Prevalence and risk factors of having antibodies to class I and II human leukocyte antigens in older haploidentical allograft candidates

The effect of donor-specific anti-human leukocyte antigen (HLA) antibodies (DSAs) has been recognized as a factor in graft failure (GF) in patients who underwent umbilical cord blood transplantation (UBT), matched unrelated donor transplantation (MUDT), or haploidentical stem cell transplantation (haplo-SCT). Presently, we know little about the prevalence of and risk factors for having anti-HLA antibodies among older transplant candidates. Therefore, we analyzed 273 older patients with hematologic disease who were waiting for haplo-SCT. Among all patients, 73 (26.7%) patients had a positive panel-reactive antibody (PRA) result for class I, 38 (13.9%) for class II, and 32 (11.7%) for both. Multivariate analysis showed that females were at a higher risk for having a PRA result for class II (P = 0.001) and for having antibodies against HLA-C and HLA-DQ. Prior pregnancy was a risk factor for having a PRA result for class I (P < 0.001) and for having antibodies against HLA-A, HLA-B and HLA-DQ. Platelet transfusions were risk factors for the following: having a positive PRA result for class I (P = 0.014) and class II (P < 0.001); having antibodies against HLA-A, HLA-B, HLA-C, HLA-DP, HLA-DQ, and HLA-DR; and having higher mean fluorescence intensity (MFI) of PRA for class I (P = 0.042). In addition, previous total transfusions were at high risk for having higher numbers of antibodies to specific HLA loci (P = 0.005), and disease course (7.5 months or more) (P = 0.020) were related to higher MFI of PRAs for class I. Our findings indicated that female sex, prior pregnancy, platelet transfusions and disease courses are independent risk factors for older patients with hematologic disease for having anti-HLA antibodies, which could guide anti-HLA antibody monitoring and be helpful for donor selection.

www.nature.com/scientificreports www.nature.com/scientificreports/ as previously reported 9 . The mean fluorescence intensity (MFI) of the anti-HLA antibodies was adjusted for the background signal using the formula described previously. Samples with an MFI of 500 or more considered to be positive.

Statistical analysis.
Descriptive statistics, including the frequency (proportions) for categorical variables and median (range) or cut-off value determined by a ROC curve for quantitative variables, were used to describe the patient demographic and clinical characteristics. A chi-square test (and Fisher's exact test, when appropriate) was used for group comparisons. Binary logistic regression tests with either univariate or multivariate analyses were performed to estimate the potential influence of the factors on anti-HLA antibodies. A two-sided P-value less than 0.05 was considered statistically significant. Statistical calculations were performed with SPSS, version 16.0 (Mathsoft, Seattle, WA, USA).

Results
Patient characteristics. The demographics and clinical characteristics are outlined in Table 1. In this study, 273 transplant candidates on waiting lists for haplo-SCT were analyzed for the presence of anti-HLA antibodies, including 165 males (60.4%) and 108 females (39.6%). There were 124 patients (45.4%) with acute myeloid leukemia (AML), 47 patients (17.2%) with acute lymphoblastic leukemia (ALL), 66 patients (24.2%) with myeloid dysplastic syndrome (MDS), 10 patients (3.7%) with chronic myelomonocytic leukemia (CMML) and 26 patients (9.5%) with other hematological diseases. The median age of the patients was 54 years (range, 50-66 years). The median MFI of panel-reactive antibodies (PRAs) for class I and class II were 2902 (range: 558-22827) and 4332 (range: 516-18363), respectively. Table 1 Table S1). The risk factors associated with anti-HLA class I PRAs were sex, the number of pregnancies, number of total transfusions, number of RBC transfusions, number of PLT transfusions and disease course. Sex, the number of pregnancies, number of total transfusions, number of RBC transfusions and number of PLT transfusions were related to having anti-HLA class II PRAs. Sex, the number of pregnancies, number of total transfusions, number of RBC transfusions, number of PLT transfusions and disease course were related to having anti-HLA class I or II PRAs. Meanwhile, sex, the number of pregnancies, the number of total transfusions, number of RBC transfusions and number of PLT transfusions were associated with having anti-HLA class I and II PRAs. Multivariate analysis showed the following: (i) both prior pregnancy and PLT transfusions (7 or more) were associated with the prevalence of PRAs for class I (P < 0.001, P = 0.014) and for class I or II (P < 0.001, P = 0.006); (ii) both female sex and PLT transfusions (7 or more) were associated with a higher prevalence of PRAs for class II (P = 0.001, P < 0.001) and for class I and class II (P = 0.002, P < 0.001) ( Table 2). Fig. 1 Table S2). Multivariate analysis showed that the following: i) female sex was associated with the development of antibodies against HLA-C (P = 0.003) and -DQ (P = 0.048); ii) prior pregnancy was independently and significantly related to the development of antibodies against HLA-A (P < 0.001), -B (P < 0.001) and -DQ (P = 0.001); iii) PLT transfusions (7 or more) were associated with having a higher prevalence of antibodies against HLA-A (P = 0.011), -B (P = 0.021), -C (P = 0.010), -DP (P < 0.001), -DQ (P < 0.001) and -DR (P < 0.001). In addition, it seemed that gender had a trend towards influencing the development of antibodies against HLA-DQ (P = 0.055) ( Table 3).

Prevalence of antibodies against HLA-A, -B, -C, -DP, -DQ and -DR loci. As shown in
Prevalence of and risk factors associated with the number of antibodies specific for HLA-A, -B, -C, -DR, -DQ and -DP. There were 79 patients with positive anti-HLA antibodies that were specific for single HLA loci, and the median number of antibodies to specific HLA loci was 44, with a range of 3 to 155 (Fig. 2). To investigate the factors associated with the number of anti-HLA antibodies, all 79 patients were classified into two groups according to the median number of antibodies. In univariate analysis, previous total transfusions, RBC transfusions and PLT transfusions were associated with the number of antibodies (Supplementary  Table S3). Meanwhile, previous total transfusions (12 or more) (HR 4.041, 95% CI 1.442-11.322, P = 0.008) was independently and significantly associated with having higher numbers of anti-HLA antibodies based on multivariate analysis.

Prevalence of and risk factors associated with MFI of PRAs for class I and class II.
There were 73 patients with positive PRAs for class I and 38 patients for class II. A prospective clinical study of our group had concluded that donor-specific anti-HLA antibodies (MFI >/ = 2000) were strongly associated with primary poor graft function (P = 0.005) 9 . So we divided patients with MFI of 500 or more into two subgroups according to the cut-off value of 2000. Univariate analysis showed that previous total transfusions, RBC transfusion and disease course were associated with MFI of PRAs for class I, meanwhile, previous total transfusions, PLT transfusions and disease course were associated with MFI of PRAs for class II. Multivariate analysis showed that previous transfusions (12 or more) (HR 3.529, 95% CI 1.119-11.133, P = 0.031) and disease course (7.5 months or more) (HR 3.389, 95% CI 1.212-9.477, P = 0.020) were related to higher MFI of PRAs for class I. In the meanwhile, PLT transfusions (7 or more) (HR 5.833, 95% CI 1.061-32.020, P = 0.042) were correlated with MFI of PRAs for class II in multivariate analysis. (Supplementary Table S4).

Discussion
In this study, we demonstrated the following: i) female sex was associated with having anti-HLA class II antibodies and antibodies to specific loci, including HLA-C and -DQ; ii) prior pregnancy was a risk factor for having anti-HLA class I antibodies and antibodies to specific loci, including HLA-A, -B and -DQ; and iii) PLT transfusions (7 or more) were related to having anti-HLA class I, and II antibodies as well as antibodies to specific loci, including HLA-A, -B, -C, -DP, -DQ and -DR. In the meanwhile, previous total transfusions were also an independent risk factor for having higher numbers of anti-HLA antibodies and higher MFI of PRAs for class I. In addition, disease course (7.5 months or more) were related to higher MFI of PRAs for class I; while PLT transfusions (7 or more) were related to higher MFI of PRA for class II. The results were consistent with the findings of previous studies [19][20][21]23 and are useful for haplo-SCT donor selection.
In this study, the prevalence of anti-HLA antibodies was 28.9% (79/273) among all older patients, which was higher than previous studies of the general population (21%) 21 and pediatric candidates (13%) 22 with hematologic disease at our center. De Clippel et al. 24 concluded that in people with a history of transfusion, the overall alloimmunization rate against class I and II HLA antigens was 20.2%. Triulzi et al. 25 found that HLA antibodies were detected in 17.3% of all female volunteer blood donors and 1.7% of transfused male volunteer blood donors. All of the abovementioned proportions of anti-HLA antibodies were lower than those observed in the results of our study. However, Yee et al. 26 found that in pediatric patients with thalassemia major, HLA antibodies were    www.nature.com/scientificreports www.nature.com/scientificreports/ detected in 10 of 19 (53%) subjects. The differences in the proportion of anti-HLA antibodies between our study and others' studies may be because of the type of disease, sex, age, and the number of previous pregnancies and transfusions.
The association between sex and having anti-HLA antibodies has been previously reported [27][28][29][30] . Gladstone et al. 30 retrospectively reviewed 957 donors to evaluate the incidence of anti-HLA antibodies and DSAs and found that the incidence of anti-HLA antibodies was higher among females than males (43.2 versus 10.8%, P < 0.0001), and DSAs were most commonly detected in female patients (30.6% versus 4.9%, P < 0.00001) among candidates for haplo-SCT. Furthermore, Hyun et al. 28 concluded that female solid organ transplantation candidates had a significantly higher PRA-positivity rate compared with that of male patients (60.3% vs 34.2%; P < 0.001). Nguyen et al. 29 also found that women were at a higher risk of antibody-mediated rejection than men, and this increased risk was additive to that of preformed donor-specific antibodies after heart transplantation. Therefore, sex can be regarded as a specific risk factor for having anti-HLA antibodies and could guide their detection before SCT.
Almost all of the recent studies have confirmed that pregnancy was a risk factor for having anti-HLA antibodies 31 ; additionally, our study also showed that pregnancy was not only associated with having PRAs for class I but was also correlated with having antibodies against different antigens encoded by the different HLA loci, including HLA-A, -B and -DQ. Masson et al. 32 showed that allo-immunized mothers presented more anti-HLA class I antibodies more frequently (86%) than they presented anti-class II antibodies (62%). This may explain why pregnancy is a risk factor for PRAs for class I rather than for class II based on the multivariate analysis of our study. In addition, our study showed that the incidences of PRAs for class I and II increased with the number of pregnancies as follows: 7.1% (zero), 14.8% (one) and 25.0% (two or more). Female patients are exposed to antigens during pregnancy 31 , and the exposure intensity increases as the number of pregnancies increases, which suggested to us that postpartum women should be monitored for anti-HLA antibodies routinely, especially among women who have had multiple pregnancies.
In the present study, the association of transfusion with a higher incidence of anti-HLA antibodies was also confirmed. First, PLT transfusions were the risk factors for having PRAs for class I and II, and were correlated with having antibodies against different antigens, including anti-HLA-A, -B, -C, -DP, -DQ, and -DR. Second, when patients had undergone more transfusions, they had higher numbers of antibodies to specific HLA loci. Huo et al. 21 found that prior transfusions (6 or more) is a risk factor associated with the prevalence of having a positive PRA either for class I or class II HLA in haploidentical transplant candidates. Hung et al. 19 observed that having undergone a recent transfusion but not having undergone a transfusion itself was independently correlated with HLA sensitization in patients with end-stage renal disease. However, most studies concluded that transfusion had little impact on anti-HLA antibodies [23][24][25]33 . In contrast to previous studies [23][24][25]33 , but in agreement with others 34 , we found that PLT transfusions were not only associated with PRAs for class I, but also correlated to PRAs for class II. The higher white blood cells contained in platelet components than that of the RBC components may account for above mentioned association 35 .
Previous studies have demonstrated that DSAs were associated with GF in both haplo-SCT 9,36,37 and MUDT 12 , and the incidence of GF increases as the MFI increased 9,36 . In this study, we found that previous total transfusions (12 or more) and disease course (7.5 months or more) were related to higher MFI of PRAs for class I, while PLT transfusions (7 or more) were related to MFI of PRAs for class II. The results in Supplementary Table S4 showed that female, multiple pregnancies, previous total transfusion, RBC transfusions or PLT transfusions and long disease courses had a tendency for high MFI, which need a larger sample size of patients to explore.
This study had several limitations. First, this was a single center study. Second, our research is limited to Chinese populations, and the possibility of extending these findings to other populations, such as white, black, and Hispanic populations, needs to be investigated. Thus, a multicenter study with a large sample size is warranted.
In summary, the results of this study suggested that female sex, prior pregnancy and PLT transfusions are independent risk factors for the generation of anti-HLA antibodies in haploidentical transplant candidates aged more than 50 years. These findings could guide clinical monitoring for HLA antibodies and may contribute to haploidentical donor selection.