In this study, we performed WES of blood DNA samples from 20 patients and identified 8 variants in 6 genes from 7 patients associated with 46,XX-CGD, including three homozygous variants in MCM9, POF1B, and PSMC3IP; compound heterozygous variants in TWNK; and three heterozygous variants in TP63 and INSRR. The five variants c.1814delT and c.-722G > T in TWNK, c.1925T > G in TP63, c.2217–2 A > C in INSRR, and c.77 A > C in PSMC3IP were first reported in 46,XX-CGD, which not only expanded the spectrum of 46,XX-CGD but also enriched the genetic architecture of the pathogenesis. Our study further supports the notion that genetic variants of several genes are important in the pathogenesis of 46,XX-CGD and may be the main reason for sporadic cases of unknown etiology.

The MCM9 protein is required for DNA repair via homologous recombination [10]. Previous studies have shown that defects in Mcm8 or Mcm9 lead to defects in homologous recombination and gametogenesis and that Mcm9(-/-) female mice are infertile and completely devoid of follicles in their ovaries [11]. In recent years, it has also been reported that homozygous variants of MCM9 can lead to primary amenorrhea and ovarian failure in women, which are accompanied by abnormal or normal height [12, 13]. P1 presented with primary amenorrhea at the age of 13. Her stature was underdeveloped at 145 cm, placing her height in the third percentile according to the height data for children and adolescents in 2023. Using WES, a previously reported splice variant (c.1151-1G > A) of the MCM9 gene was identified and assessed as LP (PVS1 + PM2_Supporting) by ACMG. The MCM9 c.1151-1G > A variant changes an intron 8 splice acceptor site conserved at the nucleotide level across species and in silico analysis predicted a new acceptor splicing site in the downstream 1 bp, which caused a frameshift mutation and premature termination (Fig. 1A). This variant is not present in the gnomAD database. Therefore, MCM9 is an attractive candidate pathogenic gene for ovarian dysgenesis.

The variants in TWNK are associated with Perrault syndrome type 5 (OMIM:616138), a rare autosomal recessive disorder characterized by sensorineural hearing loss and ovarian dysgenesis in women [14, 15]. Fewer than 30 patients with Perrault syndrome type 5 have been reported worldwide. Female patients present with severely impaired endocrine function, which is characterized by primary amenorrhea, a lack of secondary sexual characteristics and hypogonadism, and streaked ovaries [16]. In some cases, patients present with a variety of neurological symptoms, including sensorineural hearing loss, ataxia, nystagmus, hyporeflexia, and sensory axonal neuropathy with distal sensory disturbances [17]. Two novel variants of uncertain significance in the TWNK gene were identified in P2, who presented with ovarian dysgenesis. The patient has been followed up until now at the age of 18 years and is free of neurological or hearing impairment. The only symptom is ovarian dysgenesis, which requires continued follow-up and indicates a possible genotype-phenotype correlation. The nonsense variant c.1814delT (p.Val605GlyfsTer10) in the helicase structural domain, located in the last exon of TWNK, was confirmed to be of maternal origin via Sanger sequencing. The variant had no frequency in the gnomAD database population and was predicted to be deleterious in silico, which was assessed as LP by ACMG. The c.-722G > T variant located in the 5’UTR was of maternal origin. The mutation frequency in the gnomAD database was 0.021%, and the function prediction tools cannot predict the impact of the variant, which was evaluated VUS, and its pathogenicity needs to be further investigated via functional tests.

TP63 is specifically expressed in oocytes of primordial follicles and plays a decisive role in oocyte survival after DNA damage [18]. Variants in TP63 that affect the C-terminal transactivation inhibitory domain (TID) have been reported in different populations of patients with syndromic or isolated 46,XX-CGD [19, 20]. Chengzi Huang et al. identified six TP63 C-terminal TID-related variants in patients who presented with primary or secondary amenorrhea, accounting for 0.78% (8 of 1030) of isolated primary hypogonadism cases in a large Chinese population, which was confirmed by in vitro functional studies and animal experiments [21]. In this study, two adjacent variants were detected in the C-terminal TID domain of the TP63 gene in P3 and P4 with 46,XX-CGD, including a known pathogenic variant (c.1928G > A, p.Arg643Gln) and a novel variant (c.1925T > G, p.Ty642Asp). The sequence of the p.Tyr642Asp variant is highly conserved across various species, and its deleteriousness is predicted in silico. Protein structure analysis revealed significant changes in the spatial structure and surrounding hydrogen bonds, indicating that the p.Tyr642Asp variant is a promising variant of 46,XX-CGD, but further verification via functional tests is needed.

The INSRR gene encodes the insulin receptor (INSR)-related receptor. Pitetti J-L et al. described the critical role of the insulin receptor (INSR) and the IGF type I receptor (IGF1R) in adrenal development and primary sex determination in a mouse model study. The INSR/IGF1R variant gonads remain undifferentiated, regardless of genetic sex [22]. P5 carried a heterozygous c.2217–2 A > C (splicing) variant of the INSRR gene. Splice AI predicted that it might affect splicing (PVS1). The variant resulted in an 8 bp deletion downstream, which created a novel receptor splice site that resulted in a frameshift mutation and premature termination, as shown in Fig. 1B. The variant had no frequency in the gnomAD database population and was evaluated as a VUS. The variant was inherited from the father without obvious abnormalities. INSRR gene function may differ by sex or exhibit incomplete penetrance, which needs to be confirmed by further functional studies. Its potential involvement in 46,XX-CGD provides a new avenue for research.

The POF1B gene is located at Xq21.1 and encodes an actin-binding protein. This gene was first identified in 1996 in a 17-year-old primary ovarian insufficiency (POI) patient with secondary amenorrhea [23]and subsequently reported to be associated with ovarian dysgenesis, with in vitro functional experiments suggesting that POF1B may play a role in germ cell division [24]. The variant C.932 A > C of POF1B has been reported in a 21-year-old patient who presented with POI [25]. Bioinformatics analyses predicted that the variant was deleterious. The amino acid sequences are highly conserved among multiple species. The difference in our study is that P6 presented with ovarian dysgenesis characterized by primary amenorrhea. These findings reinforce the importance of POF1B in ovarian development.

PSMC3IP encodes a nuclear protein-binding α- and β-estrogen receptor as well as receptors for glucocorticoids, thyroid hormones, androgens, and progesterone. It acts as a coactivator of hormone-dependent transcriptional activation [26]. The biallelic variants in this gene have been described in four independent families with ovarian dysgenesis families whose members present with primary amenorrhea and streaked gonads [27,28,29,30]. Sirchia subsequently reported that biallelic variants in PSMC3IP are associated with secondary amenorrhea, expanding the phenotypic spectrum [31]. A novel missense variant (c.77 A > C, p.Lys26Gln) in PSMC3IP was detected in P7 presented with ovarian dysgenesis in the study. Bioinformatics analysis predicted that the variant would be deleterious. The affected amino acid residue, Lys26, is highly conserved across eight species. It is proposed to be a candidate variant for 46,XX-CGD.

Large-scale exome sequencing data have been used to explore the pathogenic genes of POI, expanding the variant spectrum and genetic architecture. POI can be syndromic, manifesting as highly variable somatic abnormalities in addition to reproductive phenotypes or nonsyndromic phenotypes, and the inheritance pattern can be monogenic, digenic, or multigenic [32]. This study is the first to focus on patients with 46,XX-CGD and identify several novel variants. Consistent with the genetic pattern of POI, the MCM9 and TWNK genes are associated with syndromic 46,XX-CGD. However, in our study, the patient with the TWNK variant has not developed neurological abnormalities to date. This presentation warrants further follow-up and could be explained by specific variant sites and different types of variants. The inheritance patterns presented in the study were only monogenic, which may be explained by the small sample size, and the continuous collection of new cases is needed. In addition, the deleteriousness of the variants was only predicted in silico, and functional studies in vivo and in vitro have not been conducted, necessitating further in-depth research.