Introduction

In hematopoietic stem cell transplantation, a close HLA match between donor and recipient correlates positively with transplantation outcome.^{1, 2, 3, 4} Since exact matching is hampered by the extreme diversity of the HLA system, large files of potential stem cell donors were established. Now, about 11 500 000 potential stem cell donors are registered worldwide⁵ and recruitment efforts are still ongoing. It is, however, well known that the efficiency of ongoing donor recruitment declines as the number of already registered donors grows.^{6, 7, 8, 9, 10} This has consequences for strategic donor center or registry planning as donor recruitment leads to considerable costs, mainly for HLA typing. Since donor centers and registries have limited resources, it is important to ask whether these resources should be used for ongoing donor recruitment or other donor center or registry operations.

Here, we focus on the effects of donor aging on the need for ongoing donor recruitment. There is a continuous loss of registered donors due to age limits for unrelated stem cell donation. However, it may not be sufficient just to replace these donors to maintain the usefulness of a donor file since there is evidence that the outcome of hematopoietic stem cell transplantation is better for younger donors.^{11, 12, 13, 14} Therefore, even aging of those donors who are far below the age limit may lead to a decrease of donor file usefulness. We verify and quantify this effect and estimate the number of donors who have to be added annually to a given donor pool to compensate for this reduction in usefulness.

Methods

Donor file usefulness

We used the number of donations that can be expected from a donor file annually as indicator for its usefulness. This number is the sum of the annual donation probabilities of all individual donors of the file. It is the central goal of our study to analyze effects of aging on donor file usefulness. However, other donor characteristics also might affect individual donation probabilities and were therefore included in the analysis. These factors are donor gender, completeness of available HLA results, and HLA phenotype frequencies.

Completeness of available HLA results

The DKMS donor file includes donors with different levels of available HLA information. All donors are at least typed at recruitment for the HLA-A and -B loci with low/intermediate resolution. Generally, there are three ways—and combinations of these ways—how additional HLA information (higher resolution and/or additional loci) is obtained for DKMS donors: (1) additional HLA typing is carried out at recruitment; (2) additional HLA typing of already registered donors is carried out without external patient-related typing requests after recruitment; (3) additional HLA typing is carried out within the scope of donor searches (for example, requests for confirmatory typing). The available HLA information of an individual donor may accumulate through combinations of these three ways over several years. It is known that the availability of more comprehensive HLA results of registered donors correlates positively with their probability of being selected to donate.^{15, 16} Here, we differentiate between donors who have been typed beyond the minimum level of available HLA information (that is, beyond low/intermediate resolution typing of the HLA-A and -B loci) without patient-related requests and donors without such 'prospective' typing. It has to be noted that the group of prospectively typed donors is heterogeneous with respect to the completeness of HLA results.

Probabilities for being typed prospectively depend on donor age, gender and HLA phenotype frequencies. These factors were used as selection criteria within several DKMS prospective typing projects during the last years. As a consequence, HLA results of DKMS donors are, on average, more complete for younger donors, male donors and donors with common HLA-AB phenotypes.

HLA phenotype frequencies of registered donors

We considered HLA-AB phenotype frequencies in order to include all DKMS donors into the analysis. All donor HLA results from DNA-based methods were converted into serological (split or broad) values. For calculations regarding phenotype frequencies, split and broad antigens were treated as independent of each other as described elsewhere.¹⁷

Regression analysis

A multiple logistic regression analysis was carried out to determine the influence of donor age and gender, completeness of HLA results and HLA phenotype frequencies on donation probabilities. Factors were dichotomized as follows: donors aged 18–29 years versus donors aged 30 years, male versus female donors, donors with some prospective HLA typing versus donors without such typing and donors with HLA-AB phenotypes that are available 11 times in the DKMS donor file versus donors whose phenotypes are available 10 times or less. SPSS 14.0 was used for calculations.

Calculation of donation probabilities by donor age

We estimated annual donation probabilities of the donors of an existing file from real data as follows: we defined a time interval T₁ of length T₁. For every donation within this interval, we determined the donor age at the day of donation. Besides, we determined donor's gender, completeness of HLA results and HLA phenotype frequency. All factors apart from donor age were dichotomized as described in Regression analysis. d(i,j,k,l) indicates the number of donations from donors of age i with gender j (1=male, 0=female), typing level k (1=some prospective typing, 0=no prospective typing) and HLA-AB phenotype frequency indicator l (1=HLA-AB phenotype frequency 11, 0=HLA-AB phenotype frequency 1–10) within T₁. We estimated n(i,j,k,l), the average number of donors of age i, gender j, typing level k and HLA-AB phenotype frequency indicator l, during T₁ by the corresponding number at the center of the interval. Then the donation probabilities p(i,j,k,l) in the time interval T₁ can be estimated as p(i,j,k,l)=d(i,j,k,l)/n(i,j,k,l). Accordingly, the annual donation probabilities were approximated by p^A(i,j,k,l)=d(i,j,k,l)/(n(i,j,k,l) T₁) where T₁ is given in years.

For the graphical representation of donation probabilities by donor age this approach led to eight donor sets G(j,k,l). G(0,1,0), for example, indicates the group of all female donors (j=0) with prospective typing (k=1) and rare HLA-AB phenotypes (l=0).

Aging-induced reduction of donor file usefulness

The expected annual donation number from the donor pool was D₀=_i,j,k,ln(i,j,k,l)p^A(i,j,k,l). Here, the summation index i for donor age ran from a₁ to a₂, where a₁ and a₂ indicate the minimum and maximum donor age, respectively. Since in Germany registered donors must be 18 years or older and are removed from the donor file at their 61st birthday, we set a₁=18 and a₂=60 in the calculations. The other summation indices ran from 0 to 1 as the corresponding factors have been dichotomized. If we assume that annual donation probabilities remain constant, only donors who reach the age limit are removed from the file, no further recruitment takes place, and no additional prospective typing are carried out, the expected annual number of donations after m years will be D_m=_i,j,k,ln(i,j,k,l)p^A(i+m,j,k,l). All p^A(i+m,j,k,l) with i+m>a₂ are equal to 0 since the corresponding donors are not in the file anymore. We neglected possible changes of HLA-AB phenotype frequencies through the loss of donors who reach the age limit.

Conditions for constant donor file usefulness

If annual donation probabilities depend on donor age, gender, completeness of HLA results and HLA phenotype frequencies, then the number of donors who must be recruited to compensate for the aging-related reduction in usefulness of a donor file depends on the composition of the group of new donors with respect to these factors. For an existing donor file, we assume that the current age, gender and HLA phenotype frequency distributions of newly recruited donors will also apply in the future. Since the completeness of HLA results of new donors depends on the future strategy of the donor center or registry, we differentiate between two scenarios: in scenario A all new donors are typed prospectively beyond an intermediate typing of the HLA-A and -B loci (k=1); in scenario B the newly recruited donors are not typed prospectively (k=0). We define a time interval T₂ of length T₂ that can but does not have to be identical with T₁. r indicates the total number of donors who are recruited in T₂, r(i,j,k,l) the number of new donors of age i, gender j, typing level k and HLA-AB phenotype frequency indicator l, and f(i,j,k,l)=r(i,j,k,l)/r the fraction of newly recruited donors with respective characteristics. The corresponding annual quantities are r^A, r^A(i,j,k,l) and f^A(i,j,k,l)=f(i,j,k,l). Based on the annual donation probabilities defined above, the recruitment of r^A new donors in a given year will lead to N_n=r^A_i,j,lf^A(i,j,k,l)p^A(i+n,j,k,l) additional donations n years after the recruitment year. Here k is no summation index but k=1 (k=0) in scenario A (B). The equation can be used to determine the number of new donors who must be recruited annually to keep the usefulness of a donor file constant over a time period of several years.

HLA phenotype frequencies of stem cell donors by age

To test the hypothesis that younger donors are preferred when several donors are available, we analyzed HLA phenotype frequencies of stem cell donors who donated in T₁. For every donation, the frequency of the donor's HLA-ABDR phenotype in the DKMS donor file was determined. For several donor age groups (18–29 years, 30–44 years, 45 years), we analyzed how many donors had HLA phenotypes with certain frequencies in the DKMS donor file (unique, 2–10, 11–100, >100). Here, we consider HLA-ABDR phenotypes as all analyzed donors have donated stem cells and are, therefore, typed at least for the HLA-A, -B and -DR loci.

Empirical data

Calculations were carried out by using real data from the donor file of DKMS German Bone Marrow Donor Center. The file consists mainly (about 99%) of Caucasian donors. Annual donation probabilities and donor HLA-ABDR phenotype frequencies are derived from donations of DKMS donors in 2004 and 2005 (T₁). All donations excluding second donations were taken into account. A 2-year period was chosen to obtain better statistics. The age and gender distribution of new donors was taken from 2005 recruitment figures (T₂). We assumed that the HLA phenotype frequency distribution of future new donors equals the current distribution of the whole donor file for every combination of donor age and gender of the new donors. Regression analysis was based on those donors who were in the file at the center of interval T₁, that is, at the end of 2004.

Results

Age distribution of registered and newly recruited donors

Figure 1 shows the age distribution of registered donors in the DKMS file (values at the center of interval T₁, that is, end of year 2004, n=1 215 807), of donors who were newly recruited in 2005 (interval T₂, n=148 511), and of the German population in 2005 (n=82 000 000).¹⁸ All curves show similar peaks at 40 years (all donors), 37 years (newly recruited donors) and 41 years (German population). The average ages for all registered donors and newly recruited donors are 39.7 years and 34.0 years, respectively. For younger ages, the curve of all registered donors is considerably below the total population curve. This is due to the fact that recruitment does not start before the age of 18. The overrepresentation of younger individuals among the newly recruited donors can be seen as result of specific recruiting efforts aimed to recruit young donors (for example, donor drives at high schools or universities). The fact that older donors are underrepresented both in the donor file and among newly recruited donors could be related to an age-related increase in health problems that preclude admission to the donor file or require exclusion from the file. Recruitment stops at age 55 as the curve for newly recruited donors shows.

Figure1.

Age distributions of registered donors (values at the center of interval T₁, that is, end of year 2004, solid line), of newly recruited donors in the interval T₂, that is, in 2005 (dashed line) and of German population in 2005 (dotted line). Curves are scaled (peaks at donor ages of 40, 37 and 41 years are set to 100) to obtain better comparability.

Full figure and legend (33K)

Analysis of factors determining donation probabilities

Results of the multiple logistic regression analysis are displayed in Table 1. It shows that all factors considered influence the probability to donate stem cells significantly. Donor gender and the accumulation of HLA information resulting from prospective typing have the highest impact on the chance to donate stem cells. It follows from these results that young male donors with prospective typing and common HLA-AB phenotypes have the highest donation probabilities.

Table 1 - Results of multiple logistic regression analysis.

Full table

Annual donation probabilities by donor age

The calculated annual donation probabilities by age for donor sets G(1,1,1), G(1,0,1), G(0,1,1) and G(0,0,1) are shown in Figure 2. Definitions of the various donor sets are given in the Methods section. The relative positions of the curves reflect differences in donation probabilities as derived from the regression analysis. For example, the curve for G(1,1,1), the group of male donors with prospective typing and common HLA-AB phenotypes, lies above the other curves for nearly all ages. The donor sets displayed in Figure 2 include only donors with common HLA phenotypes (l=1). A decrease of donation probabilities with increasing age can be seen for all four groups. Maximum donation probabilities are reached at ages 21 for donor set G(1,1,1), 26 for G(1,0,1), 21 for G(0,1,1) and 20 for G(0,0,1). The data basis for donation probabilities by age differs considerably even within curves. The peaks at ages 28 and 55 in the curve of donor set G(1,1,1), for example, are induced by 74 donations from a group of 6083 donors and 2 donations from a group of 429 donors, respectively. However, differences in donation probabilities between age groups 18–29 years and 30–44 years are significant for all groups displayed in Figure 2, the corresponding differences between age groups 30–44 years and 45 years are significant for all groups apart from G(0,1,1) (Table 2).

Figure2.

Age-dependent annual donation probabilities for four donor sets: G(1,1,1) (solid line), G(1,0,1) (dashed line), G(0,1,1) (dotted line) and G(0,0,1) (dash-dotted line). Definitions of the various donor sets are given in the Methods section.

Full figure and legend (48K)

Table 2 - Calculated annual donation probabilities per 100 donors by donor age for the donor sets G(j,k,l) and results of ²-tests.

Full table

For donors with rare HLA-AB phenotypes one would expect a lower age-dependency of donation probabilities since search coordinators should have fewer possibilities to choose between several donors. Donation probabilities between the two younger age groups do not differ significantly for donor sets G(1,0,0) and G(0,0,0). For the two other donor sets (G(1,1,0) and G(0,1,0)), the donation probabilities are even significantly higher in the age interval 30–44 years. It has, however, to be noted that these donor sets are very small (3818 donors and 3696 donors, respectively) and results are based on only few donations (13 and 7 donations in the age interval 30–44 for G(1,1,0) and G(0,1,0), respectively). The small sizes of the sets G(1,1,0) and G(0,1,0) that include donors with prospective typing results and rare HLA phenotypes follow from the preferred selection of donors with common HLA-AB phenotypes for prospective typing. Donation probabilities between the two older age groups differ significantly only for donor set G(1,0,0). A graphical representation of the annual donation probabilities of the four donor sets G(j,k,0) as for the donor sets G(j,k,1) in Figure 2 seems to be not useful. Due to considerably smaller donor set sizes, the corresponding curves are dominated by random fluctuations. Instead, Figure 3 shows pooled donation probabilities of all donors with rare HLA-AB phenotypes (donor sets G(j,k,0)) compared to donors with common phenotypes (donor sets G(j,k,1)). This representation also suggests a higher age-dependency for donation probabilities of donors with common HLA-AB phenotypes. It has, however, to be noted that these pooled donor groups also differ with respect to the factors donor gender and completeness of HLA information.

Figure3.

Age-dependent annual donation probabilities for donors with rare (donor sets G(j,k,0), solid line) and common (donor sets G(j,k,1), dotted line) HLA-AB phenotypes.

Full figure and legend (37K)

Aging-induced reduction of donor file usefulness

The effect of donor file aging on the expected annual number of donations and, therefore, on donor file usefulness is shown in Figure 4. The expected donation number decreases without ongoing donor recruitment from 1529 donations in year 0 to 1207 donations in year 5 (-21.1%) and 873 donations in year 10 (-42.9%). The annual reduction of the number of expected donations ranges from 59 (between years 4 and 5) to 70 (between years 2 and 3).

Figure4.

Number of expected donations per year for the analyzed donor file without ongoing donor recruitment.

Full figure and legend (28K)

Compensation of aging-induced reduction of donor file usefulness

The numbers of donors who need to be recruited annually during the next 10 years to keep donor file usefulness constant are shown in Figure 5. These numbers exceed the numbers of donors who reach the age limit considerably. The ratio is 3.36 (7.31) in year 1 when the new donors are (are not) prospectively typed. Though the number of donors who reach the age limit will grow for the analyzed donor file during the next years, the ratio will still be 1.35 (2.95) with (without) prospective typing of the new donors in year 10.

Figure5.

Numbers of donors who must be recruited annually during the next 10 years to keep donor file usefulness constant. Results are shown under the assumptions that new donors are (solid line) or are not (dashed line) typed prospectively. The dotted line represents the number of donors who must be recruited to only compensate for the loss of donors who are removed from the file because they have reached the age limit.

Full figure and legend (37K)

When calculating the number of donors who must be recruited to compensate for donor file aging, one must also account for aging of these new donors. Without this effect the long-term increase of the two curves of Figure 5 that are based on the concept of donor file usefulness would be considerably smaller. For example, when we assume that new donors are not prospectively typed the increase between years 1 and 10 would only be from 57 560 to 63 587 without aging of new donors instead of 79 426 when aging of new donors is considered.

HLA phenotype frequencies of stem cell donors by age

The results presented so far show that, among registered donors with common HLA phenotypes, younger donors have a higher donation probability. Therefore, it seems reasonable to assume that younger donors are preferred to older donors if more than one HLA matching donor is available. As a consequence, the fraction of real stem cell donors who have a unique or rare HLA phenotype should be higher for older than for younger donors. The results displayed in Figure 6 support this hypothesis: In 39.1% of the donations from donors below 30 years, the donor has an HLA-ABDR phenotype that is 10 times available in the DKMS donor file. The corresponding values for the age groups 30–44 and 45 are 49.6 and 67.0%, respectively. For each two of the age groups these results differ significantly (²-tests, P<0.001).

Figure6.

Age-dependent fractions of donations from donors with HLA-ABDR phenotypes that are available once (black), 2–10 times (grey), 11–100 times (streaked) or >100 times (white) in the DKMS donor file.

Full figure and legend (40K)

Discussion

Our results show that individual probabilities that registered stem cell donors will actually donate depend on several factors, including donor age. The effects of some other factors (donor gender, completeness of available HLA information) on donation probabilities are stronger than the effect of donor age. However, the special importance of the factor donor age results from the fact that donor file aging leads to a continuous reduction of donor file usefulness.

The data presented here regarding HLA phenotype frequencies of stem cell donors suggest that younger donors are preferred by transplant physicians if several HLA-matched donors are available. Such preferences are in accordance with published results indicating more favorable transplantation outcomes for transplants that use younger donors.^{11, 12, 13, 14}

The calculated donation probabilities provide the input for a model aimed to quantify the aging-induced reduction of donor file usefulness and to estimate the number R of donors who need to be recruited to compensate for this reduction. It shows that R exceeds the number of donors who reach the age limit considerably in both analyzed scenarios (intermediate typing of the HLA-A and -B loci versus more complete typing of new donors). However, substantially more new donors are needed when they are typed only for the HLA-A and -B loci at intermediate resolution. This is a strong argument for more complete typing at the time of donor recruitment. However, since such typing is generally associated with higher costs every donor center or registry has to develop its own strategy based on its specific cost structure and available funds. DKMS has decided to perform sequence-based typing of the HLA-A, -B, -C and -DRB1 loci for most of its new donors.¹⁶

One may ask if the aging-induced reduction in usefulness of the donor file can also be compensated by additional HLA typing of already registered donors instead of continued donor recruitment. Such a strategy should, similar to comprehensive typing of new donors, also have the advantage of reducing the length of unrelated donor searches.¹⁹ Besides, additional HLA typing of already registered donors could lead to similar positive short-term effects on the usefulness of the donor file as the recruitment of new donors. This strategy is, however, not sustainable without ongoing donor recruitment because of donor file aging. In combination with ongoing recruitment additional typing of already registered donors is a useful strategy to keep or enhance donor file usefulness.²⁰

Our model has some limitations that are discussed below: A central model feature is the representation of donor file usefulness through summarized individual donation probabilities p(i,j,k,l) of its registered donors. These probabilities depend on donor age and gender, HLA typing level and HLA-AB phenotype frequency but do not change over time. There are, however, scenarios in which this approach will not lead to satisfying results. This holds, for example, for a closed system of a donor pool and transplant centers that only access donors from this pool. When the donor pool is small, transplant physicians should not often be able to choose between several HLA matching donors for a specific patient. Since our results suggest that differences in donation probabilities by age are induced by the preferences of transplant physicians for younger donors if several HLA matching donors are available one would expect that differences in age-specific donation probabilities and, therefore, the aging-induced usefulness reduction as given by the model will be small while the average age at donation will be relatively high. When we look at a closed system with larger donor pool donation probabilities should differ substantially by age. However, a donor recruitment stop would lead to an increase of donation probabilities of older donors instead of a reduction of donation figures due to decreased donor file usefulness as suggested by our model since transplant physicians could not revert to younger donors from other donor registries. The donor file analyzed in this work does, however, not belong to a closed system as described above. This is shown by the fact that 1050 donors from 2094 allogeneic unrelated first transplantations in Germany in 2004 and 2005²¹ (50%) were not from the DKMS donor file. Besides, 66% (2013 of 3057) of all first donations of DKMS donors went to patients outside Germany in 2004 and 2005.

Some factors might considerably affect the empirically determined donation probabilities that are used for our model calculations: A further increase in the number of registered donors worldwide could lead to a reduction of individual donation probabilities. On the other hand, since the number of unrelated stem cell donations is growing,^{21, 22, 23} an opposite effect might also be possible. However, since the modified donation probabilities would also apply for newly recruited donors, these possible changes seem negligible for the estimation of the number of donors who need to be recruited to maintain donor file usefulness.

Our model uses further assumptions: We imply that only donors who reach the age limit are removed from the file and thus neglect the fact that registered donors can also drop out for other reasons as health status or withdrawal of informed consent. This assumption seems justified as we deal with the aging-related usefulness reduction of a donor file. Donor attrition for other reasons leads to a reduction of donor file usefulness that is not included in our analysis. As a consequence, the total number of donors who must be recruited to keep donor file usefulness constant is higher than derived from our aging-focused model. Furthermore, we assume that the age distribution of newly recruited donors will not change in the future. However, the German society is, as many Western societies, aging.¹⁸ Our results show that the age structure of newly recruited donors can be influenced by recruitment efforts that are specifically focused on young donors. Since such efforts have already been successful in the past (Figure 1), it is doubtful if the age distribution of newly recruited donors will remain as favorable as it is today. Therefore, the number of donors who need to be recruited to keep donor file usefulness constant could be underestimated through the assumption of a fixed age distribution of newly recruited donors.

In a new study based on data of the National Marrow Donor Program in the USA,⁴ survival probabilities of patients with older donors are slightly lower than for donors younger than 31 years. Hazard ratios are 1.05 and 1.06 for donors aged 31–45 and >45, respectively. Differences are, however, not statistically significant. If new results showed that effects of donor age on transplantation success were smaller than demonstrated so far,^{11, 12, 13, 14} the preference of transplant physicians for younger donors could decline. As a result, both differences in age-specific donation probabilities and the number of donors who must be recruited to compensate for donor file aging would decrease. It has to be noted in this context that not only anticipated effects on transplantation success, but also supposed differences in donor availability could induce a preference for donors of a specific age group. While it is obvious that nonavailability of donors due to medical reasons should be more common in older donors, there is some evidence from donor center practice that very young donors more often withdraw their consent to donate than older donors.

Summarized, there is no evidence that our model systematically overestimates the number of donors who must be added to the file to keep its usefulness constant. Therefore, our results indicate that a donor recruitment strategy that only aims at replacing donors who have reached the age limit will lead to a notable reduction in the usefulness of a donor file. The model presented could easily be applied to other donor pools if corresponding input data are available.

We estimate how many donors must be recruited to keep the usefulness of a donor file constant in spite of donor aging. To our knowledge, this effect has not been quantified before. The results, however, provide no guidance to whether an existing donor file with a given size, level of available HLA information, HLA frequency distribution and donor age and gender distribution should be further increased or has reached a level where investment in further improvement of file usefulness is not appropriate. Such considerations lead to the concept of optimal donor registry size.^{6, 7, 8, 9, 10} The definite assessment of optimal donor registry size remains a serious challenge for several reasons. First, the existence of donors with different levels of available HLA information in real donor pools leads to considerable methodological difficulties. Second, published analyses regarding optimal donor registry size do not consider donor age and gender distributions. Our results suggest, however, that these distributions and donor file aging are relevant factors for the determination of donor file usefulness and, therefore, for optimal registry size. Third, the negative impact of HLA mismatches for transplantation success should be included in such analyses in a quantitative way as the optimal donor registry size increases with the relevance of HLA mismatches. Fourth, the determination of optimal donor registry size raises complex ethical questions as the value of a successful stem cell transplantation must be weighed against costs for ongoing donor recruitment or additional HLA typing of already registered donors.¹⁰ While these issues demonstrate the difficulties of a quantitative assessment of optimal donor registry size, the fact that only 40–50% of all patients in need of a stem cell transplantation find a donor who matches for the HLA-A, -B, -C and -DRB1 loci at the allele level^{12, 24} suggests that efforts to increase the usefulness of the global donor pool may be indicated.

References

1. Morishima Y, Sasazuki T, Inoko H, Juji T, Akaza T, Yamamoto K 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. | Article | PubMed | ISI | ChemPort |
2. Flomenberg N, Baxter-Lowe LA, Confer D, Fernandez-Vina M, Filipovich A, Horowitz M 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. | Article | PubMed | ISI | ChemPort |
3. Petersdorf EW, Anasetti C, Martin PJ, Gooley T, Radich J, Malkki M et al. Limits of HLA mismatching in unrelated hematopoietic cell transplantation. Blood 2004; 104: 2976–2980. | Article | PubMed | ISI | ChemPort |
4. Lee SJ, Klein J, Haagenson M, Baxter-Lowe LA, Confer DL, Eapen M et al. High resolution donor-recipient HLA matching contributes to the success of unrelated donor marrow transplantation. Blood, 4 September 2007 (doi:10.1182/blood-2007-06-097386).
5. Bone Marrow Donors Worldwide. www.bmdw.org.
6. Beatty PG, Dahlberg S, Mickelson EM, Nisperos B, Opelz G, Martin PJ et al. Probability of finding HLA-matched unrelated marrow donors. Transplantation 1988; 45: 714–718. | Article | PubMed | ISI | ChemPort |
7. Sonnenberg FA, Eckman MH, Pauker SG. Bone marrow donor registries: the relation between registry size and probability of finding complete and partial matches. Blood 1989; 74: 2569–2578. | PubMed | ChemPort |
8. Beatty PG, Mori M, Milford E. Impact of racial genetic polymorphism on the probability of finding an HLA-matched donor. Transplantation 1995; 60: 778–783. | Article | PubMed | ISI | ChemPort |
9. Müller CR, Ehninger G, Goldmann SF. Gene and haplotype frequencies for the loci HLA-A, HLA-B, and HLA-DR based on over 13 000 German blood donors. Hum Immunol 2003; 64: 137–151. | Article | PubMed |
10. Kollman C, Abella E, Baitty RL, Beatty PG, Chakraborty R, Christiansen CL et al. Assessment of optimal size and composition of the US National Registry of hematopoietic stem cell donors. Transplantation 2004; 78: 89–95. | Article | PubMed | ISI |
11. Kollman C, Howe CWS, Anasetti C, Antin JH, Davies SM, Filipovich AH et al. Donor characteristics as risk factors in recipients after transplantation of bone marrow from unrelated donors: the effect of donor age. Blood 2001; 98: 2043–2051. | Article | PubMed | ISI | ChemPort |
12. Carreras E, Jiménez M, Gómez-García V, de la Cámara R, Martín C, Martínez F et al. Donor age and degree of HLA matching have a major impact on the outcome of unrelated donor haematopoietic cell transplantation for chronic myeloid leukaemia. Bone Marrow Transplant 2006; 37: 33–40. | Article | PubMed | ISI | ChemPort |
13. Mehta J, Gordon LI, Tallman MS, Winter JN, Evens AM, Frankfurt O et al. Does younger donor age affect the outcome of reduced-intensity allogeneic hematopoietic stem cell transplantation for hematologic malignancies beneficially? Bone Marrow Transplant 2006; 38: 95–100. | Article | PubMed | ISI | ChemPort |
14. Confer DL, Miller JP. Optimal donor selection: beyond HLA. Biol Blood Marrow Transplant 2007; 13: 83–86. | Article |
15. Confer DL. The national marrow donor program: meeting the needs of the medically underserved. Cancer 2001; 91: 274–278. | Article | PubMed | ISI | ChemPort |
16. Schmidt AH, Stahr A, Baier D, Ehninger G, Rutt C. Efficiency of HLA-A, -B, -C, -DRB1 high-resolution typings of newly recruited potential stem cell donors. Blood 2006; 108: 448b (abstract).
17. Schmidt AH, Stahr A, Baier D, Schumacher S, Ehninger G, Rutt C. Selective recruitment of stem cell donors with rare human leukocyte antigen phenotypes. Bone Marrow Transplant 2007; 40: 823–830. | Article | PubMed | ChemPort |
18. Statistisches Bundesamt. www.destatis.de.
19. Heemskerk MBA, van Walraven SM, Cornelissen JJ, Barge RMY, Bredius RGM, Egeler RM et al. How to improve the search for an unrelated haematopoietic stem cell donor. Faster is better than more! Bone Marrow Transplant 2005; 35: 645–652. | Article | PubMed | ISI | ChemPort |
20. Setterholm M, Morishima Y, Pepperall J, Schmidt A. Strategies for typing new volunteer donors. Tissue Antigens 2007; 69 (Suppl 1): 6–7. | Article | PubMed |
21. Jahresbericht 2006 Deutsches Register für Stammzelltransplantationen. www.drst.de.
22. 2004 Biennial Report of the National Bone Marrow Donor Registry. www.hrsa.gov.
23. Gratwohl A, Baldomero H, Frauendorfer K, Urbano-Ispizua A, Niederwieser D. Results of the EBMT activity survey 2005 on haematopoietic stem cell transplantation: focus on increasing use of unrelated donors. Bone Marrow Transplant 2007; 39: 71–87. | Article | PubMed | ISI | ChemPort |
24. Lee SJ, Klein J, Haagenson M, Baxter-Lowe LA, Confer D, Eapen M et al. Single or multiple HLA-A, B, C or DRB1 mismatches limit success of unrelated donor bone marrow transplantation. Blood 2006; 108: 55a (abstract).