of the more precise interaction model, for example, from 0.632 with the score to 0.636 with the Cox model for TRM in chronic myeloid leukemia, appears marginal compared with the simple risk score.

Impact of Risk Factors on Death From Relapse

The risk factors had concordant and discordant effects on death from TRM and death from relapse. Death from relapse increased with disease stage from hazard ratio (HR) = 1.00 in early disease to HR = 1.55 (95% CI, 1.48-1.62; P < .001) in intermediate and to HR = 3.58 (95% CI, 3.41-3.76; P < .01) in advanced disease stage. A nonsignificant increase in relapse rates was seen in older patients (HR for patients 20-40 years old vs patients <20 years old, 1.04 [95% CI, 0.99-1.09; P = .12]; for patients >40 years old, 1.06 [95% CI, 1.00-1.12; P = .06]). Death from relapse decreased with a longer time interval (HR, 0.88; 95% CI, 0.84-0.93; P < .001). Death from relapse was not affected by donor type (HR for unrelated donor vs identical sibling, 0.99; 95% CI, 0.95-1.04; P = .80) and was lower in male recipients of female transplants (HR, 0.91; 95% CI, 0.87-0.95; P < .001). The beneficial effect of this graft-versus-leukemia reaction by female cells against Y chromosome-encoded gene products, however, could never outweigh the loss from the increased graft-versus-host–associated TRM, as reflected by the overall survival (see above).

Influence of the Transplant Procedure on Impact of the Risk Score

The strictly monotonic decrease of survival and increase of TRM with increasing risk score was observed for all disease categories (Fig. 2A), although survival and TRM for the same risk score were significantly different between the different disease categories. They remained so over the whole time period despite significant improvement over time (Fig. 2B). The lower survival and higher TRM in more recent years in Figure 2B is because of the univariate comparison. In multivariate comparison, taking into account the increase in age and in high-risk patients, TRM decreased significantly in all disease categories from the first time cohort (HR, 1.0) to the last (HR, 0.38; 95% CI, 0.33-0.41).

Figure 2.

Probability of survival (left illustration) and transplant-related mortality (TRM) (right illustration) are shown at 5 years for 56,605 patients with an allogeneic hematopoietic stem cell transplantation for an acquired hematological disorder. Graphs depict outcome by increasing risk score, from 0 to 6-7. Data are shown (A) by disease, (B) by year of transplant, (C) by transplant technology, (D) by cytomegalovirus (CMV) serostatus (CMV neg = recipient CMV seronegative; CMV pos = recipient CMV seropositive), and (E) Karnofsky score (good = Karnofsky score 90 or 100; poor = Karnofsky score 80 or lower). AML indicates acute myeloid leukemia; ALL, acute lymphoblastic leukemia; CML, chronic myeloid leukemia; AA, aplastic anemia; MDS, myelodysplastic syndrome; MM, multiple myeloma; NHL, non-Hodgkin lymphoma; RIC, reduced intensity conditioning; MAC, myeloablative conditioning; BM, bone marrow as stem cell source; PB, peripheral blood as stem cell source; CB, cord blood as stem cell source; T-dep, T-cell–depleted graft; No T-dep, T-cell–replete graft; KS, Karnofsky score; EBMT, European Group for Blood and Marrow Transplantation.

This strictly monotonic decrease of survival and increase of TRM with increasing risk score followed the same pattern, whether the patient was transplanted with bone marrow, peripheral blood, or cord blood as stem cell source or was given a T-cell–depleted or T-cell–replete transplant product (Fig. 2C). It was the same with reduced intensity or standard conditioning, although TRM was generally lower and death from relapse higher with reduced-intensity conditioning (Table 4).

Table 4. Impact of Year of Transplant and Conditioning Intensity on Outcome After Allogeneic HSCT by Risk Score: Probabilities of Survival, TRM, and Death From Relapse at 5 Years
Risk ScoreSurvivalTRMDeath From Relapse
MAC (95% CI)RIC (95% CI)MAC (95% CI)RIC (95% CI)MAC (95% CI)RIC (95% CI)
  1. TRM indicates transplant-related mortality; MAC, myeloablative conditioning; CI, confidence interval; RIC, reduced-intensity conditioning (for definitions see Materials and Methods); NA, not applicable, too few patients in this group.

Entire cohort
 −00.70 (0.68-0.72)0.81 (0.74-0.89)0.16 (0.15-0.17)0.15 (0.09-0.23)0.14 (0.13-0.16)0.04 (0.01-0.09)
 −10.62 (0.61-0.63)0.69 (0.64-0.75)0.23 (0.22-0.24)0.16 (0.13-0.20)0.15 (0.15-0.16)0.14 (0.11-0.29)
 −20.54 (0.53-0.55)0.52 (0.49-0.56)0.29 (0.28-0.30)0.23 (0.20-0.25)0.18 (0.17-0.18)0.25 (0.22-0.28)
 −30.43 (0.42-0.44)0.44 (0.40-0.48)0.34 (0.33-0.35)0.27 (0.24-0.30)0.23 (0.22-0.24)0.29 (0.25-0.32)
 −40.33 (0.32-0.34)0.35 (0.31-0.39)0.41 (0.39-0.42)0.29 (0.26-0.32)0.27 (0.25-0.28)0.37 (0.33-0.41)
 −50.24 (0.22-0.26)0.32 (0.29-0.37)0.47 (0.45-0.49)0.33 (0.30-0.36)0.29 (0.28-0.31)0.34 (0.31-0.38)
 −6-70.23 (0.20-0.27)0.23 (0.18-0.30)0.52 (0.48-0.56)0.37 (0.32-0.42)0.24 (0.21-0.28)0.39 (0.33-0.45)
Patients transplanted after the year 2000
 −00.71 (0.64-0.79)NA0.12 (0.07-0.20)NA0.16 (0.12-0.21)NA
 −10.63 (0.60-0.67)0.67 (0.61-0.75)0.16 (0.15-0.28)0.19 (0.14-0.25)0.20 (0.17-0.24)0.14 (0.09-0.19)
 −20.54 (0.51-0.58)0.51 (0.46-0.57)0.26 (0.23-0.28)0.22 (0.19-0.26)0.20 (0.18-0.23)0.27 (0.22-0.32)
 −30.44 (0.40-0.48)0.45 (0.39-0.51)0.30 (0.27-0.32)0.27 (0.22-0.32)0.27 (0.24-0.30)0.28 (0.24-0.33)
 −40.33 (0.29-0.36)0.35 (0.27-0.44)0.37 (0.34-0.40)0.33 (0.25-0.41)0.30 (0.28-0.33)0.32 (0.28-0.37)
 −50.24 (0.18-0.31)0.31 (0.24-0.39)0.41 (0.36-0.46)0.36 (0.29-0.43)0.35 (0.29-0.42)0.33 (0.28-0.38)
 −6-7NA0.22 (0.13-0.37)NA0.40 (0.33-0.46)NA0.38 (0.26-0.50)

Impact of Other Previously Described Risk Factors, CMV Serostatus, and Karnofsky Score

CMV serostatus influenced outcome, with a HR of 1.01 (95% CI, 0.92-1.11; P = .77) for the seronegative recipient with a seropositive donor, of 1.18 (95% CI, 1.13-1.26; P < .001) for the seropositive recipient with a seronegative donor, and 1.15 (95% CI, 1.07-1.22; P < .001) for the seropositive recipient with a seropositive donor, compared with the seronegative recipient with a seronegative donor (HR, 1.00). TRM was lower and survival better for the CMV-seronegative recipient, but CMV serostatus did not alter the risk score (Fig. 2D). Survival was better and TRM lower in patients with a Karnofsky score of 100 or 90 compared with patients with a Karnofsky score of 80 or lower at the time of transplant, but survival decreased and TRM increased in both groups with increasing risk score (Fig. 2E).

DISCUSSION

These data confirm the validity of the previously established European Group for Blood and Marrow Transplantation risk score12 beyond patients with chronic myeloid leukemia. Five pretransplant patient and donor properties, age, stage of the disease, time interval from diagnosis to transplant, donor type, and donor-recipient sex combination, can give valuable predictive information on TRM and survival for an individual patient. Prediction is further improved by the integration of other known pretransplant risk factors, recipient CMV serostatus, or Karnofsky performance status. Decision making on a rational basis should be enhanced.

These 5 factors were not chosen at random. Impact of disease stage and patient age have been recognized early in the history of HSCT.20 Donor type has become an element with the introduction of unrelated donor registries, although better matching for HLA might change this risk factor in the near future.21 The role of the donor-recipient sex combination was first described in single patients with aplastic anemia. It has later been confirmed in all disease categories and even in kidney transplantation.22 The role of the time interval as an important factor was first seen in aplastic anemia. Later, the interval was seen as an important element in chronic myeloid leukemia, independent of the type of pretransplant therapy.12 The role of time interval in acute leukemia has long been controversial. Time from diagnosis to transplant for patients in first complete remission is split into 2 distinct periods with discordant impact. A longer time from diagnosis to first complete remission is associated with increase in TRM; a longer time from first complete remission to transplant is associated with decrease in relapse incidence and improved survival. The time interval from diagnosis to transplant alone is therefore of no predictive value in patients with acute leukemia transplanted in first complete remission. For all patients beyond first complete remission except patients with non-Hodgkin lymphoma, time interval had a significant impact. Patients with non-Hodgkin lymphoma were the only exception.

The combined effect of the risk factors was first described 10 years ago in patients with chronic myeloid leukemia.12 The value of this score was then confirmed by several independent series. It was documented again in a more recent patient population and in patients with reduced-intensity conditioning transplants as well.13 Now, the present data set validates that the pretransplant risk score can be extended, with some limitations, to all patients treated with allogeneic HSCT for a hematological disease. It shows that the individual pretransplant risks all contribute to the risk of the procedure.

The risk score was conceived to assess TRM and survival. It had also an impact on death from relapse, however, in part concordant (disease stage, time interval, age) in part discordant (donor recipient histocompatibility and sex combination). Despite these limitations and despite the discordant effects of 2 risk factors, impact of the risk score on survival remained the same. This indicates that the transplant community has not yet been able, so far, to translate the beneficial effects of a histoincompatibility-induced graft-versus-tumor reaction into a survival benefit and to overcome the negative effects of graft-versus-host disease-associated mortality. Hence, classification by the risk score gives a rapid assessment, at any place, with a raw estimate of survival probability. This risk assessment can and will be extended. Risk of TRM was constantly about 5% higher in CMV-seropositive recipients, compared with CMV-seronegative recipients, except for patients with a very high risk score. Clinical performance status, as measured by Karnofsky score, had its impact on outcome as shown in the subset of patients for whom this information was available. Patients with reduced performance score had a higher TRM compared with patients with good performance score.

These findings have implications. Other risk factors have recently been described. Donor age increases TRM in unrelated transplants recipients.23 Certain cytokine polymorphisms can be associated with increased or decreased TRM.24 Donor recipient birth order may influence the state of tolerance in HLA-identical sibling transplants and add to or decrease TRM.25 Last, comorbidity increases the likelihood of TRM, as recently shown in 2 independent series.26 Information on cytokine polymorphisms, birth order, and comorbidity was not available in the data set. It is likely that these risk factors will show the same pattern and further improve risk assessment. There was a monotonic impact of the risk score over time despite significant reduction of TRM, and the risk score was not influenced by peri- or post-transplant strategies. Survival decreased with increasing risk score in recipients of bone marrow as well as of peripheral blood or cord blood transplants. It did so whether the graft product was T-cell depleted or not. The same monotonic pattern was observed in recipients of myeloablative or reduced-intensity conditioning transplants. The study was not designed to compare outcome of HSCT with reduced intensity with outcome after standard conditioning. There are differences. Patients with reduced-intensity conditioning showed a lower TRM but a higher rate of death from relapse; survival hence was similar between the 2 categories. Still, the impact of the score remained unchanged. Patients with older age, more advanced disease, a time interval >1 year from diagnosis, an unrelated donor, and a female donor for a male recipient always had higher TRM and lower survival. These observations indicate that transplant strategies might modify outcome of HSCT less than pretransplant factors do.

What are the consequences of this report? Today, HSCT and nontransplant strategies are available for patients in all disease categories. The primary choice should be based on the risk of the disease, the chances of the nontransplant strategy succeeding, and the risk of the transplant. Risk assessment of hematological malignancies has become more precise. For example, “good risk” leukemia can be defined.11 So far, decisions have been primarily based on the general risk of the disease and the availability of a donor. This approach should be changed. HSCT could be considered, for example, even for a patient with low risk acute myeloid leukemia, if there is a low-risk transplant possible (eg, with risk score 0-1); for intermediate-risk disease, a score of 0-3 might be acceptable, and a score of 0-5 for high-risk disease. The additional elements as described above, comorbidity score, CMV status, cytokine polymorphisms, or donor age might be used to tip the balance in favor or against HSCT, or to select another more suitable donor.

In summary, the data show outcome of patients with a hematological disease and allogeneic HSCT as of today. They describe in a retrospective analysis how 5 pretransplant risk factors influence TRM and survival. They form a rational basis for risk-adapted treatment strategies.

Acknowledgements

We thank all participating teams and their staff, as well as the European Group for Blood and Marrow Transplantation Coordination office: Barcelona (F. McDonald, E. McGrath, S. M. Jones, E. J. MacHale), Paris (V. Chesnel, C. Kenzey, N. C. Gorin), London (C. Ruiz de Elvira, S. de Souza), the Austrian Registry (H. Greinix, B. Lindner), the Czech Blood and Marrow Transplantation Registry (K. Benesova, M. Trnkova), the French Registry (D. Blaise, E. Marry, F. Mesnil), the German Registry (H. Ottinger, C. Muller, K. Fuchs, S. Allgaier, A. Siebinger, H. Neidlinger), the Italian Registry (A. Bosi, R. Oneto, B. Bruno), the Dutch Registry (A. Schattenberg, A. v. Biezen), Spanish Blood and Marrow Transplantation Registry (E. Carreras, I. Espigado, J. López, A. Cedillo), the Swiss Registry (U. Schanz, H. Baldomero, E. Buhrfeind), the Turkish Blood and Marrow Transplantation Registry (G. Gurman, M. Arat, F. Arpaci, M. Ertem), and the British Registry (D. Marks, J. Cornish, K. Kirkland, R. Paul). We also thank S. Stöckli for excellent secretarial assistance.

Conflict of Interest Disclosures

This work was supported in part by European Leukemia Net LSH-2002-2.2.0-3, by a grant from the Swiss National Research Foundation, 3200B0-118176, the Swiss Cancer League, the Regional Cancer League, and the Horton Foundation. The European Group for Blood and Marrow Transplantation is supported by grants from the corporate members: Amgen Europe, F. Hoffmann-La Roche Ltd, Gilead Sciences UK, Miltenyl Biotec GmbH, Schering-Plough International Inc., Celegene International SARL, Genzyme, ViroPharma Europe, Chugai Sanofi–Aventis, Fresenius Biotech GmbH, Gambro BCT, Bayer Schering Pharma AG, Therakos, Bristol Myers Squibb, Cephalon, Pierre Fabre Médicament, Alexion Europe, Pfizer, Biosafe SA, and Merck Sharp & Dohme.

Ancillary