Allogeneic haematopoietic stem cell transplantation (allo-HSCT) is curative therapy for many malignant and non-malignant conditions including leukaemia, lymphoma, bone marrow failure syndromes, haemoglobinopathies, immunodeficiencies and inborn errors of metabolism. Stem cells for allo-HSCT can be obtained from donor bone marrow, peripheral blood, or umbilical cord blood (UCB). However allo-HSCT is an option for patients in need of a transplant only if a suitably human leucocyte antigen (HLA)-matched donor can be identified. There is a one in four chance of any sibling sharing the same HLA type as the patient, with extended family members having a much lower chance of matching. Given current average family size, this means that suitably matched related donors are available for approximately 30% of patients in need of an allo-HSCT.1 For the majority of patients who do not have a matched sibling or family member, volunteer-unrelated donors provide the only available source of stem cells for transplantation but, given the enormous variation in HLA types in the community, it may not be possible to identify such a donor.

If the need for allo-HSCT is not urgent and the patient’s parents are of reproductive age, another option may be available, and that is for parents to conceive another child in the hope that they may provide a source of stem cells for their sick sibling. In the past, parents have conceived another child in the knowledge that there was only a one in four chance that the child would be HLA matched with their sibling.2 It is now possible, using in vitro fertilisation (IVF) together with pre-implantation genetic diagnosis (PGD), to identify the HLA type of IVF embryos and, where necessary, to exclude the presence of the underlying genetic disease affecting the proband. This enables the transfer to the mother’s uterus of an embryo that is HLA matched to the sick sibling, as well as free of the genetic disorder in question (where relevant).3 Children born following the application of this technology have become known in the lay press as “saviour siblings”.4

This practice has generated considerable controversy.5 6 And unfortunately, as is often the case where the emergence of a new technology generates public and professional disagreement, much of the debate has been sensationalist and misinformed and there has been limited discussion of the scientific rationale for, and appropriate clinical application of, the use of PGD for HLA typing to produce a matched sibling donor for a child awaiting allo-HSCT.

This article examines the clinical and epidemiological evidence underpinning arguments in support of using this technology, with the aim of clarifying the circumstances in which creating a “saviour sibling” may confer a clinical advantage, and the circumstances in which this option should be discussed with parents of a child who is, or may be, a candidate for an allo-HSCT transplant.

What are the indications for using PGD for HLA typing to produce a donor?

Allo-HSCT is associated with substantial morbidity and mortality.7 Many factors, including the patient’s age, comorbidity, disease and disease status, donor histocompatibility, and the time interval between diagnosis and transplant, may influence the decision to proceed with allo-HSCT. For some diseases, the absence of a matched sibling donor may mean that allo-HSCT is not considered. This is because of the differential risks and benefits associated with sibling and unrelated donor transplantation, and in particular the impact of graft versus host disease (GvHD) on transplant outcomes and long-term disease-free survival. GvHD results from injury to the recipient’s skin, gut and liver, as a consequence of immunological differences between the donor and recipient, even between matched siblings. It is a significant cause of morbidity and mortality after transplant. The same cells that mediate GvHD can also mediate an “anti-leukaemic” effect which may counteract the disadvantage of GvHD, but only in some patients with malignancy, and not in non-malignant disease. Thus an unrelated donor transplant exposes these patients to additional risk as compared with a sibling donor transplant. So for some diseases, such as thalassaemia and sickle cell disease, a transplant may be carried out if there is a matched sibling, and is much less likely to be done if there is only an unrelated donor. In these situations, avoiding the risks of unrelated donor transplant through the use of IVF and PGD for HLA typing to produce a matched sibling donor becomes a real option. This is as long as the parents of the child requiring an allo-HSCT are of reproductive age and willing to use IVF to conceive; the IVF leads to a successful pregnancy, with a confirmed HLA-compatible sibling; and the child does not require allo-HSCT for at least 12 months. This is the minimum time needed to complete one or more cycles of IVF, perform PGD and HLA typing and gestate a pregnancy. It is important, however, not to overestimate the chance of a successful pregnancy. First, multiple cycles may be required to achieve embryo transfer of an HLA-matched embryo,8 and second, the pregnancy rate per embryo transfer is on average only 34%, though this varies depending on the IVF centre and age of the mother (range: 14.8–42.7).9 In addition, if insufficient cord blood was obtained at birth, there would be an additional delay of 6 months to allow for safe collection of bone marrow from the sibling donor (personal communication, Steve Grupp).

However, this technology can offer clinical benefit only if the following hold true. First, either there is no acceptable matched related or unrelated donor available, or allo-HSCT from a related donor is likely to result in better clinical outcomes than if an unrelated transplant is performed. Second, that outcomes following allo-HSCT using UCB (the primary stem cell source initially generated from a newborn donor) are at least equivalent to allo-HSCT performed with stem cells sourced either from bone marrow or peripheral blood. These issues will be discussed in turn in the subsequent sections of this article.

How many patients fail to find an unrelated donor?

In the past, suitably HLA-matched unrelated donors could not be located for many patients in need of allo-HSCT. Presently, however, donors may be found for the majority of patients due to the worldwide increase in participation in donor registries; the establishment of publicly accessible UCB banks; the recognition that HSCT can now be successfully performed with less than completely matched UCB donors; and the emergence of haploidentical transplantation as a genuine therapeutic possibility. In the USA, the average likelihood of finding a potential unrelated matched bone marrow or peripheral blood donor was 86% in 2005, with more than 95% of patients able to find at least one potential 4/6 HLA-matched UCB unit.10 In Australia, only 50% of patients who initiated a search at the Australian Bone Marrow Donor Registry between 2003 and 2005 found a suitable unrelated UCB, bone marrow, or peripheral blood donor (64% of cancelled searches are removed from the calculation).11

These figures fail to take into account the quality of the donated UCB unit or the amount of cells it contains, which are both critical factors for transplantation success. They also conceal that patients from non-Caucasian ethnic or indigenous populations, for example, have a much lower chance of finding a suitably matched unrelated donor. This is primarily because most volunteers on the US, European, and Australian registries are of North Caucasian descent. In the USA, for example, African-Americans have the least chance of finding a matched bone marrow, peripheral blood or UCB donor, followed by Hispanic and Asian patients.12 In Australia, Indian patients have the least chance of finding a suitable matched unrelated donor, followed by Asian patients and Aboriginal-Australians.11 What is clear, therefore, is that there may be a much stronger clinical rationale for IVF and PGD for HLA typing for patients from certain ethnic populations. Indeed, in some instances, it may be the only chance that a patient has of obtaining suitably matched stem cells for transplantation.

Does having a matched related donor confer an advantage over a matched unrelated donor?

As it is ethically and practically impossible to perform randomised, controlled studies comparing related and unrelated donor transplants, thereby precluding the possibility of Level 1 evidence, data regarding transplant outcomes are derived from other sources, including uncontrolled or single-arm studies, or, more importantly, from registry studies.

Reviews of national and international registry data show better outcomes for related donors than for unrelated donors. For example, recent data from the Centre for International Blood and Marrow Transplant Research (CIBMTR) show a superior overall survival when using an HLA-identical sibling donor compared to an unrelated matched donor for acute myeloid leukaemia (AML) and acute lymphoblastic leukaemia (ALL) in patients under 20.13 Likewise, a retrospective analysis of registry data from over 2000 leukaemia patients transplanted between 1985 and 1991 by Szydlo et al found that patients who underwent an unrelated donor transplant had higher transplant-related mortality and treatment failure.14 The difference was not as striking for patients with more advanced disease. This reflects the impact of GvHD, as described above, with patients with early disease faring worse if they did not have a sibling donor, whereas those with more advanced disease had a more comparable outcome.

Registry studies are, however, limited by the fact that they collect and report data only for patients who actually underwent transplantation, and do not include data on those who were never considered for, or never made it to transplant. In order to overcome the bias that may result from knowing whether a patient has a sibling donor, it is possible to incorporate what is known as “biological randomisation”. As there are no known inherent biases or disease-related associations in having or not having an HLA-matched sibling, rather than looking at the outcome of patients who had a related or unrelated allo-HSCT, one can follow up a cohort of patients who have a matched sibling donor available and see if their outcome is better or not. (Since this type of study is also an “intention-to-treat” analysis, it still controls for introduced biases.) So, for example, in studies where only those with a matched sibling were scheduled to undergo bone marrow transplantation, patients who did not but may still have had alternate donor (unrelated or haploidentical) transplant, are still analysed in the “no sibling donor” group. Such studies, in patients with high-risk ALL in first complete remission,15 with ALL in second complete remission,16 17 and in AML,18 have all shown that having a sibling donor offers an advantage. Furthermore, if, as was done in a study by Balduzzi, one looks at the treatment actually received, then patients with high-risk ALL in first complete remission who underwent unrelated donor allo-HSCT had an outcome inferior to those who had a matched sibling transplant.15 Prospective studies that offer an alternate donor transplant if a patient does not have a matched sibling donor are still in their infancy.15

There are less data on the different outcomes with related and unrelated donor transplants in patients with non-malignant disease, partly because, as described above, allo-HSCT is frequently only performed if there is a matched sibling donor. Some small series show comparable outcomes, for example in thalassaemia,19 but larger registry data, for Fanconi anaemia,20 severe aplastic anaemia13 and Wiskott–Aldrich syndrome21 all demonstrate significantly better outcomes in patients undergoing matched sibling transplant as compared with unrelated donor transplant.

Over the past decade considerable progress has been made towards improving the outcomes of related and unrelated donor transplants. Advances in HLA typing (particularly in allelic matching), transplant conditioning, GvHD prophylaxis and supportive care, mean that recipients of all types of allo-HSCT are more likely to survive than in previous years. Consequently there are many retrospective studies that show a similar outcome for children who undergo bone marrow transplantation with a related or unrelated donor. This, in turn, may change the rationale for PGD and HLA typing to produce a “saviour sibling”.22^–24

Are the outcomes equivalent with transplant using UCB compared to other sources of HSC?

While “saviour siblings” may, in theory, provide bone marrow and peripheral blood stem cells, for the most part, use of this technology assumes that allo-HSCT will proceed using UCB. It is therefore important to know if UCB provides outcomes comparable to other stem cell sources, particularly bone marrow, given that this is currently the major source of haematopoietic stem cells for paediatric transplant recipients.25

No randomised trials have been conducted in children comparing the outcomes of allo-HSCT using UCB versus bone marrow. Numerous uncontrolled trials, however, have found that outcomes are essentially equivalent.26^–28 Hwang performed a meta-analysis of pooled data on comparative studies of unrelated donor UCB transplantation and unrelated donor bone marrow transplantation in children requiring allo-HSCT, and concluded that there were no differences in 2-year overall survival.29 Likewise, Eapen conducted a retrospective analysis of the outcomes of transplantations in children with acute leukaemia between 1995 and 2003 and concluded that 5-year leukaemia-free survival was comparable between matched bone marrow and UCB (5/6 or 6/6 with high cell dose) transplants. Matched UCB donor transplants had the best outcome, and 4/6 UCB donor transplants had an inferior outcome, regardless of cell dose.30 Further, results of UCB transplantation for thalassaemia and sickle cell disease using HLA-matched related donors are now similar to those for bone marrow transplantation.31 Although most of these studies were for patients with malignant disease, in the series of Rocha, half of the cord blood transplants were for non-malignant disease.26

So when is PGD for HLA typing to produce a matched sibling donor a reasonable therapeutic option?

It is clear from the preceding review of the literature that there may well be situations where PGD for HLA typing to produce a matched sibling donor should be presented to parents as a therapeutic option for their ill child. In all cases, the need for allo-HSCT must be non-urgent, that is, must not be required for at least 12 months, and the mother must be of reproductive age. There are two main situations where this technology should be considered and discussed with parents. The first is where there is no suitable matched sibling donor available and transplantation using an unrelated donor carries significantly higher risks than use of a sibling donor, such as in genetic conditions including Fanconi anaemia and the inherited haemoglobinopathies such as thalassaemia. The second is where there is no suitable unrelated donor available for a child receiving an Allo-HSCT. Given that unrelated matched donors for most north Caucasian populations are represented in bone marrow donor registries and UCB banks, these situations will more likely involve children from indigenous communities or ethnic minorities.

More generally, it is arguable that PGD for HLA typing to produce a sibling donor should be discussed in relation to the care of any sick child who lacks a matched related donor and either requires a non-urgent allo-HSCT or who currently is not a candidate for HSCT but may require a transplant in the future due to progression of his or her underlying disease. The latter situation may arise in the care of children with ALL in first complete remission, as these children would generally be transplanted if they obtained a second complete remission following treatment for relapse. While up to 85% of children with ALL achieve long-term, disease-free survival with chemotherapy alone, some children still relapse. This requires that therapeutic options in the event of relapse, including transplantation, be discussed with parents. Even though the parents of a child with ALL know it is most likely that their child will remain in remission, they could be expected to want to do anything they could to maximise their child’s chance of survival should he or she relapse. As these families are usually young and may often be considering whether to have more children in any case, it would seem reasonable to at least raise the issue of PGD for HLA typing. In saying this, it is important to be clear that discussion of this technology may not constitute a recommendation to use it. Indeed, while raising the issue may lead some parents to pursue it, in many situations it may simply provide an opportunity for parents to discuss their fears and for physicians to provide reassurance, and in all situations raising this issue with parents will engender trust and explicitly acknowledge the fact that physicians and parents may attach different values to different risks and outcomes.

CONCLUSION

The care of children in need of allo-HSCT raises many complex medical, ethical and legal issues. Children are often incapable of expressing autonomous informed preferences and parents must make decisions for them. This is no small task, however, as allo-HSCT is a high-risk procedure and many of the outcomes, with and without transplant, are uncertain. In some of the conditions for which allo-HSCT is offered, the prognosis of the underlying disease without transplant is uncertain, in contrast to the known early mortality and morbidity associated with allo-HSCT. This situation is made more complex in cases where parents must make decisions for two children, one of whom is a donor and the other a transplant recipient, as this may involve trading off different risks and benefits for each child, or even a potential future child, as is the case with “saviour siblings”.

Decisions about transplantation, including the choice of donor, necessarily will be influenced not only by the best available data about clinical outcomes, but also by a range of emotional and psychological concerns, which we have not explored in this paper. These include attitude to medical uncertainty, fear for the future, avoidance of regret, and a desire to be “doing something” rather than “doing nothing”, which may extend to seeking some form of “biological insurance” against the chance of the disease recurring in a child who has already been successfully treated. As IVF and PGD for HLA typing is now a very real option in certain circumstances and is likely to become a topic of discussion with many parents whose child may be a candidate for allo-HSCT, further research is required to elucidate the clinical, social, psychological, and economic issues that might be relevant to the application of this technology and to discussions regarding its use.