Singling out genetic disorders and disease

Preimplantation genetic diagnosis (PGD) involves testing of single cells biopsied from oocytes and/or embryos generated in vitro. As only embryos unaffected for a given genetic condition are transferred to the uterus, it avoids prenatal diagnosis and termination of pregnancy. Follow-up data from PGD pregnancies, deliveries and children show an acceptable live birth rate and, so far, no detrimental effects of the procedure have been observed. Of course, the long-term health outcome is currently unknown. PGD was first performed in 1990 and remained an experimental procedure for a number of years. Now, two decades later, it is regarded as an established alternative to prenatal diagnosis: its use has expanded, the range of applications has broadened, and continuous technical progress in single-cell testing has led to high levels of efficiency and accuracy. The current gold standard methods (single-cell multiplex-PCR for monogenic diseases and interphase fluorescence in situ hybridization for chromosomal aberrations) are being replaced by single-cell whole genome amplification and array technology. These generalized methods substantially reduce the pre-PGD workload and allow more automated genome-wide analysis. The implementation of laboratory accreditation schemes brings the field at the same level of routine diagnostics. This article reviews the state of the art and considers indications, accuracy and current technical changes in the field of PGD.


Introduction
Preimplantation genetic diagnosis (PGD) is an alternative to prenatal diagnosis involving the biopsy and genetic testing of single cells from in vitro obtained oocytes and/ or preimplantation embryos. Only embryos shown to be free of the genetic defect under study are transferred to the uterus of the patient. PGD is performed for couples at high risk of transmitting a genetic condition to their children; it offers the advantage of circumventing an invasive prenatal diagnosis and therapeutic abortion.
Preimplantation genetic screening (PGS) involves the selection of euploid embryos to improve in vitro fertili za tion (IVF) results and to avoid pregnancies with chromo somal abnormalities [1]. Considering the increasing risk for aneuploidy with advanced maternal age and findings of high aneuploidy rates in spontaneous abortions, PGS has been offered to specific IVF patient groups: patients of advanced maternal age and patients with recurrent IVF failure or repeated miscarriages (not due to trans locations). For many years, PGS using fluorescence in situ hybridization (FISH) for 5 to 12 chromosomes has been applied worldwide but without real validation of its efficiency. In recent years, several randomized controlled trials have failed to show a benefit for PGS, as summar ized in the metaanalysis by Checa and colleagues [2]. Biological (embryonic mosaicism) and technical (limita tions of FISH and negative impact of cleavagestage biopsy) arguments were proposed to explain the lack of benefit. Further clinical practice now awaits new random ized controlled trials to prove that PGS with biopsy at other stages, and using new array methods with full chromosome analysis, improves live birth rates and reduces miscarriage rates [3]. PGD and PGS rely on the same technology, but since PGS is offered to (sub)fertile couples without genetic diseases, it is mostly outside the scope of this review.
PGD developed in the wake of human IVF and PCR technology. The first children born after PGD were reported by Handyside et al. in 1990 [4]. In these initial cycles, PCR was used for gender determination in families with Xlinked diseases. Later, FISH became the standard method for sexing, and also for chromosomal aberrations, while PCRbased methods were used for the detection of single gene defects. It is essential that these techniques are adapted to the singlecell level and are thoroughly validated before clinical application.
PGD requires a multidisciplinary team with a close collaboration and excellent communication between the assisted reproduction unit and the medical genetics unit. Preferentially, both units work within the same institute. Alternatively, transport PGD can be set up, and IVF

Abstract
Preimplantation genetic diagnosis (PGD) involves testing of single cells biopsied from oocytes and/ or embryos generated in vitro. As only embryos unaffected for a given genetic condition are transferred to the uterus, it avoids prenatal diagnosis and termination of pregnancy. Follow-up data from PGD pregnancies, deliveries and children show an acceptable live birth rate and, so far, no detrimental effects of the procedure have been observed. Of course, the long-term health outcome is currently unknown. PGD was first performed in 1990 and remained an experimental procedure for a number of years. Now, two decades later, it is regarded as an established alternative to prenatal diagnosis: its use has expanded, the range of applications has broadened, and continuous technical progress in single-cell testing has led to high levels of efficiency and accuracy. The current gold standard methods (single-cell multiplex-PCR for monogenic diseases and interphase fluorescence in situ hybridization for chromosomal aberrations) are being replaced by single-cell whole genome amplification and array technology. These generalized methods substantially reduce the pre-PGD workload and allow more automated genomewide analysis. The implementation of laboratory accreditation schemes brings the field at the same level of routine diagnostics. This article reviews the state of the art and considers indications, accuracy and current technical changes in the field of PGD. treatment (hormonal stimulation, oocyte retrieval and in vitro fertilization, embryo culture and embryo transfer) is carried out at a satellite assistedreproduction unit. Only the biopsied embryonic cell samples are transported, often over long distances, to a genetics unit specializing in singlecell diagnosis.
Extensive data on PGD cycles, pregnancies, deliveries and children have been collected by the European Society for Human Reproduction and Embryology (ESHRE) PGD consortium since 1997 and, although not worldwide, the data sets offer comprehensive insights into this particular field of singlecell testing [5].

Indications for preimplantation genetic diagnosis
The main indications for PGD are chromosomal abnor mali ties, Xlinked disorders and single gene or mono genic disorders. The majority of PGD cycles for chromosomal abnormalities are aimed at reciprocal and Robertsonian translocations, while cases with inversions or insertions are less frequent. For the monogenic dis orders, PGD was initially applied for the same indica tions as in prenatal diagnosis. According to the latest ESHRE PGD consortium data, the most common indications for autosomal recessive disorders are cystic fibrosis, spinal muscular atrophy and hemoglobinopathies [6]. For the autosomal dominant disorders, myotonic dystrophy type 1, neurofibromatosis and Huntington's disease are the most frequently requested indications; for the Xlinked disorders, PGD is mainly carried out for Duchenne's muscular dystrophy, hemophilia and fragile X syndrome. Sexing with FISH has been most frequently used for X linked disorders but more and more specific DNA diagnoses have been developed. Specific DNA diagnosis has important advantages: first, healthy male embryos are not discarded; and second, female carriers can be identified and excluded from transfer or not, according to the wishes of the patient and the policy of the center.
PGD is currently available for more than 200 mono genic diseases and has lately also been applied for indica tions, such as cancer predisposition syndromes and other lateonset diseases, for which prenatal diagnosis is ethically difficult [7,8]. For cancer predisposition syn dromes that are not fully penetrant and for which some form of therapeutic measures may be available, prenatal diagnosis and termination of pregnancy remain contro versial. As the preimplantation embryo is often con sidered as having less moral value than a fetus, PGD seems a more acceptable option.
Another more recent indication is human leucocyte antigen (HLA) typing of preimplantation embryos to select an embryo that is HLA compatible with an affected sibling. At birth, hematopoietic stem cells from the cord blood of the saviour baby are then used to transplant the sick sibling. HLA typing alone is carried out for acquired diseases, such as severe aplastic anemia, or HLA typing is combined with the detection of mutations underlying immunodeficiencies and hemoglobin disorders [9 11]. The major ethical objection here is that the future child may be regarded as an instrument and not as an autono mous person. This concern was studied in depth and it was concluded that considering the efforts of the parents to cure the sick child and their wish for another child, it is unlikely that they would treat the saviour child solely as a donor. In addition, it is universally accepted to rely on an existing HLAcompatible child as a donor of hemato poietic stem cells [12].
So far, few clinical cycles have been carried out for families carrying heteroplasmic mitochondrial (mt)DNA mutations [1315]. The genetics of mtDNA is quite com plex and the proportion of mutant mtDNA trans mitted from mother to offspring will vary due to a genetic bottleneck in the oocytes [16]. It is another ethically difficult indication, as the possibility exists that there are no zeromutation embryos for transfer but just embryos with a low mutation load (that is, the ratio of mutant to normal mtDNA) under a certain diseasespecific threshold, meaning that the risk for an affected child is not eliminated but only reduced. PGD for mtDNA mutations can only be offered reliably when certain criteria are fulfilled: a close correlation between the mutation load and disease severity is a first requirement; second, there should be no change in mutation load with time; and third, mutant mtDNA should be uniformly distributed over all blastomeres of the cleavagestage embryo [17]. For many mtDNA mutations the latter information is not available. Therefore, it is recommended to develop PGD within a scientific research protocol and to counsel prospective parents adequately and inform them that a first PGD cycle may be carried out merely to gather information on the reliability of PGD [18].

Assisted reproductive technology and biopsy
The first step in a PGD cycle is controlled ovarian hyper stimulation, aimed at obtaining a large cohort of mature oocytes. The aspirated oocytes are denuded of surround ing cumulus cells before IVF occurs. Intracytoplasmic sperm injection is preferred over regular IVF, regardless of the sperm quality, in order to avoid residual sperm adhering to the zona pellucida after IVF. Remaining cumulus or sperm cells may lead to contamination when PCR is used for diagnosis [19].
The biopsy procedure involving breaching of the zona pellucida and removal of the cell(s) can be carried out at different developmental stages [20]. Polar body (PB) biopsy from oocytes (first PB from metaphase II oocytes and second PB after normal fertilization) is a first possibility [21]. As PBs do not contribute to normal fertilization or embryonic development, their removal has no detrimental effect. In countries where embryo selection is forbidden, PB biopsy is the only legal option as testing can take place before syngamy, which is the moment considered as the beginning of an embryo. The main disadvantage is that PB analysis can only evaluate the maternal genetic contribution. The ESHRE PGD consortium data collection indicates that cleavagestage biopsy at day 3 is used in the majority of PGD cycles [6]. At this eightcell stage, embryo compaction has not started yet and cells are considered to be totipotent. The zona pellucida opening is mostly carried out with a non contact diode infrared laser and one or two nucleated blastomeres are aspirated [22]. The impact of the random removal of one or two embryonic cells on further development and implantation potential is highly debated. In a recent prospective analysis of a cohort of singleembryo transfers, the authors demonstrated that the live birth rate after onecell removal from eightcell embryos (37.4%) is similar to the rate of a control intracytoplasmic sperm injection group without biopsy (35.0%), but significantly higher than the rate of the two cell biopsy study group (22%). It is therefore recom mended to biopsy one cell, provided that an accurate and reliable diagnosis method is in place [23]. Another limita tion of cleavagestage biopsy is the high chromosomal mosaicism rate that seems inherent at this developmental stage and probably evolves to lower rates at the blastocyst stage via selfcorrection [24]. Trophectoderm (TE) biopsy at day 5 is a fast emerging biopsy approach. Some reports suggest that it may yield substantially higher implantation and live birth rates than cleavagestage biopsy [2527]. About 4 to 20 extraembryonic TE cells are aspirated and the inner cell mass, from which the fetus will develop, is kept intact. This is one advantage compared with cleavagestage biopsy; another advantage is that multiple cells are available for genetic testing, and this may improve diagnostic accuracy. The drawbacks of TE biopsy are that it requires a successful embryo culture system with high blastocyst rates and it leaves limited time for genetic analysis. The time problem may be solved by freezing the blastocysts, relying on efficient vitrification and thawsurvival protocols, and transferring them later in a natural cycle.
After biopsy, (single) cells are washed and either fixated for FISH analysis or tubed and lysed for amplification. Amplification reaction components are added directly to the lysed cells without prior DNA purification.

Single-cell genetic testing for monogenic disorders
In principle, any monogenic disease for which the chromosomal locus has been identified can be diagnosed at the singlecell level. The golden standard today is fluorescent multiplex PCR in which one of each primer pair is fluorescently labeled, allowing subsequent detection of PCR fragments on an automated sequencer. In indirect tests, multiple short tandem repeat markers are coamplified together in one reaction (linkagebased strategy); for direct testing, markers are combined with specific mutation(s). The main advantage of linkage based testing over mutationspecific testing is that the singlecell protocols can be used for several couples, independent of the mutation they carry. This saves time, resources and manpower in prePGD workups for diseases for which many private mutations have been identified. The use of polymorphic markers implicates that during prePGD work up, informativity and segre gation tests are performed on DNA samples of the couple and family members to establish which alleles of the informative markers segregate with the mutation. Different strategies of PCR and allele discrimination have been developed for mutation detection over recent years, and the most important ones are: amplification refractory mutation system [28], endonuclease restriction [29], minisequencing [30] and quantitative realtime PCR [26]. PCRbased protocols with short tandem repeat markers have recently also been applied in PGD cycles for structural chromosome abnormalities, in which FISH has been traditionally the preferred method [31,32]. A major bottleneck with PCR assays is that the singlecell adap tation and validation part of the prePGD work up has to be repeated with every new DNA locus. In recent years, the use of singlecell whole genome amplification (WGA) as a universal step has been demonstrated to be a practical and efficient alternative to singlecell PCR. The first WGA methods were PCRbased and suffered from incomplete genome coverage and amplification bias, but the more recent methods are markedly better [33]. One method is multiple displacement amplification that relies on isothermal strand displacement amplification with Phi29 DNA polymerase. This method is very straight forward but requires highquality DNA as the template, and yields relatively high allele dropout (ADO) and preferential amplification rates of 25% on average [34]; this is about five times higher than with fluorescent multiplex PCR. Other WGA methods involve somewhat more complex protocols with DNA fragmentation and library formation prior to amplification; they even amplify lowquality DNA, and ADO rates are approxi mately 10% [35]. Singlecell WGA generates micrograms of amplified DNA, which is sufficient for several down stream applications. Multiple standard PCR assays may be performed for haplotyping in case of monogenic diseases [36]. Haplotyping can also be combined with array comparative genomic hybridization (aCGH) for the detection of chromosomal imbalances. These strategies are already clinically applied in some centers. Further improvement and generalization will come from the intro duction of highdensity single nucleotide polymorphism (SNP) arrays, which enable evaluation of DNA haplo typing and chromosomal constitution on the same platform. The validation of these SNP arrays is ongoing [37,38]. The implementation of these new technologies will reduce the procedures substantially and introduce automation. On the other hand, these arrays will generate a tremendous amount of genetic data for instance, on disease susceptibility genes and it is expected that this will entail many ethical discussions and challenges for genetic counseling.

Single-cell genetic testing for chromosomal aberrations
The majority of PGD cycles for chromosomal aberrations are performed for translocations. Balanced translocation carriers have a high risk for producing chromosomally unbalanced gametes and present with infertility, repeated spontaneous abortions and unbalanced offspring. With FISH, fixated cells are hybridized using chromosomespecific DNA probes, which are labeled with different fluorochromes. Probe selection is based on the break points of the specific translocation and chromosomes involved, and should allow detection of all possible segregation patterns. FISH has a number of technical limitations and it is gradually replaced by aCGH. Here, whole genome amplified DNA of the test singlecell(s) is labeled with a green fluorochrome, while a red fluoro chrome is used for the control sample. Both samples are mixed and hybridized to either a normal metaphase chromosome spread or, in case of aCGH, to DNA sequences specific to human chromosomes spotted on an array. A computerized system allows analysis of the ratios between both fluorochromes for each chromosome and detection of imbalances. As metaphase comparative genomic hybridization (CGH) at the singlecell level requires several days for analysis, groups that have presented clinical application of CGH in PGD have to resort to either polar body analysis [39] or cryo preservation of the embryos [40,41]. As aCGH has a lower hybridization time, it can be performed within the time frame of PGD. Other advantages over metaphase CGH are the higher resolution and the important auto mation [42]. aCGH following multiple displacement amplification has been preclinically validated in single lymphoblasts, fibroblasts and blastomeres of trans loca tion carriers [43], and clinical data have been published for PGS applications [44]. Compared with FISH, aCGH provides a generalized platform, circumventing prePGD testing of FISH probes with every new translocation. aCGH, as well as SNP arrays, yields information on genomewide copy number variation, but only SNP arrays offer simultaneously genotype information and present a universal platform for both chromosomal aberrations and monogenic diseases. Therefore, SNP arrays are regarded as the most promising future strategy for PGD.

Accuracy
Singlecell DNA amplification is technically demanding because of the small amount of nonpurified DNA at the start, and inherent pitfalls of contamination and ADO, which can lead to misdiagnosis. ADO is defined as the random failure of amplification for one of two alleles in a heterozygous cell. The use of optimized cell lysis and DNA amplification conditions, along with sensitive detection systems, should reduce ADO to a minimum. Contamination is minimized by taking a number of prevention measures (see best practice guidelines). The application of multiplex PCR protocols further assists in monitoring problems of contamination and ADO in addition to mutation detection, making these assays highly accurate.
A recent theoretical study on the accuracy of FISH showed that the technology has the potential of high accuracy for sexing and for translocations, but not in the case of aneuploidy screening [45]. Apart from technical errors such as ADO and contamination, other possible causes for misdiagnosis involve intrinsic sample quality, such as chromosomal mosaicism, and human errors, such as mislabeling, incorrect embryo transfer or erroneous segregation analysis. According to the PGD consortium data, low error rates have been observed for both DNAamplificationbased cycles (0.5%) and FISH based cycles (0.1%) [46].

Quality control and assurance
PGD is still relatively unregulated compared with routine genetic testing. In order to better standardize singlecell testing and to achieve high quality levels, guidelines for best practice have been designed by the ESHRE PGD consortium and by the Preimplantation Genetic Diag nosis International Society [4749]. In addition, four new extensive guidelines on different aspects of PGD (organi za tion of a PGD center, FISHbased testing, amplification based testing and biopsy) are in preparation.
A powerful method for quality assurance is accredita tion. In 2008, only 33% of 53 European PGD centers had achieved or were preparing for accreditation [50]. Many countries and international authorities now recommend PGD centers to initiate the accreditation process; that is, to be committed not only to reach the standards set by the accrediting body but also to continuously enhance the quality of the services.
Two programs of external quality assessment (EQA), a key element of quality assurance [51,52], have been initiated recently. The UK National External Quality Assess ment Service [53], in collaboration with the ESHRE PGD consortium, has set up EQA programs for PGD for monogenic diseases, while the FISHbased PGD EQA program is run by the Cytogenetics European Quality Assessment [54].

Conclusions
In the past two decades, PGD has evolved from an experimental procedure to a widely accepted alternative for prenatal diagnosis that is applied to an expanding range of indications. Nonstop technical improvements have provided reliable and accurate singlecell assays for both chromosomal abnormalities and monogenic dis orders. As protocols are often family specific, their develop ment is labour intensive and time consuming and can be carried out in specialized laboratories. The introduction of powerful array technologies following universal whole genome amplification will reduce the workload and allow automation. Together with the imple mentation of accreditation schemes, this will lead to improved standardization and uniformity in the complex process of PGD.

Competing interests
The author declares that she has no competing interests.