Pregnancy loss is one of the most common pregnancy complications occurring in 12–14% of clinically recognized pregnancies. Cytogenetic abnormalities are present in the majority (50–70%) of spontaneous abortions.1 These abnormalities are highest in losses occurring early in gestation and become less common as gestation advances. For instance, losses that occur in the preembryonic period (5 weeks of gestation or less) are more likely to be associated with cytogenetic abnormalities than losses in the fetal period (more than 10 weeks of gestation).2 However, it is possible that some cases of pregnancy loss are associated with genetic abnormalities that are not detected by traditional karyotype.
One such type of genetic abnormality is small deletions or duplications, termed copy number changes. These can be detected by high resolution assays, such as array comparative genomic hybridization. Innumerable submicroscopic copy number changes can be detected with array comparative genomic hybridization. Some represent benign variations not associated with disease, others are pathogenic variations known to be associated with disease states, and others are of uncertain clinical significance.3 One of the best examples is neurodevelopmental delay. Array comparative genomic hybridization technology has identified several subtelomeric microdeletions thought to be the etiology of unexplained mental retardation in individuals with normal karyotypes.4–6
Another advantage of array comparative genomic hybridization is that it does not require viable cells that survive in culture. This technique provides almost all of the information provided by cytogenetic analysis as well as data regarding abnormalities in smaller regions of chromosomes than are identified by traditional karyotype. Accordingly, the technique has been successfully used to provide genetic information in cases of pregnancy loss when cell culture fails making cytogenetic evaluation difficult.7,8
We hypothesize that small cytogenetically cryptic de novo copy number changes are present in some cases of otherwise unexplained pregnancy loss. We chose to evaluate losses between 10 and 20 weeks of gestation because a large proportion of losses at 10 weeks of gestation or less are associated with aneuploidy. Our objective was to estimate genomic copy number changes in pregnancy losses occurring between 10 and 20 weeks of gestation using array comparative genomic hybridization.
MATERIALS AND METHODS
A prospective case series was performed to obtain preliminary information regarding the presence of genomic copy number changes in fetal loss specimens. This study was approved by the Institutional Review Boards of the University of Utah and Intermountain Health Care in Salt Lake City, Utah. Women with pregnancy loss greater than 10 weeks and less than 20 weeks of gestation were screened for enrollment in this study. Participants were enrolled from May 2007 through May 2008. They were recruited from obstetric clinics, ultrasound units, and clinician’s offices. Fifty women were approached to participate in the study, and 41 (82%) gave written informed consent. Medical and obstetric histories were obtained from all participants. Gestational age was determined by last menstrual period and ultrasound data. Clinical dating was changed if gestational age by sonographic criteria differed from that anticipated by menstrual dating by more than 10 days. This was an important consideration because the fetus often dies days or weeks before clinical evidence of pregnancy loss. Four pregnancies were determined to be less than 10 weeks of gestation, thereby excluding them from our analysis. Two additional fetuses were aneuploid based on cytogenetic testing, excluding them from further analysis. Thirty-five women remained who experienced pregnancy loss between 10 and 20 weeks of gestation with either a normal karyotype (n=9) or no conventional cytogenetic testing (n=26) and were included for analysis using array comparative genomic hybridization (Fig. 1).
In some cases, fetal and placental tissue was collected at the time of pregnancy loss when dilation and curettage was performed (n=30). In others, tissue was obtained from stored paraffin blocks in pathology (n=5). In addition, maternal and paternal blood samples were collected in sodium heparin (dark green top) and anticoagulant citrate dextrose (yellow top) tubes. DNA was isolated from fetal and placental tissue and parental whole blood with the Puregene Genomic DNA Purification Kit (Gentra Systems, Minneapolis, MN). All protocols for DNA isolation were followed according to manufacturer’s recommendations with minor modifications. DNA concentration was measured using a NanoDrop Spectrophotometer (NanoDrop Technologies, Inc, Wilmington, DE).
Array comparative genomic hybridization was performed on DNA from fetal tissue using a Spectral 2600 whole genome BAC array chip (SGI/PE 1Mb) and reagents (PerkinElmer, Turku, Finland). The test DNA was labeled with fluorescent dye and compared with normal male DNA from Promega (Madison, WI), used as a control for assay quality control. The control DNA was labeled with a different fluorescent dye. The BAC microarrays were run in duplicate using a dye-reversal strategy. A copy number change was considered present only if confirmed on both hybridizations. Chips were scanned using either the Gene Pix 4000B or the Agilent Microarray scanner and the Gene Pix Pro 6.0 software package (Axon Instruments, Union City, CA). The test and control DNA then were compared to determine whether there was a gain or loss of specific DNA sequences when hybridized to the BAC microarray. The data were analyzed using the SpectralWare (Waltham, MA) software program (SGI/PE) and ratio plots were generated based on the dye ratios at each BAC clone site on the Spectral chip. Clones that demonstrated copy number changes in the fetal tissue were compared against known copy number change regions in the Database of Genomic Variants (http://projects.tcag.ca/variation), a publicly available database that accumulates copy number change data on hundreds of healthy individuals, as well as the internal database of apparently benign copy number changes maintained by the University of Utah CGH laboratory since 2005.9 Parental DNA was analyzed using the same BAC array in cases that contained suspected pathogenic copy number changes. Copy number changes that were present in fetal tissue, but not in parental DNA were defined as de novo copy number changes.
Samples that showed de novo copy number changes in fetal tissue based in the BAC array data set were analyzed on the Agilent 244K oligonucleotide microarray chip (Agilent Technologies, Santa Clara, CA) for confirmation and to more accurately delineate the size of the copy number changes. The Agilent array has an average resolution of 6.4 Kb. Normal male or female DNA from Promega (Madison, WI) was used as a sex-matched control. Sample and control DNA was labeled with Cy5 or Cy3, respectively, using the Agilent DNA labeling kit. Following manufacturer recommended hybridization and washes, the arrays were scanned at 5 micrometers resolution using the Agilent microarray scanner and analyzed using Feature Extraction 18.104.22.168, DNA Analytics 4.0 (Agilent Technologies). Dye reversal is not used as a part of the Agilent protocol. Ratio plots were generated and areas of DNA gain or loss or both were compared with those seen on the Spectral array.
Thirty-five women were included in this study (Fig. 1). Demographics for our study population are summarized in Table 1. One participant had a history of venous thrombosis (no known thrombophilia) for which she was receiving prophylactic doses of enoxaparin. Eight women had recurrent pregnancy loss (defined as greater than or equal to three pregnancy losses); of these, one had recurrent fetal death. One patient with recurrent pregnancy loss was empirically treated with prophylactic enoxaparin, two with low dose aspirin, and one with a combination of these two medications. Four of the study participants had a family history of recurrent pregnancy loss.
DNA was successfully isolated in 30 of 35 (86%) cases, and array comparative genomic hybridization was performed in all of these. We were unable to successfully isolate DNA from paraffin embedded fetal tissue as the DNA quality was poor and therefore unreliable for this technology. Aneuploidy was detected by array comparative genomic hybridization in three of 26 cases that did not have initial cytogenetic analyses performed (13%). Therefore, a total of five cases of aneuploidy (13%) were detected out of the 37 women who agreed to participate and who were not excluded based on gestational age (see Fig. 1). Table 2 describes details regarding the cases of aneuploidy in our study population.
De novo copy number changes were detected in six of the 30 (20%) cases using the Spectral 1 MB BAC array and confirmed in four of 30 (13%) cases using the Agilent 244K oligonucleotide array. The characteristics of the de novo copy number changes detected by the Spectral array are summarized in Table 3. The confirmed de novo copy number changes on the Agilent array ranged in size from 93–289 Kb and mapped on 5p, 13q and Xp22 (Table 4 and Fig. 2). Cytogenetic testing indicated normal karyotypes in three of these cases; one failed to grow adequate cells. In the six cases with de novo copy number changes seen on the Spectral array, the higher-density Agilent oligonucleotide array detected additional copy number changes not identified with the Spectral chip and not present in the databases of benign copy number changes (Table 5). The copy number changes detected with the Agilent array ranged in size from 20 to 1,310 kb and involved multiple gene regions.
Array comparative genomic hybridization detected copy number changes in 13% of cases where routine cytogenetic testing was normal or not performed. These copy number changes involved large regions of DNA and may provide novel explanations for some cases of otherwise unexplained pregnancy loss. These findings are consistent with previous postulation that genetic abnormalities are underestimated in cases of fetal death. For instance, pregnancies affected by fetal malformations, deformations, syndromes, and dysplasias account for up to 35% of fetal deaths,7 and there is an aneuploidy rate of up to 25% in these pregnancies. The remaining 75% have phenotypes suggestive of genetic abnormalities despite their normal karyotypes.7,8 The large size of these copy number changes increases their likelihood of being clinically relevant. Indeed, two were 289 kilobases.
Three of the four de novo copy number changes detected on both arrays were duplications. When determining the disease-causing potential of copy number changes, deletions are more commonly thought to be pathogenic when compared with duplications. However, there are well-established circumstances where duplications are associated with abnormal phenotypes, such as is seen with the 22qter duplication.10 Numerous other clinical disorders have been associated with partial gene duplications.11
In addition to the copy number changes that were confirmed on both arrays, the higher density Agilent oligonucleotide array detected additional copy number changes that included multiple gene regions. Many of these genes have known function (Table 5), and some are known to be associated with disease states. For instance, copy number changes at 1p36.11, involving the RHD gene region, were seen in four of six patients. The RHD gene is known to be associated with cellular sodium transport. Another copy number change detected in four of six patients mapped to 6p21.32, a gene region implicated in the progression of multiple sclerosis.12 One particular copy number change mapped to 17q25.3, a locus including the mafG gene, has been associated with severe hematopoietic abnormalities and perinatal lethality in mouse models.13 The potential still exists for the copy number changes to be associated with alteration in gene function and disease processes when the copy number change lies outside the gene region. Even if it is not in a coding sequence, it can still affect gene function by altering the stability, splicing or localization of the mRNA.14 As more genome information is characterized, there is considerable potential to identify previously unrecognized places in the genome that may be pertinent to successful and unsuccessful pregnancy.
Array comparative genomic hybridization technology is well suited for this exploratory approach to the identification of new genetic mechanisms of pregnancy loss. The technique allows comprehensive assessment of the entire genome at very high resolution. Traditional cytogenetic techniques for analyzing abortus specimens require successful culture and evaluation of metaphase cells. Karyotype may be unsuccessful in up to half of cases, which may lead to underestimation of the frequency of genetic abnormalities in pregnancy loss.7 The presence of maternal tissue can lead to cell overgrowth by maternal cells, contributing to culture failure. In addition, cells from fetal tissue will sometimes fail to grow, especially if there is a long time interval from the demise until the culture is performed. Array comparative genomic hybridization is often useful to determine karyotype when cell culture fails. This technique does not require live cells, and can be used on DNA from macerated tissues, which are not acceptable for cell culture. Most, but not all (ie, balanced rearrangements), information obtained from traditional karyotype also can be derived by this technique.15 However, much more detailed information regarding smaller copy number changes also is obtained. Because of these advantages, many genetics laboratories are considering changing from cytogenetics to array comparative genomic hybridization as their primary method of chromosome assessment.
The wealth of information provided by array comparative genomic hybridization creates a dilemma because it is unclear which copy number changes are clinically relevant. Copy number changes are present in both patients with medical disorders and healthy individuals.16 On one extreme, copy number changes can represent clear genomic imbalances that cause well-known microdeletion/duplication syndromes.17 On the other extreme, copy number changes may be small, clinically meaningless polymorphisms. When analyzing array data for clinical purposes, copy number changes should be categorized into those that are likely to be benign, those that are likely to be pathogenic, and those that are of unknown clinical significance.3 If the copy number change is seen both in the affected patient and in a healthy parent, it is more likely to be benign. This criterion for determining the pathogenicity of a genomic variation is the standard for karyotype analysis and has thus been applied to array comparative genomic hybridization.4,18,19 One of the main strengths of our study compared with other studies of pregnancy loss was the analysis of both maternal and paternal DNA to help determine the potential pathogenicity of the copy number changes detected in the fetal specimens. It is important, however, to recognize that copy number changes, even those that are not de novo, may not represent benign changes. For example, normal individuals with copy number changes may have subtle phenotypes that are difficult to detect on clinical examination. In addition, a copy number change that is heterozygous in a parent and homozygous in offspring may affect the offspring more severely. Finally, heterozygous copy number changes have the potential to uncover recessive mutations on the other chromosome. Another criterion for determining the clinical relevance of a copy number change is the size.3 The large copy number changes noted in our cases are more likely to be associated with clinical problems than smaller ones.
De novo copy number changes in an affected patient should subsequently be compared with catalogs of copy number changes found in affected as well as healthy individuals to further assess whether the genomic imbalance has the potential to be pathogenic. Publicly available databases containing data on hundreds of healthy individuals are now available. One such resource is the Database of Genomic Variants (http://projects.tcag.ca/variation), which was used in our study. Care must be taken when utilizing these databases as only a small proportion of the reported copy number changes identified are validated, and some are reported on single platforms in very small numbers of patients. Thus, even if copy number changes in our cases were not present in either parent or in these public databases of genomic variants we cannot be absolutely certain that they are the cause of the pregnancy loss.
In our study population, the de novo copy number changes detected on the Spectral array were uniformly single clone variations. Although such a change could be a rare variant, it is more likely to be a false positive and thus could lead to incorrect interpretation of its clinical significance. For this reason, we confirmed each of the single clone copy number changes determined on the Spectral array on the Agilent oligonucleotide array. In addition, copy number change data are lacking for individuals with similar clinical presentations to those in our study population. Therefore, for the de novo copy number changes seen in our cases of fetal loss and not present in healthy parents, several other criteria may be necessary to estimate the clinical relevance of the copy number changes identified, including gene content, nature, and size.3 This will be especially important in characterizing copy number changes that are seen with increased frequency, such as the Xp22 gain that was seen in two of the four de novo changes in our population.
The clinical implications for identifying an underlying cause in previously unexplained pregnancy loss in the fetal period are important. Many cases of unexplained pregnancy loss are treated with empiric therapies that carry some risk without proven benefit.1 Accurately diagnosing a genetic etiology of pregnancy loss will curb unnecessary treatments and enable clinicians to more accurately counsel women regarding recurrence risk in future pregnancies. In addition, copy number changes that are associated with pregnancy loss will provide clues as to which genes are critical to reproduction. The next steps will involve controlled studies to compare the frequencies of copy number changes in fetal loss with normal pregnancies and further characterization of genes in the regions of copy number changes to delineate their role in both normal and abnormal pregnancy.
1.Silver RM, Branch DW. Sporadic and recurrent pregnancy loss. In: Reece AE, Hobbins J, ed. Clinical obstetrics: the fetus and the mother. Oxford: Blackwell Publishing; 2007:143–60.
2.Geraedts JP. Chromosomal anomalies and recurrent miscarriage. Infertil Reprod Med Clin North Am 1996;7:677–88.
3.Lee C, Iafrate AJ, Brothman AR. Copy number variations and clinical cytogenetic diagnosis of constitutional disorders. Nat Genet 2007;39:S48–54.
4.Shaw-Smith C, Redon R, Rickman L, Rio M, Willatt L, Fiegler H, et al. Microarray based comparative genomic hybridization (array-CGH) detects submicroscopic chromosomal deletions and duplications in patients with learning disability/mental retardation and dysmorphic feature. J Medical Genetics 2004;41:241–8.
5.Aston E, Whitby H, Maxwell T, Glaus N, Cowley B, Lowry D, et al. Comparison of targeted and whole genome analysis of postnatal specimens using a commercially available array based comparative genomic hybridisation (aCGH) microarray platform. J Med Genet 2008;45:268–74.
6.Moeschler JB. Genetic evaluation of intellectual disabilities. Semin Pediatr Neuro 2008;15:2–9.
7.Wapner RJ, Lewis D. Genetics and metabolic causes of stillbirth. Semin Perinatol 2002;26:70–4.
8.Pauli RM, Reiser CA. Wisconsin Stillbirth Service Program: II. Analysis of diagnoses and diagnostic categories in the first 1,000 referrals. Am J Med Genet 1994;50:135–53.
9.Whitby H, Tsalenko A, Aston E, Tsang P, Mitchell S, Bayrak-Toydemir P, et al. Benign copy number changes in clinical cytogenetic diagnostics by array CGH. Cytogenet Genome Res
10.Ensenauer RE, Adeyinka A, Flynn HC, Michels VV, Lindor NM, Dawson DB, et al. Microduplication 22q11.2, an emerging syndrome: clinical, cytogenetic, and molecular analysis of thirteen patients. Am J Hum Genet 2003;73:1027–40.
11.Hu X, Worton RG. Partial gene duplication as a cause of human disease. Hum Mutat 1992;1:3–12.
12.Caillier SJ, Briggs F, Cree BA, Baranzini SE, Fernandez-Viña M, Ramsay PP, et al. Uncoupling the roles of HLA-DRB1 and HLA-DRB5 genes in multiple sclerosis [published erratum appears in J Immunol 2009;182:2551]. J Immunol 2008;181:5473–80.
13.Onodera K, Shavit JA, Motohashi H, Yamamoto M, Engel JD. Perinatal synthetic lethality and hematopoietic defects in compound mafG::mafK mutant mice. EMBO J 2000;19:1335–45.
14.Cartegni L, Chew SL, Krainer AR. Listening to silence and understanding nonsense: exonic mutations that affect splicing. Nature Rev Genet 2002;3:285–98.
15.Benkhalifa M, Kasakyan S, Clement P, Baldi M, Tachdjian G, Demirol A et al. Array comparative genomic hybridization profiling of first-trimester spontaneous abortions that fail to grow in vitro. Prenat Diagn 2005;25:894– 900.
16.Scherer SW, Lee C, Birney E, Altshuler DM, Eichler EE, Carter NP, et al. Challenges and standards in integrating surveys of structural variation. Nat Genet 2007;39: S7–15.
17.Bejjani BA, Shaffer LG. Application of array-based comparative genomic hybridization to clinical diagnostics. J Mol Diagn 2006;8:528–33.
18.De Vries BB, Pfundt R, Leisink M, Koolen DA, Vissers LE, Janssen IM, et al. Diagnostic genome profiling in mental retardation. Am J Hum Genet 2005;77:606–16.
19.Friedman JM, Baross A, Delaney AD, Ally A, Arbour L, Armstrong L, et al. Oligonucleotide microarray analysis of genomic imbalance in children with mental retardation [published erratum appears in Am J Hum Genet 2006;79:1135]. Am J Hum Genet 2006;79:500–13.