The study was approved by the Institutional Ethics Committee of Shaoxing Maternity and Child Healthcare Hospital. The family members had signed informed consent documents. Parental consent was obtained for collecting the prenatal fetal amniotic cell samples at 27+5 weeks of pregnancy. Peripheral blood samples were also collected from the proband’s parents.

Genomic DNA was extracted from the proband and parental blood using DNeasy Blood and Tissue Kits (QIAGEN, Hilden, Germany), according to the manufacturer’s procedures.

For WES, DNA libraries were constructed, which were subjected to whole-exome capture on an Agilent SureSelect Low Input Reagent Kit (Agilent, California, United States). We then performed high-throughput sequencing on the Illumina HiSeq X Ten platform (Illumina, Inc., San Diego, CA, United States) (150-bp paired-end reads) to analyze point variants and small deletions or insertions. Its bioinformatic pipeline included FastQC (v0.11.9) for filtering low-quality reads (Phred <20) and adapters, alignment to the GRCh38/hg38 reference genome via BWA-MEM (v0.7.17); Picard Tools (v2.27.5) for duplicate removal and sorting; and GATK (v4.3.0.0) for base recalibration. GATK HaplotypeCaller was used for SNV/Indel calling (joint genotyping for family samples), followed by filtering (DP ≥ 10, AB ≥ 0.2 for heterozygotes, MAF <0.01 in gnomAD/ExAC, and ≥2 tools predicting pathogenicity). The quality control of sequencing data showed a total of 20,858 genes, 34962037 bp of target area length, 99.87% of target area coverage, 111.277 × of average sequencing depth in the target region, and 99.09% of proportion in the target area with an average depth of >20 ×, suggesting that the results of WES sequencing were reliable.

For CNV-seq, genomic DNA was fragmented using Hieff NGS® Fast-Pace^TM DNA Fragmentation Reagent and prepared for the PCR-free library using the Hieff NGS® Complete Adapter Kit for Illumina®. DNA was fragmented using Hieff NGS® Fast-Pace^TM Reagent (Yeasen), and libraries were built with the Hieff NGS® Complete Adapter Kit (Yeasen), then sequenced on the Illumina NovaSeq 6000 Platform (150-bp paired-end reads). Its pipeline included FastQC filtering (same as WES), alignment to GRCh38/hg38 via BWA-MEM, Picard Tools for duplicate removal; SAMtools (v1.15.1) for 100-bp bin read counting, CNVkit (v0.9.10) for GC/depth normalization and CNV calling (minimum size 10 kb), and ANNOVAR (v20210802) for annotation against DGV/ClinVar. CNV-seq quality control showed an average genome depth of 30 ×, >98% mapping rate, and GC content consistent with the reference genome.

The QC (PMID: 28361673) was used to evaluate the sequencing quality of the original data, and the low-quality and contaminated reads were removed. The variant was evaluated by dbSNP (http://www.ncbi.nlm.nih. gov/projects/SNP/), 1,000 Genomes (http://www.1000genomes. org/), ExAC (http://exac.broadinstitute.org/), gnomAD (http://gnomad.broadinstitute.org/), ESP6500 (http://evs.gs.washington. edu/EVS/), the Human Gene Mutation Database (HGMD; http://portal.biobase-international.com/hgmd/pro/start.php), and Mutation Taster (http://mutationtaster.org). The dbNSFP database was used to predict the pathogenicity of missense mutations and splice mutations. Novel variants were recognized using the HGMD and ClinVar (PMID:29165669; PMID:32596782). Sanger sequencing was used to validate the candidate variants in all family members. All mutation sites were classified according to ACMG genetic variation classification criteria and guidelines.

For this study, we employed E. coli DH5α for all cloning procedures. Using the pcMINI and pcMINI-C vectors as backbones, we constructed both wild-type and mutant fragments containing CEP152 exons and introns. Initial target fragment amplification was performed through nested PCR from normal human genomic DNA using primer pairs 52228-F/54461-R and 52436-F/54180-R. The 1,284 bp pcMINI-CEP152 wild-type fragment was directly amplified with pcMINI-CEP152-KpnI-F/XhoI-R primers, while the mutant version was generated via overlap PCR by separately amplifying left (545 bp) and right (770 bp) halves before combining them at a 1:1 ratio for full-length amplification. Similarly, the 1811 bp pcMINI-C-CEP152 wild-type was obtained by amplifying left (1,200 bp) and right (641 bp) segments, followed by 1:1 mixing, whereas the mutant required three-segment amplification (513 bp, 718 bp, and 641 bp) combined at a 1:1:1 ratio before final amplification. All primer sequences are provided in Supplementary Table S1.

The recombinant vectors were transiently transfected into HeLa and 293T cell lines, and the transfection steps were carried out according to the instructions of the liposome manufacturer. Total RNA was extracted from the cell samples after 48 h, and the extraction method was performed according to the kit instructions. After determining the concentration, cDNA was synthesized by reverse transcription using an equal volume of RNA. pcMINI-wt/mut was amplified by PCR with the primers pcMINI-F/pcMINI-R; pcMINI-C-wt/mut was amplified by PCR with the primers pcMINI-C-F/pcMINI-C-R, and the gene transcripts were detected by agarose gel electrophoresis. Each band was retrieved separately for Sanger sequencing.

The proband’s mother was a 30-year-old Chinese Han woman with regular 30-day menstrual cycles and no significant medical history. Her pregnancy was confirmed after 30 days of amenorrhea with a positive urine test. She received standard prenatal care, including 0.4 mg/day folic acid supplementation during the first trimester, with normal oral glucose tolerance test results and no exposure to teratogens. Initial prenatal screening at 13+1 weeks showed a normal nuchal translucency measurement (2.8 mm) and low-risk non-invasive prenatal testing results. However, subsequent ultrasounds revealed progressively severe fetal growth abnormalities, with head circumference measurements declining to the 0.1st percentile by 24+2 weeks and estimated fetal weight below the 1st percentile (Figures 1A,B). Fetal MRI at 30+6 weeks confirmed significant microcephaly (head circumference 266 mm, equivalent to 28 weeks’ gestation) (Figures 1C,D). Following a multidisciplinary evaluation at 31+5 weeks that confirmed severe fetal growth restriction with microcephaly, the parents elected to terminate the pregnancy after ethical committee approval. The male fetus was delivered stillborn without apparent dysmorphic features. Postmortem genetic testing was performed to determine the etiology of these findings, with the aim of guiding future pregnancies. The parents had no family history of genetic disorders or previous adverse pregnancy outcomes.

We performed whole-exome sequencing (WES) with blood total genomic DNA extracted from the proband and his parents. Furthermore, from WES and mitochondrial genome detection, two conclusions can be derived. First, no pathogenic or suspected pathogenic variants consistent with the patient’s phenotype have been detected; second, the pathogenic evidence is insufficient, but it is consistent with the patient’s phenotype, and two possible variants of pathogenesis are not excluded: one is the VPS13B gene with an variant of c.5692G>A, p. Glu1898lys (nucleotide of coding region 5,692 changed from guanine to adenine, resulting in amino acid 1898 changed from glutamic acid to lysine); this is a missense variant, and the disease correlated is Cohen syndrome. The other is the CEP152 gene at chr15:49044671 with a variant of NM_001194998.1 c.3346–5T>C relevant to two diseases—primary microcephaly type and Seckel syndrome (Figures 2A,B). Both variants were validated by Sanger sequencing to confirm their authenticity (Figure 2C).

Through CNV analysis, it was found that the subjects had a heterozygous deletion of approximately 129.6 kb in the chr15:49030446–49160066 region, which contained the CEP152 and SHC4 genes. The absence of this area was not recorded in the population database (DGV). There is no record in the ClinGen database. No record in the ClinVar database. None in the disease database deciphered the record. The CEP152 gene is associated with primary autosomal recessive microcephaly (MCPH) and Seckel syndrome (SCKS). After verification of the CNV results of the proband and the proband’s father and mother, it is found that the variant type of the proband was hemizygous, and the father harbored a heterozygous variant, while the mother was a carrier of the deletion in the relevant fragment of the CEP152 gene (Figure 2D; Supplementary Figure S1). According to the American College of Medical Genetics and Genomics (ACMG) guidelines, the NM_001194998.1 c.3346–5T>C variant was predicted to be pathogenic. In 2024, the patient achieved a spontaneous pregnancy. Prenatal diagnosis confirmed recurrence of the same genetic syndrome, with the fetus harboring compound heterozygous CEP152 NM_001194998.1 variants: a paternally inherited hemizygous c.3346–5T>C splice-site variant and a maternally inherited heterozygous deletion represented as 46,XN,del (15q11.2). seq [GRCh37/hg19](22748952–23281864)*1, which was approximately 129.6 kb and spanned the entire CEP152 locus. After multidisciplinary consultation at a tertiary genetics center, the family elected to terminate the pregnancy.

To evaluate the pathogenic potential of the CEP152 NM_001194998.1 c.3346–5T>C variant, initially classified as a VUS in ClinVar (accession: VCV003382300.1), we performed minigene splicing assays in vitro to obtain functional evidence complying with the ACMG guidelines, specifically criterion PS3 (Figure 3A). Minigene constructs were designed in two distinct configurations to ensure result reliability: the pcMINI-CEP152-WT/mut construct contained partial intron 20, exon 21, and partial intron 21 of CEP152 inserted into the pcMINI vector (Figures 3B,E). The pcMINI-C-CEP152-wt/mut construction inserted part of intron 20, exon 21, part of intron 21, and exon 22 of CEP152 into pcMINI-C vector (Figure 3D). Two sets of pcMINI-CEP152-wt/mut vectors were transfected into HeLa and 293T cells, respectively. RT-PCR results showed that the wild type produced a single band ‘a’ with a size of 510 bp in HeLa and 293T cells, while the mutant type displayed multiple bands labeled ‘a,’ ‘b,’ and ‘c.’ Single bands in the wild-type and multiple bands in the mutant in different cell lines were subjected to Sanger sequencing. (Figures 3C,F).

Sanger sequencing was further conducted to confirm the sequence of each band (Figures 3D,G). The sequencing results showed that the wild-type band ‘a’ was a normal cut band, and the cut pattern was exon A (192 bp)–exon 21 (121 bp)–exon B (57 bp). The mutant band ‘a’ was a normal cut band, and the cut pattern was exon A (192 bp)–exon 21 (121 bp)–exon B (57 bp). The mutant band ‘b’ was an abnormal splicing band, with 14 bp retention on the right side of intron 20, and the splicing pattern was exon A (192 bp)–▽intron 20 (14 bp)–exon 21 (121 bp)–exon B (57 bp). The mutant band ‘c’ was an abnormal splicing band, exon 21 was skipped, and the splicing pattern was exon A (192 bp)–Exon B (57 bp).

The consistent aberrant splicing results across the two cell lines and two vector constructs clearly demonstrate that the CEP152 NM_001194998.1 c.3346–5T>C variant directly impairs normal mRNA splicing. Importantly, this functional evidence (ACMG PS3) is not merely a “validation” of preliminary bioinformatic predictions, which only suggested possible pathogenicity, but rather serves as a core independent criterion. When combined with clinical phenotype co-segregation, it fundamentally supports the upgrade of the c.3346–5T>C variant from VUS to pathogenic (P) in line with the ACMG guidelines.