Variant Curation of the Largest Compendium of FOXL2 Coding and Noncoding Sequence and Structural Variants in BPES Authors: Charlotte Matton, Julie Van De Velde, Marieke De Bruyne, Stijn Van De Sompele, Sally Hooghe, Hannes Syryn, Miriam Bauwens, Eva D′haene, Annelies Dheedene, Martine Cools, Shoko Komatsuzaki, Ewelina Preizner-Rzucidło, Alison Ross, Christine Armstrong, Wendy Watkins, Andrew Shelling, Andrea L. Vincent, Catherine Cassiman, Sascha Vermeer, David J. Bunyan, Hannah Verdin, Elfride De Baere Published: 2026 ABSTRACT Heterozygous FOXL2 (non)coding sequence and structural variants (SVs) lead to blepharophimosis, ptosis and epicanthus inversus syndrome (BPES), a rare, autosomal dominant developmental disorder characterized by a completely penetrant eyelid malformation and incompletely penetrant primary ovarian insufficiency (POI). We collected variants from our in‐house database, generated via clinical genetic testing and downstream research testing in the Center for Medical Genetics Ghent, Belgium (2001–2024) and via literature and other resources in the same period. All retrieved variants were categorized using ACMG/AMP classifications to increase the knowledge of pathogenicity. We collected 413 unique genetic defects of the FOXL2 region, including 76 novel variants, in 864 index patients. Of these, 87% of patients were identified with a coding FOXL2 sequence variant. The polyalanine tract is a known mutational hotspot of FOXL2, illustrated here by the high percentage of pathogenic polyalanine expansions (24%). Furthermore, the molecular spectrum in typical BPES index patients is characterized by 8% coding deletions and 3% deletions located up‐ and downstream of FOXL2. The remaining 2% carry translocations along with chromosomal rearrangements of 3q23. This uniform and structured reclassification, incorporating the largest dataset of variants implicated in FOXL2‐associated disease so far, will improve both the diagnosis as well as genetic counselling for individuals with BPES. 1. Introduction FOXL2 is a 2.9‐kb single‐exon gene located on the long arm of chromosome 3 (3q22.3). The gene encodes a 376‐amino‐acid transcription factor with a highly conserved DNA‐binding forkhead domain. In addition to this forkhead domain, a 14‐amino‐acid polyalanine tract can be distinguished, of which the function remains largely unknown (Figure 1a) [3–5]. (a) The FOXL2 protein comprises 376 amino acids and contains two key domains: the forkhead domain in blue (p.54–164) and the polyalanine tract in red (p.221–234), adapted from Beysen et al. [1]. (b) The transcription factor FOXL2 is essential for ovarian development. At 5–6 weeks of gestation, the bipotential gonad differentiates towards either testis or ovaries, depending on the chromosomal sex. This process is tightly regulated by antagonistic interactions between key factors. FOXL2 inhibits SOX9, thereby repressing the protestis pathway and, in concert with other regulators, promoting ovarian differentiation. In adulthood, FOXL2 remains crucial for ovarian maintenance and function. Figure adapted from Tucker et al. [2]. (a) (b) The FOXL2 transcription factor, like other members of the forkhead transcription factor family, is involved in key developmental processes and requires tightly controlled spatiotemporal gene regulation [6]. During embryonic development, FOXL2 is expressed in the developing periocular mesenchyme, where it activates α smooth muscle actin (α‐SMA), facilitating the formation of the levator smooth muscle [7]. In developing ovaries, FOXL2 functions alongside other profemale factors, promoting the upregulation of ovarian‐specific genes while simultaneously suppressing testis‐specific factors such as SOX9 [2, 8, 9] (Figure 1b). Postnatally, FOXL2 is expressed in the granulosa cells (i.e., supporting cells of the oocyte), pituitary gland and uterus [2, 10–13]. In adult granulosa cells, FOXL2 is critical for ovarian maintenance and folliculogenesis, ensuring the integrity of ovarian function throughout reproductive life [14] (Figure 1b). In line with other highly conserved transcription factor genes, impairment of FOXL2 results in developmental disease and cancer. Heterozygous germline variants in FOXL2 are known to cause blepharophimosis, ptosis and epicanthus inversus syndrome (BPES; OMIM #110100), a rare, autosomal dominant, developmental disorder that can arise de novo or be inherited [4, 5, 15]. BPES is characterized by a fully penetrant eyelid malformation and an incompletely penetrant and variable primary ovarian insufficiency (POI). The eyelid malformation is described according to four typical features including (1) horizontal (blepharophimosis) and (2) vertical (ptosis) narrowing of the palpebral fissures, (3) a skin fold running inwards and upwards from the lower eyelid (epicanthus inversus) and (4) a lateral displacement of the inner canthi (telecanthus). POI is defined as the absence of menses for over 4 months before the age of 40, resulting in early menopause and female subfertility or infertility due to impaired folliculogenesis in adult granulosa cells. BPES is associated with additional features including thick eyebrows, displacement of the lacrimal puncta, nasolacrimal drainage problems, a broad nasal bridge and low‐set ears [11, 15, 16]. At the somatic level, FOXL2 variants are strongly implicated in adult‐type granulosa cell tumours, with a recurrent missense variant c.402C > G (p.Cys134Trp) detected in approximately 90%–95% of cases [17]. In this study, we compiled a large collection of in‐house variants identified through clinical and research genetic testing at our centre (Center for Medical Genetics Ghent, Ghent University Hospital, Ghent, Belgium) and elsewhere (Wessex Regional Genomics Laboratory, Salisbury District Hospital, Salisbury, United Kingdom; Institute of Human Genetics, University of Würzburg, Würzburg, Germany; Laboratory of Cytogenetics and Molecular Genetics, University Children′s Hospital, Krakow, Poland; Clinical Genetics Centre, Aberdeen Royal Infirmary, Aberdeen, United Kingdom; and Centre of Human Genetics, University Hospital Leuven, Louvain, Belgium) in index patients with a typical BPES phenotype. This dataset was complemented with previously published FOXL2 sequence and structural variants (SVs) affecting the FOXL2 region, all identified in patients with typical BPES. Together, this resulted in a comprehensive compendium of 413 distinct heterozygous constitutional FOXL2 variants, ranging from coding sequence changes and copy number variants (CNVs) to translocations and noncoding CNVs disrupting putative cis‐regulatory elements (CREs). All sequence variants were subsequently classified according to the most recent American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG‐AMP) standard guidelines. Notably, this large study enabled us to expand the variant spectrum with 76 novel FOXL2 genetic defects. 2. Material and Methods 2.1. Literature Search on Patients With BPES and a Confirmed FOXL2 Genetic Defect We collected publications up to December 2024 that reported typical BPES patients with a genetically confirmed heterozygous FOXL2 variant, using the following PubMed search terms: ‘FOXL2’ [All Fields] AND (‘mutation’ [All Fields] OR ‘variant’ [All Fields] OR ‘translocation’ [All Fields] OR ‘deletion’ [All Fields]). The literature search was extended via Google Scholar to not exclude publications that were not listed in PubMed. Moreover, we mined the ClinVar and Leiden Open‐source Variation Database (LOVD) databases to include variants that were only submitted to the databases. Likewise, to ensure that unpublished SVs were not overlooked, the Genomics England and Decipher databases were consulted. For all included variants, variant annotations, gnomAD allele frequencies (gnomAD V4.1.0), disease phenotype and patient information (including age and sex) were collected, if available. Duplicates were excluded from the variant collection based on overlap in patient information and author names of the included publications. In addition, studies for which the full text was unavailable or not accessible through the Ghent University Library, or not available in English, were excluded. Although a comprehensive search was performed, we cannot exclude that certain FOXL2 variants associated with BPES were missed due to the limitations of literature search tools (e.g., FOXL2 variants solely mentioned in publications′ supporting information might have been overlooked). 2.2. Unreported Patients With BPES and a Confirmed Molecular Cause From the In‐House Patient Cohort In addition to variants obtained from literature and public databases, FOXL2 variants identified through clinical genetic testing at the Center for Medical Genetics Ghent (Ghent University Hospital, Ghent, Belgium), Wessex Regional Genomics Laboratory (Salisbury District Hospital, Salisbury, United Kingdom), Institute of Human Genetics (University of Würzburg, Würzburg, Germany), Laboratory of Cytogenetics and Molecular Genetics (University Children′s Hospital, Krakow, Poland), Clinical Genetics Centre (Aberdeen Royal Infirmary, Aberdeen, United Kingdom) and Centre of Human Genetics (University Hospital Leuven, Louvain, Belgium) were included. Given the specificity of the BPES phenotype and FOXL2 being the only known disease locus for this rare syndrome, Sanger sequencing was performed across the FOXL2 open reading frame in all cases with BPES. The initial polymerase chain reaction (PCR) to amplify the region of interest was performed using three overlapping primers pairs (Supporting Information 4: Table S4) and Kapa2G Robust Master Mix (2x, Kapa Biosystems, MA, United States), followed by Sanger sequencing using the BrilliantDye kit (V3.1, NimaGent, Nijmegen, the Netherlands) and ran on an ABI3730XL DNA analyzer (Applied Biosystems, MA, United States). Due to the GC‐rich nature of the FOXL2 region, 5% dimethyl sulfoxide (DMSO) was added to increase sequencing signal intensity. Multiplex ligation‐dependent probe amplification (MLPA) was performed using a commercially available probe mix (P054, MRC Holland, Amsterdam, the Netherlands) to detect CNVs. The MLPA kit contains three probes covering the coding FOXL2 region, four probes covering the ATR region (Exon 1, 4, 22 and 47) and five probes overlapping with the PISRT1 region, a long noncoding RNA (lncRNA) located in the regulatory region upstream of FOXL2. Apart from deletion identification itself, it is also essential to size the deletion, which was previously done using targeted arrays and qPCR [18–20]. 2.3. Nomenclature Variant annotations in the FOXL2 coding region were (re)assessed using Alamut (GR Ch38) for coding DNA (c.), protein (p.) and genome (g.) nomenclature. When necessary, annotations were revised to align with the latest Human Genome Variation Society (HGVS) nomenclature guidelines (http://hgvs-nomenclature.org). 2.4. ACMG/AMP Variant Pathogenicity Classification All collected coding FOXL2 variants were (re)classified using an in‐house variant classification tool (VCT) (Center for Medical Genetics Ghent, Ghent University Hospital, Ghent, Belgium), developed in accordance with ACMG/AMP guidelines described by Richards et al. and Tavtigian et al. [21, 22]. The VCT categorizes variants into five classes based on their calculated p value, namely benign (p < 0.001), likely benign (0.001 ≤ p < 0.1), unknown significance (VUS) (0.1 ≤ p < 0.9), likely pathogenic (0.9 ≤ p < 0.99) or pathogenic (p ≥ 0.99). 2.5. LOVD and ClinVar Submission Following the (re)classification of all coding FOXL2 sequence variants, they were uploaded to the LOVD and ClinVar database to ensure comprehensive documentation of all collected FOXL2 variants. When available, patient information and in silico predictions were also included. This work was conducted according to the tenets of the Declaration of Helsinki and was approved by the data access committee of Ghent University Hospital. 3. Results 3.1. Collection of BPES Index Cases and Unique FOXL2 Genetic Defects Through a comprehensive literature and database search, we collected 497 index cases with a typical BPES phenotype that have been reported previously in international, peer‐reviewed literature and curated databases, including ClinVar, Genomics England and DECIPHER [4, 5, 15, 18–20, 23–117]. In addition, we report 367 BPES index patients with a genetic defect of the FOXL2 region identified via routine diagnostic pipelines, resulting in a total collection of 864 index cases included in this study (Table 1 and 2, Supporting Information 1–3: Tables S1–S3). This large BPES patient cohort revealed 413 unique FOXL2 genetic defects, both in the coding and noncoding regions of FOXL2. Notably, we report 76 novel variants associated with a typical BPES phenotype in this compendium (Table 1 and 2). TABLE 1(A) We identified 54 novel FOXL2 coding variants, including missense, nonsense, frameshift and in‐frame indels, in this study. All variants are annotated with their corresponding ACMG/AMP variant classifications.A. Novel coding FOXL2 variantsDNA variantPredicted protein effectACMG‐AMP classificationPhenotypec.75_78delp.(Lys25Asnfs∗124)LPPM2, PP4_PM, PVS1_PS1, PP1BPESc.119delp.(Gly40Valfs∗110)LPPM2_PP, PP4_PM, PVS1_PS1BPESc.136_137delp.(Pro46Glyfs∗49)PPM2, PP4_PM, PVS1_PS1, PP1BPESc.328delp.(Glu110Serfs∗40)LPPM2, PP4_PM, PVS1_PS1BPESc.395delp.(Pro132Argfs∗18)LPPM2, PP4_PM, PVS1_PS1, PP1BPESc.467_473delp.(Pro156Argfs∗113)PPM2, PP4_PM, PVS1_PS1, PS2BPESc.495dupp.(Leu166Alafs∗73)LPPM2, PP4_PM, PVS1_PS1BPESc.501delp.(Phe167Leufs∗104)PPM2, PP4_PM, PVS1_PS1, PM6BPESc.593delp.(Gly198Alafs∗73)LPPM2, PP4_PP, PVS1_PS1BPESc.675_705delp.(Ala226Profs∗35)LPPM2, PP4_PM, PVS1_PS1BPESc.735delp.(Lys246Argfs∗25)LPPM2, PP4_PM, PVS1_PS2, PP1BPESc.749delp.(Gly250Alafs∗21)PPM2, PP4_PM, PVS1_PS2, PS2BPESc.770dupp.(Tyr258Valfs∗276)PPM2, PP4_PM, PVS1_PS2, PS2BPESc.806dupp.(Val270Argfs∗264)LPPM2, PP4_PM, PVS1_PS2BPES with POIc.839delp.(Pro280Argfs∗76)LPPM2, PP4_PM, PVS1_PS2BPESc.846dupp.(Ala283Argfs∗251)LPPM2, PP4_PM, PVS1_PS2BPESc.858_873dupp.(Pro292Serfs∗247)LPPM2, PP4_PM, PVS1_PS2BPESc.866_867dupp.(Pro290Thrfs∗67)LPPM2, PP4_PM, PVS1_PS2BPESc.871_895delp.(His291Cysfs∗57)LPPM2, PP4_PM, PVS1_PS2BPESc.965_978dupp.(Pro327Serfs∗34)LPPM2, PP4_PM, PVS1_PS2BPESc.1078delp.(Tyr360Thrfs∗91)LPPM2, PP4_PM, PVS1_PMBPESc.1084delp.(Asp362Thrfs∗89)LPPM2, PP4_PM, PVS1_PMBPESc.19G > Tp.(Glu7∗)LPPM2, PP4_PM, PVS1_PS1BPESc.177C > Gp.(Tyr59∗)LPPM2, PP4_PM, PVS1_PS1BPESc.249C > Ap.(Tyr83∗)LPPM2, PP4_PM, PVS1_PS1BPESc.328G > Tp.(Glu110∗)LPPM2, PP4_PM, PVS1_PS1BPESc.331_332insAp.(Cys111∗)LPPM2, PP4_PM, PVS1_PS1BPESc.403G > Tp.(Glu135∗)LPPM2, PP4_PM, PVS1_PS1, PM6BPES with POIc.528C > Ap.(Cys176∗)LPPM2, PP4_PM, PVS1_PS1BPESc.564C > Gp.(Tyr188∗)LPPM2, PP4_PM, PVS1_PS1BPESc.582C > Ap.(Tyr194∗)LPPM2, PP4_PM, PVS1_PS1BPESc.622C > Tp.(Gln208∗)LPPM2, PP4_PP, PVS1_PS1BPESc.736A > Tp.(Lys246∗)LPPM2, PP4_PM, PVS1_PS2BPESc.765C > Ap.(Tyr255∗)PPM2, PP4_PP, PVS1_PS2, PS2BPESc.163C > Tp.(Pro55Ser)LPPM2, PP4_PM, PP3, PM1_PPBPESc.167C > Ap.(Pro56Gln)LPPM2, PP4_PM, PP3, PM1_PPBPES with mental retardationc.175 T > Cp.(Tyr59His)LPPM2, PP4_PM, PP3, PM1_PPBPES with POIc.184C > Gp.(Leu62Val)LPPM2, PP4_PM, PP3, PM1_PPBPESc.196G > Ap.(Ala66Thr)LPPM2, PP4_PM, PP3, PM5_PM, PM1_PPBPESc.238A > Tp.(Ile80Phe)LPPM2, PP4_PM, PP3, PM1BPESc.271 T > Cp.(Tyr91His)PPM2, PP4_PM, PP3, PM1, PS2BPESc.272A > Gp.(Tyr91Cys)LPPM2, PP4_PM, PP3, PM1_PPBPESc.298A > Cp.(Asn100His)LPPM2, PP4_PM, PP3, PM1, PP1BPESc.300 T > Ap.(Asn100Lys)LPPM2, PP4_PM, PP3, PM1_PP, PP1BPESc.308G > Cp.(Arg103Pro)LPPM2, PP4_PM, PP3, PM5_PM, PM1_PPBPESc.334 T > Ap.(Phe112Ile)LPPM2, PP4_PM, PP3, PM1BPESc.377A > Tp.(Asn126Ile)LPPM2, PP4_PM, PP3, PM1_PPBPESc.142_173delinsGCGCTp.(Lys48_Ser58delinsAlaLeu)LPPM2, PP4_PM, PM4_PM, PM6BPESc.191_233delins43p.(Ala64_Ser78delins15)LPPM2, PP4_PM, PM4_PMBPESc.214_216delp.(Glu72del)LPPM2, PS4_PP, PP4_PM, PM4_PP, PM6BPESc.666_698dupp.(Ala224_Ala234dup)PPM2, PP4_PM, PM4_PM, PS3, PM6BPESc.701_702ins30p.(Ala225_Ala234dup)LPPM2, PP4_PM, PM4_PMBPESc.1130G > Cp.(∗377Serext∗31)LPPM2, PS4_PP, PP4_PM, PM4_PMBPES with POIc.1331A > Gp.(∗377Trpext∗31)LPPM2, PP4_PM, PM4_PM, PM6BPES (A) We identified 54 novel FOXL2 coding variants, including missense, nonsense, frameshift and in‐frame indels, in this study. All variants are annotated with their corresponding ACMG/AMP variant classifications. TABLE 2(B) We identified 22 SVs affecting the FOXL2 locus in this study, including 11 deletions encompassing the entire FOXL2 gene, nine deletions located upstream of FOXL2, one combined upstream and downstream deletion flanking FOXL2 and one novel chromosomal translocation involving the FOXL2 genomic region.B. Novel structural variants affecting FOXL2 Type of SVCoordinates (hg38)PhenotypeTotal FOXL2 delBPESTotal FOXL2 delBPESTotal FOXL2 delBPESTotal FOXL2 delBPESTotal FOXL2 delBPES and cryptorchidismTotal FOXL2 delChr3 (138945645–142436590)BPESTotal FOXL2 and ATR delChr3 (136422338–148529357)BPESTotal FOXL2 and ATR delBPESTotal FOXL2 and ATR delChr3 (129785551–148834776)BPESTotal FOXL2 delBPESTotal FOXL2 delBPESDeletion upstream of FOXL2 PISRT1 BPESDeletion upstream of FOXL2 Chr3 (139230577–139234837)BPESDeletion upstream of FOXL2 Chr3 (139230577–139236822)BPESDeletion upstream and downstream of FOXL2 Deletion 1: chr3 (138640184–138819066) Deletion 2: chr3 (139003469–139366791)BPESDeletion upstream of FOXL2 Chr3 (139170001–139560000)BPESDeletion upstream of FOXL2 PISRT1 BPESDeletion upstream of FOXL2 PISRT1 BPESDeletion upstream of FOXL2 PISRT1 BPESDeletion upstream of FOXL2 Chr3 (139046001–139519000)BPESDeletion upstream of FOXL2 Chr3 (139189424–139251840)BPES46,XY,t(3,18)(q23;q21.3)BPES (B) We identified 22 SVs affecting the FOXL2 locus in this study, including 11 deletions encompassing the entire FOXL2 gene, nine deletions located upstream of FOXL2, one combined upstream and downstream deletion flanking FOXL2 and one novel chromosomal translocation involving the FOXL2 genomic region. 3.2. FOXL2 Coding Sequence Variants and Their Variant Classification A total of 752 index patients have been identified with a FOXL2 coding sequence variant making up 87% (752/864) of the total solved BPES index cases and revealed 54 unique novel variants in the coding region of the FOXL2 gene. The majority of these index patients are heterozygous for a frameshift variant (39%; 295/752), followed by in‐frame indels (31%; 231/752), missense variants (18%; 137/752), nonsense variants (11%; 82/752) and stop‐loss variants (1%; 7/752) (Supporting Information 1: Table S1, Figure 2). (a) In 87% of the BPES patient cohort (n = 752), variants were identified in the coding region of FOXL2. These include frameshift (39%), in‐frame indel (31%), missense (18%), nonsense (11%) and stop‐loss (1%) variants. (b) Coding FOXL2 variants are distributed across the entire protein. Missense variants cluster in the forkhead domain, while in‐frame indels predominantly affect the polyalanine tract, both known mutational hotspots of the FOXL2 region. Novel variants identified in this study are highlighted in yellow. (a) (b) 3.2.1. Frameshift Variants Our compendium contains 135 unique frameshift variants, including 22 novel variants, of which 24% (32/135) are classified as pathogenic and 76% (103/134) are classified as likely pathogenic. Frameshifts of the FOXL2 coding region result in either a predicted truncated protein (100/135) or an elongated protein (35/135), leading to loss‐of‐function. Notably, the most recurrent frameshift variants, including c.843_859dup (p.[Pro287Argfs∗75]) (n = 66), c.855_871dup (p.[His291Argfs∗71]) (n = 38) and c.804dup (p.[Gly269Argfs∗265]) (n = 20), are clustered downstream of the polyalanine tract (Supporting Information 1: Table S1, Figure 2). 3.2.2. In‐Frame Indels In addition to the 23 previously reported unique in‐frame indel variants, we report three novel in‐frame variants located outside the polyalanine tract (c.142_173delinsGCGCT, p.[Lys48_Ser58delinsAlaLeu]; c.191_233delins43, p.[Ala64_Ser78delins15] and c.214_216del, p.[Glu72del]) and two novel in‐frame duplications located within the polyalanine tract (c.666_698dup, p.[Ala224_Ala234dup] and c.701_702ins30, p.[Ala225_Ala234dup]) in this study. Of all unique variants, 32% are classified as pathogenic (9/28) and 68% as likely pathogenic (19/28). To date, 12 unique duplications and one deletion in the polyalanine region have been described. Polyalanine expansions are the most frequently reported type of in‐frame indels and, to a larger extent, the most frequently reported FOXL2 variants in BPES, as they are identified in 209 BPES index patients in this large cohort (24%; 209/864). The most prevalent polyalanine expansion is c.672_701 (p.[Ala225_Ala234dup]), observed in at least 156 BPES index cases, including 69 from our patient cohort. Another common polyalanine expansion, c.663_692dup (p.[Ala225_Ala234dup]), has been reported 13 times in literature and is also found in 21 index cases within our patient cohort (Supporting Information 1: Table S1, Figures 2 and 3). Polyalanine expansions were identified in 24% of the BPES patient cohort, making this the most frequent variant type. In most cases, the polyalanine tract is expanded from 14 to 24 alanine residues. Additionally, one contraction of the polyalanine tract was observed. The blue bar indicates deleted amino acids, the red bar represents the original repeat that was duplicated and the black bar denotes the inserted repeat. Interestingly, a small polyalanine expansion of 5+ alanine residues (c.684_698dup; p.[Ala230_Ala234dup]) was reported in an Indian, multigenerational family. Heterozygous family members remained (seemingly) unaffected, whereas homozygous individuals had a typical BPES phenotype. One of the homozygous female family members was also diagnosed with POI [78]. Notably, the same variant was identified in an unrelated, sporadic case in a heterozygous state with a typical BPES phenotype [104] (Supporting Information 1: Table S1). 3.2.3. Missense Variants Next, 84 unique missense variants were found of which the majority are classified as pathogenic (33%; 28/84) or likely pathogenic (61%; 51/84), and 6% as VUS (5/84). Most (n = 74) are located in the forkhead domain, which is known to be a mutational hotspot for missense variants in FOXL2. We have identified 13 novel missense variants, all located within the forkhead domain (Supporting Information 1: Table S1, Figure 2). Single nucleotide polymorphisms (SNPs) have previously been reported in the FOXL2 region, including c.501C > T (p.Phe167=), c.536C > G (p.Ala179Gly), c.647C > T (p.Ala216Val) and c.869C > A (p.Pro290His) [60, 64, 74, 85, 90]. These variants have occasionally been detected in patients with BPES; however, their presence has also been documented in unaffected individuals, and they are catalogued in public SNP databases (e.g., dbSNP: rs7432551, rs61750361 and rs565208053). This suggests that they represent population polymorphisms rather than pathogenic variants. According to ACMG classification criteria, their relatively high allele frequencies in gnomAD, together with consistent in silico predictions of benign impact, further support that these variants should be considered (likely) benign. 3.2.4. Nonsense Variants We identified 51 unique nonsense variants in a total of 82 index patients, of which 12 are novel. All variants were classified as pathogenic (35%; 18/51) or likely pathogenic (65%; 33/51). Most of these variants are observed only once, with the notable exception of the recurrent variant c.655C > T (p.[Gln219∗]), reported 10 times. Another nonsense variant, p.(Trp128∗), has been reported six times, albeit resulting from different nucleotide changes (c.383G > A, c.384G > A and c.384del) (Supporting Information 1: Table S1, Figure 2). 3.2.5. Stop‐Loss Variants Finally, a total of three unique stop‐loss variants, all classified as likely pathogenic, have been identified in seven BPES index cases. Amongst these, two are novel stop‐loss variants, c.1130G > C (p.(∗377Serext∗31)) and c. 1331A > G (p.(∗377Trpext∗31)), found in two and one BPES index cases, respectively (Supporting Information 1: Table S1, Figure 2). 3.3. SVs Affecting the FOXL2 Region 3.3.1. Copy Number Variants of the FOXL2 Region CNVs affecting the FOXL2 gene were identified in 8% (71/864) of this BPES patient cohort and were detected using MLPA FOXL2‐specific probes (Supporting Information 2: Table S2, Figure 4). All but one of these CNVs encompassed the entire FOXL2 gene. A single case involved a partial FOXL2 deletion (BPES_10). The ATR gene, also covered by the MLPA assay, was codeleted in approximately 23% of all detected deletions. Notably, all reported CNVs exclusively comprised deletions, with no instances of duplications associated with BPES documented to date. Furthermore, literature review revealed an additional 23 cases with FOXL2 whole gene deletions, albeit lacking specific genomic coordinates. Via DECIPHER, we were able to identify six more patients with BPES harbouring a FOXL2 (and ATR) deletion, with detailed genomic coordinates provided. In‐house, 31 previously reported cases with BPES were molecularly diagnosed using MLPA followed by qPCR and/or microarray‐based copy number analysis, enabling precise delineation of the deletions [18–20, 85]. Amongst the deletions that were further delineated, nearly all extended beyond FOXL2 to encompass one or more neighbouring genes. Additionally, we report 11 novel deletions spanning the entire FOXL2 region in this study. (a) SVs were identified in 13% of the BPES patient cohort (n = 112). These include whole‐gene deletions of FOXL2 (64%), deletions located up‐ or downstream of FOXL2 (22%) and translocations (14%). (b) Translocation breakpoints cluster upstream of FOXL2, disrupting the regulatory landscape and physically separating the gene from its regulatory elements. Moreover, upstream deletions share an SRO (indicated in orange) of 4.5 kb, encompassing putative FOXL2 enhancers, including the lncRNA PISRT1. The topologically associated domain (TAD) of FOXL2 is highlighted with a red bar, marking the boundaries of its regulatory domain. Deletions for which no further delineation was performed; bars indicate whether a FOXL2 or PISRT1 deletion was detected by MLPA, although the exact genomic coordinates of these deletions remain undetermined. (a) (b) Deletions disrupting the FOXL2 cis‐regulatory region, including the lncRNA PISRT1, were identified in 3% (24/864) of genetically solved BPES index patients, ultimately leading to aberrant FOXL2 expression (Supporting Information 2: Table S2, Figure 4). In‐house, 10 typical patients with BPES harbouring a cis‐regulatory deletion could be identified, as previously published by us [18–20, 85]. One of the deletions is located downstream of FOXL2, whereas the other nine deletions are located upstream of FOXL2. In Bouman et al, a noncoding 54‐kb deletion (234 kb upstream of FOXL2) is described that encompasses the upstream cis‐regulatory region including PISRT1 in a 9‐month‐old female patient and her father, both displaying a typical BPES phenotype. Based on this regulatory deletion, the shortest region of overlap (SRO) of all noncoding deletions, located upstream of FOXL2, could be delineated to a region of 4573 bp (chr3: 139230309–139234882 [Hg38]) [35]. Additionally, we describe 10 cases with BPES in this study with novel deletions upstream of FOXL2, all of which include the PISRT1 region. Importantly, one exceptional case (BPES_89) initially appeared to involve only a PISRT1 deletion, as revealed by MLPA. However, subsequent CNV‐sequencing analysis revealed the presence of two distinct deletions, located up‐ and downstream of FOXL2, indicating a more complex structural rearrangement of the FOXL2 locus (Supporting Information 2: Table S2, Figure 4). Another interesting case described a Chinese family with BPES in which all affected males additionally exhibited polydactyly (BPES_96, Supporting Information 2: Table S2). The causative variant was a deletion within the FOXL2 promoter overlapping a CRE, and functional luciferase assays demonstrated reduced promoter activity, supporting a regulatory loss‐of‐function mechanism [118]. 3.3.2. Translocations Lastly, cytogenetic rearrangements of 3q23 were found in 2% (16/864) of typical BPES index patients. Fifteen translocations had been described in literature before [4, 58, 75, 94, 106–111, 113–115, 119, 120] (Supporting Information 3: Table S3, Figure 4). Additionally, we identified one novel de novo translocation t(3,18)(q23; q21.3) in a male patient with BPES. None of the identified translocations disrupted the FOXL2 coding sequence itself. Instead, all mapped breakpoints clustered upstream of FOXL2 within its cis‐regulatory domain, consistent with a position‐effect mechanism in which disruption of long‐range enhancers leads to FOXL2 dysregulation, as previously shown [94, 109, 112, 113, 115]. 4. Discussion 4.1. Biological Relevance FOXL2 is a crucial transcription factor in both eyelid and ovarian development, but also postnatally FOXL2 remains critical for maintenance of the ovaries. Pathogenic variants in the FOXL2 gene underlie BPES, associating an eyelid malformation with POI. In this study, we compiled a comprehensive compendium of (virtually) all reported FOXL2 genetic defects associated with BPES and of our in‐house patient cohort, encompassing both (non)coding sequence variants and SVs. Across a total of 864 index cases, we distinguished 413 unique genetic defects of the FOXL2 region, including 76 novel variants. The majority of FOXL2 genetic defects are coding sequence variants, as they are identified in 87% of molecularly solved index patients with BPES. As FOXL2 is a single‐exon gene, aberrant transcripts are expected to escape nonsense‐mediated decay (NMD). Consequently, null variants, introducing premature stop codons, are predicted to give rise to truncated FOXL2 proteins, missing the forkhead domain and/or the polyalanine tract [85]. In contrast to the wild‐type FOXL2, which displays exclusive nuclear localization, truncated FOXL2 proteins tend to form nuclear aggregates [26, 121, 122]. Missense variants represent 18% of all sequence variants and are predominantly clustered within the forkhead domain, a well‐established mutational hotspot (Figure 2). Alterations in this domain are compromising DNA binding and transcriptional regulation, a mechanism that has been extensively studied previously [30, 122–125]. A second mutational hotspot within the FOXL2 gene is the polyalanine tract. In 24% of all BPES index patients, an expansion of the polyalanine tract is found, making this the most recurrent FOXL2 variant associated with typical BPES. The WT polyalanine tract is located between position p.221 and p.234, consists of 14 alanine residues and is strictly conserved amongst mammals [124, 126]. However, its function remains largely unknown. Up to date, 12 unique duplications and one deletion of the polyalanine tract have been reported (Supporting Information 1: Table S1, Figure 3), demonstrating that BPES can be caused by both expansion and contraction of the polyalanine tract, encoded by imperfect trinucleotide repeats. These expansions are most likely caused by slippage of DNA polymerase during replication of these trinucleotide repeats. It was shown previously that these expansions result in the misfolding of the protein and as such impair FOXL2 protein function in a length‐dependent manner, resulting in protein mislocalization from the nucleus to the cytoplasm with cytoplasmic aggregation of the altered FOXL2 protein [126]. The remainder of FOXL2 genetic defects consisted of deletions affecting the gene (8%) or its cis‐regulatory region (3%), as well as upstream translocations separating FOXL2 from its cis‐regulatory domain (2%). These SVs are also expected to result in FOXL2 loss‐of‐function through reduced or absent gene expression. Interestingly, these noncoding SVs represent unique natural human models for studying the cis‐regulation of FOXL2. We have previously delineated the SRO of regulatory deletions to a 7.4‐kb region upstream of FOXL2 [20]. Recently, a novel deletion has been reported that delineates the SRO even further to a region of 4.5 kb [35], comprising a cluster of candidate CREs and a lncRNA PISRT1. Interestingly, the SRO delineated in human cases with BPES has its counterpart in a natural animal model, namely the Polled Intersex Syndrome (PIS) goat [127–129]. First, the syntenic region affected by a complex noncoding SV in the Polled Intersex goat overlaps with the 4.5‐kb SRO delineated in human BPES. Second, the Polled Intersex phenotype is characterized by the dominant absence of horns in both sexes and recessive sex reversal in XX animals. Another animal model for aberrant cis‐regulation of FOXL2 is the piggyBac mouse (Foxl2PB/PB ), caused by an insertion 160 kb upstream of Foxl2, leading to reduced Foxl2 expression and causing an eyelid phenotype mimicking BPES [130]. Although the syntenic regions in human, goat and mouse are not ultraconserved, the occurrence of cross‐species noncoding defects leading to similar phenotypes supports the hypothesis of a functional conservation of the upstream cis‐regulatory region of FOXL2. However, the precise function of this cis‐regulatory region has remained elusive so far and requires further investigation. 4.2. Clinical and Diagnostic Relevance The BPES phenotype is characterized by a typical eyelid malformation, including blepharophimosis, ptosis, epicanthus inversus and telecanthus, sometimes associated with POI. The phenotype can be distinguished from other syndromes that share one or more overlapping traits, such as Say‐Barber‐Biesecker variant of Ohdo syndrome (OMIM #603736) or Kaufman oculocerebrofacial syndrome (OMIM #244450), caused by variants in KAT6B and UBE3B, respectively. Also, one case has been described with loss‐of‐function of NR2F2, resulting in a BPES‐like phenotype [131]. However, to date FOXL2 remains the only known disease gene underlying typical BPES. Importantly, penetrance for the eyelid phenotype in individuals carrying a heterozygous FOXL2 variant is considered complete, as it is observed in all molecularly solved BPES patients. The only reported exception is a small FOXL2 polyalanine expansion of +5 residues behaving as a recessive allele in a consanguineous Indian family [78]. This variant likely represents the milder end of a length‐ and dosage‐dependent pathogenic spectrum of FOXL2 polyalanine expansions, as it was also found in a heterozygous state in a young, male BPES patient [104]. In contrast, the penetrance of POI in BPES patients overall is considered reduced and age related. Molecular diagnosis remains necessary to genetically confirm BPES, particularly for clinical decision‐making and accurate genetic counselling. Given the autosomal dominant inheritance pattern, genetic counselling is essential in BPES families, both for recurrence risk assessment and reproductive planning. This updated FOXL2 variant database, including previously published and novel variants (Supporting Information 1–3: Tables S1–S3), curated according to ACMG/AMP criteria, provides a key tool for clinical laboratories and clinicians to aid interpretation of variants identified in patients with BPES. Because of the highly specific phenotype and complete penetrance of the eyelid malformation, we propose tailored rule specification of ACMG criterion PP4 for FOXL2 variant classification in BPES (see also Supporting Information 1: Table S1):-PP4: when BPES is clinically well described and assessed by a physician (distinguished from nonspecific ptosis).-PP4 moderate: when the phenotype is explicitly confirmed by an expert ophthalmologist.-PP4 strong: when, in addition, biological parameters such as FSH and AMH levels support a diagnosis of POI. -PP4: when BPES is clinically well described and assessed by a physician (distinguished from nonspecific ptosis). PP4: when BPES is clinically well described and assessed by a physician (distinguished from nonspecific ptosis). -PP4 moderate: when the phenotype is explicitly confirmed by an expert ophthalmologist. PP4 moderate: when the phenotype is explicitly confirmed by an expert ophthalmologist. -PP4 strong: when, in addition, biological parameters such as FSH and AMH levels support a diagnosis of POI. PP4 strong: when, in addition, biological parameters such as FSH and AMH levels support a diagnosis of POI. This phenotype‐driven refinement will facilitate more accurate and consistent variant classification across diagnostic laboratories. In the (Supporting Information 1: Table S1), we propose additional FOXL2‐specific rule specifications for variant classification in BPES. 4.3. Genotype–Phenotype Correlations Previous studies have attempted to define a clear genotype–phenotype correlation in BPES. Truncating variants located upstream of the polyalanine tract tend to be linked to BPES Type I (combining the eyelid malformation with POI), whereas polyalanine expansions appear more frequently associated with BPES Type II (presenting only eyelid malformations) [85, 132]. Missense variants within the forkhead domain have generally been correlated with BPES Type II [123, 132]. However, Li et al. suggested that missense variants situated in the central region of the forkhead domain may instead lead to the more severe BPES Type I phenotype [26]. Despite these observations, a consistent genotype–phenotype correlation across FOXL2 variants has not yet been established, making the prediction of POI risk in young female patients with BPES challenging. Also, in this study, no clear genotype–phenotype correlations could be established, mainly due to the lack of appropriate clinical follow‐up on fertility and endocrinological outcomes in female patients. For many variants, only a single index patient has been identified, making it difficult to determine whether these variants are associated with an ovarian phenotype. This uncertainty often stems from the fact that some variants were found only in male patients or in female patients who are still too young to evaluate ovarian (mal)function. Notably, for several recurrent variants observed in multiple female BPES patients, including c.843_859dup [p.(Pro287Argfs*75)], c.655C>T [p.(Gln219*)], c.157C>T [p.(Gln53*)], c.650C>T [p.(Ser217Phe)] and c.672_701dup30 [p.(Ala225_Ala234dup)] (Supporting Information 1: Table S1), the phenotype ranged from no ovarian involvement to primary or secondary amenorrhea, sometimes leading to infertility, demonstrating that this variability is probably not restricted to a specific variant type. In addition, all variant types and positions appeared capable of causing both BPES combined or not with POI, with varying ages of onset of POI. We therefore propose that BPES should rather be considered a single condition displaying a continuum ranging from a fully penetrant eyelid phenotype to BPES with POI, which shows age‐related penetrance in affected females. Accordingly, we emphasize that all female BPES patients, irrespective of their underlying variant, should receive appropriate endocrinological and fertility follow‐up to enable timely detection and management of ovarian dysfunction. 4.4. Diagnostic Perspective for Patients With BPES Currently, the diagnostic strategy for BPES relies on targeted FOXL2 sequencing in patients with a typical eyelid phenotype, complemented by CNV analysis of the coding region and upstream noncoding regulatory region. This approach allows the majority of affected individuals to be molecularly confirmed. Nevertheless, approximately 12% of clinically diagnosed BPES patients remain without a molecular diagnosis [15]. One striking example in our cohort was a multigenerational family of Polynesian descent that remained unsolved for many years. As no female family members had offspring, BPES with POI was suspected. Standard diagnostic testing with targeted Sanger sequencing of the FOXL2 coding region and MLPA failed to identify a pathogenic variant. Whole‐genome sequencing (WGS) of multiple family members eventually revealed the most recurrent polyalanine expansion, c.672_701dup (p.[Ala225_Ala234dup]) to cosegregate with disease. The earlier negative results were likely explained by allelic drop‐out in tested individuals. As this expansion represents a major fraction of FOXL2 variants, the case illustrates the limitations of current first‐tier approaches and highlights the need for more specifically targeted methodologies. Bunyan et al. described a fluorescent PCR‐based assay for detecting FOXL2 polyalanine expansions, providing a reliable and cost‐effective solution for identifying polyalanine expansions [31]. Beyond repeat expansions, additional classes of pathogenic variation may contribute to the unresolved fraction of BPES cases. Short‐read and long‐read WGS, as well as optical genome mapping, offer the ability to detect cryptic and complex SVs, including balanced rearrangements and long‐range regulatory disruptions affecting the FOXL2 cis‐regulatory landscape, which are not captured by current targeted assays. Moreover, most diagnostic workflows do not specifically interrogate the FOXL2 promoter, raising the possibility that pathogenic deletions or regulatory alterations in this region remain undetected and may also account for part of the missing heritability. Together, the integration of repeat‐sensitive assays, long‐read sequencing and genome‐wide SV detection approaches has the potential to substantially improve diagnostic yield, reduce the proportion of unsolved BPES cases and further refine the understanding of FOXL2 cis‐regulatory mechanisms, ultimately improving patient care and genetic counselling. 5. Conclusion In this study, we present a comprehensive overview and ACMG/AMP‐based classification of virtually all reported FOXL2 variants associated with BPES up to December 2024 and expand the variant spectrum by reporting 76 novel genetic defects. We showed that genotype–phenotype correlations for POI are not straightforward, making prediction of POI risk in young female patients with BPES particularly challenging. This is partly due to the lack of appropriate longitudinal clinical follow‐up, which is particularly relevant for prepubertal female patients. Here, we also propose that BPES should not be subclassified in two distinct subtypes but rather as a condition with a fully penetrant eyelid malformation, consistently present in all patients, with reduced penetrance and variable expressivity of POI in affected females. Looking forward, improved diagnostic strategies, including (targeted) long‐read sequencing and optical genome mapping to resolve variants in the polyalanine tract and short‐read and long‐read WGS to capture complex SVs and noncoding alterations, will be essential to increase the diagnostic yield and ultimately improve precision medicine in BPES. Author Contributions Charlotte Matton: conceptualization, data curation, formal analysis, investigation, visualization, writing of the original draft. Julie Van De Velde: data curation, formal analysis, investigation, review and editing. Marieke De Bruyne: data curation, formal analysis, review and editing. Stijn Van De Sompele: data curation, formal analysis, review and editing. Sally Hooghe: data curation, formal analysis, review and editing. Miriam Bauwens: review and editing. Hannes Syryn: review and editing. Eva D′haene: review and editing. Annelies Dheedene: formal analysis, review and editing. Martine Cools: review and editing. Wendy Watkins: formal analysis, review and editing. Shoko Komatsuzaki: formal analysis, resources, review and editing. Ewelina Preizner-Rzucidło: formal analysis, resources, review and editing. Alison Ross: formal analysis, resources, review and editing. Christine Armstrong: data curation, formal analysis, review and editing. David J. Bunyan: formal analysis, resources, review and editing. Andrew Shelling: formal analysis, resources, review and editing. Andrea L. Vincent: formal analysis, resources, review and editing. Sascha Vermeer: formal analysis, resources, review and editing. Hannah Verdin: data curation, formal analysis, review and editing. Elfride De Baere: conceptualization, funding acquisition, supervision, review and editing. Funding This research was supported by the Research Foundation Flanders (FWO/11Q3D24N to C.M. and 1802220N to E.D.B.). Conflicts of Interest The authors declare no conflicts of interest. Supporting Information Additional supporting information can be found online in the Supporting Information section. Supporting information Supporting Information 1 Table S1: All identified frameshift, nonsense, missense, in‐frame indel, and stop‐loss variants within the FOXL2 coding region that are associated with a blepharophimosis, ptosis, epicanthus inversus syndrome (BPES) phenotype.For each variant, the table provides the genomic coordinates, predicted protein effect, and variant classification according to the ACMG/AMP framework. The applied evidence criteria and corresponding arguments supporting each classification are also included. Supporting Information 2 Table S2: Overview of all structural variants (deletions) identified in the FOXL2 locus and its flanking regions, including intragenic FOXL2 deletions, combined FOXL2‐ATR deletions, and deletions located up‐ or downstream of FOXL2, including genomic coordinates (when available) and the associated phenotype. Supporting Information 3 Table S3: Overview of all translocations that disrupt or reposition the FOXL2 locus and are associated with a blepharophimosis, ptosis, epicanthus inversus syndrome (BPES) phenotype. These data highlight the contribution of structural chromosomal rearrangements to BPES and illustrate how disruption of FOXL2 or its regulatory landscape can lead to variable or more severe clinical manifestations. Supporting Information 4 Table S4: Overview of the primers that are used for Sanger sequencing in BPES, covering the entire FOXL2 gene. Supporting Information explains how genetic variants were classified using an ACMG/AMP‐based framework that considers seven types of evidence (population, genotype/phenotype, databases, computational, functional, segregation and allelic data) to determine pathogenicity.