POF is a puzzling disorder as its aetiology is very heterogeneous and most cases are still idiopathic. However, the incidence of familial cases among POF patients is estimated to exceed 30% [29, 30], suggesting a genetic basis for some cases of idiopathic POF. In particular, the association between POF and X chromosome abnormalities has been extensively described . Chromosomal anomalies occur in 8.8-33% of women with POF  and 10-15% of cases are X chromosome abnormalities, such as numerical and structural aberrations (deletions, inversions and X;autosome translocations) [1, 7, 8]. This study characterizes the chromosomal abnormalities identified in four patients affected by POF and included in a cohort of 269 patients [8, 18, 19]. Specifically, we identified 4 chromosomal abnormalities involving the X chromosome with 4 different breakpoint localizations: two de novo balanced translocations 46,X,t(X;2)(q21.33;q14.3) in case 2 and 46,X,t(X;13)(q13.3;q31) in case 4 and two maternal inherited unbalanced translocations, 46,X,der(X)t(X;19)(p21.1;q13.42) in case 1 and 46,X,der(X)t(X;Y)(q26.2;q11.223) in case 3. This genetic heterogeneity highlights both the importance of the X chromosome in POF aetiology and the complexity of the POF disorder. However, it is not possible to exclude the involvement of autosomal genes in the onset of ovarian insufficiency as some genes located on autosomes have been associated with the POF phenotype, for example INHA, FSHR and FOXL2 [32–34]. Theoretically, fertility impairment in patients with chromosomal abnormalities might be explained in various ways. First of all, chromosomal anomalies might disrupt a gene that is important for gonadal function ; alternatively the breakpoint may fall in gene-poor regions and, in this case, the translocation might induce a long-range position effect in the expression of genes flanking the breakpoint, suggesting an epigenetic control [16, 35]. Moreover, structural rearrangements involving the X chromosome may disrupt the normal pairing at meiosis, leading to meiotic arrest . However, the pattern of chromosomal aberrations is still not clearly comprehensible: chromosomal alterations with breakpoints spanning on chromosome X have also been identified in females with normal ovarian function [16, 36].
In case 1 the bioinformatic analysis of genes in the breakpoint region on chromosome X did not identify any candidate gene. However, haploinsufficiency for the ZFX gene (X-linked zinc finger protein at Xp21.2) may be an important factor as it has been identified as a candidate gene for ovarian failure [36, 37]. The gene content analysis on chromosome 19 raised some interesting points for discussion. Firstly, the breakpoint fell in 19q13.42 and this locus has been mapped as being associated with the age of natural menopause by two independent research groups through genome-wide association studies using SNP analysis [24, 26]. In case 1 the breakpoint fell near the BRSK1 gene that might influence the secretion of gonadotropin releasing hormone (GnRH) affecting the menstrual cycle since it is highly expressed in the human brain and is associated with vescicle transport and release at the axonal terminals . Additionally, case 1 might experience partial trisomy for the additional material of chromosome 19 (19q13.42 →19qter) on the derivative X chromosome due to incomplete inactivation of the derivative chromosome [38, 39]. Also the MATER gene (Maternal Antigen that Embryos Require) was mapped at 19q13.43 and it is a maternal oocyte protein essential for early embryonic development in mice and an autoantigen associated with autoimmune oophoritis, a mouse model of autoimmune POF . The MATER was identified as a causative gene in a POF patient with a psudic(1;19)(q10;q13.42) by Northup and coworkers . Moreover, the Mater gene in mice is specifically transcribed in oocytes  and human and mouse MATER genes are conserved and share several structural similarities. Although the exact mechanism of action of the MATER gene product is still unknown, knockout mice show female infertility  and so a change in gene dosage (in this case trisomy) might influence fertility . Considering that we have described the second patient as having a chromosomal aberration that involves a putative role for MATER, this gene may be a real candidate in POF aetiology and further investigations may be helpful.
Two isoforms of HS6ST gene were identified in 2 chromosomal breakpoints in our patients: HS6ST1 gene at 2q21 (−216Kb) in case 2 and HS6ST2 gene at Xq26.2 (+720Kb) in case 3. HS6ST1 and HS6ST2 are members of the heparan sulfate sulfotransferase gene family that catalyse the transfer of sulfate to heparan sulfate. Heparan sulfate proteoglicans are ubiquitously expressed on the cell surface and interact with various ligands influencing cell growth, differentiation, adhesion and migration . In 2000 Davison and Conway identified HS6ST as a possible candidate gene for POF aetiology by analyzing the breakpoint on the X chromosome in a family with POF . HS6ST2 in particular is expressed preferentially in the ovary  and it might influence oocyte development by inhibiting a proper interaction with follicular growth factors . Moreover, HS6ST1 gene mutations have recently been associated with idiopathic hypogonadotrophic hypogonadism , thereby increasing the evidence for a possible role of these two isoforms in gonadal fertility. Taking into account that HS6ST2 was identified as a putative gene responsible for POF phenotype in a family by Davison and co-workers , and that we identified the possible involvement of both isoforms in two more cases, the suggestion that this gene family may play a role in POF aetiology is reasonable and should indeed be further investigated.
Additionally, in case 2 the breakpoint fell in Xq21.33, near the DIAPH2 gene (−680Kb), a well-known gene in POF aetiology . The breakpoint was quite distant, but it was not possible to exclude the disruption of some regulatory element in cis, or a long-range effect, that would influence gene expression .
The case 4 karyotype revealed the breakpoint as being located at Xq13.3 in the POF2 critical region [14, 16], but no candidate genes were identified. However, the mechanism underlying the POF phenotype could be due to a direct effect of the chromosomal rearrangement itself without the involvement of specific genes, suggesting a sort of epigenetic control of gene-poor critical regions in patients with X chromosome aberrations [35, 46].
The X chromosome inactivation (XCI) pattern is another important feature in unbalanced translocations involving X chromosome. In case 3 RBA banding revealed the complete and preferential inactivation of der(X), whereas in case 1, 73% of metaphases showed a complete inactivation of derivative X chromosome, but 27% of metaphases evidenced an incomplete and discontinuous inactivation of autosomal material, leading to a mosaic for a partial trisomy of chromosome 19 . Furthermore, 15% of genes on the X chromosome escape X inactivation  and so the translocation might have caused an improper inactivation of derivative X chromosome and haploinsufficiency of genes involved in ovarian function [19, 38]. In case 3 the presence of Y heterochromatic regions may affect the inactivation of der(X) [1, 48]. Considering the breakpoint localizations, the XIST (X inactivation-specific transcript) region (Xq13.2) is located on the specific derivative X chromosome of each case. Cases 2 and 4 are balanced translocations and the patients showed no phenotypic abnormality except for ovarian disfunction. Thus, we may assume that XCI in these two cases was skewed, with the derivative X chromosome typically remaining active and the normal X chromosome being inactivated . Indeed, atypical XCI would result in monosomy of autosomal genes, probably leading to a more severe phenotype [50, 51].
Cases 1 and 3 are both maternal inherited translocations but the respective mothers experienced menopause at a later age than the daughters. The difference in the age of onset could have several causes. Different X inactivation patterns may influence the age of menopause onset [13, 19, 52], but also the effect of the genetic background, such as predisposing polymorphisms in the affected individuals, plays a crucial role. In these cases, identical aberrations might cause no apparent symptoms in mothers but severe clinical presentations in the offspring [53–55]. In addition, environmental factors influence the phenotype and, consequently, also the age of menopause onset .
aCGH was performed on case 4 and the data analysis revealed no major chromosomal alterations. All the observed CNVs overlapped with described polymorphic CNVs. The comparison with the literature data [49, 57–59] showed 2 CNVs significantly associated with the POF phenotype that overlapped with 2 CNVs found in the case 4 molecular karyotype: Xq13.3  and 14q32.33 . However, in our patient these overlapping CNVs were smaller and did not include genes. Moreover, 3 other CNVs described by Aboura and co-workers overlapped with the case 4 CNVs, but these variants were described as not statistically significant compared to CNVs in control populations: 1p36.13, 8p23.1 and 15q11.2 . Although a partial overlapping was found, further studies are required to asses whether there really is an association between these CNVs and the POF phenotype.