Use of cell-free DNA in miscarriage research

Emily Colley ^1,2,3,*, Adam J. Devall ^1,2, Helen Williams ^1,4 , Susan Hamilton ³, Paul Smith ^1,2, Neil V. Morgan ⁵ , Siobhan Quenby ^6,7, Arri Coomarasamy ^1,2 and Stephanie Allen ³

1. Tommy’s National Centre for Miscarriage Research, Birmingham Women’s and Children’s Hospital, Birmingham B15 2TG, UK; A.J.Devall@bham.ac.uk (A.J.D.); H.M.Williams.1@bham.ac.uk (H.W.); paul.smith@doctors.org.uk (P.S.); A.Coomarasamy@bham.ac.uk (A.C.)
2. Institute of Metabolism and Systems Research, College of Medical and Dental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
3. West Midlands Regional Genetics Laboratory, Birmingham Women’s and Children’s Hospital, Birmingham B15 2TG, UK; susan.hamilton15@nhs.net (S.H.); stephanie.allen13@nhs.net (S.A.)
4. Institute of Clinical Sciences, College of Medical and Dental Sciences, University of Birmingham, Edgbaston B15 2TT, UK
5. Institute of Cardiovascular Sciences, College of Medical and Dental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK; N.V.Morgan@bham.ac.uk
6. Division of Biomedical Sciences, Warwick Medical School, University of Warwick, Coventry CV4 7HL, UK; S.Quenby@warwick.ac.uk
7. Tommy’s National Centre for Miscarriage Research, University Hospitals Coventry & Warwickshire NHS Trust, Coventry CV2 2DX, UK

* Correspondence: emily.colley1@nhs.net

Received: 1 October 2020; Accepted: 22 October 2020; Published: 26 October 2020

Overview:

Approximately one in four pregnancies ends in miscarriage, and 50% of miscarriages are due to chromosomal abnormalities. After three consecutive miscarriages, genetic testing using product of conception (POC) tissue is recommended. Cell-free DNA (cfDNA) has been used for prenatal screening, but it is rarely performed in non-pregnancies. We investigated the use of cfDNA from maternal blood to identify chromosomal abnormalities in miscarriages. We collected and stored 100,2 blood samples from women who had experienced a first-trimester miscarriage. The mean gestational age was 7.1 weeks (range: 5-11 weeks). Samples without POC genetic test results were not analyzed in this study.

CfDNA was extracted and analyzed using a modified commercially available genome-wide non-invasive prenatal test. Results were not provided to the patient. In 57 samples, cytogenetic results were obtained by POC analysis. Chromosomal abnormalities were identified in 47% (27/57) of POC analyses and correctly in 59% (16/27) of cfDNA analyses. In total, 75% (43/57) of results were correctly identified. The average cfDNA fetal fraction was 6% (2-19%). In conclusion, cfDNA can be used to detect chromosomal abnormalities in miscarriages when the “fetal fraction” is sufficiently high, but more studies are needed to identify variables that influence the overall outcome.

keyword:

Miscarriage; Cell-free DNA; Cytogenetic analysis; Chromosomal abnormalities

1. Introduction

First trimester miscarriage is the most common complication during pregnancy [1] and is defined as a miscarriage. One in five pregnancies ends in spontaneous abortion [2], 50% of which are due to chromosomal abnormalities [3]. Identifying whether a chromosomal abnormality is the underlying etiology of pregnancy loss is important as it may affect the prognosis of future pregnancies. If a sporadic chromosomal abnormality is the cause of pregnancy loss, the prognosis for future pregnancies is better than if the chromosomal complement is normal. In that case, the cause of the miscarriage may be non-chromosomal. In cases of unbalanced chromosomal rearrangements, it is possible that one of the parents has a balanced chromosomal rearrangement. This means that future pregnancies will be susceptible to the same or other unbalanced chromosomal rearrangements. In such cases, it is important to take blood samples for parental karyotyping to assess the risk of recurrence.

The Royal College of Obstetricians and Gynaecologists (RCOG) Green-top Guideline No.17 [4] recommends cytogenetic analysis of pregnancy tissue after the third or subsequent miscarriage, or karyotyping of parental samples if pregnancy tissue is not available. Traditionally, cell culture and G-banded chromosome analysis have been used to detect abnormalities in pregnancy tissue. However, there is often a high failure rate due to poor quality of tissue received, the difficulty of culturing cells from such tissue, and limited resolution to detect microdeletion and duplication syndromes. Therefore, molecular-based approaches such as quantitative fluorescent PCR (QF-PCR) and microarrays have been implemented beyond the laboratory.

Currently, genetic testing for miscarriage is performed using pregnancy tissue consisting of placental and fetal components, called products of conception (POC). This tissue must be fresh, uncontaminated, and unfixed to be able to identify the fetal tissue and perform DNA extraction and cell culture. This carries the risk of maternal cell contamination (MCC), which may lead to misdiagnosis of the sample. POC samples contain maternal tissue intertwined with fetal tissue. Maternal cells may be carried over during the selection of fetal tissue that yielded maternal DNA during DNA extraction, or maternal cells may overgrow during cell culture. Furthermore, in many cases, POC is not available or is not returned by the patient.

Cell-free DNA (cfDNA) was first identified by Dennis Lo [5], who demonstrated that a small piece of cfDNA in the plasma of a pregnant woman represents the entire genome of the fetus. Although cfDNA is already used for prenatal screening, to date, few studies have been conducted in non-pregnant women. Only two studies, by Clark-Ganheart et al. and Yaron et al. [6,7], have evaluated the use of cfDNA in the setting of miscarriage.

A prospective cohort study by Clark-Ganheart et al. [6] analyzed cfDNA samples from 50 non-pregnant women. Gestational age by ultrasound ranged from 6.1 to 38.4 weeks. Of these samples, 38 of 50 had reportable results, of which 8 had confirmed trisomy. A study by Yaron et al. [7] tested cfDNA to analyze pregnancy loss before 14 weeks. In total, 86 pregnancies had cfDNA results with equivalent POC (from CVS sampling). The median fetal fraction was 5%. Of the 86 samples, 55 (64%) had chromosomal abnormalities, of which 30 (55%) were detected using standard noninvasive prenatal testing (NIPT) log-likelihood ratio (LLR) cutoffs. To increase sensitivity, a pregnancy loss-specific threshold was developed using a “training set” of 50 samples. This improved the detection rate to 82%.

CfDNA is highly useful to confirm the chromosomal cause of miscarriage at the time of miscarriage diagnosis by a simple blood test. In this study, we investigate how cfDNA can be used to detect chromosomal abnormalities in miscarriage and compare the results with those of POC testing.

2. Materials and Methods

2.1. Ethical approval

This study was carried out at the Tommy’s National Centre for Miscarriage Research and received research ethics approval (REC reference. 16/WM/0423, 23/11/2016, West Midlands-South Birmingham Research Ethics Committee) and approval from the Health Research Authority (HRA) under IRAS project ID, 215646.

2.2. Patient samples

Informed consent was obtained from patients who had experienced early miscarriage and visited the Tommy’s National Center for Miscarriage Research, hosted by Birmingham Women’s and Children’s Hospital.

The NHS Trust and University Hospital Coventry & Warwickshire NHS Trust conducted a series of randomised controlled trials between February 2017 and July 2019. Consent explicitly included consent to work with patients’ POC and genetic material. Samples were collected as donations for medical research and the tissue(s) were handled in accordance with the Human Tissue Act (HTA). Donors maintained the ability to withdraw consent for further use but did not retain any rights over their samples once obtained.

Eligibility criteria included maternal age ≥16 years and gestational age <12 weeks as confirmed by ultrasound at the time of miscarriage diagnosis with pregnancy tissue remaining in situ. Samples were included in the analysis if cytogenetic results were available in the corresponding POC assay, with the exception of seven known triploid cases.

Blood samples were collected for cfDNA analysis and assessment of βhCG levels. Up to 10 mL of maternal blood was collected for cfDNA in cell-free DNA BCT (STRECK) tubes, and crown-rump length (CRL) was measured by ultrasound, if available, to assess fetal gestation. Chromosomal abnormalities found on POC testing were communicated to patients via standard patient care. CfDNA results were not shared with patients.

2.3. Sample processing

Plasma was separated from whole blood by double centrifugation and transferred in 1 mL aliquots into DNA LoBind tubes (Eppendorf). These aliquots were stored at -80°C until use.

2.4. Cytogenetic analysis

For POC after the third or subsequent consecutive miscarriage, products of conception (POC) were collected as routine clinical specimens and subjected to targeted QF-PCR and chromosomal microarray analysis (CMA) according to the RCOG Green-top guideline No. 17 [4].

First, trisomy screening by QF-PCR was performed on DNA from POCs to test for trisomies 13, 18 or 21, triploidy, and sex chromosome aneuploidy. Abnormal QF-PCR was reported. If normal, CMA testing was performed using the OGT CytoSure 8 × 60 k Constitutional v3 design, exon/gene-level resolution of ~500 DDD/ClinGen-curated developmental genes and syndrome regions, stratified backbone resolution of ~120–500 kb, and analysis with CytoSure v4.9 and build GRCh37 using the CBS algorithm. Microarray analysis detected copy number imbalances >1 Mb and in some cases with higher resolution.

2.5. Cell-free DNA testing

Plasma (1 mL) from patients who consented to external laboratory testing was submitted to the Illumina Laboratory in Cambridge and processed in 24-sample batches using a modified Illumina VeriSeq™ NIPT Solution v2 workflow as previously described [8,9] but using a modern analytical platform [10] with some minor modifications.

3. Results

A total of 102 cfDNA samples were collected after miscarriage was confirmed. All 102 samples were analyzed by the VeriSeq™ NIPT v2 Solution assay on the NextSeq500. 85 matched POC samples were received. In total, 64 pregnancies had matched cytogenetic results by POC assay, 21 POC samples were not suitable for analysis, and 17 cfDNA samples did not receive a matched POC sample. The 17 unreceived POC samples and 21 POC samples not suitable for analysis were excluded from cfDNA analysis, and known triploid pregnancies were excluded (Figure 1).

Chromosomal abnormalities were confirmed by POC analysis in 27/57 cases (47%). Patient baseline characteristics are summarized in Table 1. From the 57 cases with corresponding cytogenetic analysis, the mean age was 34 years (range 20-43 years), clinical gestational age was 7.1 weeks (range 5-11 weeks), and fetal rate was 6% (range 2-19%). A total of 70% (40/57) of samples were correctly identified using VeriSeq™, including 16/27 (59%) with genetic abnormalities and 27/30 (90%) with genetically normal samples.

This corresponds to a sensitivity of 59% (16/27), specificity of 90% (27/30), and accuracy of 75% (43/57), although the sample cohort was relatively small.

Table 1.Characteristics of cfDNA samples with corresponding product of conceptus (POC) results suitable for analysis (excluding triploid pregnancies)

	Total (n = 57)	Chromosomes are normal (n = 30)	Chromosome abnormalities (n = 27)
Maternal age (years) (mean and range)	34 (20-43)	31 (20-41)	37 (24-43)
Past loss (Average value and range)	3.8 (0-14)	4.1 (0-14)	3.3 (0-6)
Gestational age (weeks) (Mean and range)	7.1 (5-11)	7.4 (5-11)	6.4 (5-9.3)
βhCG (mIU/mL) (Mean value and range)	38,356 (69-263,766)	50,632 (69-263,766)	21,538 (491-100,638)
Fetus proportion (%) (mean and range)	6 (2-19)	7 (2-19)	5 (3-12)

CfDNA analysis correctly identified 43/57 cases (75%), including 16 abnormalities and 27 normals. Table 2 compares POC results with cfDNA test results for cases with abnormalities detected by POC. Abnormalities detected by POC were common trisomies (3), monosomy X (2), common trisomy 45 combined with X (1), monosomy 21 (1), complete rare trisomies (14), mosaic rare trisomies (2), and copy number variations (4). Among the rare trisomies, trisomy 22 and trisomy 15 were the most common. The fetal fraction ranged from 3 to 12% (mean 5%). CfDNA results were completely concordant with POC results in 40/57 samples. CfDNA results produced normal results in 27/30 cases and discordant results in 3/30 known normal cases. Two mosaic samples (sample IDs 51 and 586) were not correctly identified by cfDNA testing, but other imbalances were detected in these samples. Sample ID 228 had a monosomy 21 result at POC, but cfDNA testing identified CNVs in several other chromosomes.

Four samples were miscarriages with subchromosomal deletions and duplications identified by POC analysis (sample IDs 133, 202, 303, and 319). A 56 Mb duplication at 7q22.1q36.3 and a 21 Mb terminal duplication at 19q13.12q13.43 were detected by cfDNA analysis. A 70 Mb deletion at 13q13.3q34, a 6 Mb terminal deletion at 7q36.2q36.3, a 9 Mb terminal duplication at 4q34.3q35.2, and a 30 Mb terminal deletion at 5q33.1q35.3 were not detected by cfDNA analysis.

To see if we could improve the result call between cfDNA analysis and POC cytogenetic analysis, we grouped into three categories with different gestation, βhCG levels and fetal fraction cutoffs (Table 3). Gestational age was divided into four groups: <7 weeks, 7-8 weeks, ≥8 weeks and unknown gestation. In these groups, the number of chromosomal abnormalities correctly identified by the cfDNA test increased with increasing gestational age. βhCG levels were divided into three groups: <8000, 8000-35000 and ≥35000mIU/mL. As βhCG levels increased in these groups, the number of chromosomal abnormalities correctly identified by the cfDNA test also increased. The fetal fraction group was divided into three groups: <5, 5-8 and ≥9. Again, the accuracy rate of chromosomal abnormalities by the cfDNA test increased with increasing fetal fraction group.

Table 3. CfDNA vs. Karyotype of POC

		CfDNA results
		Recognized normally (%)	Unrecognized (%)
	Total	43 (75.4)	14 (324.6)
Pregnancy period (weeks)	Less than 7 weeks	14 (66.7)	7 (33.3)
	7～8週	10 (76.9)	3 (23.1)
	8 or more	11 (100.0)	0 (0.0)
	Unknown	8 (53.3)	7 (46.7)
βhCG (mIU/mL)	Less than 8000	9 (60.0)	6 (40.0)
	8000～35,000	14 (66.7)	7 (33.3)
	More than 35,000	19 (95.0)	1 (5.0)
Fetal percentage (%)	Less than 5	13 (59.1)	9 (40.9)
	5～8	19 (79.2)	5 (20.8)
	9 or more	10 (100.0)	0 (0.0)

4. Discussion

Our cfDNA study cohort was recruited through the Tommy’s National Centre for Miscarriage Research at Birmingham Women’s Hospital and University Hospital Coventry and Warwickshire. A total of 102 samples were assessed using a modified VeriSeq™ NIPT V2 (Illumina) and 57 samples were analysed with corresponding POC cytogenetic analysis.

cfDNA analysis was performed in three categories (Table 3). In the low fetal fraction group, with a fetal fraction of less than 5%, only 60% of samples were identified with chromosomal abnormalities, most of which were from dizygotic pregnancies. On the other hand, with a fetal fraction of 9% or more, 100% of cytogenetic results were correctly identified. In our study, we note that a fetal fraction of 5% or more would allow the majority of abnormalities to be detected. However, due to the small sample size and biological variability, it is difficult to define an exact cutoff.

We believe that the discrepancies observed between POC genetic test results and cfDNA testing may be due to confined placental mosaicism. CfDNA analysis tests DNA derived only from placental/cytoblast cells, whereas POC testing may consist of fetal and whole placental tissues. This may lead to discrepancies in results. In two cases, mosaic genetic abnormalities were identified in POC analysis that could not be identified by cfDNA testing. These results may be due to placental mosaicism confined to trisomal cells or due to limitations of current methods. Mosaicism is difficult to diagnose by any method, and cfDNA analysis may be able to play an adjunct role to current POC testing in detecting biologically relevant abnormal cell lines.

Tommy’s National Center for Miscarriage Research specializes in the care of families experiencing recurrent miscarriage. These families are very conscious of when to conceive and would benefit from careful monitoring during the first trimester. As a result, miscarriages in our study cohort were diagnosed earlier than in other studies; Clark-Ganheart et al. [6] recorded 16.9 weeks (6.1-37.2), and Yaron et al. [7] 9.6 weeks (5.1-13.6) (Figure 2).

This study and others have demonstrated that in the majority of pregnancy loss cases where the pregnancy tissue is still in situ, it is possible to detect chromosomal abnormalities using cfDNA. In this study, 59% of chromosomal abnormalities were correctly identified, with a concordance rate of 75% with POC results. In comparison, Clark-Ganheart et al. [6] showed an 87.5% concordance rate when cytogenetic results were available, and Yaron et al. [7] showed an 82% concordance rate when a pregnancy loss-specific LLR threshold was used. Yaron et al. [7] used 50 cases as a training set and set a pregnancy loss-specific LLR threshold. The overall detection rate was 82% in 86 non-mosaic cases. This was achieved after identifying pregnancy loss-specific LLRs based on the training set. This indicates that the LLRs required for this cohort need to be different from singleton pregnancies. For comparison, our study analyzed cfDNA using standard NIPT LLR cutoffs, and having a pregnancy loss-specific LLR may improve detection rates. The next step in this work will be to conduct testing of this cohort of patients using a modified LLR-based pipeline to optimize the detection rate for all autosomal trisomies, using an algorithm similar to that proposed by Yaron et al. [7].

This study and others [6,7] have shown that cfDNA can be used to assess the genetic contribution to miscarriage. However, there are still genetic abnormalities that may be missed depending on the assay used (e.g. triploid and mosaic samples, autosomal trisomies at low fetal fraction/low gestational age). Triploid cases were excluded in both this study and Yaron et al. [7] because they could not be detected by the modified Illumina VeriSeq™ NIPT solution v2 workflow. However, single nucleotide polymorphism-based platforms for the analysis of cfDNA should be able to identify triploid cases.

cfDNA cannot completely replace current cytogenetic testing. Yaron et al. [7] proposed a recurrent pregnancy loss algorithm that utilizes cfDNA testing in recurrent pregnancy loss.

When a third or subsequent pregnancy loss is diagnosed, the current guideline by the Royal College of Obstetricians and Gynaecologists (RCOG) Green-top Guidelines No.17 [4] is to test pregnancy tissue for fetal aneuploidies. In parallel to this routine testing, a maternal blood sample can be taken to complete a cfDNA test. If the cfDNA test detects an aneuploidy and explains the reason for the miscarriage, no further work is necessary as the numerical error is usually sporadic and the chances of a successful subsequent pregnancy are not negatively affected. As cfDNA only detects unbalanced chromosomal abnormalities, if the cfDNA test does not detect a chromosomal abnormality, a POC cytogenetic analysis should be recommended to see if there are chromosomal abnormalities that cannot be detected by cfDNA testing (e.g. CNV, triploidy, mosaic samples). This could reduce the number of POC tests required and provide results for many patients for whom POC is not available. It is important to note that some chromosomal abnormalities will still be missed if POC is not available. If an unbalanced translocation is identified, parental karyotyping should be offered to assess whether one (or both) of the parents are carriers of the translocation.

5. Conclusion

Knowing the genetic consequences of pregnancy loss can be applied in counseling patients to inform their prognosis for future pregnancies and may also help provide psychological support to relieve them of the guilt associated with pregnancy loss.

The use of cfDNA to identify whether miscarriage is due to a chromosomal abnormality would have a significant clinical impact in patients for whom testing pregnancy tissue is unavailable or where traditional cytogenetic testing is not available due to patient preference, however, cfDNA testing is only available if the pregnancy remains ongoing at the time of miscarriage diagnosis.

We have demonstrated that by collecting maternal plasma samples when the pregnancy tissue is still in situ and with sufficient fetal fraction, it is possible to detect genetic abnormalities in miscarriages using cfDNA in some cases. Further work is needed to improve this test and identify variables that may affect the overall outcome so that it can be applied clinically.

Author contributions

EC–data analysis and writing; AJD–study design and ethical approval; HW–study design, ethical approval, grant administration and editing; SH–data analysis and critical appraisal; PS–study design and critical appraisal; NVM–supervision and critical appraisal; SQ–critical appraisal; AC–study design, supervision and critical appraisal; SA–study design, supervision, critical appraisal, data analysis, writing and editing. All authors read the manuscript and agreed to the published version of the manuscript.

資金提供

The research has been funded by Tommy’s Charity, which funds the Tommy’s National Centre for Miscarriage Research.

Acknowledgements

The authors would like to thank Lieve Page-Christians, Philippa Burns, Patrizia Di Pietro, Cosmin Deciu, Sarah Kinnings, Eoin Brown, Victoria Corey, Kathryn Robinson, Sven Schaffer, Sucheta Bhatt, Barbara Baggiani, Matt Smith, Marion Burdin and Jasmin Dehnhardt. The authors thank the Tommy National Miscarriage Research Centre at Birmingham Women’s and Children’s Hospital and the research teams at University Hospital Coventry & Warwickshire for patient recruitment and sample collection.

Conflicting interests

The authors declare no conflicts of interest.

References

1. Roberts, C.; Lowe, C. Where have all the conceptions gone? Lancet 1975, 305, 498-499. [CrossRef]
2. Macklon, N.S.; Geraedts, J.P.M.; Fauser, B.C.J.M. Conception to ongoing pregnancy: The ‘black box’ of early pregnancy loss. Hum. Reprod. Updat. 2002, 8, 333-343. [CrossRef] [PubMed]
3. Choi, T.Y.; Lee, H.M.; Park, W.K.; Jeong, S.Y.; Moon, H.S. Spontaneous abortion and recurrent miscarriage: A comparison of cytogenetic diagnosis in 250 cases. Obstet. Gynecol. Sci. 2014, 57, 518-525. [CrossRef] [PubMed]
4. Regan, L.; Backos, M.; Rai, R. The Investigation and Treatment of Couples with Recurrent First-Trimester and Second-Trimester Miscarriage; RCOG Green-top Guideline No.17; Royal College of Obstetricians and Gynaecologists: London, UK, 2011.
5. Lo, Y.M.D.; Corbetta, N.; Chamberlain, P.F.; Rai, V.; Sargent, IL; Redman, C.W.; Wainscoat, JS Presence of fetal DNA in maternal plasma and serum. Lancet 1997, 350, 485-487. [CrossRef]
6. Clark-Ganheart, C.A.; Fries, M.H.; Leifheit, K.M.; Jensen, T.J.; Moreno-Ruiz, N.L.; Ye, P.P.; Jennings, J.M.; Driggers, R.W. Use of Cell-Free DNA in the Investigation of Intrauterine Fetal Demise and Miscarriage. Obstet. Gynecol. 2015, 125, 1321-1329. [CrossRef]
7. Yaron, Y.; Pauta, M.; Badenas, C.; Soler, A.; Borobio, V.; Illanes, C.; Paz-Y-Mi, F.; Martinez-Portilla, R.; Borrell, A. Maternal plasma genome-wide cell-free DNA can detect fetal aneuploidy in early and recurrent pregnancy loss and can be used to direct further workup. Hum. Reprod. 2020, 35, 1222-1229. [CrossRef] [PubMed]
8. Pertile, M.; Flowers, N.; Vavrek, D. P1-86. Performance of a genome-wide PCR-free, paired-end sequencing-based noninvasive prenatal screening test, VeriSeq™ NIPT Solution v2, in the detection of fetal chromosomal anomalies. Poster Abstracts of the ISPD 23rd International Conference on Prenatal Diagnosis and Therapy, Singapore, 7-11 September 2019. Prenat Diagn. 2020, 40 (Suppl. 1), 74-75.
9. Andrews, D.; Padhukasahasram, B.; Kim, S. P1-66. Validation of VeriSeq™ NIPT Solution v2, a paired-end PCR-free WGS-based methodology for noninvasive prenatal screening of fetal aneuploidies and CNVs on a genome-wide scale. Poster Abstracts of the ISPD 23rd International Conference on Prenatal Diagnosis and Therapy, Singapore, 7-11 September 2019. Prenat Diagn. 2020, 40 (Suppl. 1), 62-63.
10. Leinfnger, P.; Lohmann, L.; Luscan, A.; Trost, D.; Bidat, L.; Debarge, V.; Castaigne, V.; Senat, M.-V.; Brechard, M.-P.; Guilbaud, L.; et al. Strategy for Use of Genome-Wide Non-Invasive Prenatal Testing for Rare Autosomal Aneuploidies and Unbalanced Structural Chromosomal Anomalies. J. Clin. Med. 2020, 9, 2466. [CrossRef] [PubMed]

Publisher’s Note

MDPI remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.