Classification of HLA-Matching for Retrospective Analysis of Unrelated Donor Transplantation: Revised Definitions to Predict Survival

Article Outline

• Abstract
• Introduction
• Methods
• Patients
• Impact of HLA-Matching Cohorts on Survival
• Discussion
• Acknowledgments
• References
• Copyright

Abstract 

The best unrelated donors (URD) for hematopoietic cell transplantation (HCT) are alleles matched at HLA-A, -B, -C, and DRB1. Earlier studies mostly used incomplete or lower resolution HLA typing for analysis of transplant outcome. To understand the impact of incomplete HLA characterization, we analyzed 14,797 URD HCT (1995-2006) using multivariable regression modeling adjusting for factors affecting survival. Of 21 matching cohorts, we identified 3 groups with significantly different outcomes. Well-matched cases had either no identified HLA mismatch and informative data at 4 loci or allele matching at HLA-A, -B, and -DRB1 (n = 7477, 50% of the population). Partially matched pairs had a defined, single-locus mismatch and/or missing HLA data (n = 4962, 34%). Mismatched cases had ≥2 allele or antigen mismatches (n = 2358, 16%). Multivariate adjusted 5-year survival estimates were: well-matched: 54.1 (95% confidence interval), 52.9-55.4), partially matched: 43.7 (42.3-45.2), and mismatched: 33.4 (32.5-36.5), P < .001. A better matched donor yielded 10%-11% better 5-year survival. Importantly, intermediate resolution -A, -B, and -DRB1 alleles matched “6/6 antigen matched” HCT had survival outcomes within the partially matched cohort. We suggest that these proposed HLA subgroupings be used when complete HLA typing is not available. This improved categorization of HLA matching status allows adjustment for donor-recipient HLA compatibility, and can standardize interpretations of prior URD HCT experience.

Key Words: HLA matching, Unrelated donor transplantation

Back to Article Outline

Introduction 

Refinements in histocompatibility testing methodology have allowed clarification of critical genetic loci important in regulating alloimmune responses between donor and recipient surrounding unrelated donor (URD) hematopoietic cell transplantation (HCT). As histocompatibility testing has improved, expanded matching criteria have been applied to define the optimal donor and to identify critical loci to match to achieve superior patient outcome 1, 2, 3, 4. The quality of donor-recipient matching can modify graft rejection and GVHD as well as protection against neoplastic relapse and the likelihood of long-term survival 5, 6, 7, 8, 9, 10. Numerous important studies have defined the critical loci for matching to include HLA-A, -B, -C, and -DRB1 to maximize chances for the patients' best long-term outcome 11, 12, 13, 14, 15, 16, 17, 18.

The available, searchable donor file, however, is neither fully typed at all loci nor at allele-level resolution to allow expeditious selection of the optimal donor. Currently, only approximately 9% of the U.S. National Marrow Donor Program (NMDP) and Bone Marrow Donors Worldwide (BMDW) have intermediate resolution typing at HLA-A and -B plus allele typing at -DRB1 available for immediate online searching. Allele typing has become more readily available and more widely applied for donor search and identification over the 20 years of active URD HCT activity. Nonetheless, in 2006, only 83% of transplants (not searches) had high-resolution typing at all 4 loci and 35% of URD transplants had 1 or more missing or mismatched loci. Importantly, since the founding of the NMDP and other international donor registries, many transplants have been performed using donors selected by incomplete typing with methodologies insufficient to adequately delineate the histocompatibility existing between donor and recipient. The NMDP and Center for International Blood and Marrow Transplant Research (CIBMTR) maintain an extensive database of previous transplants and prospectively collected long-term outcomes. Improved matching reclassification that incorporates the completeness and resolution of previous typings can better utilize this invaluable long-term data without attrition because of incomplete HLA typing. We analyzed the best available typing compiled from initial pretransplant histocompatibility testing, from transplant centers' later retrospective high-resolution typing, and from the retrospective NMDP high-resolution Donor-Recipient Pair project utilizing stored pretransplant blood samples from the NMDP Research Sample Repository 10, 14 in conjunction with clinical outcomes data in 14,797 URD transplants performed between 1995 and 2006.

To better classify the match grades of prior transplants and thereby guide retrospective analyses of their outcome where HLA typing may be incomplete or at lower resolution, we need to understand the potential impact of that unknown HLA data on outcome in any retrospective analyses performed. We classified the number of HLA loci tested (or unknown) and the typing resolution (antigen level equivalent to intermediate resolution DNA typing or allele level, high-resolution DNA-based typing) based on a previously described standard definition [19] for all transplants and examined both matching and clinical criteria relevant to predict their outcome. Multivariate modeling was then used to describe clusters of matching subsets with similar outcome and thereby define the outcome for HCT patients whose HLA-match grade was predicted by the regression models. We propose a new classification of HLA-matching into 3 groups to be used for all retrospective analyses of URD HCT. We urge the transplant community to replace the widely used, but inadequate, classification of “MUD” (matched URD transplantation) in all future analyses where allele-level typing is incomplete.

Back to Article Outline

Methods 

All patients whose URD HCT was facilitated by the NMDP or reported to the CIBMTR between 1995 and 2006 were eligible to be included in the study (n = 15,867). Surviving patients who did not provide signed, informed consent to allow analysis of their clinical data were excluded. All surviving recipients included in this analysis were retrospectively contacted and provided informed consent for participation in the NMDP research program. Informed consent was waived by the NMDP institutional review board for all deceased recipients. To adjust for the potential bias introduced by exclusion of nonconsenting surviving patients, a corrective action plan (CAP)-modeling process randomly excluded the same percentage of deceased patients using a biased coin randomization with exclusion probabilities based on characteristics associated with not providing consent for use of the data in survivors. This CAP modeling left 14,887 (93.9%) cases to include a broad assessment of HCT recipients. Thirty-five cases were deleted because of rare diagnoses (breast cancer, other diseases, or an inherited abnormality of platelets), 7 with missing follow-up data on their first transplant, 25 missing patient race, and 23 in the HLA groups where there were fewer than 20 cases in the group. HLA data was compiled from the “best available typing” data table in the NMDP/CIBMTR database using either transplant center-reported HLA-typing available prior to transplant, retrospectively submitted refined or higher resolution histocompatibility typing plus data derived from the NMDP-sponsored Donor-Recipient Pair Project, which retrospectively performed multiple locus allele level typing for 5708 patients within the data file.

Clinical outcome data was extracted from the NMDP/CIBMTR outcomes database. Survival was modeled using a Cox multivariable regression with the following potentially relevant factors: patient age, donor age, disease, and stage, Karnofsky performance score, patient CMV serology, cell source (bone marrow versus filgrastim-primed peripheral blood [PBSC]), first or subsequent transplant, conditioning regimen intensity, patient and donor race, GVHD prophylaxis, and year of transplant. The model was stratified on the 21 defined HLA-matching groups derived from the patient population grouped by number of loci tested (from HLA-A, -B, -C, and -DRB1), level of resolution (intermediate/low or high) at each locus, and extent of mismatch (Table 1). HLA-DQ and -DP were not included, as previous analyses had shown them of lesser importance 10, 18. The final Cox model stratified on the HLA matching group was used to estimate the baseline survival function and 95% confidence interval (CI). One-year survival estimates were based on Breslow's estimator and a log-log transform CI [20].

Table 1. Patients and HLA Matching Data

B. Patient Characteristics	No. Eval	No. (%)
Number of patients		14,797
Number of centers		166
Age, median (range), years	14,796	37 (<1-79)
Age at transplant	14,796
< 18 years		3155 (21)
18-30 years		2688 (18)
31-45 years		3929 (27)
Over 45 years		5024 (34)
Male sex	14,796	8691 (59)
Karnofsky prior to transplant ≥90	13,671	9437 (69)
Recipient CMV status	14,619
Negative		6891 (47)
Positive		7728 (53)
Recipient race	14,797
White		12,854 (87)
African American		837 (6)
Hispanic		477 (3)
Asian/Pacific Islander		294 (2)
Other		335 (2)
Graft type	14,797
Bone marrow		9075 (61)
PBSC		5722 (39)
Conditioning regimen∗	14,797
Traditional myeloablative		9166 (62)
Nontraditional myeloablative		2607 (17)
Reduced intensity		1433 (10)
Nonmyeloablative		1180 (8)
Other		411 (3)
Disease at transplant†	14,797
AML early		1225 (8)
AML intermediate		1245 (8)
AML late		1687 (11)
AML unknown		125 (1)
ALL early		713 (5)
ALL intermediate		1118 (8)
ALL late		525 (4)
ALL unknown		58 (<1)
Other leukemia		675 (5)
CML early		1511 (10)
CML intermediate		708 (5)
CML late		142 (1)
CML unknown		101 (1)
MDS early		339 (2)
MDS late		654 (4)
MDS unknown		699 (5)
Non-Hodgkin lymphoma (NHL)		1372 (9)
Hodgkin lymphoma		353 (2)
Myeloma, plasma cell disorders		311 (2)
Other malignancies		54 (<1)
Severe aplastic anemia (SAA)		600 (4)
Inherited abnormalities of erythrocytes		28 (<1)
SCID and other immune system disorders		229 (2)
Inherited disorder of metabolism		215 (1)
Histiocytic disorders		110 (1)
Disease status at transplant‡	14,797
Early		3792 (26)
Intermediate		7413 (50)
Late		3010 (20)
Other		582 (4)
First transplant	14,797
Yes		12,563 (85)
No		2234 (15)
GVHD prophylaxis	14,797
Tac or CSA ± MTX ± MMF ± other		10,122 (68)
Tac or CSA ± other (no MTX)		2046 (14)
MMF or MTX ± other (No Tac or CSA)		84 (1)
T cell depletion		1818 (12)
Other		727 (5)
Donor/recipient sex match	13,983
Male/male		5502 (39)
Male/female		3271 (24)
Female/male		2710 (19)
Female/female		2500 (18)
Donor/recipient CMV serostatus	14,797
Negative/negative		4687 (32)
Negative/positive		4582 (31)
Positive/negative		2131 (14)
Positive/positive		3056 (21)
Unknown		341 (2)
Donor age, median (range), years	13,983	35 (18-61)
Year of transplant	14,797
1995-1998		3503 (24)
1999-2002		4450 (30)
2003-2006		6844 (46)
Median follow-up of survivors (range), months	5428	37 (1-140)

PBSC indicates peripheral blood stem cells; MTX, methotrexate; MMF, mycophenylate mofetil; Tac, tacrolimus; MM, mismatched.

Groups were clustered based upon similarity of these regression model-defined 1-year survival estimates and the presence of 0, 1, or ≥2 identified missing or mismatched loci.

Back to Article Outline

Patients 

The patient demographics including patient, disease, and transplant technique characteristics potentially affecting transplant outcomes are shown in Table 1B. Table 2 shows the estimated 1-year survival for an example, baseline patient (age <18 years, acute myetogenous leukemia (AML) early disease, Karnofsky performance status (KPS) 90-100, cytomegalovirus (CMV) negative, marrow cell source, 1st HCT, myeloablative conditioning, tacrolimus/methotrexate graft-versus-host disease (GVHD) prophylaxis, performed in 1995-1998) stratified on HLA matching group. Figure 1 depicts the baseline survival curves (adjusted as if for this example patient) for the HLA subgroups. Table 3 shows the relevant clinical factors modifying survival based on the final Cox model.

Table 2. Survival after URD HCT by HLA Match Classifications

High-res indicates high resolution, allele level typing; Low-res, low/intermediate (antigen level) resolution; MM, mismatch; CI, confidence interval.

Shown for each donor:recipient matching cohort are the adjusted 1-year survival estimates ± SE for an example patient assumed to be receiving a first transplant who is a CMV negative, <18-year-old patient with a 25-year-old donor. Groups are numbered as in Table 1A. The example patient is assumed to be White with a KPS of 90-100, transplanted during 2003-2006 using myeloablative conditioning and tacrolimus plus methotrexate as GVHD prophylaxis.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Estimated 1-year survival for a first transplant, under 18 years old, CMV-negative patient with a 25-year-old donor. The patient is assumed to be White with a KPS of 90-100, transplanted in 2003-2006 using myeloablative conditioning and tacrolimus/methotrexate as GVHD prophylaxis. The 21 groups cluster (per Table 2) as: well matched (thin/blue lines); partially matched (thick/red lines); and mismatched (medium/green lines). Right panel in color for website display.

Table 3. Risk Factors Adjusted for in Estimating 1-Year Survival

Factor	RR	95% CI		P-value
Patient age
<18	1.00
18-30	1.33	1.24	1.44	<.0001
30-45	1.48	1.38	1.60	<.0001
>45	1.78	1.65	1.92	<.0001
Disease/stage
AML early	1.00
AML intermediate	0.95	0.86	1.05	.3097
AML late	1.49	1.37	1.62	<.0001
AML unknown	1.43	1.13	1.81	.0028
ALL early	0.82	0.73	0.93	.0015
ALL intermediate	1.14	1.03	1.26	.0140
ALL late	1.86	1.66	2.08	<.0001
ALL unknown	1.37	0.97	1.94	.0773
CML early	0.62	0.57	0.69	<.0001
CML intermediate	0.87	0.78	0.98	.0187
CML late	1.37	1.13	1.66	.0014
CML unknown	0.95	0.74	1.22	.6995
MDS early	0.88	0.76	1.03	.1192
MDS late	1.13	1.01	1.26	.0387
MDS unknown	1.01	0.90	1.14	.8359
NHL	1.14	1.04	1.26	.0050
Hodgkin's lymphoma	1.04	0.88	1.30	.6238
Myeloma plasma cell disorder	1.09	0.93	1.27	.2792
Other malignancies	2.01	1.48	2.73	<.0000
SAA	0.91	0.80	1.05	.1950
Inherited abnormalities of erythrocytes	0.41	0.18	0.84	.0276
SCID and immune system disorders	0.67	0.53	0.87	.0007
Inherited disorders of metabolism	0.70	0.57	0.86	.0010
Histiocytic disorders	0.95	0.71	1.27	.7411
Karnofsky
90-100	1.00
<90	1.38	1.32	1.46	<.0001
Missing	1.02	0.94	1.11	.6279
Patient CMV
Negative	1.00
Positive	1.15	1.10	1.20	<.0001
Cell Source
Marrow	1.00
PBSC	1.00	0.94	1.06	1.00
First Transplant
Yes	1.00
No	1.28	1.19	1.37	<.0001
Conditioning
Traditional myeloablative	1.00
Nontraditional myeloablative	0.92	0.85	1.01	.0646
Reduced intensity	0.85	0.80	0.91	<.0001
Nonmyeloablative	0.73	0.66	0.80	<.0001
Other	0.86	0.75	0.99	.0419
Race
White	1
African American	1.28	1.17	1.40	<.0001
Hispanic	1.12	1.00	1.40	.0495
Asian/Pacific Island	1.03	0.88	1.20	.7439
Other	0.93	0.80	1.08	.3459
Donor age
Relative risk/change in year	1.00	1.00	1.01	.0005
GVHD prophylaxis
Tacrolimus ± MTX ± MMF ± Steroids ± Other	1.00
Tacrolimus ± other	1.03	0.91	1.17	.6312
CSA + MTX ± other	1.01	0.95	1.07	.8057
CSA ± other (no MTX)	1.22	1.13	1.33	<.0001
MMF ± other	1.07	0.71	1.62	.7394
MTX ± other (no CSA)	1.65	1.19	2.29	0.0027
T cell depletion	1.14	1.06	1.23	0.0006
Other	1.43	1.23	1.67	<0.0001
Year of HCT
1995-1998	1.00
1999-2002	0.87	0.82	0.92	<.0001
2003-2006	0.74	0.69	0.80	<.0001

CSA indicates cyclosporine; MMF, mycophenolate mofetil; MTX, methotrexate.

Shown are the relative risks (RR), 95% confidence interval (CI), and P value from Cox regression modeling of 1-year survival.

As shown, the 1-year survival estimates derived from the multivariable regression identified 3 clusters of HLA subgroups defined by available high resolution data, number of loci typed, and absence of known mismatches (Table 2 and Figure 1). The best group, identified as “well-matched,” had 1-year survival estimates from 62.0% to 68.3%, and included 7477 patients, 50% of the whole population. The intermediate group, “partially matched,” were missing either high-resolution or HLA-C data or had a defined single-locus mismatch. These included 4962, 34% of the population, and had estimated 1-year survival from 55.0% to 61.3%. The “mismatched” group with the poorest outcome included those with 2 or more known mismatches or missing data. This group included 2358, 16% of the population, and estimated 1-year survival between 34.4% and 56.7%. Although 1-year survival for group 12 overlaps in outcome with group 11, the 2 mismatched alleles in this group suggest it is more appropriately classified with the mismatched cohort. As shown in Figure 1, these survival estimates defined 3 HLA subgroup clusters with survival at 1 year for a patient with an average set of covariates estimated as 67.7 (95% CI 66.6-68.9) for the well-matched, 56.1 (95% CI 54.6-57.5) for the partially matched, and 33.4 (95% CI 32.5-36.5) for the mismatched URD HCT recipients.

Back to Article Outline

Impact of HLA-Matching Cohorts on Survival 

Identifying these 3 HLA subgroup clusters we sought to estimate the survival for hypothetical patients with these 3 match grades after adjusting for the important and statistically significant clinical risk factors including patient age, race, KPS, year of transplant, conditioning intensity, and GVHD prophylaxis. In the figures following (Figure 2, Figure 3), we illustrate 1-year (Figure 2) and 5-year (Figure 3) survival for a first transplant, CMV seronegative patient with a 25-year-old marrow donor of match grade shown. The patient examples are assumed to be White, with a KPS of 90% to 100%, transplanted in 2003-2006 using tacrolimus plus short course methotrexate as GVHD prophylaxis. The example transplant recipients all received myeloablative conditioning intensity except the 55-year-old MDS patients, who are assumed to have received a reduced-intensity conditioning (RIC) regimen.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Estimated survival for different patients undergoing a first transplant. Patients are all assumed to be CMV negative, White, with a KPS of 90-100, transplanted in 2003-2006 receiving tacrolimus/methotrexate as GVHD prophylaxis and a 25-year-old donor. All transplants are assumed myeloablative except the 55-year-old late MDS patient who was assumed to have a reduced-intensity transplant. Patients' age, disease, and disease stage are as shown in each panel.

• 
  • View Large Image
  • Download to PowerPoint

• Figure 3. 

  Estimated 5-year survival for different patients undergoing a first transplant. Patients are assumed to be CMV negative, White, with a KPS of 90-100, transplanted in 2003-2006 receiving tacrolimus/methotrexate as GVHD prophylaxis and a 25-year-old donor. All transplants are assumed myeloablative except the 55-year-old late MDS patient who was assumed to have a reduced-intensity transplant. Patients' age, disease, and disease stage are as shown in each panel.

As shown, for all six example patients (panels A-F, Figure 2, Figure 3), the defined matched grade groupings identify clearly distinct survival at 1 year and 5 years for all patient cohorts including those with either early or late disease stage, who were either older or younger. HLA-matching groups resulted in approximately 10% to 11% decrements in 5-year survival between the well-matched, partially matched, and mismatched groups for each modeled patient cohort.

Greater differences were recognized within the first year posttransplant, suggesting that the impact of HLA-matching has a more profound effect on early posttransplant mortality presumptively because of graft failure, GVHD, and infection, than on later posttransplant (5 year outcomes), which may be dominated by both relapse and chronic GVHD (cGVHD) and thus possibly less affected by HLA disparity.

Examining the HLA-matching groups in detail (Table 2) identifies best outcomes associated with no recognized mismatch and informative, although not necessarily high-resolution data at all 4 loci (HLA-A, -B, -C, and -DRB1) or with allele matching at HLA-A, -B, and -DRB1. In group 4, high-resolution typing at HLA-A, -B, and -DRB1 overcomes the potentially unrecognized mismatches at the untested HLA-C locus because of strong linkage disequilibrium with HLA-C in high-resolution HLA-B matched pairs. Although outcomes overlap with group 5, the lack of any defined mismatch suggests this subgroup should be in the well-matched group. Well-matched transplants, therefore, are characterized by informative data at the HLA-A, -B, -C and -DRB1 loci and no defined mismatch, even if high-resolution typing is not available at Class I loci (groups 1-4).

In contrast, the partially matched group (Table 2, groups 5-11) includes defined, single-locus mismatches at any of the 4 loci using either high-resolution or low/intermediate-resolution (allele or antigen) typing. Most surprisingly, the group traditionally called “matched” and referred to in the casual parlance as “MUD” (matched URD) is group 9. These are matched at low/intermediate resolution for HLA-A and -B and high resolution for -DRB1 with unknown data at HLA-C. This largest (in the partially matched) subgroup (1742 patients, 12% of the whole population) has projected 1-year survival, 9% inferior to the well-matched group. This clearly indicates that the widely used term “6 of 6 matched URD” is both insufficiently precise and a definitive misnomer. In this group, many unrecognized mismatches exist that contradict that this MUD group is “matched” and explaining the poorer outcomes in this group. Similarly, 8 of 8 low-resolution matching (group 10), although uncommon, also leads to worsened survival at 1 year. This emphasizes the importance of examining all 4 loci and the level of resolution to understand the impact of matching on outcome.

The third, mismatched cohort (Table 2, groups 12-21) has greater heterogeneity and includes either mismatch or missing information at 2 or more loci in all groups. Additionally, either single low-resolution mismatch (group 14) or high-resolution mismatch (groups 17 and 21) with unknown C typing yields poorer survival as does 6 of 6 low resolution matching with unknown data at HLA-C (group 13).

Back to Article Outline

Discussion 

These data strongly emphasize that classification and analysis of previous URD HCT is best performed and most informed when it considers data at the 4 most critical loci, HLA-A, -B, -C, and -DRB1. In current donor selection, this has become widely adopted. During 2006, 83% of transplants performed had high-resolution typing at all 4 loci. Of these transplants, only 65% were well matched, whereas 26% were partially matched and 6% mismatched using this new classification schema (CIBMTR/NMDP data, unpublished). This new classification suggests that in retrospective analyses, pairs with no defined mismatch and informative data at all 4 loci can be analyzed together as the best donor:recipient pairs for transplantation among previous typed transplants. This new schema does not imply that the subgroups within the 3 cohorts are equivalent or should be used for current donor selection where informative, allele-level typing at all 4 loci is preferred.

Recognizing these 3 defined HLA-matching groups, the CIBMTR and NMDP propose that all subsequent retrospective analyses of URD HCT include the available informative data at all these 4 loci. We also propose that these 3 match grade classifications be used in all retrospective analyses of outcome. Reanalysis of data sets previously published that defined MUD by the older, and insufficiently precise term “6 of 6 matching” with low/intermediate resolution at Class I and missing data on HLA-C combines nearly 25% of patients in the intermediate, partially matched group with the other, truly well-matched cohort, and thus confounds interpretation of all older URD transplant reports. A previous NMDP analysis identified allele level mismatches in 50% of “6/6 antigen matched” pairs; 31% were mismatched at HLA-C [14]. Earlier clinical reports, therefore, may have obscured differences in outcomes that might have been attributable to incomplete HLA matching.

We urge the transplant community to adopt this schema for all retrospective and comparative analyses, and we anticipate its validation in future studies of alternative donor transplantation including umbilical cord blood and RIC cohorts. Although allele-level precision is, of course, preferred and should be utilized for current donor selection 10, 18, this classification maximizes the use of available information to better define interpretations of prior experience and to inform physicians and their patients for future decision making about the application and outcomes of URD HCT.

Back to Article Outline

Acknowledgments 

This project has been supported by funding from the NMDP and the Department of the Navy, Office of Naval Research Cooperative Agreement #N00014-99-2-0006 and grant #N00014-05-1-0859 to the NMDP. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the Office of Naval Research or the NMDP.

Back to Article Outline

References 

1. Hurley CK, Baxter Lowe LA, Logan B, et al. National Marrow Donor Program HLA-matching guidelines for unrelated marrow transplants. Biol Blood Marrow Transplant. 2003;9:610–615
   • View In Article
   • Full Text
   • Full-Text PDF (96 KB)
   • CrossRef
2. Petersdorf EW, Longton GM, Anasetti C, et al. The significance of HLA-DRB1 matching on clinical outcome after HLA-A, B, DR identical unrelated donor marrow transplantation. Blood. 1995;86:1606–1613
   • View In Article
   • MEDLINE
3. Petersdorf EW, Longton GM, Anasetti C, et al. Definition of HLA-DQ as a transplantation antigen. Proc Natl Acad Sci USA. 1996;93:15358–15363
   • View In Article
4. Petersdorf EW, Longton GM, Anasetti C, et al. Association of HLA-C disparity with graft failure after marrow transplantation from unrelated donors. Blood. 1997;89:1818–1823
   • View In Article
   • MEDLINE
5. Petersdorf EW, Gooley TA, Anasetti C, et al. Optimizing outcome after unrelated marrow transplantation by comprehensive matching of HLA class I and II alleles in the donor and recipient. Blood. 1998;92:3515–3520
   • View In Article
   • MEDLINE
6. Petersdorf EW, Hansen JA, Martin PJ, et al. Major-histocompatibility-complex class I alleles and antigens in hematopoietic-cell transplantation. N Engl J Med. 2001;345:1794–1800
   • View In Article
   • MEDLINE
   • CrossRef
7. Petersdorf EW, Anasetti C, Martin PJ, et al. Limits of HLA mismatching in unrelated hematopoietic cell transplantation. Blood. 2004;104:2976–2980
   • View In Article
   • MEDLINE
   • CrossRef
8. Sasazuki T, Juji T, Morishima Y, et al. Effect of matching of class I HLA alleles on clinical outcome after transplantation of hematopoietic stem cells from an unrelated donor. Japan Marrow Donor Program. N Engl J Med. 1998;339:1177–1185
   • View In Article
   • MEDLINE
   • CrossRef
9. Morishima Y, Sasazuki T, Inoko H, et al. The clinical significance of human leukocyte antigen (HLA) allele compatibility in patients receiving a marrow transplant from serologically HLA-A, HLA-B, and HLA-DR matched unrelated donors. Blood. 2002;99:4200–4206
   • View In Article
   • MEDLINE
   • CrossRef
10. Flomenberg N, Baxter-Lowe LA, Confer D, et al. Impact of HLA class I and class II high-resolution matching on outcomes of unrelated donor bone marrow transplantation: HLA-C mismatching is associated with a strong adverse effect on transplantation outcome. Blood. 2004;104:1923–1930
    • View In Article
    • MEDLINE
    • CrossRef
11. Petersdorf EW, Smith AG, Mickelson EM, et al. The role of HLA-DPB1 disparity in the development of acute graft-versus-host disease following unrelated donor marrow transplantation. Blood. 1993;81:1923–1932
    • View In Article
    • MEDLINE
12. Petersdorf EW, Gooley T, Malkki M, et al. The biological significance of HLA-DP gene variation in haematopoietic cell transplantation. Br J Haematol. 2001;112:988–994
    • View In Article
    • MEDLINE
    • CrossRef
13. Shaw BE, Marsh SG, Mayor NP, Russell NH, Madrigal JA. HLA-DPB1 matching status has significant implications for recipients of unrelated donor stem cell transplants. Blood. 2006;107:1220–1226
    • View In Article
    • MEDLINE
    • CrossRef
14. Hurley CK, Fernandez-Vina M, Hildebrand WH, et al. A high degree of HLA disparity arises from limited allelic diversity: analysis of 1775 unrelated bone marrow transplant donor-recipient pairs. Hum Immunol. 2007;68:30–40
    • View In Article
    • MEDLINE
    • CrossRef
15. Petersdorf EW, Kollman C, Hurley CK, et al. Effect of HLA class II gene disparity on clinical outcome in unrelated donor hematopoietic cell transplantation for chronic myeloid leukemia: the US National Marrow Donor Program Experience. Blood. 2001;98:2922–2929
    • View In Article
    • MEDLINE
    • CrossRef
16. Ho VT, Kim HT, Liney D, et al. HLA-C mismatch is associated with inferior survival after unrelated donor non-myeloablative hematopoietic stem cell transplantation. Bone Marrow Transplant. 2006;37:845–850
    • View In Article
    • MEDLINE
    • CrossRef
17. Petersdorf EW, Malkki M, Gooley TA, Martin PJ, Guo Z. MHC haplotype matching for unrelated hematopoietic cell transplantation. PLoS Med. 2007;4:e8
    • View In Article
    • CrossRef
18. Lee SJ, Klein J, Haagenson M, et al. The contribution of high resolution donor-recipient HLA matching to the success of unrelated donor marrow transplantation. Blood. 2007;110:4576–4583
    • View In Article
    • CrossRef
19. Spellman S, Maiers M, Setterholm M, et al. A standard definition of HLA matching for hematopoietic stem cell transplant outcomes. Biol Blood Marrow Transplant. 2006;12(Suppl 1):211
    • View In Article
20. Klein JP, Moeschberger ML. Survival Analysis: Statistical Methods for Censored and Truncated Data. New York: Springer-Verlag; 2003;
    • View In Article