The genetic etiology of spontaneous abortion: insights from chromosomal microarray analysis and whole-exome sequencing

Abstract

Spontaneous abortion is one of the most common adverse pregnancy outcomes, with an incidence of approximately 15–25% in clinically recognized pregnancies. Chromosomal abnormalities are recognized as a primary genetic etiology, yet the specific underlying molecular mechanisms and pathways remain incompletely understood. Embryonic tissue samples from 1585 patients with spontaneous abortion were collected from January 2019 to June 2025. Chromosomal microarray analysis (CMA) was used for whole-genome detection of chromosomal abnormalities, including aneuploidy and copy number variations (CNVs). Functional enrichment analysis of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) were conducted on CNV-related genes. 34 cases with negative CMA results and a history of recurrent pregnancy loss (≥ 2 episodes) were randomly selected to undergo whole-exome sequencing (WES) for the screening of monogenic variants. Statistical analysis included chi-square tests for inter-group comparisons, restricted cubic splines (RCS), and multivariate logistic regression for assessing risk factors. The detection rate of chromosomal abnormalities was 54.44%, mainly numerical abnormalities, with trisomy 16 and X monosomy being the most common subtypes. Structural variants include CNVs, loss of heterozygosity (LOH) and Runs of homozygosity(ROH). The results of the GO enrichment analysis indicated that genes associated with CNVs are primarily involved in hemostasis regulation and monocyte migration; KEGG enrichment analysis indicated that the Toll-like receptor signaling pathway and the neuroactive ligand-receptor interaction pathway may play a significant role in the pathogenesis of miscarriage. Multivariate Logistic regression analysis revealed that a mother’s age of ≥ 35 years was an independent risk factor (OR = 1.72, 95%CI: 1.31–2.26), while a gestational age of ≥ 12 weeks had a protective effect (OR = 0.27, 95%CI: 0.21–0.34). The RCS model revealed a U-shaped relationship between maternal age and the risk of chromosomal abnormalities. The risk was lowest at ages 28–29 (OR = 0.99) and increased gradually after age 30; the risk of chromosomal abnormalities peaked between 8 and 10 weeks of gestation. The diagnostic yield of WES in recurrent miscarriage cases with normal CMA results reached 26.47%. This approach identified a variant in the MOS gene associated with oocyte maturation defects, along with clinically significant variants in possible miscarriage-implicated genes such as TTC7A. Chromosomal abnormalities, particularly trisomy 16 and monosomy X, along with dysregulation of hemostatic and immune pathways, are central to spontaneous abortion etiology. Advanced maternal age, defined as 30 years or older, and early gestational loss are key risk factors, challenging the conventional threshold of 35 years. The use of a combined CMA and WES strategy has effectively improved the genetic diagnosis rate, demonstrating the complementary value and advantages of combining chromosomal and single-gene testing technologies in diagnosing the causes of miscarriage.

Introduction

Spontaneous abortion(SA) refers to the spontaneous termination of a pregnancy occurring before 28 weeks of gestation, with the fetus weighing less than 1,000 g. Based on the gestational period, pregnancies terminated before 12 weeks are defined as early miscarriages, while those occurring at 12 weeks or later are classified as late miscarriages¹. SA is a common adverse pregnancy outcome², with approximately 15%-25% of clinically identified pregnancies ending in miscarriage³. Its etiology is complex and multifactorial, including genetic, endocrine, infectious, coagulation, immune, environmental, and unhealthy lifestyle factors, among others⁴. Most current studies suggest that embryonic chromosomal abnormalities are the primary cause^5–7. Chromosomal abnormalities can disrupt the normal developmental process of the embryo or lead to abnormal implantation in the uterus, subsequently leading to miscarriage^8–11. However, research into the molecular mechanisms of SA remains limited, and the specific signaling pathways underlying embryonic developmental arrest remain unclear.

Currently, genetic testing methods for fetal tissue have shifted from traditional chromosomal karyotyping to chromosomal microarray analysis (CMA). As one of the main technologies of CMA, single nucleotide polymorphism microarray technology (SNP-array) not only identifies aneuploidy and copy number variations (CNVs) but also provides information on loss of heterozygosity (LOH), uniparental disomy (UPD), and other chromosomal abnormalities. It offers advantages such as operational simplicity, rapid turnaround time, and high efficiency^12–14. Whole-exome sequencing (WES) performed on fetal products of conception (POC) tissue can identify single-gene causes associated with coagulation disorders, immune regulation abnormalities, and embryonic developmental defects. It not only identifies pathogenic variants in known Mendelian disease genes but also reveals new candidate genes through rare variant enrichment analysis and pathway mining. It compensates for the technical limitations of CMA in the diagnosis of monogenic variants and recessive genetic etiologies.

This study employed CMA combined with WES technology to analyze miscarriage tissues, aiming to clarify the types of chromosomal abnormalities associated with SA, their related genes, and signaling pathways. Additionally, data on maternal age, gestational week at miscarriage, and number of miscarriages were collected to assess potential associations with chromosomal abnormalities, exploring the application value and clinical significance of combined testing in the genetic etiology analysis of SA.

Subjects and methods

Research subjects

This study included patients who experienced spontaneous abortion and were treated at the Urumqi Maternal and Child Health Hospital between January 2019 and June 2025. A total of 1,585 embryonic tissue samples were collected, of which 3 were from in vitro fertilization and embryo transfer (IVF-ET) pregnancies, while the remainder were from naturally conceived pregnancies. Inclusion criteria: (1) Spontaneous abortion was diagnosed by transvaginal color Doppler ultrasound, and embryonic tissue samples were successfully obtained; (2) Age 20–45 years; (3) Gestational age ranges from 4 to 27^+ 6 weeks; (4) Patients who had signed an informed consent form for SNP-array testing; (5) Complete clinical data. Exclusion criteria: (1) Induced abortion (including therapeutic termination due to maternal medical conditions or social factors); (2) Termination of pregnancy beyond 28 weeks of gestation; (3) Miscarriage caused by uterine malformations, reproductive system diseases, endocrine disorders, or infections. All embryonic tissue samples were obtained from chorionic villi or fetal muscle tissue collected during uterine curettage or induced abortion surgery for spontaneous abortion in the Department of Obstetrics and Gynecology, Urumqi Maternal and Child Health Hospital. This study was approved by the Ethics Committee of the Urumqi Maternal and Child Health Hospital (Approval Number: XJFYLL2023037).

Methods

This study employed a two-phase design. In the first phase, CMA technology was used to detect chromosomal variations at the whole-genome level in 1,585 samples of miscarried fetal tissue to identify genetic causes at the chromosomal level. In the second phase, 34 samples were selected from patients with negative CMA results and a history of recurrent miscarriage (≥ 2 episodes) for WES analysis to identify potential pathogenic mutations at the monogenic level.

Sample preparation

Under sterile conditions, approximately 50 mg of chorionic villi tissue or muscle tissue from a formed fetus was collected and thoroughly washed with sterile saline. Tissue and blood samples were prepared for genomic DNA extraction using the Qiagen DNeasy Blood & Tissue Kit. DNA quantity and quality were determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Madison, Georgia, USA).

Genomic DNA extraction

Genomic DNA was extracted from the aborted tissue using the Qiagen Kit (QIAamp DNA Blood MiniKit). DNA concentration and purity were determined using the NanoDrop 2000 spectrophotometer or Qubit fluorometer, with the OD260/OD280 ratios maintained between 1.8 and 2.0 and the OD260/OD230 ratio greater than 2.0.

Analysis of short tandem repeats

Short tandem repeat (STR) analysis was used to rule out maternal cell contamination and assist in determining ploidy. Fluorescently labeled multiplex PCR was used to amplify target STR loci, followed by capillary electrophoresis for genotyping. Contamination levels were qualitatively and semi-quantitatively assessed by comparing allele peak profiles and peak area ratios between fetal (POC) samples and mothers. A contamination rate of < 5% was set as the acceptance threshold. Positive and negative controls were incorporated throughout the process, sterile procedures were implemented in designated zones, and sample quality control was strictly enforced to ultimately screen out samples contaminated with maternal cells.

CMA

Miscarriage tissue samples were analyzed using the Affymetrix CytoScan™ 750k chip. Labeled DNA fragments hybridized with probes on the chip, and hybridization conditions (temperature and humidity) were strictly controlled to ensure specific binding between the probes and DNA fragments. Unbound DNA fragments are removed using a washing buffer. After washing and drying, the chip is scanned using an Affymetrix GeneChip scanner to read the fluorescence signals. Copy number variations (CNVs) were called with the following thresholds: ≥50 kb for deletions and ≥ 100 kb for duplications in autosomes, and ≥ 100 kb for deletions and ≥ 200 kb for duplications in sex chromosomes. This array enables reliable detection of chromosomal mosaicism when the abnormal cell fraction is ≥ 20–30%, while mosaicism with a lower abnormal cell fraction may remain undetected due to technical constraints. All variants were annotated using the OMIM, DGV, and ISCN databases.

WES

Genomic DNA was extracted from placental villi tissue samples using a Qiagen DNA extraction kit. After passing quality control, library construction was performed. The DNA was randomly fragmented using sonication, with target fragment sizes of approximately 200–350 bp. Subsequently, the DNA fragments were hybridized and captured using the Berry Genomics exome capture kit. Following elution, enrichment, PCR amplification, and purification steps, a WES library was obtained. Next-generation sequencing was performed using a NovaSeq 6000 sequencer (Illumina, USA) with a paired-end (PE) read length of 150 bp. The detailed experimental and bioinformatics analysis pipeline was standardized for all samples as follows:

Proband-only analysis

WES analysis was performed in a proband-only manner, with only genomic DNA from embryonic villi tissues included in the sequencing and analysis; parental samples were not collected for this study.

Reference genome

The reference sequence for the nuclear genome is GRCh37/hg19, and the reference sequence for the mitochondrial genome is rCRS.

Alignment, variant calling and annotation

Target regions were captured and sequenced. Reads were aligned to the reference genome, and variants (SNVs, Indels, CNVs, ROH, UPD) were called and annotated using public databases (gnomAD, ClinVar, OMIM) and in silico tools.

Quality control (QC) metrics

Strict quality control was implemented throughout the WES process, and samples failing to meet the following criteria were excluded: Base quality Q30 ratio ≥ 90%; target region coverage at ≥ 20× ≥90%; mean sequencing depth of target regions ≥ 100×; mean mitochondrial genome sequencing depth ≥ 100×.

Variant filtering strategy

Variants were filtered based on quality, frequency, region, and phenotype. Samples with MCC ≥ 5% and low-quality variants were excluded. Rare coding and splice-site (± 10 bp) variants were retained, prioritizing those associated with spontaneous abortion or fetal demise for pathogenicity evaluation.

ACMG classification

Genetic variants were classified according to ACMG guidelines into five categories: Pathogenic (P), Likely Pathogenic (LP), Variant of Uncertain Significance (VUS), Likely Benign (LB), and Benign (B).

Sanger confirmation

All pathogenic (P) and likely pathogenic (LP) variants identified by WES were validated by Sanger sequencing. PCR amplicons spanning the variant sites were amplified, and Sanger sequencing was performed on a capillary electrophoresis platform; sequencing results were analyzed to confirm the presence of the variants, ensuring the accuracy of WES detection results.

Statistical analysis

Initial data were entered and organized using Excel spreadsheets, and analyzed using SPSS 26.0 statistical software. Count data were described using counts and percentages (%). Intergroup comparisons were performed using the chi-square test, with a significance level set at P<0.05. Variables showing statistically significant differences in the univariate analysis were included in subsequent logistic regression analyses and restricted cubic spline (RCS) models. Using R version 4.4.0, logistic regression analyses were performed separately for each factor; based on the rms package included in the software, RCS models were used to construct dose-response curves for the relationship between maternal age, gestational age at miscarriage, and the risk of chromosomal abnormalities.

Functional enrichment analysis

The ClusterProfiler package in RStudio was used to perform GO and KEGG enrichment analyses on CNV-related genes. Pathways with a P-value < 0.05 were selected as significantly enriched. Based on GO functional classification and KEGG annotation results, related genes were classified and sorted in descending order according to gene proportion. The top ten results were selected and visualized using the circlize and ggplot2 packages and presented in bubble chart form.

Results

Chromosomal abnormalities in tissues resulting from SA

Among 1585 cases of SA embryo tissues, CMA detected a total of 863 cases of chromosomal abnormalities, with a chromosomal abnormality detection rate of 54.44% (863/1585). Among the cases of chromosomal abnormalities detected, numerical abnormalities were the most common, totaling 784 cases, with a detection rate of 49.46% (784/1585), accounting for 90.85% (784/863) of the abnormal composition ratio. Among them, there were 538 cases of trisomy, with a detection rate of 33.94% (538/1585), monosomy was detected in 131 cases (8.26%), triploidy in 85 cases (5.36%), and mosaicism in 30 cases (1.89%). A total of 79 cases of structural variants were detected, with a detection rate of 4.98% (79/1585), accounting for 9.15% (79/863) of the abnormal composition ratio. Among them, 59 cases (3.72%) were CNVs, 4 cases (0.25%) were LOH, and 16 cases (1.01%) were Runs of homozygosity(ROH). The types and distribution of chromosomal abnormalities detected in all cases of SA were listed in Table 1.

Table 1.

Frequency and distribution of chromosomal abnormalities in abortion tissues.

Chromosomal abnormality	n	Detection rate(%)	Proportion of abnormal composition(%)
Numerical chromosomal abnormalities	784	49.46	90.85
Trisomy	538	33.94	62.34
Monosomy	131	8.26	15.18
Triploid	85	5.36	9.85
Mosaicism	30	1.89	3.48
Structural chromosomal abnormalities CNVs	79 59	4.98 3.72	9.15 6.84
LOH ROH	4 16	0.25 1.01	0.46 1.85

Characteristics of abnormal chromosome number

The most common numerical chromosomal abnormalities in this study (49.46%, 784/1585), including trisomy, monosomy, triploid and mosaics (Fig. 1). Except for chromosomes 1 and 19, numerical abnormalities were found in all other chromosomes. Among them, the incidence of trisomy was significantly higher than that of other types, with a total of 538 cases, accounting for 62.34% (538/784) of the number of abnormal chromosomes. Among them, trisomy 16 accounted for 19.0%, being the most common chromosomal abnormality site, followed by X monosomy (15.9%) and trisomy21 (10.8%). These data indicate the specific distribution of chromosomal abnormalities in abortion tissues, revealing the high incidence of specific types of chromosomal abnormalities in abortion.

Fig. 1.

Distribution and incidence of numerical chromosomal abnormalities (n = 784).

GO and KEGG annotations of CNVS-related genes

In this study, genes were extracted from the CNV variant sites detected in SA tissues, and GO and KEGG enrichment analysis were conducted to compare their biological functions and pathways. The CNVs variant sites covered a total of 4,668 genes. After screening (P < 0.05), 932 GO annotations and 51 KEGG pathways were obtained.

Through GO enrichment analysis, it was found that the related genes were significantly enriched in three major types of functional annotations: Molecular Function (MF), Biological Process (BP), and Cellular Component (CC). After visual analysis of the enrichment results, the top ten significant items (ranked in descending order of p-value) included hemostasis, mononuclear cell migration, epigenetic regulation of gene expression, among others (Fig. 2A). The KEGG enrichment analysis results show that Genes are mainly enriched in pathways such as Toll-like receptor signaling pathway, Neuroactive ligand-receptor interaction, and Virion-Hepatitis viruses(Fig. 2B).

Fig. 2.

Results of enrichment analysis of CNVS-related genes in spontaneous abortion. A represents GO enrichment analysis, and B represents KEGG enrichment analysis. The X-axis represents the ratio of the number of target genes in the GO annotation or KEGG pathway to the total number of genes (gene ratio), the Y-axis represents the GO annotation entry or KEGG pathway, the bubble color indicates enrichment significance (-log10 p value), and the bubble size indicates the number of genes enriched on the pathway or annotation.

Univariate analysis of chromosomal abnormalities in embryos

In this study, there were statistically significant differences in the rates of chromosomal abnormalities in embryos among patients of different ages, gestational weeks of abortion, and the number of abortions (P < 0.001). The risk of chromosomal aberrations in embryos in the group of elderly pregnant women (≥ 35 years old) is significantly higher than that in younger pregnant women (< 35 years old). From the perspective of the gestational weeks of miscarriage, the proportion of chromosomal abnormalities in embryos in early pregnancy cases < 12 weeks was significantly higher than that in cases ≥ 12 weeks. In addition, the detection rate of chromosomal abnormalities in embryos of patients with recurrent miscarriage (≥ 2 episodes) is significantly higher than that of patients with sporadic miscarriage (1 time). Please refer to Table 2 for details.

Table 2.

Univariate analysis of chromosomal abnormalities [Cases (%)]

Classification	n	Abnormal (n = 863)	Normal (n = 722)	χ2	P
Age(years) <35	1270	653(51.42)	617(48.58)	23.665 144.062 5.683	<0.001 <0.001 0.017
≥ 35 Gestational weeks (weeks) <12 ≥ 12 Number of miscarriages(episodes) 1 ≥ 2	315 1074 511 885 700	210(66.67) 696(64.80) 167(32.68) 463(54.15) 400(57.14)	105(33.33) 378(35.20) 344(67.32) 422(45.85) 300(42.86)

Multivariate analysis of influencing factors of chromosomal abnormalities in embryos of patients with miscarriage

The results of Logistic multivariate regression analysis showed that age ≥ 35 years old (OR = 1.72, 95%CI: 1.31–2.26) was positively correlated with chromosomal abnormalities, while gestational weeks ≥ 12 weeks was negatively correlated with chromosomal abnormalities (OR = 0.27, 95%CI: 0.21–0.34), please refer to Table 3 for details.

Table 3.

Results of multivariate analysis of influencing factors of chromosomal abnormalities in spontaneous abortion.

Variables	Group	β	SE	Z	P	OR(95%CI)
Age	<35
	≥ 35	0.54	0.14	15.19	< 0.001	1.72(1.31–2.26)
Gestational weeks	<12
	≥ 12	– 1.32	0.12	128.05	< 0.001	0.27(0.21–0.34)
Number of miscarriages	1
	≥ 2	– 0.08	0.11	0.52	0.47	0.92(0.75–1.15)

Dose-response relationships between maternal age, gestational age, and chromosomal abnormalities

This study employed the Restricted Cubic Spline(RCS) model to analyze the dose-response relationship between maternal age and chromosomal abnormalities. The number of nodes was selected based on the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). When the number of nodes was set to 3, the model exhibited the lowest AIC value (1998.58) and the smallest BIC value (1998.58), indicating that this model achieved the optimal balance between goodness of fit and complexity. Therefore, three nodes were selected for the final analysis, and the two factors—gestational age and number of previous miscarriages—which were significant in the univariate analysis, were included in the model.

The results of the nonlinear test using the RCS model indicate a significant nonlinear association between maternal age and chromosomal abnormalities (χ² = 6.73, P = 0.003). Overall, the relationship between maternal age and chromosomal abnormalities follows a U-shaped pattern. When maternal age ranges from 20 to 28 years, the risk of chromosomal abnormalities decreases gradually with increasing age. Within this age range, the risk is highest at age 20 (OR = 1.32, 95%CI: 0.77–2.27) and reaches OR=1at age 28; at age 29 (OR = 0.99, 95%CI: 0.87–1.14), the risk of chromosomal abnormalities was slightly lower than OR = 1, indicating a slight protective effect; When the mother’s age ranges from 30 to 45 years, the risk of chromosomal abnormalities rises rapidly with age. Within this age range, the risk is OR = 1 at age 30, and the highest risk is observed at age 45 (OR = 4.01, 95%CI: 2.14–7.50), with age 45 representing the highest risk of chromosomal abnormalities across the entire age range in this study. See Fig. 3A.

Fig. 3.

Curve of the restricted cubic spline model for the relationship between the age and gestational weeks of pregnant women and the risk of chromosomal abnormalities. The blue solid line represents the corrected OR value of the restricted cube spline model, the light blue area represents the 95%CI of the OR value, and the black dashed line represents the OR.

In addition, the RCS model analyzed the nonlinear relationship between gestational age and the risk of fetal chromosomal abnormalities. To achieve the optimal balance between model fit and complexity, four nodes were selected for the final analysis; this configuration yielded the lowest AIC value (1981.76) and the smallest BIC value (2180.39). The model was constructed by adjusting for the two factors—age and number of previous miscarriages—that were found to be statistically significant in univariate analyses.

The results indicate a significant nonlinear association between gestational age and the risk of chromosomal abnormalities (overall association test: P < 0.001; nonlinearity test: P < 0.001). The curve plotted by RCS clearly reveals the dynamic trend of this association: the risk of chromosomal abnormalities follows an inverted “U”-shaped pattern, rising initially and then declining with gestational age. Starting at 8 weeks, the odds ratio (OR) for fetal chromosomal abnormalities shifts from less than 1 to greater than 1, and the risk peaks around 9 weeks of gestation (OR = 1.07, 95%CI: 1.02–1.12). Thereafter, as gestational age increased, the risk of chromosomal abnormalities decreased significantly, reaching a low level and stabilizing around 20 weeks of gestation. See Fig. 3B.

WES results of recurrent miscarriage

A total of 34 cases with negative CMA results and a history of recurrent pregnancy loss (≥ 2 episodes) were subjected to WES analysis, with an overall variant detection rate of 26.47% (9/34). The clinically relevant incremental diagnostic rate reached 5.88% (2/34), which included the MOS and F13A1 variants, both of which have been associated with pregnancy loss in the literature. 9 out of the 34 probands were identified with genetic variants, and a total of 18 variants involving 13 genes were detected in these 9 positive probands. Among these variants, pathogenic (P) and likely pathogenic (LP) variants were defined as clinically significant diagnostic findings, including 1 P variant and 7 LP variants that mapped to 7 genes (FLCN, MOS, DDX41, MYO7A, TTC7A, HSPG2, RYR1); the remaining variants were classified as VUS, involving 6 genes (ERBB3, OSTM1, F13A1, MAGEL2, KDM6B, GAS2L2) .

The coding impacts of the identified variants were diverse, including 11 missense variants, 5 truncating variants (3 frameshift variants, 1 nonsense mutation, 1 large fragment deletion), 1 synonymous variant, and 1 splice-site variant. Detailed information on clinically significant P/LP variants and VUS is summarized in Tables 4 and 5, respectively.

Table 4.

Summary of clinically significant variants.

No.	Gene	Variant	Coding impact	ACMG classification	Disease(Inheritance)	Zygosity
1	MOS	NM_005372.1:c.285 C > A (p.N95K)	Missense	LP	Oocyte/fertilized egg/embryo maturation disorder type 20 (AR)	Het
2	FLCN	NM_144997.7:c.1285dup (p.H429Pfs*27)	Frameshift	P	Birt-Hogg-Dube syndrome (AD); Primary spontaneous pneumothorax (AD)	Het
3	DDX41	NM_016222.4:c.1621 + 2T > G	Splice-site	LP	Polymorphic familial predisposition to myeloproliferative/lymphoproliferative neoplasms (AD)	Het
4	MYO7A	NM_000260.4:c.2558G > A (p.R853H)	Missense	LP	Autosomal dominant deafness type 11(AD); Autosomal recessive deafness type 2(AR); Usher syndrome type 1B(AR)	Het
5	TTC7A	NM_020458.4:c.910_911del (p.L304Vfs6); NM_020458.4:c.1621 C > T (p.Q541)	Frameshift; Nonsense	LP; LP	Gastrointestinal dysfunction-Immunodeficiency Syndrome type 1 (AR)	Het; Het
6	HSPG2	NM_005529.7:c.7159 − 314_7399del	Deletion	LP	Silverman-Handmaker type segmental dysplasia of cartilage (AR); Schwartz-Jampel syndrome type 1 (AR)	Het
7	RYR1	NM_000540.3:c.14204G > A (p.R4735Q); NM_000540.3:c.5209dup (p.L1737Pfs*4); NM_000540.3:c.5209dup (p.L1737Pfs*4)	Missense; Frameshift; Frameshift	VUS; LP; LP	Autosomal recessive congenital myopathy type 1B(AR); Autosomal dominant congenital myopathy type 1 A with susceptibility to malignant hyperthermia(AD); King-Denborough syndrome(AD); Susceptibility to Malignant Hyperthermia Type 1(AD)	Het; Het; Het

Table 5.

Summary of other variants of uncertain significance.

No.	Gene	Variant	Coding impact	ACMG Classification	Disease(Inheritance)	Zygosity
8	ERBB3	NM_001982.4:c.2864G > A (p.R955H)	Missense	VUS	Autosomal recessive visceral neuropathy type 1 (AR)	Het
9	OSTM1	NM_014028.4:c.917 A > G (p.H306R)	Missense	VUS	Autosomal recessive osteosclerosis type 5 (AR)	Het
10	F13A1	NM_000129.4:c.2156 A > G (p.Y719C); NM_000129.4:c.2041T > G (p.F681V); NM_000129.4:c.720G > T (p.L240=)	Missense; Missense; Synonymous	VUS; VUS; VUS	Coagulation factor XIII-A subunit deficiency (AR); Thrombosis tendency (AD)	Het; Het; Het
11	MAGEL2	NM_019066.5:c.2639G > T (p.R880L)	Missense	VUS	Schaaf-Yang syndrome (AD)	Het
12	KDM6B	NM_001348716.2:c.3496G > A (p.V1166M)	Missense	VUS	Stolerman neurodevelopmental syndrome (AD)	Het
13	GAS2L2	NM_139285.4:c.817 C > T (p.R273C); NM_139285.4:c.701G > T (p.R234L)	Missense; Missense	VUS; VUS	Primary ciliary dyskinesia type 41 (AR)	Het; Het

Potential compound heterozygous variants for autosomal recessive conditions were detected in individual probands; parental segregation analysis was not performed due to the unavailability of parental genomic samples. (AD, autosomal dominant; AR, autosomal recessive; Het, heterozygous.)

Discussion

Miscarriage is a common adverse pregnancy outcome. This adverse pregnancy outcome not only seriously affects the physical and mental health of women but may also trigger a series of family, economic, and social problems. Recent epidemiological investigations have shown¹⁵ that the incidence of natural loss in the early stage of pregnancy has been on a continuous upward trend. It is crucial to accurately identify its genetic causes. This study adopted a combined detection strategy integrating CMA and WES, which significantly enhanced the sensitivity and coverage of genetic abnormality detection compared to traditional karyotyping. CMA can efficiently identify chromosomal aneuploidy, polyploidy, single diploidy and genomic CNVs, while WES can deeply detect single nucleotide variations (SNVs) and small indels and other point mutations, effectively supplementing the deficiency of CMA in the detection of pathogenic mutations in single-gene diseases¹⁶. By integrating the abnormal results detected by the two techniques, the chromosomal abnormalities and single-gene inheritance mechanisms that lead to embryonic development disorders can be systematically revealed, providing more comprehensive genetic evidence for in-depth exploration of the molecular causes of SA.

The association between numerical chromosomal abnormalities and SA

The results of this study showed that the detection rate of numerical chromosomal abnormalities was 49.46%, predominantly in the form of trisomy. Among these, trisomy 16 and monosomy X were the most common abnormalities. The primary mechanism leading to aneuploidy is the non-disjunction of homologous chromosomes in maternal meiosis stage I, which is consistent with most studies^17,18. In particular, trisomy 16 is associated with carrying multiple key genes for placental development and abnormal fetal head circumference, which can easily lead to early embryonic growth arrest^19,20. In addition, the detection rate of triploids in this study was also relatively high. The formation of triploids often results from errors in fertilization, and these errors typically occur stochastically, implying a relatively low recurrence risk. However, considering that potential parental germ cell mosaics may affect the risk of recurrence, it is recommended that couples undergo genetic counseling to screen for potential risks.

The association between structural chromosomal abnormalities and SA

Among the 79 cases of structural variants detected in this article, 59 cases were deletions or duplications of CNVs, including 47 pathogenic cases, 3 potentially pathogenic cases, and 9 cases of CNVs of unknown significance. CNVs were significantly enriched in GO terms related to hemostasis, monocyte migration, and epigenetic regulation of gene expression. The hemostatic mechanism is critical for maintaining normal pregnancy. Factor VIII, thrombin and fibrinolysis, which are crucial for placental formation and maintaining pregnancy, affect blood flow and prevent bleeding²¹. If the hemostatic mechanism becomes imbalanced, it may lead to a pre-thrombotic state, resulting in the formation of placental microthrombi, which in turn causes ischemia, hypoxia and abnormal development of the embryo. In addition, this study also suggests that abnormal monocyte migration is closely related to miscarriage. After monocytes differentiate into macrophages during normal pregnancy, they participate in angiogenesis, tissue remodeling and immune regulation through their phagocytic function and cytokine secretion to maintain maternal and fetal immune tolerance²². Appropriate monocyte migration and macrophage differentiation are necessary conditions for a successful pregnancy. However, when monocyte migration is abnormal or macrophage dysfunction occurs, it affects their ability to clear apoptotic cells, leading to the accumulation of apoptotic cells and secondary necrosis, causing excessive local inflammatory response or disruption of immune tolerance, resulting in failed embryo implantation or early miscarriage. This is consistent with the research of True et al.²³. Meanwhile, the disorder of DNA methylation patterns in epigenetic regulatory abnormalities is also an important mechanism leading to SA. Hemberger et al.²⁴ confirmed that hypermethylation of the CpG island of a specific gene can ultimately hinder normal embryo implantation and placental development by disrupting genetic imprinting, interfering with the proliferation and invasion of trophoblast cells, and damaging endometrial receptivity. KEGG enrichment analysis revealed that Toll-like receptor signaling pathways and abnormal activation of neuroactive ligand-receptor interactions could both activate NF-κB signaling by binding to the corresponding receptors, leading to excessive secretion of pro-inflammatory cytokines, disrupting the Th1/Th2 cytokine balance required for maintaining pregnancy, and tilting the immune microenvironment towards a pro-inflammatory direction. Triggering an immune attack on the embryo can disrupt the balance of the maternal and fetal immune microenvironment and promote the occurrence of recurrent miscarriage^25,26. These biological processes are closely related to the pathogenesis of SA and together form a complex regulatory network that affects embryo implantation and pregnancy maintenance.

Among the cases of structural variants, 4 cases of LOH and 16 cases of ROH were detected. Not all LOH/ROH events involved imprinted chromosomes; only 9 cases were located on imprinted chromosomes (7, 11, 14, 15 and 20), with the remaining 11 on non-imprinted autosomes. Parental blood sample testing confirmed UPD for chromosomes 7, 11, and 16. However, complete parental origin verification was not performed, so we could not definitively distinguish whether these UPD events were maternal or paternal. The imprinted region on chromosome 7 is associated with Silver-Russell syndrome (SRS), while the imprinted region on chromosome 11 is linked to Beckwith-Wiedemann syndrome (BWS) and SRS. Chromosome 16 contains imprint-related genes such as NAA60 and CMTM1. UPD(16) has a high incidence and carries multiple autosomal recessive (AR) disease-related genes, suggesting an increased risk of fetal AR genetic disorders.

Notably, definitive diagnosis of imprinting disorders requires confirmation of UPD parental origin, as only UPD from the specific parent associated with the syndrome can cause the corresponding methylation disorder. These findings are therefore reported as suspected UPD rather than definitive diagnoses.

For imprinted chromosomes, such as chromosomes 7, 11, and 15, there are important imprinted genes that may lead to severe clinical phenotypes, and the risk of subsequent pregnancies may be relatively high. It should be noted that WES was not performed on these UPD cases to rule out unmasked autosomal recessive variants; therefore, the potential risk discussed here primarily stems from imprinting disorders. It is recommended to undergo preimplantation genetic testing (PGT) to reduce the risk of serious genetic diseases and the probability of recurrent miscarriage.

Analysis of influencing factors of SA

This study revealed the key risk stratification of genetic causes of SA through univariate and multivariate analyses. Advanced age and early abortion were significantly associated with chromosomal abnormalities, while the impact of the number of abortions on genetic defects in embryos remains controversial. In this study, the rate of chromosomal abnormalities in elderly women was as high as 66.67%.

The results of the RCS model showed that there was a nonlinear dose-response relationship between the age of pregnant women and chromosomal risk in this study, and there was a clear “U”-shaped relationship between them. The curve reached its lowest point (OR = 0.99) at 29.09 years of age and intersected the reference line (OR = 1) at 28.33 years and 30.10 years, indicating a relatively lower risk of chromosomal anomalies within this age interval, whereas the risk increased on either side of this interval. This pattern suggests that distinct biological mechanisms may underlie the risk in different age groups. A slightly increased risk at a younger age may be related to incomplete ovarian function, unstable ovulation quality, or related social behavioral factors²⁷. The sharp increase in the risk of higher age is mainly attributed to mechanisms such as oocyte aging, an increase in meiosis error rates, and the accumulation of genetic damage²⁸. It is worth emphasizing that, unlike the traditional view that generally regards 35 years old as the threshold for a significant increase in risk, the fitting curve of this study clearly shows that the risk of chromosomal abnormalities has shown a definite upward trend since the age of 30 (OR > 1). This finding suggested that the current clinical screening strategy, which focuses on 35 years old as the cut-off point, may have a lag in clinical practice. More attention should be paid to risk assessment and genetic counseling for pregnant women aged 30 to 34.

Further exploration of the nonlinear association between gestational age and the risk of chromosomal abnormalities revealed a significant overall association between the two (P < 0.001) and a clear nonlinear trend (P < 0.001). As shown by the curve, OR for chromosomal abnormalities exhibited a marked upward trend between gestational weeks 5 and 10, peaking between weeks 8 and 10 (OR > 1). This peak coincides precisely with the gestational weeks associated with a high incidence of early miscarriage in clinical practice, suggesting that chromosomal abnormalities are the primary genetic factor driving pregnancy loss during the 8-10-week window when miscarriages occur. Since lethal numerical chromosomal abnormalities frequently cause embryonic arrest during this stage, leading to pregnancy loss, the risk of detecting chromosomal abnormalities in miscarriage cases during this gestational period is significantly elevated. After 10 weeks of gestation, the OR value gradually decreases and remains below 1, with the relative risk of chromosomal abnormalities showing a continuous downward trend. This trend indicates that miscarriages occurring between 8 and 10 weeks have the highest likelihood of being associated with embryonic chromosomal abnormalities. As gestational age progresses, the causes of miscarriage increasingly shift toward non-chromosomal factors, such as the maternal uterine environment, endocrine factors, and immune-related factors. This provides important evidence for clinicians to select more targeted diagnostic strategies when evaluating miscarriages at different gestational ages.

The rate of chromosomal abnormalities in the first trimester was significantly higher than that in the second trimester group, indicating that the tolerance threshold of early embryos to fatal aneuploidy is lower. It is particularly worth noting that univariate analysis showed a significantly increased rate of chromosomal abnormalities in the recurrent miscarriage group (P < 0.001), but after adjusting for age and gestational weeks, the multivariate Logistic regression model showed no independent association between the number of miscarriages and chromosomal abnormalities (OR = 0.81, P = 0.228). This reason is attributed to the higher proportion of elderly pregnant women (≥ 35 years old) and early pregnancy miscarriage in the recurrent miscarriage group, both of which are strong confounding factors of chromosomal abnormalities. Therefore, the main causes of recurrent miscarriage are more likely to be non-genetic factors such as maternal immunity or abnormal blood coagulation²⁹. It is recommended that elderly pregnant women consult a doctor for nutritional supplementation to improve oocyte function. If necessary, preimplantation genetic aneuploidy screening (PGT-A) should be conducted to reduce the risk of aneuploidy pregnancy. For those with normal chromosomes in their first abortion, it is necessary to turn to maternal factor screening.

This study primarily focuses on the relationship between key variables—such as age, gestational age at miscarriage, and number of miscarriages—and chromosomal abnormalities. However, other potential factors, such as parity and data on consanguineous marriages, may also influence the risk of spontaneous miscarriage. Because this study was a retrospective analysis, data for these variables were incomplete or missing and could not be included in the analysis. Future prospective studies should systematically collect more comprehensive information on maternal environmental factors and lifestyle to assess their combined effects on embryonic chromosomal abnormalities.

Genetic variations of recurrent miscarriage and their clinical significance

In this study, the WES variant detection rate in miscarriage tissues was 26.47%, and a clinically relevant incremental diagnostic yield of 5.88% (2/34) was obtained when only variants with definitive literature evidence for miscarriage association were counted as positive diagnoses, which was consistent with the findings of similar studies³⁰. Among the identified variants, pathogenic variations accounted for 5.56%, likely pathogenic variations for 38.89%, and variants of uncertain significance (VUS) for 55.55%. In the present study, the MOS gene c.285 C > A (p.N95K) variant was found to have a well-documented reproductive relevance in published literature: this gene encodes a regulatory factor for oocyte maturation, and loss of function or functional variations of this gene have been reported to induce oocyte and embryo maturation disorders³¹, which is described to be directly associated with early embryonic arrest in relevant studies. The biallelic variants of TTC7A conformed to the autosomal recessive (AR) genetic pattern, with the two variants located in exon 7 and exon 14, respectively; these variants are known to cause immune-gastrointestinal syndrome in existing research, and while the primary phenotypic manifestation is immune dysregulation, severe immune deficiency has been suggested to potentially lead to intrauterine growth restriction or fetal loss in clinical reports. Beyond the fetal phenotypes reported in most relevant studies, several studies have also indicated that pregnant women carrying RYR1 variants may have an increased risk of adverse pregnancy outcomes, including recurrent pregnancy loss³². For variants in genes such as ERBB3 and GAS2L2, which have no direct reported association with miscarriage in current literature, they might potentially affect pregnancy maintenance through indirect biological pathways, though this remains to be further verified. Most of the pathogenic variants listed in the tables lack a clear documented pathological association with recurrent miscarriage: for example, variants in FLCN and DDX41 are primarily reported to be associated with the risk of adult-onset tumors, and their heterozygous status is not described to impact embryonic development in existing research. Additionally, MYO7A (linked to hereditary deafness) and HSPG2 (associated with skeletal dysplasia) variants are in a carrier state under the AR genetic pattern and do not result in phenotypic manifestations according to current studies, thus representing incidental genetic findings detected in miscarriage tissues with no known correlation to the abortion phenotype. Nevertheless, although VUS lack direct functional evidence to confirm their pathogenicity, their potential role in the occurrence of miscarriage cannot be completely excluded, as supported by partial relevant research. For instance, Xiang et al.³³ reported that carriers of heterozygous F13A1 variants were more susceptible to placental dysfunction and pregnancy-associated thrombosis, identifying this gene as a strong risk factor for recurrent miscarriage in Han Chinese women. This well-documented association justifies including the F13A1 variant in the clinically relevant incremental diagnostic yield of WES relative to CMA. Furthermore, Schaaf-Yang syndrome, caused by MAGEL2 gene mutations, has been confirmed in molecular mechanism studies, clinical case reports and animal model experiments to be closely associated with severe embryonic developmental disorders, implying a potential link to early pregnancy loss^34,35.

Notably, all genetic variants identified in this study were detected directly in embryonic miscarriage tissues, and this study only infers their potential association with spontaneous abortion rather than a definitive causal relationship. Causal verification of genetic variants for early miscarriage requires multi-dimensional evidence including parental segregation analysis, in vitro functional experiments and in vivo animal model studies—these verification steps were not performed in the present research, which is a major limitation. The variants classified as potentially associated with miscarriage in this study were screened based on existing literature evidence for relevance to pregnancy maintenance and embryonic development, and their specific regulatory mechanisms in the pathogenesis of spontaneous abortion require further in-depth research.

Strengths and limitations

Strengths

The main strength of this study lies in the combined application of CMA and WES. This strategy establishes a comprehensive genetic testing system from chromosomal abnormalities to single-gene variants and effectively improves the genetic diagnostic yield of spontaneous abortion. Relying on a large sample size of 1585 cases, the study systematically characterized the distribution features of chromosomal abnormalities associated with spontaneous abortion and identified relevant pathogenic pathways through GO/KEGG enrichment analysis. By applying univariate analysis, multivariate regression and RCS models for in-depth analysis of risk factors, it first revealed the nonlinear correlations between maternal age, gestational age and the risk of embryonic chromosomal abnormalities, challenging the traditional clinical threshold for advanced maternal age. In addition, WES was performed on CMA-negative cases of recurrent miscarriage, which enriched the spectrum of monogenic genetic etiologies for recurrent miscarriage.

Limitations

This study has certain limitations. Firstly, as a retrospective study, clinical data collection was incomplete. Moreover, parental genomic samples were not included, making it impossible to conduct familial verification and segregation analysis of genetic variants. Additionally, there is a lack of in vitro and in vivo functional validation of the identified variants, which affects the accuracy of variant pathogenicity assessment. Secondly, limited by testing costs, the sample size for WES was relatively small, leading to certain limitations in the generalizability of the study results. Meanwhile, the combined detection strategy still cannot cover all types of genetic variations such as balanced translocations, small-fragment CNVs and epigenetic abnormalities, which may result in undiagnosed genetic etiologies in some cases. In addition, no long-term follow-up was conducted on the study subjects, so the actual clinical application effect of this combined detection strategy in reducing the recurrence risk of spontaneous abortion cannot be evaluated.

Conclusion

The occurrence of SA is closely related to chromosomal abnormalities in embryos. This study explored the biological processes and signaling pathways associated with the key genes, and provided clues and research directions for a deeper understanding of the molecular mechanisms of SA. For high-risk women of childbearing age, relevant health education should be strengthened, and pre-pregnancy check-ups should be actively promoted to reduce birth defects and avoid adverse pregnancy outcomes in subsequent pregnancies. This multi-omics combined strategy not only has increased the genetic diagnosis rate of abortion tissues by 15% to 20%, but also establishes a complete detection chain from macroscopic chromosomal abnormalities to microscopic gene defects, providing a comprehensive basis for the precise classification and genetic counseling of unexplained recurrent miscarriages in clinical practice.

Author contributions

LW: Research conception and design, data extraction and organization, participation in some data analysis, and writing some manuscripts. PL: Participated in some data analysis, wrote some manuscripts, and created charts. WH and ML: Participated in the review and editing of manuscripts. JY: Assist in literature retrieval, review and background research. SX: Validation, Writing- review and editing, Project administration, Supervision, Resources, Formal Analysis, Funding acquisition. LW and PL contributed equally to this work. All authors reviewed and approved the manuscript.

Funding

This research was supported by Xinjiang Uygur Autonomous Region Science and Technology Innovation Team (Tianshan Innovation Team) Project (2022TSYCTD0016), Prenatal Screening, Diagnosis and System Construction of Congenital Structural Abnormalities Based on Maternal-Fetal Medicine in Xinjiang Region (2023B03018-1) and Application of Copy Number Variation Detection Combined with Whole exome Sequencing in Genetic Etiological Analysis of Spontaneous Abortion (2023D01F22).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

This study was conducted in accordance with the principles of the Declaration of Helsinki. The protocol was approved by the Urumqi Maternal and Child Health Hospital (Reference No.: XJFYLL2023037]). Written informed consent was obtained from all participants prior to their enrollment in the study. For minors or participants with disabilities, consent was acquired from their parents or legal representatives. All data were anonymized to protect participants’ privacy.