Informed consents were acquired from the pregnant women and their relatives prior to their inclusion in this study. This study received approval from the Prenatal Diagnosis Ethics Committee at the Maternal and Child Health Hospital of Jiaxing, China (2025-Y-016) and followed the Helsinki Declaration of 1964 and its subsequent amendments. Participants were informed that the sequencing results from the exome sequencing cohort would be available for scientific reports after the study, and only results related to fetal abnormalities would be shared with parents.

We conducted a retrospective analysis of patients diagnosed with fetal skeletal dysplasia via prenatal ultrasound at the Fetal Medical Center of the Maternal and Child Health Hospital of Jiaxing, China. The patients were enrolled from January 2020 to December 2022. Chromosomal karyotyping and microarray analyses showed no abnormalities in the amniotic fluid or cord blood samples from the 28 fetuses. Among the cases, four involved one parent with skeletal dysplasia. One mother had a history of two pregnancies affected by skeletal dysplasia, while the remaining cases had no family history of hereditary diseases. Fig. 1 outlines the testing workflow.

Fig. 1.

The procedure of genetic testing for fetuses with suspected skeletal development abnormalities. WES, whole-exome sequencing; WGS, whole genome sequencing; CMV, cytomegalovirus.

This flowchart outlines the stepwise genetic testing pipeline for cases identified with fetal skeletal development abnormalities via ultrasound. The workflow proceeds as follows:

Initial Screening: Detection begins with ultrasound diagnosis of fetal skeletal development abnormality, which identifies cases requiring further prenatal genetic assessment.

Invasive Prenatal Sampling: For eligible cases, interventional prenatal diagnosis is performed, involving the collection of amniotic fluid or umbilical cord blood samples.

First-tier Genetic Analyses:

Two parallel initial testing pathways are employed based on clinical considerations:

Pathway 1: Karyotype analysis + cytomegalovirus (CMV) testing: Conventional cytogenetic analysis (karyotyping) assesses gross chromosomal abnormalities, while CMV testing rules out viral-associated fetal anomalies. If results from this pathway are negative (no pathogenic findings after 7 weeks of waiting for final karyotype and CMV results), the pipeline transitions to whole exome sequencing/whole genome sequencing (WES/WGS) for broader genetic interrogation. If pathogenic chromosomal changes or CMV-related pathologies (pathopoeias) are detected, this pathway concludes.

Pathway 2: Karyotype analysis + CMV + WES/WGS: This integrated pathway simultaneously performs karyotyping, CMV testing, and high-resolution exome/genome sequencing upfront, aiming to accelerate diagnosis for complex or high-risk cases.

Turnaround Time: The standard report time for results from the testing pipeline is 4 weeks for pathways incorporating concurrent WES/WGS, whereas the first pathway may extend to 7 weeks before potentially transitioning to additional sequencing.

Key transition: The move to WES/WGS occurs when initial karyotype and CMV testing (in Pathway 1) yields no definitive answers, enabling exploration of single-gene and submicroscopic genetic variants associated with skeletal dysplasias and other fetal anomalies.

All two-dimensional (2D) images were captured using the E8 ultrasound system (No.: YZB/AUS 1561-2015; GE Healthcare Austria GmbH & Co OG, Zipf, Upper Austria, Austria) with a transabdominal probe (No.: RAB2-5; GE Healthcare Austria GmbH & Co OG, Zipf, Upper Austria, Austria). All images were saved and analyzed offline with version 4.0 of the four-dimensional (4D) imaging software.

After gestational weeks had been verified by ultrasonography in the first trimester, it was found in the second and third trimesters that fetal femur lengths (FL) were less than the 5th quantile (P5) of the normal FL value in the corresponding gestational weeks. Regarding long bone morphologies, normal limb long bones exhibited straight, strong echoes with posterior sound shadows. Abnormalities were considered in cases of uniform thickening or local thickening of the long bone, bending deformation of the bone, and angular fractures. Finally, the ossification degree was determined via bone resonance and posterior bone shadows.

Routine fetal amniotic fluid or cord blood puncture was conducted. A total of 23 amniotic fluid samples (10 mL each), 5 cord blood samples (3 mL each), and 3 mL of venous blood samples from the couples were collected. The STR genotype detection revealed that all fetuses had biological parent-child relationships, and eliminated the contamination of fetal DNA by maternal genomic DNA.

Amniotic fluid, cord blood, and peripheral venous blood samples from pregnant women and their spouses were collected and subjected to G-band chromosomal karyotyping analysis. The chromosomes were prepared by conventional cell cultures, phytohemagglutinin stimulation and colchicine treatment. Cytogenetic analyses were conducted on the slides through G-band. The 25 mitotic phases were microscopically counted. Five of them were analyzed and described according to the International System for Human Cytogenetic Nomenclature (ISCN 2016), which standardizes the notation for chromosomal aberrations, copy number variants (CNVs), and genomic coordinates (Genome Reference Consortium Human Build 37 [GRCh37]/hg19 assembly).

Genomic DNA were extracted from amniotic fluid or cord blood samples and analyzed for single nucleotide polymorphisms using the Affymetrix CytoScan 750K (Affymetrix Inc., 901859, Santa Clara, CA, USA) array gene chip. Including 550,000 nonpolymorphic (NP) probes and 200,000 SNP probes (0.1Mb resolution) platform.

Extraction of DNA from cord blood and peripheral blood samples was performed using the DNA extraction kit (Tiangen Biochemical Technology Co., Beijing, China), as instructed by the manufacturer. Then, the purity and quantity of the DNA were assayed. The remaining samples were stored at –20 °C. Purification and sequencing of PCR amplification products were performed by Huada Gene Co., Ltd. (Shenzhen, Guangdong, China). Candidate mutation sites were determined after DNA extraction and quality testing, genome library construction, target region capture and sequencing, as well as bioinformatics analyses. Data analyses: sequenced fragments were aligned with the human genome version 19 (hg19) human reference genome on (University of California, Santa Cruz) UCSC through Burrows-Wheeler aligner (BWA) to remove duplicates. Then, Genome Analysis Toolkit (GATK) was used for base mass correction, single nucleotide variant (SNV), insertion-deletion (INDEL), and genotypic detection. Copy number variant (CNV) detection at the exon level was conducted by Exome Depth.

In this assay, about 500 ng of genomic DNA was broken by ultrasound into 250–300 bp fragments, mixed with the end repair buffer (T4 Polynucleotide Kinase, 10,000 U/mL, lot: 16041820; ENZYMATICS, Beverly, MA, US + Klenow fragment, 5000 U/mL, lot: 7121620; ENZYMATICS, Beverly, MA, US + TaKaRa Taq™, lot: AJ12290A; TAKARA Bio Inc., Kusatsu, Shiga, Japan) and repair enzyme mixture (10X T4 PNK Reaction Buffer, lot: EK0032; ENZYMATICS, Beverly, MA, US) (20 °C for 30 min; 65 °C for 30 min; 10 °C for 1 min). Then, the ligase, buffer and adaptor were added to the end repair system, and the mixture was kept at 22 °C for 30 min. After purification by magnetic beads, the sample was added to the PCR amplification reaction system (98 °C for 2 min; 98 °C for 20 s; 65 °C for 30 s; 72 °C for 40 s, a total of 8–10 cycles; 72 °C for 4 min). After magnetic bead purification, the quality of the library was tested. Briefly, the library had a, concentration $>$15 ng/µL in a volume of 25 µL, and the effective molecular concentration was $>$10 nM. The DNA fragment peak was located at 430 bp, ranging from 300 to 600 bp. Sequencing was conducted post-library validation.

Based on high-throughput WES, Sanger sequencing was performed to verify the mutation sites of the fetus and parents. The corresponding PCR primers were designed for potential pathogenic sites, followed by PCR amplification and product purification. Sequencing was conducted on ABI3130 BigDye Terminator v3.1 Cycle Sequencing Kit, lot: 4337455 (Applied Biosystems, Foster, CA, US). Then, sequencing results were aligned with the standard sequence in GenBank using the Lush aligner share software version 1.13.1 Huada Gene Co., Ltd. (Shenzhen, Guangdong, China) to determine the mutation sites. The pathogenicity of mutation sites was determined based on genetic mutation classification standards formulated by the American college of medical genetics and Genomics (ACMG) in 2015. Mutation pathogenicity classification was based on ACMG and the Association for Molecular Pathology (AMP) Guidelines for Interpretation of Sequence Variations. This guideline was interpreted based on the ClinGen Sequence Mutation Interpretation Working Group and Association for Clinical Genomic Science (ACGS).

This study mainly adopts descriptive analysis.

According to the next-generation sequencing (NGS) results and genetic patterns, 28 patients were divided into 4 groups. All grouping information is shown in Table 1. The grouping criteria are as follows: (1) Parental dominant (PAD) Group (n = 5): Criteria: Cases with heterozygous pathogenic variants in autosomal dominant genes (e.g., FGFR3, ACAN) inherited from an affected parent, confirmed by parental Sanger sequencing (to rule out de novo mutations). Clinical correlation: Positive family history of skeletal dysplasia (e.g., achondroplasia) and typical radiographic features (e.g., rhizomelic limb shortening). (2) De novo Mutation Dominant (DNM) Group (n = 10): Criteria: Cases with novel heterozygous variants absent in both parents, validated by trio-based Whole Exome Sequencing (WES) (variant allele frequency $\ge$20% in the proband, 1% in parents). Clinical correlation: Sporadic occurrence (no family history) and de novo mutation-specific phenotypes (e.g., thanatophoric dysplasia due to FGFR3 p.Lys650Glu). (3) Autosomal recessive (AR) Group (n = 5): Criteria: Cases with bi-allelic pathogenic variants in autosomal recessive genes (e.g., COMP, COL2A1), requiring both parents to be asymptomatic carriers (confirmed by parental heterozygosity via NGS). Clinical correlation: Consanguineous parental relationships (60% of cases) and compound heterozygous/homozygous genotypes associated with severe early-onset phenotypes (e.g., spondyloepiphyseal dysplasia). (4) Unresolved (UN) group (n = 8): Criteria: Cases with negative NGS results after multi-level analysis (targeted gene panels, WES, copy number variation analysis), excluding technical limitations (e.g., low sequencing depth, sample contamination). Clinical correlation: Phenotypically suspected skeletal dysplasia (e.g., long bone shortening on ultrasound) but no molecular confirmation, consistent with the “genetically unresolved” category (10%–15% prevalence in prior studies). Fig. 2 shows examples of common pathogenic outcomes detected in fetal bone dysplasia, as well as the appearance and ultrasound phenotype of one case of fetal bone dysplasia after delivery.

Fig. 2.

Genetic mutations and induced labor with fetal bone abnormalities. (A) shows a homozygous mutation of RAPSN (inherited from both parents). (B) shows a homozygous mutation of P3H1 (inherited from both parents). (C) shows a heterozygous mutation of FGFR3 (newly developed mutation). (D) shows the appearance and ultrasound examination of induced labor with abnormal fetal bone development. RAPSN, receptor-associated protein of the synapse; P3H1, prolyl 3-hydroxylase; FGFR3, fibroblast growth factor receptor 3; AR, autosomal recessive; AD, autosomal dominant inheritance.

Gestational weeks: Gestational weeks for the PAD group ranged from 17 to 30 weeks (median = 20 weeks); 19 to 28 weeks (median of 25.5 weeks) for the DNM group; 16 to 33 weeks (median of 20 weeks) for the AR group and from 17 to 35 weeks (median of 28 weeks (the largest among the four groups)) for the UN group. Twenty-one cases were detected by ultrasound in the second trimester (pathogenic mutations were not found in 4 cases), while 7 cases were detected by ultrasound in the third trimester (pathogenic mutations were not found in 4 cases). The gestational weeks were determined by prenatal ultrasound. Differences in mean number of gestational weeks (24.4 $\pm$ 4.8 weeks for the PAD group, 25.2 $\pm$ 4.4 weeks for the DNM group, 24 $\pm$ 6.4 weeks for the AR group, and 25.4 $\pm$ 6.6 weeks for the UN group) were not significant (analysis of variance, p = 0.8497). The maximum gestational weeks were 30 weeks (PAD group), 28 weeks (DNM group), 33 weeks (AR group), and 35 weeks (UN group), respectively.

Prenatal ultrasound performance: A total of 24 cases had shortness of long bones, 1 case had bent bones, 1 case had equinovarus, 2 cases had bone development deficiencies, 4 cases had retarded intrauterine growths and 17 cases had other abnormalities. Among the 8 cases without definite genetic etiologies, there was 1 case of multiple skeletal deformities at 18 weeks of pregnancy, 3 cases of femoral shortening with fetal growth restriction (FGR) (30–34 weeks of pregnancy), 1 case of gallbladder defects, 1 case of clubfoot (24 weeks of pregnancy), and 1 case who developed fetal edema and shortness of long bones around 18 weeks of two pregnancies. Twenty one cases developed fetal skeletal abnormalities in the second trimester (pathogenic mutations were not detected in 4 cases), while 7 cases had fetal skeletal abnormalities in the third trimester (pathogenic mutations were not detected in 4 cases).

The PAD group: One case had shortness of long bones with polyhydramnios; 1 case had shortness of long bones with hydrocephalus ($>$99%); 3 cases had simple shortness of long bones (1 case had long bone curvature, and 1 case had a strong bone echo).

The DNM group: Eight cases had shortness of long bones combined with increased head circumference ($>$99%); 2 cases had a narrow chest; 1 case had simple and severe shortness of long bones.

The AR group: One case had skeletal abnormalities with FGR; 1 case had diaphragmatic eventration and pleural effusion; 3 cases had simple shortness of long bones.

The UN group: Three cases had shortness of long bones with FGR; 1 case had edema; 1 case had cardiac malformations; 1 case had simple strephopodia; 2 cases had shorter femurs (5%), of which 1 case had a partial long bone defect.