A comprehensive review for defining cut-off values of oocyte, zygote, and embryo maturation arrest (OZEMA) due to maternal-effect genes: towards the establishment of clinical criteria

Abstract

STUDY QUESTION

Which clinical definition of oocyte, zygote, and embryo maturation arrest (OZEMA) merits genetic investigation?

SUMMARY ANSWER

The diagnosis OZEMA is considered when: only two mature oocytes are obtained from six cumulus–oocyte complexes (COCs), only one zygote is obtained from six metaphase II oocytes (MIIs), or no blastocysts are obtained from six two-pronuclei zygotes (2PNs).

WHAT IS KNOWN ALREADY

OZEMA is a cover term for various defects, including oocyte abnormalities, fertilization failure, cleavage arrest, and abnormal embryo development, often associated with specific genetic variants. While some variants in specific genes are associated with distinct phenotypes, others commonly display variable expressivity across cycles and patients. Currently, there are no standardized cut-off values defining the OZEMA disorder.

STUDY DESIGN, SIZE, DURATION

This is a pooled secondary analysis of aggregated data extracted from published studies. A PubMed search was conducted to identify peer-reviewed original studies in the human reporting genetically confirmed cases of OZEMA. Studies were included if patients had displayed OZEMA in at least two medically assisted reproduction (MAR) cycles. In total, clinical and genetic data were extracted from 132 publications, including 521 patients and 39 genes.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Clinical phenotypes from 132 publications were classified into oocyte maturation arrest (OMA), zona pellucida abnormality (ZPA), fertilization failure (FF), zygote arrest (ZA), early embryo arrest (EEA), mixed phenotypes, and other abnormalities. Data at both cycle and patient levels were evaluated using ratios corresponding to key developmental milestones (ratio of MII/COC, 2PN/MII, cleaved embryos/2PN, blastocysts/2PN). These ratios were compared with the expected values Vienna Consensus binomial modeling to assess whether the observed proportions were significantly lower than the established competency thresholds (P < 0.05).

MAIN RESULTS AND THE ROLE OF CHANCE

This review identified 39 OZEMA-associated genes. Analysis of 668 cycles from 253 patients carrying variants in 35 genes showed abnormal developmental outcomes in 93%, based on binomial modeling. Reliable diagnosis of OZEMA requires a sufficient number of COCs, MIIs, or 2PNs to distinguish true developmental impairment from normal biological variation. When these minimal thresholds are not reached, statistical power is insufficient to reliably confirm developmental arrest. Based on our findings, OZEMA is diagnosed when only two or fewer mature oocytes form from six COCs, one or no zygote forms from six MIIs, or no blastocysts form from six 2PNs.

LIMITATIONS, REASONS FOR CAUTION

This study is limited by its retrospective design and reliance exclusively on published data from women selected for poor IVF outcomes and the presence of variants in candidate OZEMA genes, which may introduce publication bias and reporting heterogeneity. Therefore, this study does not report on genetic yield from women with normal IVF outcomes. Variability in clinical protocols and genetic testing methods, limited ancestral diversity, and missing data across studies could contribute to confounding factors.

WIDER IMPLICATIONS OF THE FINDINGS

This study is an important step toward understanding the clinical and genetic spectrum of OZEMA, highlighting the variability of and overlap between affected subphenotypes. Establishing cut-off values against consensus oocyte and embryo developmental competency levels provides a benchmark for more accurate classification. Importantly, this represents the most comprehensive analysis of published OZEMA cases with genetic variants to date, strengthening the reliability of the proposed clinical thresholds. These findings may guide personalized treatment strategies and genetic counseling, ultimately improving outcomes in MAR.

STUDY FUNDING/COMPETING INTEREST(S)

No funding was received for this study. The authors declare no competing interests.

REGISTRATION NUMBER

Not applicable.

Introduction

Understanding the genetic basis of infertility is essential for accurate diagnosis and personalized treatment strategies in medically assisted reproduction (MAR). MAR laboratories predominantly receive patients with non-syndromic infertility. Non-syndromic forms of infertility, due to pathogenic variants in specific genes, disrupt gametogenesis, gonadal function, or reproductive tract physiology, but are not associated with additional systemic or syndromic phenotypes. Indeed, with rare exceptions, the severity of syndromic disorders of which infertility is just one symptom usually excludes these patients from reproduction. A notable example is cystic fibrosis, which presents with a broad spectrum of severity, ranging from isolated congenital bilateral agenesis of the vas deferens (CBAVD), leading to obstructive azoospermia, to life-threatening pulmonary and pancreatic involvement (Kaplan et al., 1968).

Furthermore, since the diagnosis of syndromic diseases is typically handled by specialized reference centers with the required expertise, genetic diagnosis with a focus on infertility is directed primarily toward non-syndromic cases. Genetic testing related to infertility is receiving increasing attention and is poised to become a routine component of modern MAR. It not only identifies the underlying causes of infertility but also enables personalized treatment strategies, ultimately improving the management of affected couples (Capalbo et al., 2021). To be relevant, genetic testing must be based on a strict clinical evaluation that defines precise inclusion and exclusion criteria for the phenotype under study. Such criteria are fundamental for a targeted approach that preselect genes potentially involved in a given phenotype.

For some infertility phenotypes, clinical diagnosis is straightforward. For example, azoospermia is defined by the World Health Organisation (WHO) as the absence of spermatozoa in the ejaculate, while primary ovarian insufficiency (POI) in females is defined as the loss of ovarian function before the age of 40 years (Panay et al., 2025). However, other phenotypes can be more challenging to define. A prime example is oocyte, zygote, or embryo maturation arrest (OZEMA). Initially, two distinct phenotypes, oocyte maturation defect (OOMD) and premature embryonic lethality (PREMBL) were considered. OOMD covered oocyte maturation arrest, fertilization abnormalities, degenerated and abnormal oocytes, while PREMBL was related to early embryo arrest. However, the Online Mendelian Inheritance in Man (OMIM; OZEMA—OMIM—(OMIM.ORG)), the reference online catalog of human genes and genetic disorders, recently reclassified OOMD and PREMBL as a single phenotype under the term OZEMA. This broader category includes developmental arrest at stages ranging from the oocyte to the blastocyst, including fertilization failure, oocyte degeneration, and morphological anomalies such as giant oocytes and abnormal polar bodies. While this classification is generally valid, mixed phenotypes are often observed, in which maturation defects, fertilization, and cleavage abnormalities can occur within the same cycle or across different cycles in the same patient, often reflecting a single underlying genetic etiology.

It is also important to recognize that fertilization defects and early embryo arrest can have a paternal genetic origin, however, these cases are recognized as a distinct male infertility phenotype in OMIM and are not included in the female OZEMA disorder (OMIM No. 620500, No. 620196, No. 617214, No. 619258, No. 620499, No. 618643). Six genes were identified for the fertilization failure (FF) phenotype in the latest systematic review for male infertility (Houston et al., 2021). Later, genetic variations in eight additional genes were reported for FF and early embryonic developmental arrest (EEA) (Supplementary Table S1). Determining whether the deficiency lies in sperm- or oocyte-related genes is crucial for tailoring the most effective treatment for patients with FF and/or EEA (Cardona Barberán et al., 2023). Notably, certain genes are associated with both spermatogenic failure and OZEMA in OMIM. For example, FBXO43 (OMIM No. 617214) illustrates this genetic overlap. While OMIM maintains distinct sex-specific entries, this classification may warrant reconsideration as additional evidence emerges.

In addition, a polygenic cause may also be plausible for the OZEMA disorder, particularly in cases lacking a single causative variant. However, polygenic traits involve several small-effect variants across multiple pathways, making it harder to elucidate mechanisms. Genome-wide association studies may help identify these polygenic forms of infertility. However, because individuals with infertility are less likely to pass on their genes, negative selection may reduce the prevalence of associated risk alleles in the population.

It is important to note that oocyte maturation involves not only nuclear maturation but also cytoplasmic maturation. This maturation process relies on the storage and fine regulation of the use of various types of RNA (mRNA, miRNA, and lncRNA) and proteins. Pathogenic variations can disrupt the synthesis, activation, modification, and/or degradation of RNA or proteins, leading to impaired function. Their consequences may not be immediate, and therefore remain undetectable during routine evaluation in a MAR cycle, yet may lead to consequent impaired embryonic development.

In some cases, defects in specific genes are exclusively associated with a certain phenotype as illustrated in our latest systematic review (Van Der Kelen et al., 2023). For example, BTG4 is linked solely to first cleavage arrest (Liu et al., 2021; Wang et al., 2023), while WEE2 is exclusively associated with fertilization failure (Sang et al., 2018; Yang et al., 2019; Zhou et al., 2019; Jin et al., 2021). Conversely, depending on the gene defect involved, the phenotype may exhibit variable expression, covering a part or even the entire OZEMA spectrum. For example: biallelic pathogenic variants in MEI1 are associated with FF and EEA (Dong et al., 2021); biallelic variants in MOS have been linked to FF, zygote arrest (ZA), EEA, mixed phenotypes, and other phenotypes (such as abnormal oocytes) (Zeng et al., 2022); and biallelic mutations in PADI6 have been implicated among others in oocyte maturation arrest (OMA), FF, and EEA (Xu et al., 2016).

To establish precise inclusion and exclusion criteria for genetic testing of this phenotype, we conducted a comprehensive review of the literature on the female origin of OZEMA with the aim of better defining phenotype–gene relationships. We focused on the monogenic female genetic contribution to the OZEMA disorder. Specifically, our objectives were: (i) to identify which genes are associated with the full-spectrum OZEMA disorder (i.e. cases showing all recognized OZEMA subphenotypes), (ii) to identify which genes are linked to partial forms involving one or two specific subphenotypes, and (iii) to define cut-off values for each subgroup within the OZEMA disorder.

This work is an initial attempt to establish defined clinical thresholds for OZEMA, providing inclusion and exclusion criteria that can be applied in clinical and laboratory settings. These findings will facilitate the work of medical geneticists in genetic diagnostic laboratories and guide clinicians in recognizing which patients are most likely to benefit from genetic testing. Moreover, the findings may help guide personalized treatments and genetic counseling tailored to the underlying molecular defect and improve research efforts.

Materials and methods

Literature search, data extraction, and curation

We filtered the genes assessed in our systematic review and listed in FeRGI database for the OZEMA disorder to include articles until 1 November 2021 (https://www.eshre.eu/Specialty-groups/Special-Interest-Groups/Reproductive-Genetics/FeRGI/Database) (Van Der Kelen et al., 2023). To capture articles published between 1 November 2021 and 5 November 2024, we performed a comprehensive literature search in PubMed (https://pubmed.ncbi.nlm.nih.gov/) using Medical Subject Headings (MeSH) terms and relevant keywords describing cases of OZEMA in which genetic defects explain the phenotype (Table 1). Only original research studies published in peer-reviewed, English language journals with full-text access and reporting on human cases were included. Review articles, studies involving only animal models, and cell-based studies were excluded (Fig. 1).

Table 1.

Search terms for OZEMA up to 1 November 2021, including Medical Subject Headings (MeSH) terms combined with keywords related to the OZEMA phenotype.

PubMed Search terms (10 643 results)	“infertility, female/genetics”[MeSH Terms] OR  “genetics”[MeSH Terms] OR “genetics”[MeSH Subheading] OR “genetics”[All Fields] OR “genetic”[All Fields] OR  “Genes”[MeSH Terms] OR “gene”[All Fields] OR “Genes”[All Fields] OR  “genetic variation”[MeSH Terms] OR “Indel”[All Fields] OR “SNV”[All Fields] OR “CNV”[All Fields] OR “VUS”[All Fields] OR “variant”[All Fields] OR “variant s”[All Fields] OR “variants”[All Fields] OR “mutation”[All Fields] OR “mutations”[All Fields] OR “mutation s”[All Fields] OR  “Chromosomes, Human, X/genetics”[MeSH Terms] OR “X chromosome”[All Fields] OR “chromosome X”[All Fields] OR “X-chromosome”[All Fields] OR  “Gene deletion”[MeSH Terms] OR “gene deletion”[All Fields] OR “gene deletions”[All Fields] OR  “Gene Duplication”[MeSH Terms] OR “Gene Duplication”[All Fields] OR “gene duplications”[All Fields] AND  “infertility, female”[MeSH Terms] OR (“infertility”[All Fields] AND “female”[All Fields]) OR “female infertility”[All Fields] OR (“infertility”[All Fields] AND “female”[All Fields]) OR “infertile female”[All Fields] OR  “Meiosis”[MeSH Terms] OR “Meiosis”[All Fields] OR  “Oogenesis”[MeSH Terms] OR “Oogenesis”[All Fields] OR  ((“Oocytes”[MeSH Terms] OR “Oocytes”[All Fields]) AND “maturation failure”[All Fields]) OR “maturation defect”[All Fields] OR “maturation arrest”[All Fields] OR  “gonads/growth and development”[MeSH Terms] OR (“gonads”[All Fields] AND “growth”[All Fields] AND “development”[All Fields]) OR  “Gametogenesis”[MeSH Terms] OR “Gametogenesis”[All Fields] OR  ((“Fertilization”[MeSH Terms] OR “Fertilization”[All Fields]) AND “failure”[All Fields]) OR “defect”[All Fields] OR  (“fertility/genetics”[MeSH Terms] OR “Fertility/genetics”[All Fields] OR “fertility”[All Fields]) AND “genetics”[All Fields] OR  “Embryo Implantation”[MeSH Terms] OR “Embryo Implantation”[All Fields] OR (“Embryo Implantation”[All Fields] AND “human”[All Fields]) OR  (“Embryonic Development”[MeSH Terms] OR “Embryonic Development”[All Fields]) AND (“defects”[All Fields] OR “defect”[All Fields] OR “defect s”[All Fields] OR “arrest”[All Fields] OR “arrests”[All Fields] OR “arrest s”[All Fields] OR “failure”[All Fields]) AND (“human s”[All Fields] OR “humans”[All Fields] OR “human”[All Fields]) OR  “oocyte”[All Fields] OR “zygote”[All Fields] OR “embryo”[All Fields]) AND “maturation arrest”[All Fields]) OR “maturation defect”[All Fields] OR  (“Genetic Testing”[MeSH Terms] OR “Genetic Testing”[All Fields]) AND (“infertility”[All Fields] OR “fertility”[All Fields]) OR  “monogenic”[All Fields] AND “infertility”[All Fields] OR  “embryo lethality”[All Fields] OR  “maternal effect genes” [All fields] NOT  “Preimplantation Diagnosis”[MeSH Terms] OR “Prenatal Diagnosis”[MeSH Terms] OR “Chromosome Aberrations”[MeSH Terms] OR “Genetic Association Studies”[MeSH Terms] OR “heart defects, congenital”[MeSH Terms] NOT  “Review”[Publication Type] AND  (“humans”[MeSH Terms] AND “female”[MeSH Terms] AND “english”[Language]) AND  2021/11/01:2024/11/05[Date—Publication] Filters : Human-Female-English-full text-medline
Inclusion criteria	- Published in peer-reviewed journal - Publications in English - Studies concentrating on female infertility OZEMA phenotype - Studies conducted in human patients - Studies concentrating on finding a genetic cause - Studies presenting a monogenic form of infertility - Studies presenting original data
Exclusion criteria Title/abstract screening	- Publication not in English - Studies investigating non-human species; i.e. animals/plants/fungi/prokaryotes - Studies includes one of the following examples:   • Male infertility   • Studies investigating polygenic or multifactorial forms of infertility   • Chromosomal anomalies   • mtDNA anomalies   • variants described in population databases   • Epigenetics/methylome   • Polymorphisms   • Gene expression
Exclusion criteria full text screening	- Studies investigating genetic risks factors or associations - Studies presenting polymorphisms - Deletion or duplications of multiple genes - Studies presenting chromosomal aneuploidies or rearrangements (if break point does not disturb a gene) - Full text not available

Figure 1.

Overview of data search, selection criteria, and analytical workflow to define clinical thresholds for OZEMA. OZEMA, oocyte, zygote, and embryo maturation arrest; COCs, cumulus–oocyte complexes; MII, metaphase II oocyte; 2PN, two-pronuclei zygote.

Data extraction was performed for all included articles, capturing information on the gene name, gene variant, and clinical MAR data. The extracted data included the numbers of cumulus–oocyte complexes (COCs) retrieved, mature oocytes (MII) obtained, two pronuclei zygotes (2PNs), cleaved embryos, and transferable blastocysts. This information was collected both at the individual cycle level (cycle-based data) and aggregated across all cycles for each patient (patient-based data). Population databases were used solely to annotate allele frequencies of previously reported variants and were not applied as exclusion criteria or for variant reclassification.

Patients were included only if OZEMA was observed in at least two MAR cycles. This criterion was established to minimize the inclusion of individuals who do not consistently exhibit the OZEMA disorder. Furthermore, cycles or patients with incomplete clinical data were excluded from the analysis.

The clinical phenotypes from all cycles and patients were evaluated and categorized into several sub-phenotypes for subsequent analysis: OMA, zona pellucida abnormalities (ZPA), FF, ZA, EEA, mixed phenotype, and others. The mixed phenotype category includes cases in which at least two of the following, OMA, FF, ZA, and EEA, occur within the same cycle or patient. The ‘other’ category includes cases such as empty follicle syndrome (EFS), abnormal oocytes (i.e. large polar body, giant oocytes), or degenerated oocytes. An arbitrary threshold of four events, based primarily on clinical experience, was used as a cut-off in order to ensure meaningful observations of maturation arrest at various developmental stages. When fewer than four follicles, COCs, MIIs, or 2PNs are available, the sample size is too small to reliably diagnose maturation arrest. For example, if only two COCs are retrieved and both are immature, it is uncertain whether this reflects true OZEMA or simply random variation. As a result, cycles or patients were excluded if fewer than four follicles were present in case of EFS, if fewer than four COCs were retrieved in cases of OMA, if fewer than four MII in cases of FF, or if fewer than four 2PNs in cases of ZA or EEA.

Data analysis

The cycle-based data and the patient-based data were analyzed separately. Patient-based ratios were derived by aggregating data across all available cycles for each individual. For each sub-phenotype, specific ratios were calculated for every cycle and for every patient. For OMA, the MII/COC ratio was determined; for FF, the 2PN/MII ratio was assessed; for ZA, the cleaved embryos/2PN ratio was observed; and for EEA, the usable blastocysts/2PN ratio was determined. In case of a mixed phenotype, all aforementioned ratios were calculated for each cycle and patient.

We applied the binomial test to the different ratios, using competency values based on the Vienna Consensus (ESHRE Special Interest Group of Embryology and Alpha Scientists in Reproductive Medicine, 2017), which defines minimum acceptable levels and benchmark values for oocyte-related, fertilization, embryo development, embryo transfer, cryopreservation, and laboratory-specific key performance indicators.

The following acceptable levels and benchmark values were applied.

• Maturation rate = number of mature oocytes (MII)/number of COCs retrieved: expected proportion is 75%.

• Fertilization rate = number of 2PN zygotes/number of MII used for IVF/ICSI: expected proportion is 60%.

• Zygote development rate = number of cleaved embryos/number of 2PN zygotes: expected proportion is 95%.

• Early embryo development rate = number of blastocysts/number of 2PN zygotes: expected proportion is 40%.

To evaluate whether the observed proportions deviated significantly from these expected benchmark values, we applied a one-sided binomial test. A significance level of 5% (alpha = 0.05) was considered statistically significant. The binomial test was used as a screening tool to identify cycles/patients whose observed proportions fell significantly below guideline reference rates (Fig. 1). These classifications were used for descriptive hypothesis-generating analyses of gene involvement rather than for formal inferential testing.

Results

The genetic basis of OZEMA in women

The establishment of causal links between gene dysfunction and human diseases requires the integration of multiple independent levels of evidence. Gene–disease relationships (GDRs) are connections between specific gene variants and the diseases or conditions they cause. The assessment of GDR validity involves a systematic evaluation of the strength, consistency, and reproducibility of the available data supporting the evidence based on the guideline (Smith et al., 2017), thereby determining the reliability of the proposed genetic association.

In our systematic review that collected data on the genetics of female infertility up until 1 November 2021, we analyzed 24 GDRs for the OZEMA disorder (Van Der Kelen et al., 2023). Among 24 GDRs scored in the systematic review, 16 had a moderate to definitive score, which is the limit to be included in a diagnostic gene panel, 8 had a limited score. After our additional literature search between 1 November 2021 and 5 November 2024, we have now identified 39 genes, including 15 novel genes from 48 new articles published after 1 November 2021. Of these genes, 28 had an autosomal recessive (AR) and 8 had an autosomal dominant (AD) mode of transmission, while 3 genes had both AR and AD inheritance (Supplementary Table S2).

Characterization of the studies included

Cycle based

A total of 786 ART cycles were initially identified. After applying exclusion criteria, only 381 cycles were included in the final analysis. There were 84 cycles excluded because they did not meet the minimum threshold of four COCs/MIIs/2PNs retrieved, 96 cycles were excluded because they concerned only a single cycle, 10 cycles lacked full-text availability or were not published in English, or were mentioned in a review article, and 215 had incomplete clinical data.

The 381 included cycles were performed in 176 patients and published in 80 distinct articles. These included 85 cycles associated with OMA, 95 cycles with ZPA, 57 with FF, 27 with ZA, 79 with EEA, 24 with mixed phenotypes, and 14 with other phenotypes (Table 2).

Table 2.

Summary of study parameters and phenotype distributions in cycle-based and patient-based analyses.

Study parameters	Cycle-based	Patient-based
Articles (n.)	80 (132)	93 (132)
Cycles (n.)	381 (786)	668 (–)
Patients (n.)	176 (416)	253 (521)
Genes (n.)	35 (39)	35 (39)
Phenotype
Oocyte maturation arrest (OMA) (n. (%))	85 (22%)	92 (36%)
Zona pellucida abnormality (ZPA) (n. (%))	95 (25%)	37 (15%)
Fertilization failure (FF) (n. (%))	57 (15%)	27 (11%)
Zygote arrest (ZA) (n. (%))	27 (7%)	13 (5%)
Early embryo arrest (EEA) (n. (%))	79 (21%)	37 (15%)
Mixed phenotype (n. (%))	24 (6%)	29 (11%)
Other phenotype (n. (%))	14 (4%)	18 (7%)

The table presents the number of articles, cycles, patients, and genes included, alongside the frequency of observed phenotypes: oocyte maturation arrest (OMA), zona pellucida abnormalities (ZPA), fertilization failure (FF), zygote arrest (ZA), early embryo arrest (EEA), mixed phenotypes, and other phenotypes. For the study parameters, two numbers are shown: the first (outside brackets) indicates values after applying exclusion criteria (included in the final analysis), and the second (in brackets) indicates values before applying exclusion criteria. The total number of cycles identified for patient-based calculation could not be given, due to incomplete and unavailable data.

Patient based

A total of 521 patients were initially considered, with 253 patients described in 93 articles, included after applying the exclusion criteria. There were 268 patients excluded for the following reasons: full-text access was lacking for 8 patients, 124 patients exhibited OZEMA in only one cycle, 129 patients had incomplete clinical data, and 6 patients did not meet the minimum threshold of four COCs/MIIs/2PNs retrieved and 1 patient was described in a non-English article.

The 253 included patients collectively underwent 668 cycles and carried variants in 35 different genes. The number of cycles was higher in the patient-based approach, as in some studies, clinical information was reported only per patient rather than per cycle. In addition, more cycles were excluded in the cycle-based approach than in the patient-based approach because the minimum threshold of four COCs, MIIs, or 2PNs was not reached.

Among the patients included in this retrospective analysis, the most prevalent subphenotype was OMA (n = 92, 36%) followed by ZPA (n = 37, 15%), EEA (n = 37, 15%), FF (n = 27, 11%), and ZA (n = 13, 5%). Interestingly, the mixed phenotype category represented 29 (11%) of patients, while 18 (7%) exhibited another subphenotype of OZEMA, underscoring the complexity of the OZEMA disorder (Table 2, Fig. 2).

Figure 2.

Overview of the main OZEMA disorder. Each box summarizes key characteristics including the number of patients affected, percentage of abnormal maturation or development based on the binomial model presented in this study, and representative genes implicated in each phenotype. The dark green box presents an overview of all OZEMA subtypes. OZEMA, oocyte, zygote, and embryo maturation arrest.

For ease of reading, we chose to present the data according to the stage of development: OMA, ZPA, FF, ZA, EEA, and mixed and other phenotypes.

Two genes, MEI1 and REC114, were excluded due to incomplete data. Both genes were reported in only one article related to OZEMA. Because the articles did not provide complete clinical information, we were unable to determine the number of COCs, MII oocytes, 2PN fertilizations, cleavage rates, or blastocyst rates for these genes. TUBA1C gene was also excluded, as it was reported in only one article for which full-text access was unavailable. In addition, for TUBA4A, although two articles were identified, one lacked sufficient clinical data and the other was not available in full-text reports; both articles were therefore excluded.

Following the application of the binomial test to patient-level data, 93% of patients met the established criteria for OZEMA. This finding supports the use of statistical thresholds to guide targeted genetic testing in infertility diagnostics and provides a basis for the phenotype-specific analyses presented below. Gene variants and related subphenotypes are listed in Supplementary Table S2.

Oocyte maturation arrest

OMA was associated with variants in 11 different genes: CCNB3, CDC20, ENSA, LHX8, MAD2L1BP, PABPC1L, PATL2, TBPL2, TRIP13, TUBB8, and ZFP36L2. Using the binomial model based on the Vienna consensus values, we determined cut-off values for the maximum number of MII oocytes relative to the total number of COCs retrieved, in order to achieve statistical significance for defining the OMA phenotype. For example, applying our statistical model, when six COCs are retrieved, OMA can only be referred to if fewer than or equal to two MII oocytes are present, whereas when 10 COCs are obtained, fewer than or equal to four MII oocytes must be present. Table 3 summarizes the maximum number of MII oocytes relative to the total number of COCs retrieved to achieve statistical significance in the case of OMA.

Table 3.

Cut-off values for the maximum number of metaphase II (MII) oocytes relative to the total number of cumulus–oocyte complexes (COCs) retrieved for the diagnosis of oocyte maturation arrest (OMA), based on the Vienna Consensus.

N. collectedoocytes	Max N. mature oocytes for a significantly low rate
1	–
2	–
3	0
4	0
5	1
6	2
7	2
8	3
9	4
10	4
11	5
12	5
13	6
14	7
15	7
16	8
17	9
18	9
19	10
20	11
21	11
22	12
23	13
24	13
25	14
26	15
27	15
28	16
29	17
30	17
31	18
32	19
33	20
34	20
35	21

Statistical analysis is not meaningful when two or fewer COCs are retrieved due to insufficient sample size.

Nearly all reported cases of OMA in the literature (98%) showed a statistically significant reduction in the ratio of MII oocytes to COCs retrieved (P < 0.05) (Fig. 2). The only exceptions were two cycles with a borderline low oocyte yield (one mature oocyte out of four, and two mature oocytes out of five). Similarly, when analyzing data at the patient level rather than by cycle, only two OMA patients (one patient with two mature oocytes out of four, and one patient with two out of five) did not meet statistical significance based on the binomial statistical model. These findings underscore the critical impact of oocyte yield on the statistical evaluation of maturation arrest and highlight the necessity of retrieving a minimum number of oocytes to avoid mislabeling of OMA due to low oocyte numbers.

Zona pellucida abnormality

ZPA is associated with variants in only four genes: ZP1, ZP2, ZP3, and ZP4. Clinically, patients with ZPA frequently present empty follicle syndrome (EFS), characterized by the failure to retrieve oocytes despite the presence of apparently mature follicles and apparently successful ovarian stimulation as well as final follicle maturation that meets expected clinical criteria. Due to the limited availability of data, particularly regarding the count and size of follicles preceding the ovarian puncture, we were unable to perform any statistical analysis (Fig. 2).

Fertilization failure

FF was associated with variants in five different genes: ASTL, CDC20, PATL2, TUBB8, and WEE2. Using the binomial model, based on the Vienna consensus values, we established statistically significant cut-off values for the maximum number of 2PNs relative to the number of MII oocytes obtained for defining the FF phenotype. For example, if six MII oocytes are obtained, fewer than two 2PNs are required, while with 10 MII oocytes, fewer than three 2PNs are required to statistically refer to FF. Table 4 summarizes the maximum number of 2PN zygotes in relation to the total number of MII collected to achieve statistical significance in the case of FF.

Table 4.

Cut-off values for the maximum number of two pronuclei (2PNs) zygotes relative to the total number of metaphase II (MII) oocytes retrieved for the diagnosis of fertilization failure (FF), based on the Vienna Consensus.

N. MII oocytes	Max N. fertilized oocytes for a significantly low rate
1	–
2	–
3	–
4	0
5	0
6	1
7	1
8	1
9	2
10	2
11	3
12	3
13	4
14	4
15	5
16	5
17	6
18	6
19	7
20	7
21	8
22	8
23	9
24	9
25	10
26	10
27	11
28	11
29	12
30	13
31	13
32	14
33	14
34	15
35	15

Statistical analysis is not meaningful when three or fewer MIIs are available due to insufficient sample size.

Out of the 57 cycles, 18 did not meet the criteria for a statistically low fertilization rate (2PNs/MIIs) (Fig. 2). However, among 27 patients classified with FF due to OZEMA, only one patient did not exhibit a statistically low fertilization rate, while 26 (96%) met the criteria.

Zygote arrest

The ZA phenotype constitutes only a small fraction of the total reported cases of OZEMA. This highlights the rarity of the ZA phenotype within the broader spectrum of OZEMA presentations. This phenotype was associated with variants in three different genes: BTG4, CHEK1, and RGS12.

Table 5 summarizes the maximum number of zygotes required, according to the binomial model, based on the Vienna consensus values, relative to the number of 2PN oocytes, to achieve statistical significance for defining the ZA phenotype. For example, if 10 2PN are obtained, statistically speaking, seven or fewer cleavage embryos are needed to be able to refer to ZA.

Table 5.

Cut-off values for the maximum number of cleavage embryos relative to the total number of two pronuclei (2PNs) zygotes retrieved for the diagnosis of zygote arrest (ZA), based on the Vienna Consensus.

N. 2PN	Max N. cleavage embryos for a significantly low rate
1	–
2	0
3	1
4	2
5	3
6	4
7	5
8	5
9	6
10	7
11	8
12	9
13	10
14	11
15	12
16	13
17	13
18	14
19	15
20	16
21	17
22	18
23	19
24	20
25	21
26	22
27	23
28	24
29	24
30	25
31	26
32	27
33	28
34	29
35	30

Statistical analysis is not meaningful when only one 2PN is available for defining zygote arrest (ZA) due to insufficient sample size.

In all published studies on cycles and patients classified with the ZA phenotype, we applied the binomial model to assess the presence of OZEMA. In every case (13 patients, 100%), the observed number of embryos was lower than the number expected from the observed 2PN count. This indicates a developmental problem in the progression from 2PN stage to cleavage stage embryos (Fig. 2).

Early embryo arrest

EEA was associated with variants in 14 different genes: FBXO43, KHDC3L, KPNA7, LHX8, MOS, NLRP2, NLRP5, NLRP7, NLRP14, OOEP, PADI6, SYPR4, TLE6, and TUBB8.

The binomial model, based on the Vienna consensus values, allowed us to determine thresholds for the maximum number of blastocysts required, based on the number of 2PN zygotes obtained, in order to achieve statistical significance for defining EEA (Table 6). For example, if six 2PNs are obtained, absence of blastocysts indicated EEA, while when 10 2PNs are obtained, one or zero blastocyst-indicated EEA.

Table 6.

Cut-off values for the maximum number of blastocytes relative to the total number of two pronuclei (2PNs) zygotes retrieved for the diagnosis of early embryo arrest (EEA), based on the Vienna Consensus.

N. 2PN	Max N. blastocysts for a significantly low rate
1	–
2	–
3	–
4	–
5	–
6	0
7	0
8	0
9	0
10	1
11	1
12	1
13	1
14	2
15	2
16	2
17	3
18	3
19	3
20	3
21	4
22	4
23	4
24	5
25	5
26	5
27	6
28	6
29	6
30	7
31	7
32	7
33	8
34	8
35	8

Statistical analysis is not meaningful when five or fewer 2PNs are available for defining early embryonic arrest (EEA) due to insufficient sample size.

Among 79 cycles, 19 did not meet the criteria for a statistically low blastocyst rate (blastocysts/2PNs) (Fig. 2). However, among the 37 patients classified with EEA due to OZEMA, only 4 patients did not exhibit a statistically low blastocyst rate based on the binomial model, while 33 (89%) met the criteria.

Mixed phenotype

A mixed phenotype is characterized by the co-occurrence of two or more developmental arrest patterns: OMA, FF, ZA, and EEA. The most common combination observed was OMA and FF, occurring in 14 of the 29 patients (48% of the mixed group). This phenotype is associated with variants in seven different genes: CCNB3, CDC20, CIP2A, PABPC1L, PADI6, PATL2, and TUBB8.

Two patients (7%) exhibited the full combination of all four phenotypes: OMA, FF, ZA, and EEA. Four patients (14%) presented with OMA, FF, and EEA, while three patients (10%) had a combination of FF and EEA. Two patients (7%) had OMA, ZA, and EEA, and one patient (3%) showed a combination of OMA and EEA.

A total of 68 different phenotype-specific ratios were calculated within the mixed phenotype group. Of these, 18 ratios (27%) did not reach statistical significance (Fig. 2), while 50 (73%) did meet the criteria. Notably, 15 of the 29 patients (52%) exhibited significantly low ratios across all included phenotypes. In contrast, 9 patients (31%) had a significantly low ratio for only one phenotype. The remaining 5 patients (17%) showed a combination of at least three phenotypes, of which at least two had significantly low ratios, thereby supporting their classification within the mixed phenotype group.

Other phenotypes

While OZEMA is often described in terms of the absence of mature oocytes, fertilization failure, or arrest during embryonic development, additional, less commonly reported phenotypes also fall within the spectrum of OZEMA. Pathogenic variants in the LHCGR gene, known to cause empty follicle syndrome, were identified in three patients. One patient carried a pathogenic variant in MOS, resulting in abnormal oocytes characterized by an enlarged polar body. Seven patients harbored pathogenic variants in PANX1, associated with degeneration of retrieved oocytes or 2PNs. Variants in PATL2 (three cases) and TBPL2 (one case) were also linked to oocyte degeneration. Finally, three patients carrying TUBB8 variants presented with morphologically abnormal oocytes (Fig. 2).

Discussion

Over the past two decades, genetic research has made significant progress in identifying monogenic causes of human infertility in both men and women. Several hundred genes have been identified (Houston et al., 2021; Van Der Kelen et al., 2023), leading to the development of potentially powerful diagnostic tools such as in silico gene panels.

In recent years, multiple laboratories have begun offering genetic diagnostic services to MAR centers, enabling them to assess etiologies with more precision. These analyses cover a wide range of infertility phenotypes in both men and women. However, as with any genetic diagnostic process, accurate diagnosis depends on a rigorous phenotypic definition.

In the context of infertility, the criteria for categorizing phenotypes remain a topic of ongoing discussion particularly for qualitative phenotypes, which are more complex to define. For instance, in cases of gamete absence, such as azoospermia, the phenotype is clearly defined. In contrast, OZEMA still lacks well-established clinical diagnostic criteria.

In the present article, we accurately define the inclusion criteria for the various subphenotypes, specifically OMA and EEA, of the OZEMA disorder in female patients through a comprehensive literature review. The OZEMA classification also includes FF, first cleavage (zygote) arrest, degeneration occurring before or after ICSI, and abnormal oocyte morphology.

The Vienna Consensus (ESHRE Special Interest Group of Embryology and Alpha Scientists in Reproductive Medicine, 2017) sets benchmarks and competency values for procedures such as oocyte retrieval, fertilization, embryo development, and cryopreservation. Based on the Vienna consensus criteria and binomial statistical modeling, we established quantitative thresholds for each OZEMA subphenotype. In particular, we identified threshold ratios below which the phenotype can be considered established beyond the role of chance. Because the classification thresholds are based on binomial tests that do not model within-cycle correlation, these findings should be interpreted as exploratory. Future studies with richer metadata should apply hierarchical or beta-binomial models for confirmatory inference.

In this study, two approaches to define the minimal sample size for a reliable OZEMA diagnosis were considered. First, we applied a pragmatic threshold of four follicles, COCs, MIIs, or 2PNs, which is mainly based on clinical experience and routine practice. This arbitrary cut-off serves to filter out chance observations: when fewer than four samples are present, the likelihood that developmental arrest is due to random variation increases, and diagnostic certainty drops. Second, we applied statistical calculations to data derived from the literature, obtaining phenotype-specific minimal sample sizes for reliable assessment. These evidence-based thresholds differ per arrest type: for example, at least three COCs for OMA, four MIIs for FF, two 2PNs for ZA, and six 2PNs for EEA must be obtained to reliable diagnose OZEMA (Table 7, Fig. 1). These numbers reflect statistical requirements to ensure diagnostic reliability for each specific phenotype and can differ from the clinically chosen threshold. Thus, while the general cut-off of four provides a practical rule for everyday use, the more nuanced statistical minima increase certainty and help tailor assessment to each arrest phenotype.

Table 7.

Cut-offs providing an evidence-based framework to standardize diagnosis and minimize misclassification due to insufficient sample sizes, thereby enhancing the accuracy of genetic and clinical evaluation in reproductive medicine.

Developmental Stage	Minimum Sample Size Required
Oocyte Maturation Arrest (OMA)	≥ 3 COCs
Fertilization Failure (FF)	≥ 4 MIIs
Zygote Arrest (ZA)	≥ 2 2PNs
Early Embryonic Arrest (EEA)	≥ 6 2PNs

Concerning exclusion criteria, and due to the considerable variability in ovarian stimulation responses, we propose that the phenotype must be observed in at least two separate stimulation cycles to ensure diagnostic accuracy.

This study aimed to identify gene variants linked to the full OZEMA disorder and to partial forms limited to one subphenotype. Our findings show that some gene variants are highly phenotype-specific, while others act across multiple OZEMA presentations (Supplementary Table S2).

Variants in ZP1, ZP2, ZP3, and ZP4 were confined to ZPA, supporting a mechanism restricted to the zona pellucida. In contrast, PATL2 variants occurred in OMA, FF, mixed forms, and degenerative oocytes, whereas TUBB8 variants spanned OMA, FF, EEA, mixed forms, and phenotypes such as large polar bodies and multiple pronuclei. These patterns indicate that PATL2 and TUBB8 variants disrupt core oocyte processes, producing broader phenotypic effects.

Given the challenges in managing patients with the OZEMA disorder, several clinical strategies have been investigated to optimize oocyte retrieval, fertilization outcomes, and embryo development. For instance, it has been shown that lower aspiration pressure (100 mmHg) facilitated oocyte retrieval, and ICSI is a preferred treatment for patients with abnormal ZP carrying ZP4 pathogenic variants (Wei et al., 2022). Similarly, dual trigger with GnRH agonist combined with a high dose of hCG and prolonging the interval between ovulation triggering and oocyte retrieval increased the chances of obtaining mature oocytes in patients with LHCGR mutations (Lu et al., 2019). Live births were reported in both cases.

Significant progress is also being made in the development of experimental therapeutic approaches. Treatment of oocytes with specific inhibitors, such as PF477736, which suppresses kinase activity, has been shown to overcome embryonic arrest in patients with CHK1 mutation (Zhang et al., 2024). In this study, PF477736 was added to the culture medium of donated frozen zygotes and fresh oocytes from patients with ZA carrying a causative CHK1 mutation. All zygotes treated with the inhibitor progressed beyond the one-cell stage, and two of them even developed into good-quality blastocysts, while the untreated ones remained at the 2PN stage. Wildtype cRNA injection to rescue oocyte maturation arrest, FF and/or EEA has also been attempted in women with causative pathogenic variants in WEE2, CDC20, and TRIP13, respectively (Sang et al., 2018; Zhang et al., 2020; Zhao et al., 2021). Complementary RNA-injected oocytes matured and formed 2PNs, and some developed into good-quality blastocysts, while in the control group, none of the oocytes could mature or fertilize. However, no blastocyst transfer after treatment with inhibitors or cRNA injection has been reported so far, and the potential risks remain largely unstudied.

FF and EEA may, in some cases, arise from paternal genetic factors. Genetic variations in six genes were identified for male FF phenotype in the latest systematic review for male infertility (Houston et al., 2021). Later, different genetic variations in eight genes were published for male FF and EEA (Supplementary Table S1). Determining whether the deficiency lies in sperm- or oocyte-related genes is crucial for tailoring the most effective treatment for patients with FF and/or EEA. ICSI combined with AOA has been reported to be the most effective treatment for patients with sperm-derived FF and/or EEA with mutations in DPY19L2, PLCZ1, ACTL7A, and ACTL9 (Kuentz et al., 2013; Kashir et al., 2012; Yan et al., 2020; Dai et al., 2021; Cardona Barberán et al., 2024). However, patients with oocyte-related factors apparently do not benefit from AOA (Cardona Barberán et al., 2023). Since the present article focuses on the female origin of OZEMA disorder, we did not analyze sperm-derived FF and/or EEA.

Germline nuclear transfer is another experimental treatment for FF and EEA after ICSI caused by oocyte factors. Theoretically, it can be performed at different stages of oocyte and embryo development; the two most commonly employed strategies are maternal spindle transfer and pronuclear transfer (Costa-Borges et al., 2023; Hyslop et al., 2025). So far, both techniques have been used for other indications than monogenic forms of female infertility.

Recently, maternal pre-pronuclear transfer from a patient with CHK1 mutation into enucleated fertilized oocytes obtained from healthy donors has been performed (Zhang et al., 2024). Two out of four reconstructed fertilized oocytes demonstrated the ability to undergo division and develop into morphologically normal blastocysts. Others did not exhibit cleavage with distinct pronuclei, even up to the third day of embryo development.

Although these results are encouraging, data in humans remain very limited as all these trials were experimental and preclinical. The effectiveness and safety should be carefully evaluated before further preclinical and clinical research is undertaken.

As the identification of gene defects paves the way for functional studies, new therapeutic approaches will likely be developed in the near future. However, as currently no effective treatment options are available in the majority of cases, patients should be advised to stop MAR attempts with their own oocytes and to consider alternatives such as oocyte donation or adoption.

It is important to note that the genetics of infertility represents a new paradigm in reproductive medicine. As in other areas of medical genetics, once a genetic variant is identified as the cause of a patient’s condition, the disorder becomes familial, as other family members may also be affected. In such cases, it is essential to consider not only the couple but also the patient’s family members. This underscores the need to offer genetic counseling not only to the patient but also to their relatives (Verpoest et al., 2023). The importance of genetic counseling and a comprehensive approach, extending beyond the individual patient to their family, marks a significant paradigm shift in the field of MAR.

This manuscript is based on published findings where pathogenic gene variants were identified for oocyte-derived OZEMA disorder. The retrospective design of this study and its reliance on published data exclusively from women selected for poor MAR outcomes as well as the presence of variants in candidate OZEMA genes, may introduce publication bias and reporting heterogeneity. The included studies differed substantially in terms of source data, regarding stimulation protocols, insemination techniques (classical IVF vs ICSI), and embryo culture systems. This heterogeneity can also introduce bias into pooled proportions and limit the generalizability of the findings, especially given the limited population diversity in the literature. Evaluating this in a different population would be challenging. Consequently, the applicability of the proposed thresholds to other populations will require validation in more diverse cohorts. In addition, exclusion of studies due to incomplete clinical or genetic reporting may not have been random and could have introduced further selection bias. Furthermore, although using an a priori pragmatic cutoff of ≥ four follicles, COCs, MIIs, or 2PNs provides a clinically convenient method for excluding low-yield cycles, this approach may also introduce selection bias by disproportionately excluding cycles associated with specific biological or clinical phenotypes. Since we based our study on published findings, the possibility of male involvement cannot be excluded totally for FF, ZA, and EEA.

Future studies should apply these OZEMA classification criteria in a prospective way to explore their predictive performance upon genetic findings. Furthermore, this study does not report data on genetic yield from women with normal IVF outcomes. Variability in clinical protocols, genetic testing methods, and data completeness across studies could contribute to confounding factors. Crucially, the criteria as they are described in this manuscript should be interpreted relative to the quality standards, laboratory conditions, and embryology results available at an individual fertility center, as published data cannot always be extrapolated.

Conclusion

This study derives statistically informed clinical thresholds to define and categorize different OZEMA sub-phenotypes that subsequently establish a framework for genetic testing guidance, while acknowledging the need for prospective validation of the genetic classification tool. This is a novel approach to address the qualitative nature of the clinical phenotype categorization of OZEMA. Genetic classification of infertility not only offers the potential to resolve previously idiopathic cases but also to personalize patient care. However, due to the novelty of this field, such personalization remains limited. That said, therapeutic options are beginning to emerge for both male and female genetic infertility. This study is the first to list a number of published monogenic variants and their associated OZEMA disorder and to comprehensively propose clinical criteria for defining OZEMA.

Supplementary data

Supplementary data are available at Human Reproduction online.

Data availability

Data underlying this manuscript can be obtained from the authors upon request.

Authors’ roles

A.V.D.K., O.O., W.V., and S.V. served as the primary investigators, and were responsible for study design, data interpretation, analysis, and manuscript writing. F.M. performed the statistical analysis. All other authors (M.S., B.J.H.D., E.J., A.C., C.L., G.K., E.K., K.S.) contributed to the final manuscript.

Funding

No funding was obtained for this work.

Conflict of interest

The authors have no conflict of interest to declare.