Monogenic Findings in Early Pregnancy Loss: Whole-Exome Sequencing Study of Euploid Products of Conception

WES was performed on 66 euploid POCs derived from 60 unrelated couples. In six families, two losses were analyzed. Gestational age ranged from 6 to 12 weeks (mean 8.23 ± 1.25), and maternal age ranged from 20 to 42 years (mean 30.9 ± 5.18). Couples reported 2–13 prior losses (mean 3.56 ± 2.14) and 0–2 live births (0.28 ± 0.57). Of the POCs, 34 were male and 32 female (M:F = 1.09:1); 32 originated from Macedonian and 34 from Albanian families. Full demographic and clinical details are provided in Supplementary Table S1.

Genomic DNA was extracted from two to three chorionic villus fragments or fetal tissue of approximately 2 mm² using the MagCore Super automated nucleic-acid extractor (RBC Bioscience). An in-house QF-PCR assay, targeting short tandem repeats on chromosomes 13, 18, 21, X, and Y, was applied to exclude maternal-cell contamination and detect triploidy [12, 13, 14]. MLPA (MRC-Holland) subtelomeric probe mixes were used to screen for numerical or large structural chromosomal anomalies [2, 3]. Samples negative by both assays proceeded to WES.

Libraries were prepared with the Twist Human Core + RefSeq + Mitochondrial panel (Twist Bioscience) and sequenced on an Illumina NovaSeq 6000 with 2 × 100 bp paired-end reads. Reads were aligned to the GRCh38 reference genome using BWA-MEM v0.7.15 [15]. The mean on-target depth exceeded 100x, with at least 98% of bases covered at >=10x. Single-nucleotide variants (SNVs) and small insertions/deletions (indels) were called using GATK v4.3 HaplotypeCaller with 25 bp exon padding and annotated with Ensembl VEP v106 [16,17]. CNV analysis was performed on exome depth-of-coverage data using the CeGaT pipeline; only high-confidence, clinically relevant CNVs were considered reportable, and clinically relevant CNVs were validated by array comparative genomic hybridization (4x180k, Oxford Gene Technologies/Agilent Technologies) and/or MLPA.

Variants were classified according to American College of Medical Genetics and Genomics ACMG/AMP guidelines as pathogenic, likely pathogenic, variant of uncertain significance (VUS), likely benign, or benign [18]. Analysis was performed exome-wide rather than being restricted to a predefined gene list, with prioritization of rare coding and splice-site variants in genes. Variant review incorporated population frequency, predicted functional consequence, known gene-disease validity, expected inheritance model, segregation data when available. Population-frequency assessment relied on gnomAD v2.1.1 and v4.0 and an internal dataset of 1513 WES samples. Variants above the allele-frequency thresholds used in our diagnostic pipeline were generally deprioritized unless previously established as pathogenic.

Clinical significance was assessed using Franklin (Genoox) and VarSome, followed by manual curation. All reported P/LP variants and selected VUS discussed as possible contributors were confirmed by Sanger sequencing. Primers designed with Primer3 and PCR conditions are listed in Supplementary Tables S2 and S3. Parental studies were performed in all cases, using Sanger sequencing, MLPA, or arrayCGH as appropriate. QF-PCR was the principal method used to exclude maternal-cell contamination and triploidy.

Whole-exome sequencing (WES) was performed on 66 euploid products of conception (POCs) from early pregnancy losses (EPLs) occurring before 12 gestational weeks. A molecular diagnosis, defined as pathogenic or likely pathogenic (P/LP) variant(s) consistent with the expected mode of inheritance, was established in 13/66 POCs (19.7%), including one large de novo 21q22.12-q22.3 duplication encompassing DYRK1A and RUNX1.

An additional 9/66 POCs (13.6%) harbored findings considered possible monogenic contributors rather than definitive diagnoses. The remaining 44/66 cases (66.7%) had no reportable variant. Figure 1 summarizes the overall classification of cases represented among reportable findings.

Figure 1.

Overview of molecular findings in 66 euploid early pregnancy losses. Diagnostic summary showing cases with a genetic diagnosis, possible monogenic contribution (VUS + P/VUS), and no reportable variant, with autosomal recessive (AR) and autosomal dominant (AD) findings indicated where applicable

To reduce overinterpretation, reportable findings were further stratified into three categories: (1) genes plausibly associated with prenatal or early embryonic lethality, (2) genes causing severe congenital disorders not typically considered embryonically lethal, and (3) genes linked to later-onset or susceptibility phenotypes. Detailed molecular data and this interpretive framework are summarized in Tables 1 and 2. Figure 2 provides an overview of the reportable findings according to interpretive category and the major developmental or organ systems represented.

Figure 2.

Overview of the identified genetic findings grouped by major developmental or organ-system association. (A) Genetic diagnoses, including definitive molecular diagnoses, and the pathogenic copy-number finding 21q22.12-q22.3 duplication. (B) Possible monogenic contributors. Genes are mapped to their principal affected systems, including neurologic, cardiac, blood/coagulation, kidney anomalies, skeletal, metabolic, ciliopathies, and multi-system involvement. CPLANE1 and RPGRIP1L are shown in both ciliopathy and multi-system categories because of their broader phenotypic effects.

Among genes plausibly associated with prenatal or early embryonic lethality, molecular diagnoses were identified in cases with biallelic pathogenic variants in CPLANE1 and DHCR7, while additional potential monogenic contributors included biallelic combinations in GBA1, PKHD1, and RPGRIP1L. Parental testing was available in all cases and confirmed biparental inheritance in recessive findings. Pedigree analysis of families with definitive molecular diagnoses and, in selected cases, recurrent affected pregnancy losses (Figure 3).

Figure 3.

Pedigrees of families with definitive molecular diagnoses in the euploid early pregnancy loss cohort, grouped according to interpretive category. Previous losses, livebirths, and affected fetuses are shown for each family, and recurrent affected fetuses are integrated into the same pedigree when present.

Among genes causing severe congenital disorders not typically considered embryonically lethal, molecular diagnoses consisted of heterozygous variants in SLC6A1, NF1, DSG2, DVL1, and one RBM8A-associated case with the characteristic combination of the low-penetrance regulatory variant and the recurrent 1q21.1-q21.2 deletion. The potential contributory subset included heterozygous VUS in MYH3, PAH, PRDM6, SCN5A, TBX18, and TSC1. All heterozygous variants were inherited from one parent, with exception of SLC6A1: c.740C>A which occurred de novo; parental phenotypic information was not available.

Among genes linked to later-onset or susceptibility phenotypes, reportable findings included homozygous F5: c.1601G>A and heterozygous VWF: c.3797C>T variants.

One de novo 21q22.12-q22.3 duplication represented a distinct pathogenic copy-number diagnosis.

This group included findings in F5 and VWF, both of which were classified as P/LP at the molecular level but are more difficult to interpret as direct causes of first-trimester loss. These variants may reflect susceptibility, maternal-fetal interaction, placental factors, or other context-dependent mechanisms rather than classic monogenic embryonic lethality.

The identification of fetal homozygosity for Factor V Leiden in a case with maternal homozygosity is noteworthy, but it should be regarded as hypothesis-generating rather than directly explanatory for EPL. Likewise, the mechanistic link between a VWF finding and early embryonic demise remains uncertain. Accordingly, these results are best interpreted as clinically relevant findings of uncertain etiologic weight in relation to EPL [34].

More broadly, the diagnostic variants in our cohort spanned multiple disease categories, mirroring patterns reported in previous studies [35, 36, 37, 38, 39, 40, 41, 42]. The repeated recovery of similar functional categories across independent cohorts supports the view that disruption of core developmental pathways contributes to a subset of euploid EPL. At the same time, the presence of susceptibility or later-presenting diagnoses in early losses highlights that some findings may reflect allele-specific severity, variable expressivity, reduced penetrance, or coincident maternal and/or environmental factors rather than direct monogenic causation.

Recurrent biallelic findings in CPLANE1 and DHCR7 in unrelated families represent some of the strongest candidates for involvement in euploid EPL because they affect genes linked to severe developmental disorders and were observed more than once in the cohort as well as in other previously published studies [7, 25]. In contrast, the RBM8A case highlights the importance of integrated SNV/CNV interpretation, because the diagnosis depends on the combination of the regulatory c.-21G>A allele with the 1q21.1 deletion in trans. Finally, de novo findings such as SLC6A1 c.740C>A and the 21q22.12-q22.3 duplication clearly establish molecular diagnoses in the conceptus, but their direct explanatory value for first-trimester loss remains more uncertain and remains to be clarified.