Abstract

BACKGROUND: β‐Thalassaemia results from co‐inheritance of two mutant β‐globin alleles. Allogeneic cord blood cell transplantation (CBT) from an HLA‐identical sibling donor is an excellent treatment option for β‐thalassaemia. In families with an affected child and willing to have another child, IVF followed by preimplantation genetic diagnosis (PGD) can be applied to exclude affected embryos. Furthermore, healthy embryos could be HLA matched with the affected child so that cord blood from the future newborn can be used to transplant the affected sibling. METHODS: We developed an indirect single‐cell HLA typing technique based on the use of a bank of seven microsatellite markers within the HLA locus from which four informative and evenly distributed markers were selected. RESULTS: The methodology was validated in three β‐thalassaemia families having six ovarian stimulation cycles in view of IVF and PGD. Six PGD cycles were performed in two families. On 58 embryos tested, the combined PCR was successful in 54 (93%). Two transfers were done and one clinical pregnancy was obtained. Using confirmatory analysis on 50 embryos, the accuracy for HLA typing was 100%. CONCLUSION: This strategy offers a new therapeutic option for patients with β‐thalassaemia and other monogenic diseases that can be cured with CBT.

Introduction

β‐Thalassaemia is one of the most common monogenic diseases in humans and is widespread throughout the Mediterranean region and parts of the African and Asian continents. The clinical and molecular aspects have been reviewed recently (Olivieri, 1999). Nearly 200 different mutations of the β‐globin gene cluster have been described, and they result in either absence or reduced synthesis of the β‐globin chains (Hardison et al., 1998). In severe cases, including β‐thalassaemia major, the clinical picture is the result of marked ineffective erythropoiesis leading to erythroid marrow expansion, osteopenia, bone deformities and iron overload. The latter is aggravated by the high transfusion requirements and can be partially controlled by rigorous chelator therapy. However, the majority of cases will ultimately develop organ damage, in particular of the heart and liver, leading to reduced quality of life and a life expectancy limited to the fourth or fifth decade.

Allogeneic haematopoietic stem cell transplantation (alloSCT) is the only curative treatment for thalassaemia major (Lucarelli et al., 1998, 1999, 2002). When an HLA‐identical sibling marrow donor is available and in patients with effective chelating therapy prior to transplantation as well as absence of hepatomegaly or portal fibrosis (low Pesaro risk score), the chance of cure is currently ∼90%. In addition, alloSCT from alternative donors or donor sources has been applied succesfully in patients with haemoglobinopathies including β‐thalassaemia (Gaziev et al., 2000; La Nasa et al., 2002; Locatelli et al., 2003), but is less favourable compared with that with an HLA‐identical donor. On the other hand, the outcome after alloSCT with HLA‐identical stem cells obtained from umbilical cord seems to be comparable with bone marrow transplantation.

Preimplantation genetic diagnosis (PGD) is an established method for the diagnosis of genetic diseases at the embryonic stage so that implantation of affected embryos can be avoided (ESHRE, 2002). Couples opting for PGD to avoid the transmission of a genetic disease have to undergo an IVF treatment with ovarian stimulation and ICSI. The embryos have to be biopsied at day 3. A genetic analysis of the blastomeres is performed in order to be able to select ‘unaffected’ embryos for transfer to the uterus. For monogenic diseases, PCR is applied at the single‐cell level. We developed a novel PGD strategy in which the detection of a particular gene mutation was combined with the selection for HLA‐identical embryos in order to obtain an unaffected child that can become an HLA‐matched donor for its sibling. This strategy was validated in three different families with β‐thalassaemia and may offer a new therapeutic perspective for thalassaemic patients as well as for those with other genetic diseases that can be cured with alloSCT.

Patients and methods

Selection of families

The approach was approved by the local ethical committee. Couples who were sent to our centres by the referring gynaecologist or geneticist had to present a report from the paediatrician‐haematologist. The indication for alloSCT was verified by an ad hoc committee including haematologists experienced in alloSCT. The PGD procedure was explained to the couple by an experienced geneticist, stressing the fact that theoretically only three out of 16 embryos would be ‘unaffected’ by the disease and HLA matched. Therefore, the success rate was going to be low and possibly several treatment cycles could be necessary before a pregnancy would be established and followed by the birth of a child. The possibility of a misdiagnosis inherent to single‐cell PCR was discussed, as well as the possibility of control by chorionic villus sampling (CVS) at 10 weeks of pregnancy. A discussion concerning the fate of the ‘unaffected’ but HLA‐non‐identical embryos also took place, the options being: (i) cryopreservation for later transfer; or (ii) research to confirm the diagnosis. Informed consent concerning the course of the PGD procedure, as well as the use of non‐transferred embryos for research (confirmation of the diagnosis), was given.

The parents were seen by an experienced psychologist who evaluated their motivation for having another child to become a donor for their affected sibling, i.e. were the parents planning to have another child in any case, or was the plan to have another child solely for the purpose of treating the affected child? Both situations occurred and were accepted.

Case descriptions

In each β‐thalassaemia family, the healthy siblings were HLA‐non‐identical to their affected sibling. Two couples were of Italian origin and one of Turkish origin. The distinct mutations in the β‐globin gene are represented in Figure 1. In the first couple, both parents are carriers of a C>T mutation in codon 39 of the β‐globin gene. They have one healthy and one affected child; the mother (age 35 years) already had two miscarriages. In the second family, the disease is caused by two mutations (IVS1+6T>C and IVS1+110G>A). The mother (age 36 years) had one miscarriage, and two affected pregnancies were terminated after prenatal diagnosis (PND). The couple have an affected child and a healthy child after PGD at our centres for which HLA typing had not been done at that time (De Rycke et al., 2001). The third couple (mother age 31 years) have one affected child and two healthy children, while two affected pregnancies were terminated after PND. The disease in this family is due to two different mutations in codon 8/9+G and IVS1+110G>A of the β‐globin gene.

HLA typing strategy

The HLA typing was performed by using short tandem repeats (STRs) as microsatellite markers present in the HLA locus as described previously (Foissac et al., 2000). From a series of 155 of such STRs, a first selection of four STRs (CA repeats) was made based on their amplification efficiency at the single‐cell level and on their localization: one telomeric but very proximate to class I‐A, one within class I (between A and B), one within class II‐DQ and one centromeric but very proximate to class II (Figure 2). The markers were then selected further per individual family based on their informativity, i.e. the possibility to distinguish the four parental HLA haplotypes through these markers. We have a bank of seven STRs to ensure informativity in all families. To select for an HLA‐identical and healthy (non‐affected) embryo, the amplification of the four family‐specific markers was combined with the disease‐specific mutations in multiplex PCR. Molecular tests were developed to use the selected combination at the single‐cell level.

HLA informativity testing in the families

Informativity testing for segregation analysis was performed by PCR on DNA samples obtained in these three families (parents and children) using 5 pmol of each primer (one of which was labelled with Cy5, indocarbocyanine), buffer 1 provided by the manufacturer, 0.2 mM dNTP and 1.25 U of AmpliTaq DNA Polymerase (Applied Biosystems, Brussels, Belgium) with the application of a standard PCR protocol [5 min 96°C, 30× (30 s 96°C, 30 s 55°C, 30 s 72°C), 5 min 72°C] in an Eppendorf Master Cycler (VWR International, Leuven, Belgium). Fluorescent fragments were analysed on an Automated Laser Fluorescence Express DNA sequencer (ALFExpress, Amersham Pharmacia Biotech, Roosendaal, The Netherlands) (De Rycke et al., 2001). Fragment analysis was carried out using AlleleLinks software provided by the manufacturer.

Mutation analysis in single cells

Mutations in the β‐globin gene (Hb) were detected using primers as previously reported (De Rycke et al., 2001): HB1/HB2, HB1/HB3 and hemi‐nested PCO5/HB3/HB9 (HB9 forward: CTGACTCCTGAG GACAAG) for the three families, respectively (Figure 1). Restriction enzyme digestion was performed on 5 µl PCR product as described by the manufacturer (New England Biolabs, Westburg, Leusden, The Netherlands) (Figure 1). The C>T mutation in codon 39 in the β‐globin gene abolishes an NlaIV restriction site, and the codon 8/9+G mutation creates a PflFI restriction site. The β‐thalassaemia family with the IVS1+6T>C (SfaNI digests the affected allele) and IVS1+110G>A (AvaII digests the normal allele) mutations has been reported previously (De Rycke et al., 2001).

Collection of single lymphoblasts

Lymphocytes isolated from blood samples from the couples and their offspring were transformed by Epstein–Barr virus (EBV) and cultured according to standard procedures (Ventura et al., 1988). The procedure for the collection of single lymphoblasts into 0.2 ml PCR tubes containing 3 µl of alkaline lysis buffer (ALB) [200 mM NaOH, 50 mM dithiothreitol (DTT)] (Li et al., 1988) and blanks has been described previously (Sermon et al., 1998a).

Multiplex PCR protocols

Prior to multiplex PCR, cells were lysed by incubation for 10 min at 65°C. The Expand High Fidelity (EHF) PCR system was used as described by the manufacturer (Roche Diagnostics, Brussels, Belgium); a hot start was included in all procedures. Optimal PCR conditions for each primer combination were set up on single heterozygous lymphoblasts. In general, two consecutive PCR rounds were performed (Table I): (i) 5 min 95°C, 10× (30 s 95°C, 30 s 55°C, 30 s 72°C), 7 min 72°C; and (ii) 5 min 95°C, 40× (30 s 95°C, 30 s 53–55°C as required, 30 s 72°C), 7 min 72°C. In the first round, the PCR mix contained all primers, EHF buffer 2, 2 mM dNTP, 1.4 U of EHF DNA polymerase and neutralization buffer (10 mM Tricine, pH 8.3). For the second round, primer combinations were split up. A 3 µl aliquot from the first round was transferred into a new PCR tube, and the neutralization buffer was omitted from the reaction mix.

ICSI procedure

Ovarian stimulation was carried out as described previously (Kolibianakis et al., 2002). On day 0, oocyte retrieval, removal of cumulus and corona cells and ICSI were carried out as previously described (Joris et al., 1998). ICSI was used in order to avoid contamination with sperm cells and surrounding cumulus cells (Lissens and Sermon, 1997; Liebaers et al., 1998), and to avoid unexpected fertilization failure (Staessen et al., 1999). Fertilization was examined 16–22 h after ICSI; embryo development (cleavage stage and fragmentation rate) was assessed 24 and 48 h later and prior to biopsy. The embryos were cultured in sequential media.

Embryo biopsy

Laser biopsy was performed on the morning of day 3 (after fertilization) according to previously described procedures (Joris et al., 2003). Briefly, embryos were incubated for 5 min in Ca2+–Mg2+‐free medium (EB10, Vitrolife, Sweden) for decompaction. Embryo biopsy was performed in HEPES‐buffered Earle’s medium supplemented with 0.5% (w/v) human serum albumin. Zona drilling was performed by fixing the embryo on a holding pipette and subsequently applying two or three pulses of 5–8 ms (1.48 µm) on the zona pellucida. The opening was made between two blastomers. Two cells were gently aspirated using a biopsy pipette (inner diameter 35–40 µm) and released into the medium. Two blastomeres were taken from each embryo with at least six cells and <50% fragmentation, except in two cycles where one cell was taken because of high workload (restrictions in time and equipment). Procedures for washing, transfer into PCR tubes containing 3 µl of ALB, and blanks have been described previously (Sermon et al., 1998b).

Results

HLA matching strategy

By using two internal STRs (one within HLA class I and one within class II) and two external STRs (one telomeric to A and one centromeric to DQ), a fingerprint from the entire HLA locus could be obtained (Figure 2). As shown in Table II for the β‐thalassaemia family 1, the segregation phase of the STRs fully corresponded to the direct HLA typing. A bank of seven STRs had to be set up to ensure informativity in the three families which could be fully covered using one telomeric STR (MOG3), three STRs within class I (between A and B: D6S1571, D6S1611 and D6S1560, respectively), one STR within class II (within DQ: D6S2443) and two centromeric STRs (D6S1610, D6S1610 and D6S1552, respectively).

Optimization of the multiplex PCR for HLA typing and β‐thalassaemia mutations at the level of single heterozygous lymphoblasts involved splitting up the primers after 10 cycles in order to perform a second PCR round (Table I). This was necessary because (i) some of the fragments overlapped in size (only one fluorescent label is available for ALFExpress detection) and (ii) to reach high amplification efficiency for all primer pairs involved. In the second PCR round, 40 cycles were done. Strict requirements were taken into account to ensure the reliability of each individual test: 94.6–97.0% efficiency, 0–3.6% contamination and 0–5.7% allelic drop out (ADO, i.e. one of the alleles is not amplified) were accepted in order to make the diagnosis efficient and accurate (Table III).

Clinical PGD cycles

So far, six clinical PGD cycles have been performed for HLA‐matched embryos in two β‐thalassaemia families (Table IV); the third family is planned for their first treatment cycle in the near future. The β‐thalassaemia family 1 had three clinical PGD cycles with a large number of oocytes, good fertilization and cleavage. In the first cycle, 16 cumulus–oocyte complexes (COCs) were retrieved, 13 mature oocytes were injected and eight became fertilized. Seven embryos were biopsied (two cells); all cells were amplified (14 of 14), there was no contamination but there was ADO for the telomeric marker MOG3, the centromeric marker D6S1610 and the mutation simultaneously in one cell (each one out of 14 or 7.1%). There were five HLA‐non‐identical healthy embryos (one homozygous normal and four carriers) and one abnormal embryo (one HLA locus could not be amplified). One compacting HLA‐compatible carrier embryo was transferred on day 4 (Figure 3). The patient did not become pregnant. The remaining embryos were retested and their diagnosis was confirmed. In the second PGD cycle, 38 COCs were retrieved and 33 oocytes were injected. Twenty‐two oocytes became fertilized and 18 embryos were biopsied. In this case, only one blastomere was removed from the embryo. One embryo had no diagnosis because of amplification failure; 17 out of 18 embryos were diagnosed (94.4%): three were abnormal (one HLA locus could not be amplified), two were HLA‐identical but affected and 12 were HLA‐non‐identical (two homozygous normal, seven carriers and three affected). There was no contamination. All embryos were retested, the HLA typing was confirmed but, due to the one‐cell biopsy, two misdiagnoses had occurred at the level of the mutation (two out of 18 or 11.1% ADO): an HLA‐non‐identical homozygous normal embryo was shown to be a carrier and an HLA‐identical affected embryo was shown to be a carrier. The latter embryo had a poor morphology and would not have been considered for transfer. In the third cycle, 33 COCs were retrieved, 31 mature oocytes were injected and 29 oocytes became fertilized. Twenty‐two embryos were biopsied; 44 blastomeres were analysed. All cells were amplified (44 of 44); there was one ADO in the telomeric marker MOG3 (one out of 44 or 2.3%) and one in the mutation (one in 44 or 2.3%), one blank was contaminated by carry‐over (one in 44 or 2.3%). Six embryos were HLA‐identical (three carriers and three affected), 16 were HLA‐non‐identical (nine homozygous normal, six carriers and one affected). Two HLA‐identical carrier blastocysts were transferred on day 5; the third selected embryo was of poor quality and could not be frozen. The remaining embryos were checked post‐PGD and their diagnosis was confirmed. The patient was pregnant, as indicated by two consecutive positive HCG levels. The ultrasound at 7 weeks of pregnancy showed two sacs and three fetuses (a singleton and a monozygotic twin of which only the twin showed heart beats); however, at 8 weeks, no heart beats were observed and the pregnancy was lost. The aborted material was analysed and confirmed to be HLA‐identical to the affected child and carrier of the codon 39C>T mutation causing the β‐thalassaemia in the family. No distinction could be made between the three fetuses.

The second β‐thalassaemia family had three clinical PGD cycles with a lower number of oocytes, but good fertilization and cleavage. In the first cycle, eight COCs were retrieved. Seven mature oocytes were injected, five became fertilized and four embryos were biopsied. All blastomeres were amplified (eight out of eight); there was no ADO, but there were two contaminations in the blanks by carry‐over (two out of eight or 25%). The four embryos were HLA‐non‐identical carriers and they were cryopreserved because the couple requested it for consideration of later transfer. In the second cycle, seven COCs were retrieved, seven oocytes were mature and were injected. Four oocytes became fertilized; only one embryo could be biopsied (two cells). It was an HLA‐non‐identical carrier (two of two cells amplified, no ADO, no contamination); it was not cryopreserved because it was of poor quality and the diagnosis was confirmed post‐PGD. In the third cycle, 15 COCs were retrieved and 12 mature oocytes were injected. Eight oocytes became fertilized and six embryos were biopsied (one cell). All cells were amplified and there was no contamination. All embryos were HLA‐non‐identical; there was no ADO at the level of the STRs. However, the genetic constitution at the level of the Hb mutation was not clear in three embryos because of unexpected low amplification, resulting in no diagnosis for the mutations. Three embryos were carriers. In the post‐PGD check, the HLA typing was confirmed; five embryos were carriers of one of the Hb mutations and one embryo was affected.

In summary, of the total number of 58 embryos tested, results of combined PCR were as follows: 54 out of 58 (93.1%) successful, one amplification failure and three inconclusive to detect the thalassaemia‐related mutations. Three embryos were transferred and 50 embryos were retested. In all these embryos, the HLA typing was confirmed (100%). In six clinical cycles, two transfers were performed and one clinical pregnancy (defined as fetal heart beats on ultrasound) ensued that ended in an early miscarriage at 8 weeks.

Discussion

We developed a novel PGD approach that combines the genetic analysis of single‐gene disorders with indirect HLA typing on single cells. The purpose of this strategy is to ascertain to the parents that their next child is not only unaffected but could also provide HLA‐identical haematopoietic stem cells to transplant and eventually cure the affected child. Our strategy was validated in three families each with a child affected by β‐thalassaemia major. This genetic disease is a suitable model for such a strategy, for several reasons. First, β‐thalassaemia is predominantly caused by point mutations in the β‐globin gene that can be readily located by DNA sequencing. The mutations can then be detected by PCR/PGD. Secondly, β‐thalassaemia is a very common single‐gene disorder which is recognized as a worldwide public health problem (Weatherall and Clegg, 1996). Therefore, the number of candidate couples requesting such a diagnostic and therapeutic option may be considerable, so that, if clinically applied, it may represent an approved practice in the management of thalassaemic patients in childhood. Thirdly, thalassaemia is an autosomal recessive genetic disorder with a high frequency of heterozygotes, called silent carriers or β‐thalassaemia minor, which is a relatively mild medical condition. As a consequence, only embryos that are affected (i.e. homozygous for the disease‐related mutation) have to be excluded for implantation, thereby increasing the likelihood of finding suitable candidate embryos for transfer. Fourthly, alloSCT using HLA‐identical donors is the best treatment option for severe forms of β‐thalassaemia.

For the majority of patients with severe β‐thalassaemia, blood transfusions and iron chelation therapy are the mainstays of medical treatment. Even in Western countries and despite the availability of good medical treatment, ∼50% of affected patients die before the age of 35 years (Modell et al., 2000). On the other hand, the cure rate of thalassaemia with allogeneic bone marrow transplantation using HLA‐identical donors is ∼90% in children at an early stage of the disease. The problems with alloSCT are the availability of an HLA‐identical sibling donor and the transplant‐related toxicity, mainly due to graft‐versus‐host disease. With the current status of international registries such as the National Marrow Donor Program (NMDP), it was recently shown that at least one 6/6 matched donor or umbilical cord blood unit was available for 80% of patients (Krishnamurti et al., 2003). However, results of alloSCT using unrelated donors are less favourable compared with the use of HLA‐identical donors, although experience is limited. Out of a recent series of 32 patients, and using extended haplotypes in the majority of cases, the survival was 69% with a median follow‐up of 30 months and the incidence of chronic graft‐versus‐host disease was 25% (La Nasa et al., 2002). Related umbilical cord blood is an excellent alternative source of stem cells for patients with haemoglobinopathies (Locatelli et al., 2003). In a recent analysis of 44 patients reported to Eurocord, the 2‐year probability of survival for patients with thalassaemia was 79% and even 89% in those belonging to a low risk category. The study also showed a very low risk of graft‐versus‐host disease and suggested that optimization of the transplant programmes was likely to improve the results further.

From the technical standpoint, we were faced with the difficulty of combining PGD for β‐thalassaemia with HLA typing on a single‐cell level. PGD for HLA typing is very complex because the locus is highly polymorphic. Taking into account the size of the region (4 × 103 kb), the different regions (class I and class II), the large number of loci (A, B, C, DR, DP and DQ) and the possible recombination within the region (Thomsen et al., 1985; Sullivan et al., 1997), the development of a reliable single‐cell PCR for each couple requesting a matched donor for their affected child would be time consuming. A direct HLA typing approach using allele‐specific primers has been reported for Fanconi’s anaemia (Verlinsky et al., 2001). Such a strategy that is based on the absence or presence of an allele may be susceptible to technical errors including ADO and contamination, resulting in misdiagnosis.

The use of microsatellite markers for PGD has already been shown to be an elegant methodology to improve PGD protocols for cystic fibrosis, an autosomal recessive disease for which >1000 distinct mutations are known and thus making the development of mutation‐specific protocols for PGD unfeasible (Dreesen et al., 2000; Goossens et al., 2000; Moutou et al., 2002). Similarily, our approach for indirect HLA typing is to ascertain the inheritance of the matching haplotypes by fingerprinting the HLA locus using informative STR markers. By selecting two outer markers (a telomeric and a centromeric marker that are preferentially closely associated with the loci because recombination could be observed) and two inner markers (one within class I‐A and ‐B and one within class II‐DR and DQ) for single‐cell multiplex PCR, we indirectly type the HLA locus by segregation analysis and control for recombination within the region. So far, we have not observed recombination in lymphocytes or embryos, indicating that selection of microsatellite markers was highly adequate. Informativity testing has now been done in 12 families using the bank of seven STRs; only in one family was recombination observed using the external centromeric markers. One particular combination of markers can be used in five families (our unpublished data). The reliability of this strategy is further ensured by carefully analysing in advance the limitations of each individual protocol on a large number of heterozygous lymphoblasts. The accuracy for the HLA typing was 100%; combined PGD was successful in 54 of 58 (93%) cases and the few technical failures were due to (i) unexpected low amplification resulting in no diagnosis; or (ii) the fact that ADO occurred when only one cell from the embryo could be analysed, resulting in misdiagnosis. In the latter case, misdiagnosis could have been avoided using two‐cell biopsy. The debate of one‐ versus two‐cell biopsy is still ongoing. Retrospectively we found that the implantation capacity of an embryo is not impaired when removing two cells (18% implantation rate in Van de Velde et al., 2000). However, Pickering et al. (2003) reported an implantation rate of 27% that may reflect their policy to biopsy only one cell. A prospective randomized controlled study has been set up at our centres to evaluate the implantation capacity after one‐ versus two‐cell biopsy in relation to the efficiency and reliability of the test (occurrence of no diagnosis or misdiagnosis by ADO, contamination or amplification failure).

Another important advantage of our strategy is its applicability to other diseases where a similar combination of excluding a genetic disorder and selection for an HLA‐identical embryo may be requested, such as sickle cell anaemia, Fanconi’s anaemia, Wiskott Aldrich’s syndrome and severe chronic granulomatosis. By extending our experience to these indications, we were able to set up a bank of seven markers to ensure informativity in all families accepted in our programme (the feasibility of the approach has been validated now in nine different families (our unpublished data). The establishment of such an STR bank enables us to substantially shorten the waiting list to start a treatment cycle to a median of 4–8 weeks because a number of STR combinations have already been worked out and can be used for other families in which informativity can be ensured by using the same STRs. It is important to keep the time to develop a family‐tailored PCR protocol as short as possible. Obviously there is a positive impact on overall cost‐effectiveness, but there are also clinical arguments. Patients with congenital anaemias can survive for many years. However, they will develop a number of time‐dependent complications such as liver fibrosis (thalassaemia) and vaso‐occlusive events (sickle cell anaemia). Such complications cause direct morbidity and mortality but may also reduce the success rate of an eventual transplant procedure. For patients with congenital immune deficiencies, the continuous risk of potentially dangerous infections will decrease if the transplant is performed earlier.

We also have to consider some limitations of our technique. First, the method is labour‐intensive and thus the financial cost is high. On the level of a health care programme, this aspect must be compared with the probably higher costs related to the use of unrelated donors or a continued programme of standard medical treatment (if that is an option) for a number of decades with a no‐cure perspective. Secondly, the lower the number of embryos, the lower the probability of having a suitable one. The number of embryos available during a cycle is dependent on the efficacy of the ovarian stimulation protocol on the one hand and the age of the mother on the other. Theoretically, one in four embryos are HLA identical and three out of four will not carry the mutation or be heterozygous so that only three out of 16 will be transferable. Thirdly, the intrinsic implantation capacity of the embryos is highly variable and the chance of a successful pregnancy may be not higher than 26% (Van de Velde et al., 2000), leaving room for further improvement. Finally, our strategy involving embryo selection is likely to raise some ethical controversy. We argue that the selection of a healthy child by means of PGD is an accepted alternative to other techniques of PND (ESHRE, 2002) for patients who wish to avoid offspring with a genetic disease. On the other hand, a selection based on HLA type may be more difficult to accept. One might reason that such a genetic background does not a priori infer any advantage or disadvantage for the newborn and that anyhow this genotype is inherited in 25% of cases, only by chance. In addition, the issue of instrumentalization of the child to be born may be raised: the child is created to cure another one. This difficult ethical issue has been discussed previously (Pennings et al., 2002) and can at least be partially addressed by careful genetic and psychological counselling of the parents so that their motivation and capacity to raise another child can be fully assessed and integrated in the selection of candidates for this strategy.

In summary, we have developed a method to combine PGD for β‐thalassaemia and a number of monogenic diseases with the selection of HLA‐identical embryos so that the unaffected child to be born can be a donor of umbilical cord blood stem cells to transplant and eventually cure an affected sibling. This novel approach that is characterized by its reliability and a relatively short work up time of development and application may represent a new and curative therapeutic option for many potential candidates, although further validation on a larger number of cases remains necessary.

Acknowledgements

The authors wish to thank their colleagues from the Centre for Reproductive Medicine and the Centre for Medical Genetics, in particular Bart Saerens for informativity testing. We are grateful to Drs Y.Gillerot (Centre de Génétique, Université Catholique de Louvain, Belgium), N.Van Regemorter (Centre de Génétique, Université Libre de Bruxelles, Belgium) and R.Frydman (Hôpital Antoine‐Béclère, Paris, France) for referring the patients. Supported by grants from the Fund for Scientific Research Flanders (FWO‐Vlaanderen) and the Research Council of the VUB (OZR).

Figure 1. Mapping within the β‐globin gene of the distinct mutations (codon 39 C>T; IVS1+6T>C and IVS1+110G>A, and codon 8/9+G and IVSI+110G>A), restriction enzymes (NlaIV; SfaNI and AvaII; PflFI and AvaII) and primers (sets HB1/HB2; HB1/HB3; hemi‐nested PCO5/HB3 and HB9/HB3) for the three β‐thalassaemia families, respectively (not to scale).

Figure 1. Mapping within the β‐globin gene of the distinct mutations (codon 39 C>T; IVS1+6T>C and IVS1+110G>A, and codon 8/9+G and IVSI+110G>A), restriction enzymes (NlaIV; SfaNI and AvaII; PflFI and AvaII) and primers (sets HB1/HB2; HB1/HB3; hemi‐nested PCO5/HB3 and HB9/HB3) for the three β‐thalassaemia families, respectively (not to scale).

Figure 2. Segregation of the STRs in the HLA locus. The telomeric marker MOG3 is closely linked to class I‐A; markers D6S1571, D6S1611 and D6S1560 are localized between class I‐A and ‐B; marker D6S2443 is located within class II‐DQ; and markers D6S1610 and D6S1552 are centromeric to class II.

Figure 2. Segregation of the STRs in the HLA locus. The telomeric marker MOG3 is closely linked to class I‐A; markers D6S1571, D6S1611 and D6S1560 are localized between class I‐A and ‐B; marker D6S2443 is located within class II‐DQ; and markers D6S1610 and D6S1552 are centromeric to class II.

Figure 3. PGD for β‐thalassaemia and HLA‐identical embryos in family 1: fragment analysis showing the result after multiplex PCR. (A) The first arm of the second reaction with markers MOG3, D6S2443 and the β‐thalassaemia (Hb) mutation in codon 39C>T (abolishes an NlaIV restriction site; the normal allele is 198 bp, the affected allele is 247 bp). (B) The second arm of the second reaction with markers D6S1610 and DS1571. Lane 1 represents the external standard molecular weight marker from 100 to 300 bp; standards of 100 and 300 bp are included in each lane. DNA of the affected child (ad haplotype) is shown in lane 2. The results of one of two cells of embryos are shown: ad identical carrier (3), ad identical affected (4), bc not‐identical carrier (5) and bd not‐identical homozygous normal (6).

Figure 3. PGD for β‐thalassaemia and HLA‐identical embryos in family 1: fragment analysis showing the result after multiplex PCR. (A) The first arm of the second reaction with markers MOG3, D6S2443 and the β‐thalassaemia (Hb) mutation in codon 39C>T (abolishes an NlaIV restriction site; the normal allele is 198 bp, the affected allele is 247 bp). (B) The second arm of the second reaction with markers D6S1610 and DS1571. Lane 1 represents the external standard molecular weight marker from 100 to 300 bp; standards of 100 and 300 bp are included in each lane. DNA of the affected child (ad haplotype) is shown in lane 2. The results of one of two cells of embryos are shown: ad identical carrier (3), ad identical affected (4), bc not‐identical carrier (5) and bd not‐identical homozygous normal (6).

Table I.

Multiplex PCR conditions in the three β‐thalassaemia families

Couple First round PCR 10 cycles annealing at 55°C Second round PCR split 40 cycles annealing at 55 or 53°Ca 
HB1F + HB2R (1.6 pmol) HB1F + HB2R (1.6 pmol) 
 MOG3F + MOG3R (15 pmol) MOG3F + MOG3R (15 pmol) 
 D6S1571F + D6S1571R (2.5 pmol) D6S2443F + D6S2443R (1.6 pmol) 
 
 D6S2443F + D6S2443R (1.6 pmol) D6S1571F + D6S1571R (5 pmol) 
 D6S1610F + D6S1610R (2.5 pmol) D6S1610F + D6S1610R (2.5 pmol) 
 
HB1F + HB3R (1 pmol) HB1F + HB3R (1 pmol) 
 MOG3F + MOG3R(7.5 pmol) MOG3F + MOG3R (7.5 pmol) 
 D6S1611F + D6S1611R (2.5 pmol) D6S2443F  +  D6S2443R (2.5 pmol) 
 
 D6S2443F + D6S2443R (2.5 pmol) D6S1611F + D6S1611R (2.5 pmol) 
 D6S1610F + D6S1610R (2.5 pmol) D6S1610F + D6S1610R (2.5 pmol) 
 
PCO5F + HB3R (2.5 pmol) HB9F + HB3Ra (15 pmol) 
 
 MOG3F + MOG3R (7.5 pmol) MOG3F + MOG3R (7.5 pmol) 
 D6S1560F + D6S1560R (2.5 pmol) D6S2443F + D6S2443R (5 pmol) 
 
 D6S2443F + D6S2443R (5 pmol) D6S1560F + D6S1560R (2.5 pmol) 
 D6S1552F + D6S1552R (5 pmol) D6S1552F + D6S1552R (5 pmol) 
Couple First round PCR 10 cycles annealing at 55°C Second round PCR split 40 cycles annealing at 55 or 53°Ca 
HB1F + HB2R (1.6 pmol) HB1F + HB2R (1.6 pmol) 
 MOG3F + MOG3R (15 pmol) MOG3F + MOG3R (15 pmol) 
 D6S1571F + D6S1571R (2.5 pmol) D6S2443F + D6S2443R (1.6 pmol) 
 
 D6S2443F + D6S2443R (1.6 pmol) D6S1571F + D6S1571R (5 pmol) 
 D6S1610F + D6S1610R (2.5 pmol) D6S1610F + D6S1610R (2.5 pmol) 
 
HB1F + HB3R (1 pmol) HB1F + HB3R (1 pmol) 
 MOG3F + MOG3R(7.5 pmol) MOG3F + MOG3R (7.5 pmol) 
 D6S1611F + D6S1611R (2.5 pmol) D6S2443F  +  D6S2443R (2.5 pmol) 
 
 D6S2443F + D6S2443R (2.5 pmol) D6S1611F + D6S1611R (2.5 pmol) 
 D6S1610F + D6S1610R (2.5 pmol) D6S1610F + D6S1610R (2.5 pmol) 
 
PCO5F + HB3R (2.5 pmol) HB9F + HB3Ra (15 pmol) 
 
 MOG3F + MOG3R (7.5 pmol) MOG3F + MOG3R (7.5 pmol) 
 D6S1560F + D6S1560R (2.5 pmol) D6S2443F + D6S2443R (5 pmol) 
 
 D6S2443F + D6S2443R (5 pmol) D6S1560F + D6S1560R (2.5 pmol) 
 D6S1552F + D6S1552R (5 pmol) D6S1552F + D6S1552R (5 pmol) 

In general, two consecutive PCR rounds were performed: in the first round (10 cycles, annealing at 55°C) the PCR mix contained all primers, and in the second round (40 cycles, annealing at 55°C or a53°C) primer combinations were split up.

Table II.

Linkage of the STRs to the loci in β‐thalassaemia family 1

Segregation Father Mother Affected child Sibling 
 
Telomeric MOG3 147 133 121 137 147 137 147 121 
A locus A32 A29 A26 A28 A32 A28 A32 A26 
A/B D6S1571 168 156 156 156 168 156 168 156 
B locus B35 B45 B38 B35 B35 B35 B35 B38 
DR locus DR11 DR01 DR13 DR10 DR11 DR10 DR11 DR13 
DQ D6S2443 169 171 169 171 169 171 169 169 
Centromeric D6S1610 135 137 135 131 135 131 135 135 
Segregation Father Mother Affected child Sibling 
 
Telomeric MOG3 147 133 121 137 147 137 147 121 
A locus A32 A29 A26 A28 A32 A28 A32 A26 
A/B D6S1571 168 156 156 156 168 156 168 156 
B locus B35 B45 B38 B35 B35 B35 B35 B38 
DR locus DR11 DR01 DR13 DR10 DR11 DR10 DR11 DR13 
DQ D6S2443 169 171 169 171 169 171 169 169 
Centromeric D6S1610 135 137 135 131 135 131 135 135 
Table III.

Results from multiplex PCR combining indirect HLA typing (four markers) and the β‐thalassaemia mutation (Hb) on heterozygous single lymphoblasts from three β‐thalassaemia families: efficiency and accuracy (ADO and contamination) of the different tests

 Family 1 Family 2 Family 3 
Efficiency (%) 46/48 (95.8) 62/64 (97.0) 53/56 (94.6) 
Accuracy    
 ADO (%)    
  Telomeric 1/46 (2.1) 2/62 (3.2) 1/53 (1.9) 
  A/B 1/46 (2.1) 1/62 (1.6) 1/53 (1.9) 
  DQ 1/46 (2.1) 3/62 (4.8) 3/53 (5.7) 
  Centromeric 1/46 (2.1) 0/62 1/53 (1.9) 
  Hb 1/32 (3.1) 2/62 (3.2) 2/53 (3.8) 
 Contamination (%)    
  Telomeric 0/48 0/64 1/56 (1.8) 
  A/B 0/48 0/64 2/56 (3.6) 
  DQ 0/48 0/64 1/56 (1.8) 
  Centromeric 0/48 0/64 1/56 (1.8) 
  Hb 0/48 0/64 2/56 (3.6) 
 Family 1 Family 2 Family 3 
Efficiency (%) 46/48 (95.8) 62/64 (97.0) 53/56 (94.6) 
Accuracy    
 ADO (%)    
  Telomeric 1/46 (2.1) 2/62 (3.2) 1/53 (1.9) 
  A/B 1/46 (2.1) 1/62 (1.6) 1/53 (1.9) 
  DQ 1/46 (2.1) 3/62 (4.8) 3/53 (5.7) 
  Centromeric 1/46 (2.1) 0/62 1/53 (1.9) 
  Hb 1/32 (3.1) 2/62 (3.2) 2/53 (3.8) 
 Contamination (%)    
  Telomeric 0/48 0/64 1/56 (1.8) 
  A/B 0/48 0/64 2/56 (3.6) 
  DQ 0/48 0/64 1/56 (1.8) 
  Centromeric 0/48 0/64 1/56 (1.8) 
  Hb 0/48 0/64 2/56 (3.6) 

ADO, allelic drop out.

Table IV.

Results from clinical PGD cycles

 Family 1 Family 1 Family 1 Family 2 Family 2 Family 2 Total 
 Cycle 1 Cycle 2 Cycle 3 Cycle 1 Cycle 2 Cycle 3  
COC 16 38 33 15 117 
MII 13 33 31 12 103 
2PN 22 29 76 
Embryos biopsied 18 22 58 
Efficiency (%) 14/14 (100) 17/18 (94.4) 44/44 (100) 8/8 (100) 2/2 (100) 6/6 (100) 91/92 (98.9) 
Accuracy        
 ADO (%)        
  Telomeric 1/14 (7.1) 0/17 1/44 (2.3) 0/8 0/2 0/6  
  A/B 0/14 0/17 0/44 0/8 0/2 0/6  
  DQ 0/14 0/17 0/44 0/8 0/2 0/6  
  Centromeric 1/14 (7.1) 0/17 0/44 0/8 0/2 0/6  
  Hb 1/14 (7.1) 2/17 (11.8) 1/44 (2.3) 0/8 0/2 0/6  
 Contamination (%)        
  Telomeric 0/14 0/18 1/44 (2.3) 0/8 0/2 0/6 1/92 (1.1) 
  A/B 0/14 0/18 0/44 0/8 0/2 0/6  
  DR/DQ 0/14 0/18 0/44 0/8 0/2 0/6  
  Centromeric 0/14 0/18 0/44 0/8 0/2 0/6  
  Hb 0/14 0/18 0/44 2/8 (25.0) 0/2 0/6 2/92 (2.2) 
Identical        
 Homozygous normal 
 Carrier 4 (7.0) 
 Affected 5 (8.8) 
Not‐identical        
 Homozygous normal 12 (21.0) 
 Carrier 25 (43.9) 
 Affected 4 (7.0) 
No diagnosis 3a 4/58 (6.9) 
Abnormal 4 /57 (7.0) 
Transferred 
HCG      
Sacs      
Fetuses      
Heart beats at 7 weeks      
 Family 1 Family 1 Family 1 Family 2 Family 2 Family 2 Total 
 Cycle 1 Cycle 2 Cycle 3 Cycle 1 Cycle 2 Cycle 3  
COC 16 38 33 15 117 
MII 13 33 31 12 103 
2PN 22 29 76 
Embryos biopsied 18 22 58 
Efficiency (%) 14/14 (100) 17/18 (94.4) 44/44 (100) 8/8 (100) 2/2 (100) 6/6 (100) 91/92 (98.9) 
Accuracy        
 ADO (%)        
  Telomeric 1/14 (7.1) 0/17 1/44 (2.3) 0/8 0/2 0/6  
  A/B 0/14 0/17 0/44 0/8 0/2 0/6  
  DQ 0/14 0/17 0/44 0/8 0/2 0/6  
  Centromeric 1/14 (7.1) 0/17 0/44 0/8 0/2 0/6  
  Hb 1/14 (7.1) 2/17 (11.8) 1/44 (2.3) 0/8 0/2 0/6  
 Contamination (%)        
  Telomeric 0/14 0/18 1/44 (2.3) 0/8 0/2 0/6 1/92 (1.1) 
  A/B 0/14 0/18 0/44 0/8 0/2 0/6  
  DR/DQ 0/14 0/18 0/44 0/8 0/2 0/6  
  Centromeric 0/14 0/18 0/44 0/8 0/2 0/6  
  Hb 0/14 0/18 0/44 2/8 (25.0) 0/2 0/6 2/92 (2.2) 
Identical        
 Homozygous normal 
 Carrier 4 (7.0) 
 Affected 5 (8.8) 
Not‐identical        
 Homozygous normal 12 (21.0) 
 Carrier 25 (43.9) 
 Affected 4 (7.0) 
No diagnosis 3a 4/58 (6.9) 
Abnormal 4 /57 (7.0) 
Transferred 
HCG      
Sacs      
Fetuses      
Heart beats at 7 weeks      

Two β‐thalassaemia families each underwent three ICSI cycles; the efficiency and the accuracy of the PGD protocol for β‐thalassemia in combination with HLA typing were highly acceptable. After diagnosis, healthy HLA‐identical embryos were transferred. In six cycles, one clinical pregnancy was obtained.

COC, cumulus–oocyte complex; MII, mature oocyte; 2PN, normally fertilized oocyte; ADO, allelic drop out.

aInconclusive for the thalassaemia‐related mutations.

References

De Rycke M, Van de Velde H, Sermon K, Lissens W, De Vos A, Vandervorst M, Vanderfaeillie A, Van Steirteghem A and Liebaers I (
2001
) Preimplantation genetic diagnosis for sickle‐cell anemia and for beta‐thalassemia.
Prenat Diagn
 
21
,
214
–222.
Dreesen JC, Jacobs LJ, Bras M, Herbergs J, Dumoulin JC, Geraeds JP, Evers JL and Smeets HJ (
2000
) Multiplex PCR of polymorphic markers flanking the CFTR gene; a general approach for preimplantation genetic diagnosis of cystic fibrosis.
Mol Hum Reprod
 
6
,
391
–396.
ESHRE (
2002
) ESHRE Preimplantation Genetic Diagnosis Consortium: Data Collection III (May 2001).
Hum Reprod
 
17
,
233
–246.
Foissac A, Salhi M and Cambon‐Thomsen A (
2000
) Microsatellites in the HLA region: 1999 update.
Tissue Antigens
 
55
,
477
–509.
Gaziev D et al. (
2000
) Bone marrow transplantation from alternative donors for thalassemia: HLA‐phenotypically identical relative and HLA‐nonidentical sibling or parent transplants.
Bone Marrow Transplant
 
25
,
815
–821.
Goossens V, Sermon K, Lissens W, Vandervorst M, Vanderfaeillie A, De Rycke M, De Vos A, Hendrix P, Van de Velde H, Van Steirteghem A and Liebaers I (
2000
) Clinical application of preimplantation genetic diagnosis for cystic fibrosis.
Prenat Diagn
 
20
,
571
–581.
Hardison RC, Chui DH, Riemer CR, Miller W, Carver MF, Molchanova TP, Efremov GD and Huisman TH (
1998
) Access to a syllabus of human hemoglobin variants (1996) via the World Wide Web.
Hemoglobin
 
22
,
113
–127.
Joris H, Nagy Z, Van de Velde H, De Vos A and Van Steirteghem, A (
1998
) Intracytoplasmic sperm injection: laboratory set‐up and injection procedure.
Hum Reprod
 
13
Suppl 1,
76
–86.
Joris H, De Vos A, Janssens R, Devroey P, Liebaers I and Van Steirteghem A (
2003
) Comparison of the results of human embryo biopsy and outcome of PGD after zona drilling using acid Tyrode medium or a laser.
Hum Reprod
 
18
,
1896
–1902.
Kolibianakis E, Bourgain C, Albano C, Osmanagaoglu K, Smitz J, Van Steirteghem A and Devroey P (
2002
) Effect of ovarian stimulation with recombinant follicle‐stimulating hormone, gonadotropin releasing hormone antagonists, and human chorionic gonadotropin on endometrial maturation on the day of oocyte pick‐up.
Fertil Steril
 
78
,
1025
–1029.
Krishnamurti L, Abel S, Maiers M and Flesh S (
2003
) Availability of unrelated donors for hematopoietic stem cell transplantation for hemoglobinopathies.
Bone Marrow Transplant
 
31
,
547
–550.
La Nasa G, Giardini C, Argiolu F, Locatelli F, Arras M, De Stefano P, Ledda A, Pizzati A, Sanna MA, Vacca A, Lucarelli G and Contu L (
2002
) Unrelated donor bone marrow transplantation for thalassemia: the effect of extended haplotypes.
Blood
 
99
,
4350
–4356.
Li HH, Gyllensten UB, Cui XF, Saiki RK, Erlich HA and Arnheim N (
1988
) Amplification and analysis of DNA sequences in single human sperm and diploid cells.
Nature
 
335
,
414
–417.
Liebaers I, Sermon K, Staessen C, Joris H, Lissens W, Van Assche E, Nagy P, Bonduelle M, Vandervorst M, Devroey P and Van Steirteghem A (
1998
) Clinical experience with preimplantation genetic diagnosis and intracytoplasmic sperm injection.
Hum Reprod
 
13
Suppl 1,
186
–195.
Lissens W and Sermon K (
1997
) Preimplantation genetic diagnosis: current status and new developments.
Hum Reprod.
 ,
12
,
1756
–1761.
Locatelli, F et al. (
2003
) Related umbilical cord blood transplantation in patients with thalassemia and sickle cell disease.
Blood
 
101
,
2137
–2143.
Lucarelli G, Galimberti M, Giardini C, Polchi P, Angelucci, E, Barociani D, Erer B and Gaziev D (
1998
) Bone marrow transplantation in thalassemia. The experience of Pesaro.
Ann NY Acad Sci
 
850
,
270
–275.
Lucarelli G, Clift RA, Galimberti M, Angelucci E, Giardini C, Barociani D, Adreani, M, Gaziev D, Erer B, Ciaroni A, D’Adamo F, Albertini F and Muretto P (
1999
) Bone marrow transplantation in adult thalassemic patients.
Blood
 
93
,
1164
–1167.
Lucarelli G, Andreani M and Angelucci E (
2002
) The cure of thalassemia by bone marrow transplantation.
Blood
 
16
,
81
–85.
Modell B, Khan M and Darlison M (
2000
) Survival in beta‐thalassaemia major in the UK: data from the UK Thalassaemia Register.
Lancet
 
355
,
2051
–2052.
Moutou C, Gardes N and Viville S (
2002
) Multiplex PCR combining deltaF508 mutation and intragenic microsatellites of the CFTR gene for pre‐implantation genetic diagnosis (PGD) of cystic fibrosis.
Eur J Hum Genet
 
10
,
231
–238.
Olivieri NF (
1999
) The beta‐thalassemias.
N Engl J Med
 
341
,
99
–109.
Pennings G, Schots R and Liebaers I (
2002
) Ethical considerations on preimplantation genetic diagnosis for HLA typing to match a future child as a donor of haematopoietic stem cells to a sibling.
Hum Reprod
 
17
,
534
–538.
Pickering S, Polidoropoulos N, Caller J, Scriven P, Ogilvie CM, Braude P, F.R.C.O.G. and the PGD Study Group (
2003
) Strategies and outcomes of the first 100 cycles of preimplantation genetic diagnosis at the Guy’s and St. Thomas’ Center.
Fertil Steril
 
79
,
81
–90.
Sermon K, De Vos A, Van de Velde H, Seneca S, Lissens W, Joris H, Vandervorst M, Van Steirteghem A and Liebaers I (
1998
) Fluorescent PCR and automated fragment analysis for the clinical application of preimplantation genetic diagnosis of myotonic dystrophy (Steinert’s disease).
Mol Hum Reprod
 
4
,
791
–796.
Sermon K, Goossens V, Seneca S, Lissens W, De Vos A, Vandervorst M, Van Steirteghem A and Liebaers I (
1998
) Preimplantation diagnosis for Huntington’s disease (HD): clinical application and analysis of the HD expansion in affected embryos.
Prenat Diagn
 
18
,
1427
–1436.
Staessen C, Camus M, Clasen K, De Vos A and Van Steirteghem A(
1999
) Conventional in‐vitro fertilization versus intracytoplasmic sperm injection in sibling oocytes from couples with tubal infertility and normozoospermic semen.
Hum Reprod
 
14
,
2474
–2479.
Sullivan KA, Wolfe MA, Lopez M, Jaspan JB and Bryer‐Ash M (
1997
) First report of recombination between the HLA‐DR and HLA‐DQ loci within a family.
Hum Immunol
 
57
,
37
–43.
Thomsen M, Cambon‐Thomsen A, Abbal M, Calot M, Arnaud J and Ohayon E (
1985
) Recombination between HLA‐A and C and between HLA‐B and complement locus C4 in the same individual.
Tissue Antigens
 
25
,
75
–78.
Van de Velde H, De Vos A, Sermon K, Staessen C, De Rycke M, Van Assche E, Lissens W, Vandervorst M, Van Ranst H, Liebaers I and Van Steirteghem A (
2000
) Embryo implantation after biopsy of one or two cells from cleavage‐stage embryos with a view to preimplantation genetic diagnosis.
Prenat Diagn
 
20
,
1030
–1037.
Ventura M, Gibaud A, Le Pendu J, Hillaire D, Gerard G, Vitrac D and Oriol R (
1988
) Use of a simple method for the Epstein–Barr virus transformation of lymphocytes from members of large families of Reunion Island.
Hum Hered
 
38
,
36
–43.
Verlinsky Y, Rechitsky S, Schoolcraft W, Strom C and Kuliev A (
2001
) Preimplantation diagnosis for Fanconi anemia combined with HLA matching.
J Am Med Assoc
 
285
,
3130
–3133.
Weatherall DJ and Clegg JB (
1996
) Thalassemia—a global public health problem.
Nature Med
 
2
,
847
–849.