Female-Versus-Male Alloreactivity as a Model for Minor Histocompatibility Antigens in Hematopoietic Stem Cell Transplantation


* Corresponding author: Martin Stern, sternm@uhbs.ch


H-Y encoded gene products were the first to be recognized as clinically relevant minor histocompatibility antigens. Compared to other gender combinations, female donor/male recipient (FDMR) transplants are associated with increased graft-versus-host disease (GvHD), increased transplant-related mortality (TRM) and reduced risk of relapse. Still, their relative impact on transplant outcome remains controversial. We analyzed donor/recipient sex combination in 53 988 patients treated with allogeneic hematopoietic stem cell transplantation (HSCT) between 1980 and 2005. We found a strong increase in chronic GvHD and late TRM and decreased survival in FDMR transplants irrespective of underlying disease. Conversely, FDMR patients had lower relapse rates. The negative effect on survival decreased with advancing disease stage as relapse protection became more important. Effects of H-Y alloreactivity were most pronounced in patients transplanted from HLA-matched donors and in those receiving transplants from an adult donor. Adjustment for acute and chronic GvHD only partially corrected the effects of H-Y alloreactivity. Analysis of the FDMR proportion over time indicated that the frequency of this gender combination has declined in unrelated transplants over the last 10 years. These data define the role of H-Y mismatching in allogeneic HSCT and support the current practice of avoiding female donors for male patients, if possible.


Allogeneic hematopoietic stem cell transplantation (HSCT) from a histocompatible donor is the treatment of choice for patients with hematopoietic malignancies at high risk of relapse after standard chemotherapy. Tumor eradication occurs through a combination of the cytoreductive properties of the conditioning regimen and immunologic graft-versus-tumor effects. In the case of a complete human leukocyte antigen (HLA) match between donor and recipient, targets recognized on malignant cells by donor T-lymphocytes may be either tumor-associated antigens or minor histocompatibility antigens (mHAg). The latter are peptides derived from cellular proteins encoded by polymorphic genes that differ between donor and recipient and that are recognized by donor T-lymphocytes when presented on recipient HLA molecules. Depending on the distribution of their tissue expression, mismatched mHAg may also causeGvHD. Among the first described and best characterized mHAg are a group of Y-chromosome-encoded proteins (H-Y) with universal tissue distribution that may be recognized in the setting of a mismatched gender combination (1,2).

H-Y antigens are an excellent model for mHAg mismatches in allogeneic HSCT: several H-Y mHAgs restricted to HLA antigens with a broad distribution in Caucasians have been described (3–12), and mHAg mismatching is therefore relatively independent from HLA. Furthermore, information on donor and recipient gender is available in virtually all allogeneic HSCT, whereas for somatic chromosome-encoded minor antigens, elaborate functional or genetic testing is required to determine whether donor and recipient are mismatched. H-Y is therefore uniquely suited to analyze the role of mHAg mismatching in allogeneic HSCT.

H-Y was first reported as a clinically relevant transplantation antigen in a female patient with aplastic anemia rejecting a male donor graft (2). An increased risk of graft rejection in the female recipient/male graft situation was recently confirmed in a large registry study (13). Studies in patients with chronic myeloid leukemia treated with allogeneic HSCT in first chronic phase have shown that female donor/male recipient (FDMR) transplants are associated with a relative protection from disease relapse, an increase in GvHD and transplant related mortality, (TRM) and a net negative effect on survival (14). Similar data have been published for patients treated with allogeneic stem cell transplantation for myeloma (15). For other malignant hematopoietic diseases only limited data are available (16,17).

In this study, we used a large cohort of transplants reported to the European group for Blood and Marrow Transplantation (EBMT) to analyze the impact of H-Y alloreactivity in the major malignant hematopoietic diseases treated with allogeneic HSCT. The large timespan over which the study extends also allowed us to analyze the evolution of donor selection patterns.

Patients, Materials and Methods

Study design

This retrospective study is based on 53 988 allogeneic transplants carried out between 1980 and 2005 and reported to the EBMT by standardized questionnaire or electronic data management system. Patients were included if information was available on: patient age, disease, disease stage, date of diagnosis, donor and recipient sex, type of donor and source of stem cells. The 53 988 transplants reported in this study represent 66% of all allogeneic first transplants reported to EBMT between 1980 and 2005 (n = 81 787). Outcome of FDMR patients was compared to outcome of transplants with another gender combinations combined. Main endpoints analyzed were rates of survival, TRM and relapse, and incidence of acute and chronic GvHD.

Patient population and definitions

Analysis was restricted to patients treated with an allogeneic HSCT for acute myeloid or lymphoblastic leukemia, chronic myeloid leukemia, (CML) lymphoma/chronic lymphocytic leukemia, myelodysplastic syndrome (MDS) or multiple myeloma. Patients were included if information on disease type, stage at transplant, date of transplant, patient age at transplant, donor and recipient gender, donor type (related vs. unrelated, HLA matched vs. mismatched) and grade of acute and chronic GvHD were available. Patients transplanted for solid tumors and bone marrow failure syndromes were excluded, as were those receiving autologous or syngeneic transplants.

Disease stage was defined as follows: in acute leukemia, patients transplanted in first complete remission (CR) were considered to have early-stage disease, those in any other CR intermediate stage and all other patients advanced disease; inCML, patients transplanted in first chronic phase were considered to have early-stage disease, those in blast crisis advanced disease and all others intermediate stage; inMDS, patients transplanted in first CR were considered to have early-stage disease, those in any other CR and those without any preceding treatment intermediate stage and all other patients advanced stage; in patients with lymphoma, those in first CR were considered to have early-stage disease, those in first partial remission and those untreated were considered to have intermediate-stage disease and all others advanced stage; and in multiple myeloma, patients in first CR were considered to have early-stage disease, those in partial or any other CR and those untreated were considered to have intermediate-stage disease and all other patients advanced-stage disease.

To study the effect of prior sensitization to H-Y antigen, we used donor age as a surrogate marker for history of pregnancy that was not available in donors. We stratified into two groups (donor < 20 years vs. donor > 20 years), assuming that the majority of donors with a history of pregnancy would fall into the second group.

Details of the 53 988 patients are listed in Table 1. Transplants with a gender combination other than FDMR were grouped, as no differences in the major endpoints were seen between the three other possible gender combinations. There were minor but statistically significant differences between FDMR patients and controls. There were fewer patients with acute myeloid leukemia (AML) and MDS, more patients with acute lymphoblastic leukemia (ALL) and lymphoma and fewer patients with early disease in the FDMR group. FDMR patients were slightly younger, had more frequently an HLA-identical sibling donor transplant, more frequently a bone marrow transplant and were more frequently transplanted in earlier years and with myeloablative conditioning.

Table 1.  Characteristics of 53 988 patients with an allogeneic HSCT for a hematological malignancy
 Female donor/male recipientOther gender combinationProportion FDMR within subgroupp-Value
Disease (n,% total) < 0.001
  AML 3701(28.5%)13 358(32.6%)21.7% 
  ALL 3405(26.2%) 9205(22.5%)27.0% 
  CML 3284(25.3%)10 369(25.3%)24.1% 
  MDS 1007 (7.7%) 3766 (9.2%)21.1% 
  Lymphoma 1264 (9.7%) 3245 (7.9%)28.0% 
  Multiple myeloma342 (2.6%) 1042 (2.5%)24.7% 
Stage (n,% total) < 0.001
  Early 6547(50.6%)21 459(52.4%)23.5% 
  Intermediate 3639(28.4%)11 131(27.2%)24.9% 
  Advanced 2650(20.4%) 8208(20.0%)24.4% 
Age at transplant < 0.001
  Median (range)32.5(0.4–74)  34.5(0.2–77) 
Donor type (n,% total) < 0.001
  Identical sibling 9673(74.4%)27 706(67.6%)25.9% 
  Matched unrelated 1821(14.0%) 8695(21.2%)17.3% 
  Mismatched related854 (6.6%) 2191 (5.3%)28.0% 
  Mismatched unrelated655  (5.0%)  2393 (5.8%)21.5% 
Source (n,% total) < 0.001
  Bone marrow 8272(63.6%)25 042(61.1%)24.8% 
  Peripheral blood 4543(34.9%)15 397(37.6%)22.5% 
  Cord blood188 (1.4%)546 (1.3%)25.6% 
Conditioning regimen (n,% total)  0.005
  Myeloablative10 095(87.3%)31 403(86.3%)24.3% 
  Reduced intensity 1465(12.7%) 4982(13.7%)22.7% 
Year of transplant < 0.001
  Median (range)1998(1980–2005)   1999(1980–2005) 

Statistical analysis

The primary endpoint was overall survival; secondary endpoints were cumulative incidences of relapse and TRM (defined as death in remission), and the cumulative incidence of acute and chronic GvHD.

Outcome data were compiled from the date of transplantation through the date of death or date of last contact. Overall survival curves were calculated according to Kaplan and Meier and compared using the log-rank test. Incidence of relapse and TRM were treated as competing risk outcomes and cumulative incidences were calculated accordingly and compared using the Gray test. Similarly, death from any cause was treated as a competing risk for acute and chronic GvHD. All cumulative incidence and survival estimates are at 5 years after transplantation unless otherwise indicated.

Adjusted relative risks for pretransplant risk factors were calculated using Cox proportional hazards models. Pretransplant factors included in the multivariable models were age of patient at transplant, year of transplant, donor type (related vs. unrelated, HLA matched vs. mismatched), donor age, stem cell source and type of conditioning regimen (myeloablative vs. reduced intensity). To separate early from late effects, donor-recipient gender combination was coded as two time-dependent covariates (one from day of transplant up to 6 months, the other covering the time period after 6 months posttransplant). Within these two intervals, the influence of donor-recipient sex combination was verified to be proportional over time. The impact of acute and chronic GvHD was assessed by including acute and chronic GvHD as time-dependent covariates in the Cox models.

To compare the magnitude of the influence of donor/recipient sex matching in different patient populations, we incorporated pretransplant risk factors including FDMR status and the interaction product of FDMR status and the variable of interest (e.g. FDMR status multiplied by diagnosis) into Cox models. If the p-value of such a product was <0.05, we considered the interaction significant.

Patient and transplant characteristics of FDMR transplants were compared to other gender combinations using Pearson's chi-square test and Mann–Whitney U-test where appropriate (Table 1).


Univariate analysis of survival, TRM and relapse

Figure 1 depicts unadjusted overall survival and cumulative incidences of relapse and TRM in the entire cohort. Increase in TRM in FDMR transplants (35.1% vs. 30.2%, p< 0.001) was greater than protection from relapse (25.7% vs. 27.7%, p < 0.001), leading to a net negative effect on survival (43.2% vs. 46.7%, p < 0.001). Survival curves for FDMR and other gender combinations were not distinct in the first 6 months but then separated. Unadjusted rates of survival, TRM and relapse rates in patients stratified by disease, stage, donor age, type of conditioning regimen, HLA compatibility and T-cell depletion are presented in Table 2.

Figure 1.

Overall survival, transplant-related mortality and relapse incidence in female donor/male recipient transplants compared to other gender combinations. Unadjusted survival and cumulative incidences of relapse and transplanted mortality in 53 688 HSCT recipients grouped by donor/recipient sex combination (female donor/male recipient [FDMR] vs. other combinations).

Table 2.  Unadjusted probability of survival, TRM and relapse at 5 years after allogeneic HSCT depending on donor/recipient sex combination
 SurvivalTransplant-related mortalityRelapse
  1. FDMR = female donor/male recipient; Other = other gender combinations.

  Multiple myeloma32.6%35.7%0.0747.7%36.6%0.00331.0%39.3%0.006
Donor age
  <20 years54.0%53.9%0.7220.8%20.4%0.5532.2%32.4%0.58
  >20 years40.2%44.9%< 0.00138.3%31.8%<0.00127.2%30.6%<0.001
HLA match
  Fully matched44.7%48.7%<0.00134.1%28.6%<0.00127.2%29.9%<0.001
Partially mismatched31.6%30.0%0.4145.7%45.2%0.8227.8%29.1%0.19
T-cell depletion

Multivariable analysis of survival,TRM and relapse impact of underlying disease

Multivariable time-dependent Cox models were used to compare the magnitude of FDMR alloreactivity effects in different patient populations, and in the early and late posttransplantation period. Excess TRM in FDMR patients was mostly confined to patients surviving more than 6 months and FDMR showed no interaction with the type of disease patients underwent transplantation for (p = 0.75, Table 3). Underlying disease did modify effects on relapse rates with borderline statistical significance (p = 0.07): Protection was greatest in patients with CML and myeloma (RR 0.79 and 0.82), whereas late relapse rates were not reduced in patients transplanted for lymphoma (RR 1.08). We further subdivided lymphoma patients into those transplanted for chronic lymphocytic leukemia (CLL, n = 627) and other non-CLL lymphoma (n = 2028): no protection from late relapse could be detected in either subgroup (RR 1.14 and 1.06, respectively). To assess disease susceptibility to graft-versus-tumor reactions in patients with lymphoma, we analyzed the impact of acute and chronic GvHD on relapse rates. Development of both acute and chronic GvHD significantly reduced relapse risk in patients with lymphoma (p = 0.003 and p < 0.001), suggesting that the lack of an effect of female-versus-male alloreactivity was not due to general resistance of tumor cells to graft-versus-tumor reactions in patients with lymphoma.

Table 3.  Adjusted hazard ratios of female donor/male recipient transplants in 53 988 HSCT recipients stratified by disease, stage, donor age, conditioning regimen, donor type and graft manipulation
 Survival earlyp-ValueSurvival latep-ValueTRM earlyp-ValueTRM latep-ValueREL earlyp-ValueREL latep-Value
  1. TRM= transplant-related mortality; REL= relapse; early = first 6 months after transplant; late = period starting 6 months after transplant relative risks adjusted for disease, stage, patient age, donor type, year of treatment and type of conditioning (myeloablative vs. reduced intensity).

  2. 1 Interaction between risk factor (e.g. disease type) and gender mismatch.

Disease 0.061 0.0011 0.411 0.261 0.751 0.071
 AML0.94 (0.87–1.01)0.091.16 (1.07–1.26)<0.0011.00 (0.91–1.08)0.901.50 (1.33–1.70)<0.0010.93 (0.84–1.02)0.130.92 (0.84–1.02)0.10
 ALL1.00 (0.93–1.08)0.991.09 (1.01–1.18)0.041.03 (0.94–1.12)0.541.52 (1.32–1.75)<0.0010.89 (0.80–0.99)0.040.85 (0.76–0.96)0.007
 CML1.10 (1.03–1.19)0.011.33 (1.22–1.46)<0.0011.10 (1.02–1.20)0.011.65 (1.47–1.86)<0.0010.95 (0.82–1.10)0.490.79 (0.70–0.88)<0.001
 MDS1.05 (0.93–1.18)0.431.04 (0.88–1.21)0.671.08 (0.94–1.24)0.301.26 (0.98–1.61)0.110.90 (0.74–1.10)0.300.87 (0.73–1.04)0.13
 Lymphoma0.99 (0.87–1.11)0.821.37 (1.17–1.60)<0.0010.98 (0.85–1.13)0.751.52 (1.22–1.90)<0.0011.00 (0.83–1.20)0.961.08 (0.91–1.28)0.36
 Multiple myeloma1.12 (0.90–1.39)0.291.22 (0.96–1.54)0.091.18 (0.94–1.49)0.161.76 (1.24–2.51)0.0020.93 (0.59–1.46)0.760.82 (0.63–1.05)0.11
Stage 0.181 <0.0011 0.721 <0.0011 0.441 0.701
 Early1.03 (0.97–1.10)0.351.32 (1.24–1.41)<0.0011.05 (0.99–1.12)0.131.73 (1.59–1.89)<0.0010.91 (0.82–1.01)0.090.94 (0.88–1.01)0.08
 Intermediate1.07 (0.99–1.14)0.071.10 (1.02–1.19)0.021.08 (1.00–1.17)0.051.45 (1.28–1.63)<0.0010.96 (0.86–1.07)0.430.89 (0.81–0.97)0.008
 Advanced0.96 (0.90–1.03)0.231.03 (0.93–1.13)0.601.01 (0.93–1.09)0.861.18 (1.01–1.38)0.040.92 (0.84–1.00)0.060.92 (0.83–1.01)0.08
Donor age 0.021 0.161 0.011 0.881 0.551 0.591
 <20 years0.86 (0.75–0.99)0.041.11 (0.98–1.25)0.110.86 (0.72–1.01)0.071.65 (1.31–2.09)<0.0010.96 (0.81–1.13)0.610.96 (0.85–1.09)0.53
 >20 years1.03 (0.98–1.10)0.271.21 (1.14–1.29)0.0011.08 (1.01–1.15)0.031.55 (1.41–1.70)<0.0010.90 (0.82–0.99)0.030.92 (0.86–0.98)0.01
Conditioning 0.0061 0.291 0.041 0.051 0.031 0.111
 MAC1.00 (0.96–1.04)0.981.20 (1.14–1.26)<0.0011.03 (0.99–1.08)0.171.64 (1.53–1.76)<0.0010.91 (0.85–0.98)0.0080.90 (0.85–0.95)<0.001
 RIC1.19 (1.06–1.35)0.0031.11 (0.97–1.27)0.351.21 (1.05–1.40)0.0081.25 (1.00–1.57)0.051.06 (0.92–1.21)0.040.99 (0.87–1.14)0.94
Donor type 0.0041 0.0091 0.171 0.0011 0.501 0.471
 Matched1.04 (1.00–1.09)0.041.20 (1.15–1.26)<0.0011.07 (1.03–1.13)0.0021.63 (1.53–1.75)<0.0010.94 (0.88–1.00)0.040.91 (0.87–0.96)<0.001
 Mismatched0.89 (0.82–0.99)0.030.99 (0.86–1.17)0.500.96 (0.81–1.26)0.580.99 (0.75–1.31)0.940.88 (0.75–1.03)0.120.97 (0.83–1.14)0.79
T-cell depletion 0.281 0.701 0.121 0.071 0.061 0.011
 No1.05 (0.88–1.01)0.081.19 (1.12–1.27)<0.0011.09 (1.03–1.16)0.0031.57 (1.44–1.72)<0.0010..93 (0.86–0.99)0.040.90 (0.84–0.96)0.001
 Yes1.00 (0.91–1.09)0.971.19 (1.07–1.31)0.0011.00 (0.90–1.10)0.951.35 (1.16–1.58)<0.0011.04 (0.94–1.15)0.461.06 (0.95–1.18)0.30

Effects varied with disease stage. The relative risk of late TRM in FDMR patients was greatest in early-stage patients and gradually declined with advancing disease (p< 0.001). In contrast, the relative protection from relapse remained independent from disease stage (p = 0.70). Due to the declining excess in TRM, and due to a greater absolute risk of disease relapse in patients with more advanced disease, the net negative effect on survival gradually decreased with advancing disease stage. In the highest risk patients, survival was no longer different as the effects of female-versus-male alloreactivity on TRM and relapse rate compensated each other.

Impact of donor age and conditioning regimen

Donor age—used as surrogate marker for pregnancy history that was unfortunately not available—affected transplant in a time-dependent manner: In recipients of adult donor grafts, FDMR transplants were associated with a significantly increased early TRM (RR 1.08) compared to other gender combinations. No such excess early TRM was observed in FDMR transplants from female donors younger than 20 years (p = 0.01 vs. adult donors, Table 3). In contrast, hazard ratios for late TRM, relapse rate and survival in FDMR patients were not significantly different between recipients of transplants of donors younger or older than 20 years. Although not statistically significant (p = 0.08), in the adult donor transplant population excess late TRM in FDMR transplants gradually decreased with increasing donor age (20–25 years 2.39, 25–30 years 1.76, 30–40 years 1.61, >40 years 1.44). No such trend was seen for relapse and no differences were seen within the group of patients receiving transplants from donors younger than 20 years of age.

Type of conditioning regimen also influenced female-versus-male alloreactivity. Early TRM was increased significantly more in FDMR patients after reduced intensity conditioning (RR 1.21 vs. 1.03 for myeloablative conditioning, p = 0.04), whereas late excess TRM was more frequent in FDMR patients after myeloablative than after reduced intensity conditioning (RR 1.64 vs. 1.25, p = 0.05, Table 3).

Increasing use of reduced intensity conditioning regimens was also responsible for a decrease in overall hazard ratios for TRM and survival during the observation period. However, when analysis was stratified by intensity of conditioning regimen, no interaction was seen between FDMR status and year of transplant for all endpoints (data not shown).

Impact of HLA matching and T-cell depletion

Stratification of patients by degree of HLA match showed that excess TRM and protection from relapse in FDMR patients was confined to patients receiving fully HLA-matched grafts. No differences were seen between matched related and matched unrelated donors. Data on ex vivo T-cell depletion were available 41 112 patients. Within this cohort, FDMR patients receiving T-cell depleted grafts did not benefit from late protection from relapse (HR 1.06 vs. 0.90 for T-cell replete grafts, p = 0.01). Late excess TRM was also reduced (HR 1.35 vs. 1.57 for T-cell replete grafts), resulting in similar hazard ratios for late mortality (1.19 for FDMR vs. other patients in both T-cell depleted and T-cell replete transplants).


Both acute and chronic GvHD were more frequent in FDMR patients than in control patients. Cumulative incidence of grade II–IV was 29% in FDMR transplants compared to 26% in other gender combination transplants (p < 0.001). Cumulative incidence of chronic GvHD was 43% in FDMR transplants compared to 35% in controls (p < 0.001). The excess chronic GvHD associated with FDMR transplants was almost exclusively graded as extensive, whereas limited chronic GvHD was only slightly increased (limited 21% vs. 20%, extensive 21% vs. 15% for FDMR and other gender combinations, respectively). In multivariable analysis, adjusting for disease, stage, year of transplant, donor type and type of conditioning regimen, the relative risk of chronic GvHD associated with FDMR transplants (RR 1.55, 95% CI 1.47–1.64) exceeded the relative risk of grade II–IV acute GvHD (RR 1.12, 95% CI 1.07–1.17).

Risk of acute GvHD in FDMR patients was modified by the age of the donor: male patients with a female donor less than 20 years of age had a relative risk of only 1.03 compared to other gender combinations, those receiving grafts from a female older than 20 years of age had a relative risk of 1.14 (p = 0.005 vs. younger donor FDMR transplants). In contrast, donor age played no role in the excess chronic GvHD seen in FDMR patients (RR donor < 20 years 1.56 vs. 1.56 for donors > 20 years, p = 0.99).

Excess acute GvHD associated with FDMR transplants was more frequent after reduced intensity than after myeloablative transplants (RR 1.21 vs. 1.03, p = 0.04). In contrast, chronic GvHD in FDMR patients was increased more significantly after myeloablative than after reduced intensity conditioning (RR 1.59 vs. 1.26, p = 0.01). Both early and late TRM were increased in FDMR transplants to a higher degree after HLA matched than after HLA-mismatched transplantation. T-cell depletion had a protective effect on development of acute and chronic GvHD (RR 0.60 and 0.75, respectively). However, within the population of patients receiving T-cell depleted grafts, acute and chronic GvHD were more frequent in FDMR patients, and the relative risk versus other gender combinations was comparable to that found in T-cell replete grafts (p = 0.67 and 0.53 for acute and chronic GvHD).

To assess the causal relationship among excess GvHD, increased TRM and protection from relapse in FDMR patients, we used Cox models integrating acute and chronic GvHD as time-dependent covariates. After adjusting for acute and chronic GvHD, relative risks for late TRM remained highly elevated whereas those for late relapse partially corrected (Table 4).

Table 4.  Hazard ratios of transplant-related mortality and relapse in female donor/male recipient patients adjusted for acute and chronic graft-versus-host disease
 TRM latep-ValueREL latep-Value
  AML1.44 (1.24–1.67)<0.0011.05 (0.95–1.16)0.36
  ALL1.27 (1.06–1.52)0.0090.92 (0.83–1.03)0.14
  CML1.62 (1.41–1.86)<0.0010.90 (0.81–1.00)0.05
  MDS1.21 (0.91–1.63)0.190.85 (0.68–1.04)0.12
  Lymphoma1.46 (1.12–1.91)0.0061.03 (0.84–1.27)0.76
  Multiple myeloma1.55 (1.01–2.39)0.040.88 (0.66–1.18)0.39

Donor selection trends

Analysis of the frequency of FDMR fraction of all transplants showed that in unrelated transplants, the relative frequency of FDMR started to decline from a baseline of around 25% (which would be expected if donor-recipient sex combination were governed by chance only) in the second half of the 1990s. In the period since 2000, FDMR transplants accounted for 11.8% of unrelated donor (UD) transplants for AML, 17.8% of UD transplants for ALL, 15.2% of UD transplants for lymphoma, 24.3% of UD for multiple myeloma, 13.7% of UD transplants for MDS, and 14.8% of UD transplants for CML. In contrast, in patients transplanted from sibling donors, frequency of FDMR transplants was near 25% (AML 23.4%; ALL 27.7%; lymphoma 29.3%; multiple myeloma 24.8; MDS 24.4%; CML 24.8%). Figure 2 graphically depicts the changing pattern in UD selection trends for acute leukemia and chronic myeloid leukemia along with stable rates of FDMR frequencies in related donor transplantation.

Figure 2.

Evolution of the proportion of female donor/male recipient transplants over time. Proportion of female donor/male recipient pairs among all gender combinations in HLA-identical sibling and unrelated donor (UD) transplantation for chronic myeloid and acute leukemia. A proportion of 25% is expected if the gender combination is governed by chance only. The proportion of the FDMR combination in HLA-identical sibling transplants is stable throughout the observation period, whereas rates of FDMR transplant have continually declined since 1995 in UD transplantation.


This retrospective analysis in a large cohort of patients receiving allogeneic HSCT defines the role of gender mismatching in the outcome after allogeneic HSCT for hematological malignancies. Previous data have shown an increase in TRM, a relative protection from relapse and a negative effect on survival in patients with CML transplanted in first chronic phase. Our study extends these findings to all hematological malignancies treated with allogeneic HSCT today. In a cohort of more than 50 000 transplants reported to the EBMT, we show that the female donor/male patient gender combination is universally associated with an increase in late TRM that coincides with an increase in chronic GvHD. The increased risk for TRM must be weighed against a relative protection from relapse. However, in the cohort studied, we were not able to identify a population of patients where protection from relapse outweighs the increased risk of TRM. Even in patients with advanced stage, the survival gained by protection from relapse at most compensates the excess TRM. If the excess risk of extensive chronic GvHD in FDMR patients is additionally taken into account, it is apparent that such a gender combination should be avoided, if possible, irrespective of underlying disease and disease stage.

The large number of patients included in this study allowed us to compared effects of female-versus-male alloreactivity in different subpopulations. Underlying disease had no appreciable effect on the excess risk of TRM associated with FDMR transplants. In contrast, relative risks of relapse—which may be interpreted as degrees of susceptibility of underlying diseases to female-versus-male alloreactivity—varied considerably. In agreement with previous data, diseases most susceptible to female-versus-male alloreactivity were chronic myeloid leukemia and myeloma. We were not able to detect an effect of gender mismatching on the relapse rate in patients transplanted for lymphoma. This appeared not due to refractoriness to graft-versus-tumor reactions in lymphoma patients, as relapse rates were strongly associated with incidence and severity of acute and chronic GvHD. To our knowledge, the question whether H-Y minor antigens are expressed on malignant lymphoid cells has not been investigated.

Effects of H-Y graft-versus-host reactions seemed mainly confined to patients surviving more than 6 months after transplant. In parallel, the relative risk of chronic GvHD was increased more profoundly than that of acute GvHD. While T-cell depletion reduced the overall incidence of both acute and chronic GvHD, the relative risk of GvHD in FDMR patients receiving T-cell depleted grafts was comparable to that of FDMR patients after T-cell replete allogeneic HSCT. Rates of excess TRM and overall mortality associated with FDMR transplants were not statistically correlated with T-cell depletion. This may seem unexpected, as both GvHD and graft-versus-leukemia effects have traditionally been understood as driven primarily by T lymphocytes. The absence or lower intensity of pharmacological GvHD prophylaxis in T-cell depleted transplants may partially explain why significant effects of female-versus-male alloreactivity are also detected in T-cell depleted transplants. Alternatively, humoral responses against H-Y antigens—which have recently been shown to correlated with both chronic GvHD and protection from relapse in FDMR patients (18)—might be implicated in female-versus-male alloreactivity in T-cell depleted allogeneic HSCT.

In contrast, transplantation from an HLA-mismatched donor abolished measurable effects of female-versus-male alloreactivity on transplant outcome. Alloreactive T lymphocytes against mismatched HLA antigens can be detected at a much higher frequency than those directed against mismatched minor antigens, which may explain why appreciable effects of female-versus-male alloreactivity were confined to patients receiving HLA-identical grafts. Alternatively, the minor histocompatibility-driven immune response might depend on matching for major histocompatibility antigens.

Pregnancy history was not available in the donors of this cohort. We therefore used donor age as a surrogate marker, assuming that the vast majority of donors with a history of pregnancy would be 20 years of age or older at donation, and that the frequency of female donors sensitized to H-Y would be low in donors younger than 20 years of age. Excess acute GvHD and early TRM in FDMR transplants was indeed confined to patients grafted from an adult female donor, whereas chronic GvHD and late TRM was not affected by donor age, suggesting that sensitization increases early alloimmune effects but is not necessary for late effects. These data are consistent with a small study in patients with aplastic anemia that showed an increase in acute GvHD in patients transplanted from female donors with a history of pregnancy (19).

In contrast to patients receiving myeloablative transplants, where effects of female-versus-male alloreactivity were almost exclusively seen in the period starting 6 months after transplantation, an increase in TRM was already detected in the first 6 months after transplantation in reduced-intensity transplants. This difference is likely explained by the fact that immunosuppressive treatment is tapered slowly and carefully after myeloablative transplantation, whereas immunosuppressive agents are typically discontinued earlier and more abruptly after reduced-intensity conditioning in order to induce graft-versus-leukemia reactions.

To assess whether excess TRM and the relative protection from relapse in FDMR transplants were driven exclusively by the higher incidence of acute and chronic GvHD, we used multivariable models integrating acute and chronic GvHD as time-dependent covariates. Adjusting for GvHD only partially corrected the FDMR hazard ratios for TRM and relapse, indicating that excess TRM and protection from relapse were at least partially independent from clinically recognized GvHD. Similar data for the GvHD-adjusted risk of relapse have been reported earlier (17) and graft-versus-host reactions that specifically target host hematopoietic cells have been suggested to account for this. However, our data indicating that H-Y mismatching also remains a strong predictor of TRM after adjusting for acute and chronic GvHD would rather suggest that clinically significant graft-versus host reactions occur that are not recognized by the current grading criteria.

In conclusion, we show in a large cohort of patients treated with allogeneic HSCT that H-Y mismatching in FDMR transplants impacts on rates of relapse, TRM and survival. The size of the population analyzed allowed us to define the magnitude of H-Y-associated effects. We were not able to identify a population where the protection from relapse outweighed excess TRM. Survival is at best equal. If incidence of chronic GvHD is additionally taken into account, there is no population where preferentially choosing a female donor for a male patient seems justified. Analysis of the proportion of FDMR transplants over time indicates that in unrelated transplantation, the proportion of FDMR transplants is significantly lower than would be expected by chance. No such trend was seen for identical sibling transplants, where typically at most one HLA-matched sibling is available per patient. These data suggest that transplant physicians actively try to avoid an FDMR situation in the unrelated donor setting whenever possible. The results of the current analysis support this strategy. It remains open, whether a male matched unrelated donor might be the preferred choice for a male patient with a female sibling donor only.


The cooperation of all participating teams and their staff (listed in the Appendix), the EBMT Co-ordination office; Barcelona (F. McDonald, E. McGrath, S. M. Jones, E. J. Mac Hale), Paris (V. Chesnel, C. Kenzey, C. Durand, N.C. Gorin), London (C. Ruiz de Elvira, S. Hewerdine, S. de Souza, N. Fortin-Robertson), the Austrian Registry (H. Greinix, B. Lindner), the Czech Registry (K. Benesova, M. Trnkova), the French Registry SFGM (D. Blaise, C. Raffoux, Z. Chir), the German Registry (H. Ottinger, K. Fuchs, C. Müller, S. Allgaier, A. Müller), the Italian Registry (A. Bacigalupo, R. Oneto, B. Bruno), the Dutch Registry (A. Schattenberg, A. V. Biezen, M. Sneets, R. Brand), the Spanish Registry (J. Rifon, A. Cedillo, J. López), the Swiss Registry (U. Schanz, H. Baldomero, E. Buhrfeind), the Turkish Registry (G. Gurman, M. Arat, F. Arpaci, M. Ertem) and the British Registry (C. Craddock, J. Cornish, K. Towlson, M. Wilson) is greatly appreciated. The authors also thank S. Stöckli for excellent secretarial assistance, as well as L. John for technical assistance with data management.

The work was supported in part by the 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. EBMT is supported by grants from the corporate members: Amgen Europe GmbH, F. Hoffmann-La Roche, Gilead Sciences, Miltenyl Biotec GmbH, Schering-Plough International Inc., Celegene International SARL, Chugai sanofi—aventis SNC, Fresenius Biotech GmbH, Gambro BCT, Genzyme, Pfizer, Berlex AG (Schering AG Germany), Therakos, Bristol Myers Squibb, Novartis, Cephalon, Laboratoires Pierre Fabre and GE Healthcare.


Supporting Information

The following supporting information is available in the online version of this article:

List of Participating Centers

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.