The introduction of arrays in prenatal diagnosis: A special challenge

Authors

  • Annalisa Vetro,

    1. Medical Genetics, University of Pavia, Pavia, Italy
    Search for more papers by this author
  • Katelijne Bouman,

    1. Department of Genetics, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Ros Hastings,

    1. CEQA and UK NEQAS for Clinical Cytogenetics, Women's Centre, John Radcliffe Hospital, Oxford University Hospitals Trust, Oxford, United Kingdom
    2. Also on behalf of the Genetic Services Quality Committee of the European Society of Human Genetics
    Search for more papers by this author
  • Dominic J. McMullan,

    1. West Midlands Regional Genetics Laboratory, Birmingham Women's Hospital NHS Trust, Birmingham, United Kingdom
    Search for more papers by this author
  • Joris R. Vermeesch,

    1. Centre for Human Genetics, University Hospital Leuven, Katholieke Universiteit Leuven, Belgium
    Search for more papers by this author
  • Konstantin Miller,

    1. Institute of Human Genetics, Hannover Medical School, Hannover, Germany
    2. Also on behalf of the Genetic Services Quality Committee of the European Society of Human Genetics
    Search for more papers by this author
  • Birgit Sikkema-Raddatz,

    1. Department of Genetics, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • David H. Ledbetter,

    1. Genomic Medicine Institute, Geisinger Health System, Danville, Pennsylvania
    Search for more papers by this author
  • Orsetta Zuffardi,

    1. Medical Genetics, University of Pavia, Pavia, Italy
    2. IRCCS C. Mondino National Institute of Neurology Foundation, Pavia, Italy
    3. Also on behalf of the Genetic Services Quality Committee of the European Society of Human Genetics
    Search for more papers by this author
  • Conny M.A. van Ravenswaaij-Arts

    Corresponding author
    1. Department of Genetics, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands
    2. Also on behalf of the Genetic Services Quality Committee of the European Society of Human Genetics
    • Department of Genetics, University of Groningen, University Medical Center Groningen, P.O. Box 30.001, 9700 RB Groningen, The Netherlands.
    Search for more papers by this author

  • For the Focus on CNV Detection with Diagnostic Arrays

Abstract

Genome-wide arrays are rapidly replacing conventional karyotyping in postnatal cytogenetic diagnostics and there is a growing request for arrays in the prenatal setting. Several studies have documented 1–3% additional abnormal findings in prenatal diagnosis with arrays compared to conventional karyotyping. A recent meta-analysis demonstrated that 5.2% extra diagnoses can be expected in fetuses with ultrasound abnormalities. However, no consensus exists as to whether the use of genome-wide arrays should be restricted to pregnancies with ultrasound abnormalities, performed in all women undergoing invasive prenatal testing or offered to all pregnant women. Moreover, the interpretation of array results in the prenatal situation is challenging due to the large numbers of copy number variants with no major phenotypic effect. This also raises the question of what, or what not to report, for example, how to deal with unsolicited findings. These issues were discussed at a working group meeting that preceded the European Society of Human Genetics 2011 Conference in Amsterdam. This article is the result of this meeting and explores the introduction of genome-wide arrays into routine prenatal diagnosis. We aim to give some general recommendations on how to develop practical guidelines that can be implemented in the local setting and that are consistent with the emerging international consensus. Hum Mutat 33:923–929, 2012. © 2012 Wiley Periodicals, Inc.

Introduction

Genome-wide arrays are rapidly replacing conventional karyotyping in postnatal cytogenetic diagnostics. Using arrays, an extra 15% of causally related chromosomal abnormalities are being detected over routine microscopic and multiplex ligation-dependent probe amplification (MLPA) or fluorescent in-situ hybridization (FISH) for subtelomeric screening in patients with developmental delay (DD) and/or multiple congenital anomalies (MCA) [Stankiewicz and Beaudet, 2007]. Not surprisingly, there is a growing request for arrays in the prenatal setting as well, especially in cases with ultrasound abnormalities. Conventional karyotyping has demonstrated abnormal karyotypes in 8–35% of fetuses with ultrasound abnormalities [Faas et al., 2010; Kleeman et al., 2009; Valduga et al., 2010]. It is to be expected that more genomic imbalances will be detected by using high-resolution genome-wide array. Several array studies, using various inclusion criteria, have documented 1–3% extra prenatal submicroscopic abnormalities above those detected by karyotyping [Coppiger et al., 2009; Kleeman et al., 2009; Shaffer et al., 2008; Van den Veyver et al., 2009]. In the presence of a combination of several ultrasound abnormalities, the additive value of array was reported to be 8% by Le Caignec et al. (2005). Publications based on retrospective evaluations, that is, postnatal arrays in neonates, aborted fetuses, and stillborns with prenatal ultrasound abnormalities but normal karyotype, report 5–10% pathogenic copy number variants (CNVs). A recent meta-analysis of prenatal samples demonstrated that 5.2% extra submicroscopic diagnoses can be expected if benign CNVs are excluded [Hillmann et al., 2011].

The difference in additional diagnostic yield over microscopic karyotyping between the prenatal and postnatal setting may be explained by the fact that only a minority of the submicroscopic abnormalities that are detected postnatally in patients with developmental delay are accompanied by congenital abnormalities detectable by prenatal ultrasound. Thus, one might argue that the use of genome-wide array in prenatal diagnostics should not be restricted to pregnancies with ultrasound abnormalities [Strassberg et al., 2011]. On the other hand, the interpretation of array results is challenging, especially in the absence of ultrasound abnormalities, as pathogenic CNVs that do not result in a phenotype detectable by ultrasound may sometimes be hard to distinguish from benign CNVs. The presence of large numbers of CNVs with no major phenotypic effect, as found in healthy individuals, impedes the interpretation of array results in DD/MCA patients [Iafrate et al., 2004; Sebat et al., 2004]. In the prenatal setting, this is even more complicated and raises questions as to what and what not to report.

Another question is whether parents should be analyzed at the same time to facilitate the interpretation of CNVs. The general rule that de novo chromosomal imbalances are more likely to be clinically significant, while familial CNVs are less likely does not always hold true and can never be the sole criterion for clinical interpretation. Several studies have shown the clinical relevance of inherited CNVs and therefore the de novo origin of a CNV is not by itself a good indicator of its clinical relevance [Bisgaard et al., 2007; De Ravel et al., 2006; Mencarelli et al., 2008].

A more reliable way of determining the clinical relevance of a CNV in a patient is to compare it with CNVs gathered in large databases with data of healthy controls. The Database of Genomic Variants (http://projects.tcag.ca/variation) is a well-known database, which is used worldwide. Several laboratories also have an in-house or national reference database available for this purpose. Most recently, large-scale association studies have enabled the determination of the added risk of recurrent genomic disorders [Cooper et al., 2011; Kaminsky et al., 2011]. Despite these helpful databases and analyses, the clinical significance of the majority of CNVs remains unknown.

Although it can be envisaged that in the future genome-wide arrays will replace karyotyping as a first-line test for detecting chromosomal abnormalities in the prenatal setting [Strassberg et al., 2011], it is important to collect critical points of view and process these in recommendations for guidelines on how to introduce this technique in routine prenatal practice. This article is the result of a working group meeting organized by the Department of Genetics, University Medical Centre Groningen, and the Genetic Services Quality Committee of the European Society of Human Genetics (ESHG). It was held prior to the ESHG 2011 Conference in Amsterdam. Our aim is to explore the introduction of genome-wide arrays in routine prenatal diagnosis and to give some general recommendations on how to develop practical guidelines that comply with national and international guidelines where they exist, and that can be implemented in the local setting.

Arrays in the Prenatal Setting: For what Indication?

Currently, at least in Europe, microarrays are not used systematically in all pregnancies requiring fetal karyotyping (for advanced maternal age, maternal anxiety, positive biochemical/ultrasound markers, etc.), as conventional cytogenetics is still the most commonly used approach. There are three reasons for this: (1) the interpretation difficulties of array results in the absence of ultrasound abnormalities, and also the chance of unsolicited findings (see the following section), (2) the paucity of highly trained cytogeneticists or laboratory scientists experienced enough to analyze array data in the more complex context of a prenatal setting, (3) the high cost of array analysis, as in many cases it is necessary to examine both the fetus and the parents. Moreover, in some countries, health policies are driving prenatal diagnosis from the use of direct, invasive techniques for detecting Down syndrome in women of advanced age toward noninvasive prescreening techniques (serological investigations and ultrasound screening) to improve the ratio of detection rate to miscarriage risk.

Since genome-wide array analysis has a 5.2% higher detection rate than karyotyping when applied to pregnancies with ultrasound abnormalities [Hillman et al., 2011], an increasing number of genetic laboratories are offering this technique. The implementation restricted to ultrasound abnormalities is relatively easy: less economical and organizational burden and fewer difficulties with the interpretation of the array results. However, given the higher and more accurate yield of chromosomal imbalances, it is attractive to replace conventional karyotyping by array analysis in all women who undergo invasive prenatal testing. Ogilvie et al. (2009) estimated that there is a 1-in-300 to 1-in-600 chance of detecting a known disability-causing CNV through genome-wide arrays that would have remained undetected by karyotyping. The extension of array analysis to all invasive samples will largely depend on the increase in knowledge and the capability of laboratories to be confident with the technique, and on the different health care systems.

The abnormalities that are exclusively detected by array are not age-related and considering the frequency of some disorders detectable by array, one could argue that this technique should be offered to all pregnant women, irrespective of age and the results of the screening programmes. This could be more convenient from an efficiency and cost–utility point of view [Harris et al., 2004]. However, the US ACOG statement that invasive testing should be available to all women is a position that is not commonly shared in Europe as the increased detection rate will be accompanied by a less favorable miscarriage/detection ratio and by counseling dilemmas when unsolicited CNVs or CNVs of unknown clinical relevance are found.

The discussion on whether all pregnant women should be offered an invasive procedure to detect as many potentially pathogenic imbalances as possible may soon be superseded. Noninvasive prenatal diagnosis (NIPD), based on cell-free fetal DNA in maternal serum, enables the detection of a selected set of diagnoses (i.e., common aneuploidies) with a high sensitivity and without risk for miscarriages [Chiu et al., 2011; Chiu and Lo, 2011; Papageorgiou et al., 2011]. Thus, in the near future, it is imaginable that NIPD will replace all the current biochemical screening tests or be the first-tier test after biochemical screening indicates an increased risk for Down syndrome, while genome-wide array investigations will be used only for those cases with abnormal ultrasound results [Lichtenbelt et al., 2011].

Which Array Platform and fetal material should be used in Prenatal Diagnosis?

Analysis of directly isolated DNA, either from amniotic fluid cells or from the mesenchymal core cells of chorionic villi, is preferable to DNA from cultured cells since the former allows for the great majority of results to be available within one week and avoids culture artefacts. Analysis for maternal cell contamination should be performed.

Most laboratories use oligonucleotide or SNP arrays with a high resolution, detecting CNVs with a threshold of 100–150 kb, although a higher resolution is often achieved. In general, a backbone with a minimal resolution of 200 kb is recommended [Vermeesch et al., 2012]. Many platforms are enriched in gene-dense regions and regions with known microdeletion/microduplication syndromes. To avoid interpretation problems in the prenatal situation, a targeted array design has been suggested [Sahoo et al., 2006; Park et al., 2010]. Such a platform will cover all the known pathogenic CNVs and have a low-resolution backbone for the detection of relatively large CNVs. The idea of a targeted array is that the detected number of CNVs of unknown clinical significance will be kept to a minimum, thus facilitating the interpretation. However, knowledge of CNVs is expanding rapidly and new relevant microdeletion/duplication syndromes are continuously being reported, necessitating a frequent update of such a platform with a targeted design. This is not cost effective and more importantly, pathogenic imbalances may be missed. In our workshop's opinion, it is preferable to use the same platform for both prenatal and postnatal settings. A high-resolution array guarantees a minimum of false-negative results. Moreover, experience is very important in the interpretation process. Therefore, the cytogeneticist or laboratory scientist should be familiar with the platform used and know its strengths and weaknesses. Experience is best gained by analyzing a sufficient number of patients, thus for most laboratories this means pre- and postnatal samples. The interpretation process is easier if the laboratory has an in-house control dataset, analyzed with the same array platform [Vermeesch et al., 2012].

The Interpretation of Prenatally Detected Copy Number Variants

A major challenge when using genome-wide arrays in a prenatal setting is interpreting CNVs of unknown or uncertain clinical significance. Helpful tools to interpret CNVs are provided by several public databases that collect data from healthy subjects, such as the Database of Genomic Variants (http://projects.tcag.ca/variation), or from patients with multiple congenital anomalies and/or developmental disabilities, such as DECIPHER (http://decipher.sanger.ac.uk/), ECARUCA (http://www.ECARUCA.net), and ISCA (https://www.iscaconsortium.org) [de Leeuw et al., 2012]. In the near future, the growing amount of data collected by such resources will allow a much better correlation between a certain CNV and a possible pathogenic condition. The clinical characterization of a proband is crucial for the use of these databases. Obviously, this is difficult in the prenatal situation, where phenotyping is limited to what is visible by ultrasound. Even if an ultrasound abnormality is present, this is almost always nonspecific, thus impairing our ability to state whether it is linked to the detected CNV.

As public databases become more useful in distinguishing between the clinically relevant and benign CNVs, they may increasingly allow the interpretation of a case before extending the investigation to the couple. However, at present, blood samples from both parents should be taken at the same time as prenatal sampling to allow the laboratory to analyze them in a short time and to provide the correct interpretation of novel CNVs. There is a general agreement in the literature about the importance of the availability of parental DNA in prenatal array testing [Bui et al., 2011; Coppinger et al., 2009; D'Amours et al., 2012; Faas et al., 2010; Friedman, 2009; Zuffardi et al., 2011].

In all guidelines, a rule of thumb is that de novo CNVs, not occurring in normal individuals, are considered supportive evidence for likely causality of an abnormal phenotype. However, the frequency of de novo CNVs is not trivial. The CNV mutation rate has been estimated to be in the range of 1.2 × 10 CNVs per haploid genome per transmission at a median resolution of 150 kb, amounting to about 2.5 CNVs/100 live births [Conrad et al., 2010; Sebat et al., 2007; Xu et al., 2008]. Thus, the tendency to call de novo CNVs pathogenic has led to a possible overestimation of the causality of de novo CNVs [Vermeesch et al., 2011], especially in the case of small CNVs (<500 kb). Moreover, when a parent carrying the same CNV as the fetus is phenotypically normal, as occurs in most cases, the pathogenicity of the CNV cannot be excluded, since (1) a gene within the deletion region located on the homologous nondeleted chromosome can be mutated, thus leading to an autosomal recessive condition [Lesnik Oberstein et al., 2006]; (2) the CNV can contain an imprinted gene that has a pathogenic effect only when inherited from the mother and not from the father, or vice versa [Kagami et al., 2010]; or (3) the CNV itself is a risk factor promoting a given disease in a specific but unknown genomic context, in other words, the CNV has its own pathogenic burden but exhibits incomplete penetrance. The last point is well demonstrated by several recurrent and frequently inherited deletions/duplications associated with different neurological phenotypes, for which a second hit is considered necessary for the full manifestation of the syndrome [Girirajan and Eichler, 2010; Girirajan et al., 2010]. Thus, the inheritance status of a CNV should never be the sole criterion to assume that the CNV is benign when inherited, but it may suggest its pathogenicity when de novo, although it will require reevaluation on a case-by-case basis. This evaluation will depend on the genomic region involved and on the clinical context, keeping in mind that it may be hard and sometimes impossible to trace the link between the fetal phenotype and the function of the genes within the CNV, especially when the gene content is not directly related to a known disease or a mammalian developmental pathway. Even CNVs falling in gene-desert regions cannot be disregarded a priori, since it has been proven that they can have a pathogenic impact, depending on the genomic context, and whether they are inherited or de novo. As such, the example of the cryptic duplicated region upstream to SOX9, which can be inherited from a normal father but leads to SRY-negative XX males, is paradigmatic [Cox et al., 2011; Vetro et al., 2011].

The analysis of parental arrays may be restricted to the CNVs found in the fetus, or parental analysis might be performed by FISH or qPCR, thus avoiding finding CNVs of uncertain significance in the couple (possibly related to late-onset disorders, cancer predisposition or so on), partly reducing the ethical issues generally related to a genome-wide analysis (see section on pretest counseling). If one of the parents is not available, or in case of nonpaternity, interpreting the fetal array results can be difficult [Zuffardi et al., 2011]. In this situation, array may still be preferred over conventional karyotyping because of the detection of fully penetrant CNVs with a severe phenotype.

When a mosaic result is detected, it is important that the laboratory and clinician are aware of the possibility of pseudomosaicism or confined placental mosaicism. A confirmatory sample is needed, especially if ultrasound findings are expected and are either not present or are inconsistent with the result. The clinical interpretation of prenatal mosaics is hampered by the fact that most clinical data are obtained postnatally and thus are liable to postnatal ascertainment bias.

It is not easy to gain a high level of experience in a short time. As previously mentioned, experience is best gained when a sufficient number of cases are analyzed and the interpretation process is easier if the laboratory has an in-house control dataset. The formation of a network of laboratories with high expertise in the field, available for consultation or data exchange, can be useful in this regard.

Is Additional Conventional Karyotyping Still Needed?

In case an imbalance is detected by array, conventional karyotyping or FISH on metaphases may be helpful to unravel the underlying mechanism of the aberration and thus reveal information that is important for recurrence risk. However, these analyses will seldom affect the management of the current pregnancy. More importantly, parental analysis should be performed to rule out a balanced translocation or insertion for an accurate estimation of the recurrence risk.

On the other hand, if no imbalance is detected by array, it is up to the laboratories whether to perform conventional karyotyping, to exclude a de novo balanced rearrangement in case of ultrasound abnormalities. However, the chance of finding such a rearrangement that causes the ultrasound abnormalities by disrupting or otherwise influencing the expression of a gene is extremely low. Missing an inherited balanced rearrangement is of minor importance, since such a finding will not explain the ultrasound abnormalities in the current pregnancy.

Is There a Place for Rapid Aneuploidy Detection?

In case of ultrasound abnormalities, laboratories may perform rapid aneuploidy detection (FISH, MLPA, or QF-PCR) prior to array analysis to detect a common aneuploidy such as trisomy 21, −18, −13 or a monosomy X. If rapid aneuploidy detection indicates the presence of an aneuploidy, conventional karyotyping should be performed to exclude an unbalanced translocation. After excluding common aneuploidies, the laboratory can proceed to array analysis. Depending on the local situation this two-step strategy may be cost effective.

Pretest Counseling and Prenatal Parental Consent

Genetic counseling for prenatal analysis, either karyotyping and/or genome-wide array should be performed according to the professional guidelines in medicine: do good and do no harm, while respecting the patient's autonomy. Counselors should provide clear information, which is easy to understand for the future parents and in a nondirective way, to help them to make their own decision based on the ethical concept of the fetus as a patient. Although there are no autonomy-based obligations to the fetus, it will later become a child and thereby achieve an independent moral status [Chevernak and McCullough, 2011; Dondorp et al., 2012]. Given these ethical considerations, counseling for genome-wide array analysis in the prenatal setting is difficult and challenging. When ultrasound abnormalities are present, performing pretest counseling directly after diagnosing the fetal anomalies is almost impossible: psychological stress makes it difficult for the parents to take in new information and they will struggle to make a well-balanced decision with the information they obtain [Evans, 2006]. Therefore, information should not only be given orally, but also as a written summary to be read again later.

Parental consent for prenatal array analysis should be obtained after genetic counseling by a genetic counselor, a clinical geneticist or a genetically trained fetal medicine specialist. The most important issues that should be included in the counseling session are summarized in Table 1. The counseling has to be carried out in a clinical setting and should include documentation of the medical history of both parents and a three-generation family pedigree. This extended documentation may reveal one or more genetic disorders in the family and, obviously, such information is crucial in interpreting the array results.

Table 1. Important Pretest Counseling Issues for Prenatal Genome-Wide Array Analysis
  1. a Depending on the tests used and the agreements between the laboratory and clinicians on what to report (see the section “Communication between Laboratory and Clinicians”).

  2. b In case of ultrasound abnormalities.

General, obtain:
Medical history of both parents
Medical history of the pregnancy, including ultrasound findings
Family pedigree of both parents, to three generations
Common aneuploidies, explain:
Risk for common aneuploidies in the fetus
Most important clinical features of the common aneuploidies
Prenatal array, explain:
Why parental blood samples are needed (or may be needed if not drawn immediately)
The chances of finding an abnormality and the different types of abnormalities that can be detected by genome-wide arraya:
• CNVs explaining or likely to result in fetal (ultrasound) anomalies
• De novo variations of unknown clinical significance
• Inherited variations of unknown significance
• Unsolicited findings—CNVs that are health debilitating
• Apparently balanced chromosomal rearrangements with normal array results
Residual risk, explain:
Differential diagnosis if all cytogenetic genetic test results are normalb
Provide written information and:
• Offer psychosocial support and record needs and preferences of the couple
• Explain expected time frame for different test results
• Agree upon the manner of communicating the test results
• Achieve consent on what and what not to report to the parents

Parents should be counseled about the chances of detecting chromosome aberrations and genomic imbalances by prenatal fetal testing and about the limitations of the tests. If genome-wide analysis is performed in the parents, they should be counseled appropriately, including informed consent on what information they want to obtain. If applicable, the parents need to be informed that analysis of the parental array results may be restricted to imbalances detected in the fetus. It is of prime importance to consider what the couple wants to know about their current pregnancy, about themselves, and about possible future pregnancies. The mother should be made aware, either in private or in written information, that nonpaternity can influence the interpretation of the results and that nonpaternity may be disclosed if SNP arrays are used. Parents should also be made aware that the array technique does not detect every genetic disease or well-known syndrome. In a cohort of 141 fetuses with ultrasound anomalies and normal array results (Agilent 180k oligo array, Agilent Technologies Inc., Santa Clara, CA, USA) that were reviewed postnatally, a clinical syndrome was diagnosed in 15% of the cases, of which 30% were due to a single gene mutation (unpublished data, KB, BS-R).

The parents should have the possible outcomes of any prenatal array test explained with the options available. The fetal result can be either normal or abnormal, with the latter having different options: the observed copy number variations (CNVs) may (1) explain the fetal ultrasound anomalies, (2) be a de novo variation of unknown clinical significance, (3) be an inherited variation of unknown significance, or (4) result in an unsolicited finding, that is, an increased risk for diseases of known or unpredictable severity for the fetus, now or later in life, but not related to the ultrasound anomalies.

Parents should be informed about the possibility of health-debilitating unsolicited findings. This can involve a late-onset inherited disease that occurs de novo or is inherited in the family. The family pedigree might have already led to this suspicion, demonstrating the importance of recording the family history. In our experience, such an unsolicited finding is detected in about 1–2 per 1,000 analyses by genome-wide array. Pichert et al. recently reported CNVs of cancer-related genes in 29 of 4,805 (0.6%) patients examined by array [Pichert et al., 2011]. Five, or 0.1%, were deletions (MSH2, MSH6, CDKN1B, PRKAR1A, SMARCB1) that would result in an increased nonsyndromic cancer risk and may give rise to a counseling dilemma. A distinction can be made between treatable and nontreatable late-onset diseases, for example, hereditary cancer versus hereditary dementia. The current tendency in Europe is to ask parents whether they want to be informed about treatable late-onset diseases. Some laboratories have a policy of not communicating unsolicited CNVs related to nontreatable diseases. A recent publication showed that 55% of 61 future parents want to be informed about adverse health effects at an adult stage [Srebniak et al., 2011]. Unfortunately the authors made no distinction between treatable and nontreatable late-onset diseases.

There is no consensus on whether to report deletions of recessive genes with a high carrier frequency in the respective population.

A lot of information has to be communicated to and understood by the parents, which is especially challenging when they are under stress because of ultrasound abnormalities (Table 1). The time frame available in the prenatal setting to discuss and consider the findings is limited, given the boundaries set by different laws for abortion [Sijmons et al., 2011]. Counseling for prenatal array analysis is therefore challenging and more socioethical research is needed to develop best practice guidelines [Zuffardi et al., 2011].

Communication between Laboratory and Clinicians

As many different results can be generated by prenatal array, it is important that the counselor has a close collaboration with the genetics laboratory. Most importantly, there should be an agreement on what to report to the clinician prior to testing, for example, whether nonpathogenic CNVs or CNVs of unknown clinical significance should be reported. The counselor should know not only which tests are available, but also in what order they are performed in the laboratory, for example, whether the prenatal array is preceded by rapid aneuploidy testing and whether additional karyotyping will be performed or the array is done as a first-line diagnostic test. Both counselor and laboratory should, at short notice, be able to communicate any additional information obtained after sampling by counseling or ultrasound, or about complex results that may arise from the prenatal array analysis. It is important to be aware that for a correct prenatal molecular array diagnosis the clinical diagnosis of the absence or presence of ultrasound abnormalities is indispensable.

Counselors and laboratory geneticists should not only agree on what and what not to report to the counselor and other clinicians directly involved in the prenatal care, but also what to report to the couple. This may depend on the indication for which the array was performed: it is obvious that de novo CNVs, not occurring in normal individuals and with a high likelihood of being causal for the ultrasound anomalies detected should be reported. However, one should not report CNVs of unknown clinical significance in the absence of ultrasound abnormalities and carefully weigh the benefits and drawbacks of reporting if the CNV is unlikely or unknown to be causative for a detected ultrasound abnormality. Reporting these CNVs may cause unnecessary anxiety in the couple, and especially de novo CNVs may be prone to overestimation of their causality [Vermeesch et al., 2011]. It is important that the laboratory informs the clinician when a pathogenic CNV is not consistent with the ultrasound findings or when the origin of the CNV may have a contributing effect, for example, when an imprinted gene is involved. The laboratory and clinicians should also reach consensus on whether to report unsolicited findings resulting in an increased risk for diseases of known or unpredictable severity for the fetus or the parents, now or later in life. If the agreement is that parents can choose whether to be informed about unsolicited findings, the laboratory should be unequivocally informed of their choice to prevent communication of undesirable results.

The maximum reporting time of the prenatal array results should be equivalent to that for karyotyping amniotic fluid cells, that is, 17 days [Hastings et al., 2012]. For timely patient management, especially in case of ultrasound abnormalities and a gestational age approaching the legal limit for termination of pregnancy, a maximum reporting time of two weeks is preferred.

Existing Guidelines and Initiatives

In France, the Association des Cytogénéticiens de Langue Française (ACLF) has established guidelines for genome-wide array [Association des Cytogénéticiens de Langue Française, 2010]. The test is currently not reimbursed by the health care system.

In Germany, in a statement published in 2010, the German Society of Human Genetics did not see any indication for prenatal genome-wide array analysis [Deutsche Gesellschaft für Humangenetik, 2010]. The new German S2-guidelines on human genetic diagnosis do not give indication criteria for prenatal arrays, although testing is not excluded explicitly [Deutsche Gesellschaft für Humangenetik, 2011]. Currently, there is no reimbursement of the costs of the test.

In the Netherlands, prenatal diagnostics is only performed in university medical centers. They perform genome-wide array analysis in the case of prenatal ultrasound abnormalities, found by either routine screening or upon request, after common aneuploidies have been excluded by QF-PCR. Pretest counseling and blood sampling of the parents are a prerequisite. National guidelines are not yet available.

In the United Kingdom, the Association of Clinical Cytogenetics (ACC) Best Practice Guidelines for constitutional genome-wide array (v1.0 2009) do not include specific recommendations for use of arrays in the prenatal setting and introduction of arrays in cases with anomalies detected by ultrasound has so far only been implemented sporadically. Of note, a large, nationally funded, prospective multicenter study in the UK called EACH (Evaluation of Array Comparative genomic Hybridization in prenatal diagnosis of fetal anomalies) is about to commence with the aim of comparing array CGH with karyotyping in fetuses with one or more structural anomalies identified at 11–14 or 18–20 weeks screening, or with isolated nuchal translucency >3 mm at 11–14 week screening. This study of over 1,000 pregnancies in the next two years will include qualitative measurements to explore and understand patient, health professional and commissioner preferences for array CGH.

In the United States, in the case of prenatal diagnosis, conventional karyotyping is, at this point, considered the gold standard by the American College of Obstetricians and Gynaecologists, with array considered an optional, adjunctive test but not as a stand-alone diagnostic [ACOG committee, 2009]. The National Institute of Child Health and Human Development (NICHD) has funded a large, multicenter study comparing whole-genome chromosomal microarray to conventional karyotyping with the initial results expected shortly (an abstract was presented at the 2011 ISPD meeting).

Conclusions

This article gives an overview of all the issues that should be considered when performing genome-wide arrays in prenatal diagnosis. The most important recommendations that will be helpful when establishing local or national guidelines are: (1) establish the indications for the use of genome-wide array analysis in the prenatal setting. The value of arrays has been proven for ultrasound abnormalities. (2) An array platform with a minimal resolution of 200 kb is recommended. (3) Laboratory specialists should have sufficient experience with the interpretation of array results. (4) Parental blood sampling is highly recommended. (5) Pretest counseling, including providing written information, and parental consent are a prerequisite. (6) The laboratory and the clinicians should agree on what to report and what not to report before offering array diagnostics. (7) There should always be optimal communication between the laboratory specialists and the clinicians.

Ancillary