Proteomic and metabolomic alterations in follicular fluid associated with embryological outcomes in women with isolated ANA-positive infertility: a prospective multi-omics cohort study.

A total of 32 ANA-positive infertility patients (posANA group) and 38 ANA-negative infertility patients (negANA group) were included in the study, as detailed in Table  1 . The median baseline FSH level in the posANA group was approximately 1.5 IU/L higher than in the negANA group [7.505 (6.473, 8.293) vs. 6.195 (5.093, 7.117), P =  0.001]. The median TSH level was also higher by approximately 0.5 mIU/L in the posANA group [2.455 (1.912, 3.084) vs. 1.960 (1.585, 2.560), P =  0.034]. No significant difference was observed in AMH levels between the two groups [3.075 (2.110, 5.223) vs. 3.165 (2.388, 6.550), P =  0.243]. Table 1 Comparison of clinical characteristics between infertile patients with positive ANA titers and those with negative ANA titers posANA negANA Z/χ 2 value P Patients, n 32 38 - - Female age, years (IQR) 34.000 (30.750, 36.250) 33.000 (29.000, 35.000) −1.688 0.091 Male age, years 33.500 (31.750, 35.250) 32.000 (29.000, 34.000) −1.439 0.150 Duration of infertility, years 2.000 (1.000, 3.020) 2.000 (1.000, 3.000) −0.686 0.492 BMI, kg/m 2 22.200 (20.200, 25.100) 23.350 (21.500, 24.450) −0.724 0.469 bFSH, IU/L 7.505 (6.473, 8.293) 6.195 (5.093, 7.117) −3.254 0.001 bLH, IU/L 4.590 (3.360, 6.285) 3.760 (2.808, 5.385) −1.827 0.068 bE 2 , pg/mL 35.500 (28.000, 43.250) 30.500 (22.857, 42.877) −1.009 0.313 bP, ng/mL 0.455 (0.318, 0.632) 0.450 (0.305, 0.640) −0.071 0.944 AMH, ng/mL 3.075 (2.110, 5.223) 3.165 (2.388, 6.550) −1.167 0.243 bPRL, ng/mL 16.420 (11.575, 20.635) 14.200 (10.995, 16.795) −0.552 0.581 bT, ng/mL 0.360 (0.310, 0.440) 0.380 (0.220, 0.470) −0.253 0.801 TSH, mIU/L 2.455 (1.912, 3.084) 1.960 (1.585, 2.560) −2.116 0.034 Cr,μmol/L 49.000 (45.125, 54.250) 49.600 (44.725, 56.950) −0.678 0.498 BUN, mmol/L 4.230 (3.538, 5.400) 4.185 (3.775, 4.540) −0.590 0.555 TG, mmol/L 0.945 (0.552, 1.520) 0.960 (0.665, 1.520) −0.500 0.617 TC, mmol/L 4.745 (4.130, 5.338) 4.800 (4.420, 5.190) −0.487 0.626 FPG, mmol/L 5.300 (4.975, 5.645) 5.200 (5.050, 5.420) −0.480 0.631 AFC, n 12.000 (9.750, 14.000) 14.500 (12.000, 18.000) −2.371 0.018 Duration of Gn, days 9.000 (8.000, 9.000) 9.000 (8.000, 9.000) −0.411 0.681 Total Gn dose, IU (IQR) 2250.000 (1912.500, 2437.500) 2162.500 (1812.500, 2400.000) −0.969 0.333 E 2 on trigger day, pg/mL 2866.000 (2178.250, 3921.750) 3776.500 (2469.250, 4999.250) −1.644 0.100 LH on trigger day, IU/L 2.410 (1.662, 3.373) 2.190 (1.353, 2.960) −1.138 0.255 P on trigger day, ng/mL 0.770 (0.525, 1.215) 1.290 (0.720, 1.667) −2.187 0.029 Sperm concentration, 10 6 /mL 60.200 (46.700, 98.870) 77.825 (51.392, 101.170) −0.679 0.497 Spermatozoa with normal morphology, % 3.830 (3.325, 5.345) 3.920 (3.380, 4.688) −0.241 0.810 Progressive motile sperm, % 44.570 (32.165, 56.440) 47.865 (37.555, 56.915) −0.440 0.660 Infertility type, % 1.161 0.281  Secondary 17 (53.125) 25 (65.789)  Primary 15 (46.875) 13 (34.211) Insemination method, % - 0.313 *  ICSI 6 (18.750) 8 (21.053)  IVF 23 (71.875) 30 (78.947)  IVF + Rescue ICSI 2 (6.250) 0 (0.00) Rescue ICSI 1 (3.125) 0 (0.00) Data are presented as median (interquartile range), or number (percentage) P <  0.05: significant statistical difference Abbreviations : BMI Body Mass Index, AMH Anti-Mullerian hormone, AFC Antral follicle count, bFSH basic level of follicle-stimulating hormone, bLH basic level of luteinizing hormone, bE 2 basic level of estradiol, bP basic level of pregestational hormone, bPRL basic level of prolactin, bT basic level of testosterone, TSH Thyroid stimulating hormone, Cr Creatinine, BUN Blood urea nitrogen, TC Total cholesterol, TG Triglycerides, FPG Fasting plasma glucose, Gn Gonadotropin, IQR Interquartile range * Fisher’s exact test Comparison of clinical characteristics between infertile patients with positive ANA titers and those with negative ANA titers Data are presented as median (interquartile range), or number (percentage) P <  0.05: significant statistical difference Abbreviations : BMI Body Mass Index, AMH Anti-Mullerian hormone, AFC Antral follicle count, bFSH basic level of follicle-stimulating hormone, bLH basic level of luteinizing hormone, bE 2 basic level of estradiol, bP basic level of pregestational hormone, bPRL basic level of prolactin, bT basic level of testosterone, TSH Thyroid stimulating hormone, Cr Creatinine, BUN Blood urea nitrogen, TC Total cholesterol, TG Triglycerides, FPG Fasting plasma glucose, Gn Gonadotropin, IQR Interquartile range * Fisher’s exact test The AFC in the negANA group was significantly higher than in the posANA group [14.500 (12.000, 18.000) vs. 12.000 (9.750, 14.000), P =  0.018]. Correspondingly, the negANA group exhibited higher P levels on trigger day [1.290 (0.720, 1.667) vs. 0.770 (0.525, 1.215), P =  0.029], and a trend towards higher peak estradiol (E 2 ) levels [3776.500 (2469.250, 4999.250) vs. 2866.000 (2178.250, 3921.750), P =  0.100]. There were no significant differences between the groups in total Gn dose [2250.000 (1912.500, 2437.500) vs. 2162.500 (1812.500, 2400.000), P =  0.333] or Gn stimulation duration [9.000 (8.000, 9.000) vs. 9.000 (8.000, 9.000), P =  0.681]. However, the posANA group had significantly fewer oocytes retrieved [8.500 (6.000, 10.000) vs. 12.500 (9.000, 15.000), P <  0.001], fewer 2PN embryos [6.000 (4.000, 7.250) vs. 11.000 (8.000, 13.750), P <  0.001], and lower IVF/ICSI fertilization rates [147/194 (75.77%) vs. 302/352 (85.80%), P =  0.003; 50/62 (80.65%) vs. 117/121 (96.69%), P <  0.001]. The posANA group also had fewer cleavage embryos [6.000 (4.000, 8.250) vs. 11.000 (8.000, 13.750), P <  0.001], fewer 2PN cleavage embryos [6.000 (4.000, 7.250) vs. 11.000 (8.000, 13.750), P <  0.001], and fewer available embryos [5.500 (3.750, 7.000) vs. 10.000 (7.250, 13.000), P <  0.001]. Additionally, the proportion of available embryos was significantly lower in the posANA group [178/201 (88.56%) vs. 396/421 (94.06%), P =  0.016]. The number of high-quality embryos was significantly lower in the posANA group [3.000 (2.000, 5.250) vs. 7.000 (5.000, 11.000), P <  0.001], and the proportion of high-quality embryos was also lower [112/197 (56.85%) vs. 303/416 (72.84%), P <  0.001]. The posANA group had fewer blastocysts [2.000 (1.000, 4.000) vs. 6.000 (4.000, 8.000), P <  0.001] and a lower proportion of blastocyst formation [74/159 (46.54%) vs. 238/388 (61.34%), P =  0.001]. Furthermore, the number of high-quality blastocysts was significantly lower in the posANA group [0.000 (0.000, 2.000) vs. 3.500 (2.250, 5.750), P <  0.001], and the proportion of high-quality blastocysts was also lower [33/74 (44.59%) vs. 163/238 (68.49%), P <  0.001] (Table  2 ). Table 2 Characteristics of controlled ovarian stimulation cycles and oocyte pick-up outcomes between posANA and negANA groups posANA negANA Z/χ 2 value P Patients, n 32 38 - - No. of oocytes retrieved, n 8.500 (6.000, 10.000) 12.500 (9.000, 15.000) −4.131 <.001 No. of 2PN, n 6.000 (4.000, 7.250) 11.000 (8.000, 13.750) −4.786 <.001 Fertilization rate, %  IVF 147/194(75.77%) 302/352(85.80%) 8.599 0.003  ICSI 50/62(80.65%) 117/121(96.69%) - <.001 *  No. of cleavage embryos, n 6.000 (4.000, 8.250) 11.000 (8.000, 13.750) −4.763 <.001  No. of 2PN cleavage embryos, n 6.000 (4.000, 7.250) 11.000 (8.000, 13.750) −4.786 <.001  No. of available embryos, n 5.500 (3.750, 7.000) 10.000 (7.250, 13.000) −4.953 <.001  Proportion of available embryos, % 178/201(88.56%) 396/421(94.06%) 5.788 0.016  No. of High-quality embryos, n 3.000 (2.000, 5.250) 7.000 (5.000, 11.000) −5.254 <.001  Proportion of High-quality embryos, % 112/197(56.85%) 303/416(72.84%) 15.772 <.001  No. of High-quality blastocysts, n 0.000 (0.000, 2.000) 3.500 (2.250, 5.750) −5.845 <.001  Proportion of High-quality blastocysts, % 33/74(44.59%) 163/238(68.49%) 13.797 <.001  No. of blastocysts, n 2.000 (1.000, 4.000) 6.000 (4.000, 8.000) −5.324 <.001  Proportion of blastocysts, % 74/159(46.54%) 238/388(61.34%) 10.080 0.001 Data are presented as median (interquartile range), or number (percentage) P <  0.05: significant statistical difference Abbreviations : IVF In vitro fertilization, ICSI Intracytoplasmic sperm injection * Fisher’s exact test Characteristics of controlled ovarian stimulation cycles and oocyte pick-up outcomes between posANA and negANA groups Data are presented as median (interquartile range), or number (percentage) P <  0.05: significant statistical difference Abbreviations : IVF In vitro fertilization, ICSI Intracytoplasmic sperm injection * Fisher’s exact test We further used multivariable linear regression models to control for potential confounding factors and evaluated the independent effect of ANA status on various embryological outcomes (Table  3 ). The results showed that ANA positivity was independently associated with multiple embryological outcomes, including oocytes retrieved, 2PN embryo number, cleavage embryo number, 2PN cleavage embryo number, available embryo number, high-quality embryo number, blastocyst formation, and high-quality blastocyst formation ( P <  0.05). However, ANA status did not significantly affect the available embryo rate (β = −0.076, 95% CI: −0.168 – 0.017, P =  0.114). Table 3 Multivariate analysis of laboratory outcomes between negANA and posANA group Outcomes β (95%CI) S.E P a No. of oocytes retrieved −2.313 (−4.087 ~ −0.539) 0.905 0.013 No. of 2PN −3.024 (−4.799 ~ −1.248) 0.906 0.001 No. of cleavage embryos −2.881 (−4.611 ~ −1.150) 0.883 0.002 No. of 2PN cleavage embryos −2.934 (−4.673 ~ −1.194) 0.887 0.002 No. of available embryos −3.187 (−4.754 ~ −1.619) 0.800 <.001 Proportion of available embryos −0.076 (−0.168 ~ 0.017) 0.047 0.114 No. of High-quality embryos −3.047 (−4.395 ~ −1.699) 0.688 <.001 Proportion of High-quality embryos −0.148 (−0.276 ~ −0.020) 0.065 0.026 No. of blastocysts −2.756 (−4.041 ~ −1.471) 0.656 <.001 Proportion of blastocysts −0.188 (−0.320 ~ −0.057) 0.067 0.007 No. of High-quality blastocysts −2.548 (−3.650 ~ −1.447) 0.562 <.001 Proportion of High-quality blastocysts −0.396 (−0.558 ~ −0.235) 0.082 <.001 a The outcomes of the multivariate linear regression analysis conducted with adjusted for bFSH, TSH, AFC and P on trigger day Multivariate analysis of laboratory outcomes between negANA and posANA group a The outcomes of the multivariate linear regression analysis conducted with adjusted for bFSH, TSH, AFC and P on trigger day To ensure the reliability of the untargeted metabolomics data, the base peak chromatogram (BPC) of the QC samples showed overlapping peak intensities, indicating that the instrument stability was satisfactory (Additional file 1: Figure S1 A). PCA analysis demonstrated tight clustering of the QC samples, further supporting the excellent reproducibility of the LC–MS analysis over the experimental period (Additional file 1: Figure S1 B). A total of 1852 metabolites were identified, covering lipids, organic acids and their derivatives, heterocyclic organic compounds, benzimidazoles, etc. (Fig.  2 A). PCA and OPLS-DA analyses were conducted to explore differences in FF metabolomic profiles between the posANA and negANA groups (Fig.  2 B and C). The PCA score plot showed that most samples were distributed within the 95% confidence interval, with an apparent tendency toward separation between the two groups. The supervised OPLS-DA score plot showed a clearer separation of FF metabolomic profiles between the posANA and negANA groups. The permutation test of the OPLS-DA model demonstrated that the R 2 Y(cum) and Q 2 (cum) values were 0.997 and 0.723, respectively (Fig.  2 D), indicating that the model was not overfitted and exhibited high predictive ability, making it suitable for subsequent optimization analyses. Fig. 2 Comprehensive untargeted metabolomic profiling of follicular fluid in posANA and negANA infertile patients. A Chemical classification and proportional distribution of all identified metabolites in follicular fluid. B PCA score plot showing the overall distribution of metabolomic profiles in the negANA and posANA groups. Each point represents one sample, and the shaded ellipses indicate the group distribution. C OPLS-DA score plot showing the separation of metabolomic profiles between the negANA and posANA groups. Model performance parameters (R 2 X, R 2 Y, and Q 2 ) are indicated in the panel. D Permutation test of the OPLS-DA model used to assess model stability and the potential risk of overfitting. The intercepts of R 2 and Q 2 are shown in the panel. E Volcano plot of DEMs between the posANA and negANA groups. The x-axis shows log 2 FC, and the y-axis shows − log10 of the nominal, unadjusted P value. Metabolites with VIP ≥ 1 and P <  0.05 were considered DEMs. F KEGG pathway enrichment analysis of DEMs. Bubble size represents the number of mapped metabolites in each pathway, and color indicates the enrichment P value. The top 20 pathways ranked by –log10(P value) are shown. G Heatmap with hierarchical clustering of DEMs across all samples. Rows represent metabolites and columns represent individual samples. Relative metabolite abundance is shown after normalization, and the color scale indicates the relative abundance level Comprehensive untargeted metabolomic profiling of follicular fluid in posANA and negANA infertile patients. A Chemical classification and proportional distribution of all identified metabolites in follicular fluid. B PCA score plot showing the overall distribution of metabolomic profiles in the negANA and posANA groups. Each point represents one sample, and the shaded ellipses indicate the group distribution. C OPLS-DA score plot showing the separation of metabolomic profiles between the negANA and posANA groups. Model performance parameters (R 2 X, R 2 Y, and Q 2 ) are indicated in the panel. D Permutation test of the OPLS-DA model used to assess model stability and the potential risk of overfitting. The intercepts of R 2 and Q 2 are shown in the panel. E Volcano plot of DEMs between the posANA and negANA groups. The x-axis shows log 2 FC, and the y-axis shows − log10 of the nominal, unadjusted P value. Metabolites with VIP ≥ 1 and P <  0.05 were considered DEMs. F KEGG pathway enrichment analysis of DEMs. Bubble size represents the number of mapped metabolites in each pathway, and color indicates the enrichment P value. The top 20 pathways ranked by –log10(P value) are shown. G Heatmap with hierarchical clustering of DEMs across all samples. Rows represent metabolites and columns represent individual samples. Relative metabolite abundance is shown after normalization, and the color scale indicates the relative abundance level In this study, a total of 369 differential metabolites (VIP ≥ 1, P <  0.05) were detected by OPLS-DA and univariate analysis, showing significant differences between the posANA and negANA groups. These DEMs were visualized using a volcano plot, with upregulated and downregulated metabolites distinguished by different colors (Fig.  2 E). Compared to the negANA group, the posANA group had 214 downregulated metabolites and 155 upregulated metabolites in FF. Detailed information on the DEMs can be found in Additional file 2A. Further KEGG pathway enrichment analysis revealed that upregulated metabolites in the posANA group were mainly enriched in the Retrograde endocannabinoid signaling pathway and GnRH signaling pathway (Fig.  2 F), while downregulated metabolites were predominantly involved in Steroid hormone biosynthesis, Bile secretion, and Tryptophan metabolism (Fig.  2 F). A clustering heatmap of the differential metabolites between groups is shown in Fig.  2 G, which demonstrates a clear clustering trend of these metabolites in both the posANA and negANA groups. Detailed information on the KEGG pathway enrichment analysis can be found in Additional file 2B. As shown in the figure, the peptide length distribution of the mass spectrometry data met standard quality-control criteria, and the peptide number, protein coverage, and intensity values were all within the expected ranges (Additional file 1: Figure S2A–C). Although a proportion of proteins were identified by single peptides, multiple quality-control metrics supported the overall robustness of the proteomic dataset, and downstream analyses were interpreted cautiously (Additional file 1: Figure S2D). PCA and OPLS-DA were then performed to compare proteomic differences between the posANA and negANA groups (Additional file 1: Figure S3A, B). The PCA score plot showed that most samples fell within the 95% confidence interval, with a modest separation trend between groups. The OPLS-DA score plot suggested distinguishable protein expression patterns in FF between the two groups. However, the permutation test demonstrated that although the OPLS-DA model exhibited a relatively high goodness-of-fit (R 2 Y = 0.898), it did not reach statistical significance ( P =  0.14), and the predictive ability was limited (Q 2  = 0.235) (Additional file 1: Figure S3C). These observations indicate that the proteomic differences between the two groups may contain an element of randomness. To comprehensively explore the potential impact of ANA on the FF proteome, a threshold of P <  0.05 was applied for DEP identification. In total, 50 significantly altered proteins were identified (Additional file 3 A). KEGG pathway classification showed that these DEPs were mainly associated with the immune system, signal transduction, and signaling molecules and interactions (Fig.  3 A). Among them, 23 proteins were downregulated and 27 were upregulated in the posANA group (Fig.  3 B). Hierarchical clustering revealed distinct expression patterns of these DEPs between groups (Fig.  3 C). Fig. 3 Analysis of DEPs between posANA and negANA groups. A KEGG pathway classification of the DEPs. The x-axis represents the number of proteins annotated to each pathway category. B Volcano plot of DEPs between the posANA and negANA groups. The x-axis shows log 2 FC, and the y-axis shows –log10 P value. Proteins with P <  0.05 were considered DEPs. C Heatmap with hierarchical clustering analysis of DEPs across all samples. Rows represent proteins and columns represent individual samples. Relative protein abundance is shown after normalization. D GO enrichment network of DEPs generated using the Cytoscape plug-in ClueGO. Nodes represent enriched GO terms, and edges indicate functional relationships between terms. Different colors denote distinct functional groups. E Enriched biological process terms of DEPs identified using Metascape. The x-axis shows –log10(P value). F Protein–protein interaction (PPI) network of DEPs constructed using the STRING database. Nodes represent proteins, and edges represent known or predicted protein–protein associations. Only interactions with a confidence score > 0.7 (high confidence) were included for visualization and network analysis. Node colors indicate the direction of expression change between groups Analysis of DEPs between posANA and negANA groups. A KEGG pathway classification of the DEPs. The x-axis represents the number of proteins annotated to each pathway category. B Volcano plot of DEPs between the posANA and negANA groups. The x-axis shows log 2 FC, and the y-axis shows –log10 P value. Proteins with P <  0.05 were considered DEPs. C Heatmap with hierarchical clustering analysis of DEPs across all samples. Rows represent proteins and columns represent individual samples. Relative protein abundance is shown after normalization. D GO enrichment network of DEPs generated using the Cytoscape plug-in ClueGO. Nodes represent enriched GO terms, and edges indicate functional relationships between terms. Different colors denote distinct functional groups. E Enriched biological process terms of DEPs identified using Metascape. The x-axis shows –log10(P value). F Protein–protein interaction (PPI) network of DEPs constructed using the STRING database. Nodes represent proteins, and edges represent known or predicted protein–protein associations. Only interactions with a confidence score > 0.7 (high confidence) were included for visualization and network analysis. Node colors indicate the direction of expression change between groups Further pathway and network enrichment analyses revealed that these proteins were predominantly enriched in the platelet degranulation pathway (Fig.  3 D, E). Detailed results of pathway and biological process enrichment analyses for the differentially expressed proteins are provided in the Additional file 4A-C. To visualize protein–protein interactions, a high-confidence (score > 0.7) PPI network was constructed using STRING (Fig.  3 F; Additional file 4D). Most DEPs clustered around PLG as a central node, and PLG and its associated proteins were generally upregulated in the posANA group, suggesting enhanced platelet degranulation activity in the FF of ANA-positive patients. To construct an integrated molecular landscape of ANA-associated infertility and to elucidate the interplay between metabolites and proteins, we performed a multi-omics analysis by integrating proteomic and untargeted metabolomic data obtained from the same FF samples. Sample distribution in the DIABLO model demonstrated a clear separation between posANA and negANA individuals at both the proteome and metabolome levels (Fig.  4 A). Among the 369 DEMs, 38 were annotated with defined KEGG functional pathways (Additional file 5 A). These 38 metabolites, together with differentially expressed proteins, were subjected to joint KEGG pathway enrichment analysis. The integrated analysis revealed that co-regulated proteins and metabolites were primarily involved in phenylalanine, tyrosine and tryptophan biosynthesis, phenylalanine metabolism, and glycerophospholipid metabolism (Fig.  4 C, D; Additional file 5B, 5 C). Fig. 4 Integrated multi-omics network analysis of proteomic and untargeted metabolomic data in follicular fluid from ANA-associated infertile patients. A Sample distribution based on the DIABLO model using the untargeted metabolomic dataset (left) and proteomic dataset (right). Each point represents one sample, and colors indicate the negANA and posANA groups. B Correlation between the first latent components derived from the metabolomic and proteomic datasets in the DIABLO model. The scatter plot shows the sample distribution according to the first components of the two omics blocks, and the correlation coefficient is displayed in the panel. C Joint KEGG pathway enrichment analysis of key co-regulated metabolites and proteins identified from the integrative multi-omics analysis. The x-axis represents pathway impact, and the y-axis shows –log10(P value). Bubble size and color reflect pathway enrichment characteristics. D Distribution of proteins and metabolites within the major enriched KEGG pathways identified in the integrative analysis. Bar length indicates the number of features mapped to each pathway, and colors distinguish metabolites from proteins. E Circos-style network showing cross-omics associations between key proteomic and metabolomic features. Nodes represent metabolites or proteins, node size reflects degree, and edges indicate significant correlations (correlation coefficient ≥ 0.4 or ≤ –0.4, P <  0.05). Edge colors denote positive or negative correlations Integrated multi-omics network analysis of proteomic and untargeted metabolomic data in follicular fluid from ANA-associated infertile patients. A Sample distribution based on the DIABLO model using the untargeted metabolomic dataset (left) and proteomic dataset (right). Each point represents one sample, and colors indicate the negANA and posANA groups. B Correlation between the first latent components derived from the metabolomic and proteomic datasets in the DIABLO model. The scatter plot shows the sample distribution according to the first components of the two omics blocks, and the correlation coefficient is displayed in the panel. C Joint KEGG pathway enrichment analysis of key co-regulated metabolites and proteins identified from the integrative multi-omics analysis. The x-axis represents pathway impact, and the y-axis shows –log10(P value). Bubble size and color reflect pathway enrichment characteristics. D Distribution of proteins and metabolites within the major enriched KEGG pathways identified in the integrative analysis. Bar length indicates the number of features mapped to each pathway, and colors distinguish metabolites from proteins. E Circos-style network showing cross-omics associations between key proteomic and metabolomic features. Nodes represent metabolites or proteins, node size reflects degree, and edges indicate significant correlations (correlation coefficient ≥ 0.4 or ≤ –0.4, P <  0.05). Edge colors denote positive or negative correlations Using a Circos plot, we further visualized the cross-omics correlations and identified a panel of strongly co-regulated molecules highly associated with the latent components in the DIABLO model (Fig.  4 E). These molecules likely represent key molecular signatures in ANA-associated infertility. Collectively, our findings highlight substantial crosstalk in amino acid metabolism and disturbances in glycerophospholipid metabolism within the FF of ANA-positive patients, providing potential mechanistic links and therapeutic targets for improving oocyte and embryo developmental competence in this population. To identify metabolites and proteins closely associated with oocyte and embryo quality in ANA-related infertility, correlation analyses were performed between the differential metabolites/proteins and embryological outcomes in the posANA and negANA groups. Candidate key molecules were first screened using LASSO regression, and those appearing in ≥ 80 out of 100 repeated LASSO runs were retained as robust predictors of oocyte and embryo developmental competence (Additional file 1: Figure S4 and Table  4 ). Table 4 Proteins and metabolites closely associated with embryological outcomes identified by LASSO selection Laboratory outcomes Candidate proteins Candidate metabolites No. of oocytes retrieved HSP90AB1 5-HETE,Indoleacetic acid,cAMP,Dehydroepiandrosterone sulfate,Testosterone glucuronide,Uridine 5'-monophosphate(UMP),Hydrocortisone,Etiocholanolone glucuronide,3-Methyloleind No. of 2PN - 3-Methylindole,Hydrocortisone,L-phenylalanine No. of cleavage embryos - 3-Methylindole,Hydrocortisone No. of 2PN cleavage embryos - 3-Methylindole,Hydrocortisone No. of available embryos HS3ST1,IGF2,SERPINE1,PRSS1,SERPINF2,LAMP2,PRSS3,ATXN1,TTN L-phenylalanine,3-Methylindole,Hydrocortisone,Testosterone glucuronide Proportion of available embryos - PC(20:4/20:4) No. of High-quality embryos HS3ST1,KNG1,IGF2,PROC,SERPINE1,PRSS1,HSP90AB1,SERPINF2,LAMP2,PRSS3,ATXN1,TTN L-phenylalanine,3-Methylindole,SM(d18:1/16:0),PC(18:0/22:6) Proportion of High-quality embryos SERPINF2,PROC,CLU SM(d18:1/16:0),PC(20:4/20:4) No. of High-quality blastocysts HS3ST1,KNG1,LCAT,SERPINE1,HSP90AB1,PRSS3,ATXN1,TTN LPA(18:2),Thyroxine,3-Methylindole,Dehydroepiandrosterone sulfate,SM(d18:1/16:0) Proportion of High-quality blastocysts SERPINE1,KNG1 cAMP,3-Methyloleind,5-HETE,LPA(18:2),PC(20:4/20:4) No. of blastocysts HS3ST1,KNG1,LCAT,HSP90AB1,PRSS3,ATXN1 3-Methylindole,SM(d18:1/16:0) Proportion of blastocysts PROC,LCAT,KNG1,LAMB2 - Proteins and metabolites closely associated with embryological outcomes identified by LASSO selection Hierarchical clustering analysis revealed that six differentially expressed molecules ( P <  0.05, |r|≥ 0.4) were significantly correlated with oocyte retrieval number and multiple embryo development parameters. Specifically, thyroxine, L-phenylalanine, 3-methylindole, and hydrocortisone showed strong positive correlations with oocytes retrieved, the number of 2PN embryos, cleavage embryos, 2PN cleavage embryos, available embryos, high-quality embryos, blastocysts, and high-quality blastocysts. Among these, thyroxine, L-phenylalanine, and 3-methylindole were positively correlated with the proportion of high-quality blastocyst, and 3-methylindole also showed a significant positive association with the proportion of blastocyst formation ( P <  0.05). In contrast, SM(d18:1/16:0) exhibited significant negative correlations with all of the above outcomes as well as the proportion of high-quality embryos ( P <  0.05). PC(18:0/22:6) was negatively associated with oocytes retrieved, 2PN embryos, cleavage embryos, 2PN cleavage embryos, available embryos, high-quality embryos, and high-quality blastocysts ( P <  0.05). With respect to proteins, ATXN1 and others showed significant negative correlations with multiple embryo developmental outcomes, whereas PRSS3 and several proteins exhibited positive associations with oocyte number and embryo quality ( P <  0.05) (Fig.  5 ). However, all protein–outcome correlations had relatively low effect sizes (|r|< 0.4), limiting their reliability as biomarkers for embryo developmental potential (Additional file 6). Fig. 5 Heatmap of Spearman correlation analysis between candidate metabolites, proteins, and clinical parameters. Rows represent individual metabolites or proteins, and columns represent clinical and embryological variables. Each colored cell indicates the Spearman correlation coefficient for a given feature–parameter pair. Purple denotes a positive correlation, green denotes a negative correlation, and greater color intensity indicates a stronger correlation. The annotation bar on the right indicates feature type (metabolite or protein). Feature names shown in bold black highlight metabolites or proteins with an absolute correlation coefficient of |r|≥ 0.4 with embryo laboratory outcome-related parameters. Statistical significance is indicated within the cells as follows: * P <  0.05; ** P <  0.01; *** P <  0.001 Heatmap of Spearman correlation analysis between candidate metabolites, proteins, and clinical parameters. Rows represent individual metabolites or proteins, and columns represent clinical and embryological variables. Each colored cell indicates the Spearman correlation coefficient for a given feature–parameter pair. Purple denotes a positive correlation, green denotes a negative correlation, and greater color intensity indicates a stronger correlation. The annotation bar on the right indicates feature type (metabolite or protein). Feature names shown in bold black highlight metabolites or proteins with an absolute correlation coefficient of |r|≥ 0.4 with embryo laboratory outcome-related parameters. Statistical significance is indicated within the cells as follows: * P <  0.05; ** P <  0.01; *** P <  0.001 Previous studies have suggested that ANA titers do not correlate directly with autoimmune disease activity, and whether the impact of ANA on oocyte and embryo quality is titer-dependent remains unclear [ 17 , 52 , 53 ]. To further investigate whether different ANA titers influence laboratory embryonic outcomes—and whether such effects could be reflected by associated metabolites and proteins—we stratified patients according to their serum ANA titers. No significant differences in laboratory outcomes were observed between the high-titer ANA group (≥ 1:160) and the low-titer ANA group (1:80). In contrast, negANA patients exhibited significantly better outcomes than both ANA-positive groups (Table  5 ; Additional file1: Table S2). These findings suggest that the presence of ANA, rather than its titer, may be the primary determinant of impaired embryonic development. Table 5 Comparison of embryological laboratory results in infertile patients with varying ANA titers Variables negANA ≧1:160 1:80 H/χ 2 value P Patients, n 38 14 18 - - No. of oocytes retrieved, n 12.500 (9.000,15.000) a,b 7.000 (5.000,9.000) 9.000 (7.000,10.000) 18.323 <.001 No. of 2PN, n 11.000 (8.000,13.750) a,b 4.500 (4.000,7.000) 6.500 (5.000,7.750) 23.562 <.001 Fertilization rate, %  IVF 302/352(85.80%) b 45/57(78.95%) 102/137(74.45%) 9.156 0.010  ICSI 117/121(96.69%) a,b 35/43(81.40%) 15/19(78.95%) - 0.004 *  No. of cleavage embryos, n 11.000 (8.000,13.750) a,b 4.500 (4.000,7.000) 7.000 (5.000,8.750) 23.559 <.001  No. of 2PN cleavage embryos, n 11.000 (8.000,13.750) a,b 4.500 (4.000,7.000) 6.500 (5.000,7.750) 23.563 <.001  No. of available embryos, n 10.000 (7.250,13.000) a,b 4.000 (3.000,7.000) 6.500 (4.250,7.750) 25.751 <.001  Proportion of available embryos, % 1.000 (1.000,1.000) 1.000 (0.740,1.000) 1.000 (0.808,1.000) 3.919 0.141  No. of High-quality embryos, n 7.000 (5.000,11.000) a,b 3.000 (1.250,3.750) 4.000 (2.250,6.000) 28.354 <.001  Proportion of High-quality embryos, % 0.800 (0.573,0.871) 0.583 (0.278,0.750) 0.586 (0.425,0.729) 7.720 0.021  No. of High-quality blastocysts, n 3.500 (2.250,5.750) a,b 0.000 (0.000,0.750) 1.000 (0.000,2.000) 35.577 <.001  Proportion of High-quality blastocysts, % 0.732 (0.511,0.851) a,b 0.000 (0.000,0.375) 0.292 (0.000,0.625) 19.265 <.001  No. of blastocysts, n 6.000 (4.000,8.000) a,b 1.000 (0.250,2.750) 3.000 (1.250,4.000) 29.364 <.001  Proportion of blastocysts, % 0.592 (0.500,0.830) a 0.450 (0.036,0.500) 0.422 (0.333,0.656) 11.820 0.003 a negANA vs. ≧1:160, P value < 0.05 b negANA vs. 1:80, P value < 0.05 c ≧1:160 vs. 1:80, P value < 0.05 * Fisher’s exact test Comparison of embryological laboratory results in infertile patients with varying ANA titers a negANA vs. ≧1:160, P value < 0.05 b negANA vs. 1:80, P value < 0.05 c ≧1:160 vs. 1:80, P value < 0.05 * Fisher’s exact test Given that previously identified metabolic alterations between posANA and negANA patients were more clinically meaningful, PCA was applied to visualize overall metabolic differences among the three groups (Fig.  6 A). High- and low-titer ANA groups were not clearly separated, indicating highly similar metabolomic profiles, whereas both ANA-positive groups were distinctly separated from the negANA group. Fig. 6 Expression trends of metabolites across different clinical ANA titers and their correlation with embryological outcomes. A PCA of metabolomic profiles in the control group (negANA), low-titer ANA group (1:80), and high-titer ANA group (≥ 1:160). Each point represents one sample, and the shaded ellipses indicate the group distribution. B Selection of the optimal number of Mfuzz clusters based on the within-cluster sum of squares (WSS) method. The elbow plot shows WSS values for cluster numbers ranging from 1 to 10. The vertical dashed line indicates the selected cluster number (k = 4), corresponding to the point at which the WSS curve begins to plateau after a marked decline. C Mfuzz clustering of 281 metabolites into four distinct expression clusters. For each cluster, the left panel shows the membership profiles of metabolites, and the right panel shows the corresponding expression heatmap across the control, low-titer ANA, and high-titer ANA groups. D UpSet plot showing the overlap of embryological outcome-related metabolites identified from different Mfuzz clusters. The upper bar plot indicates the number of overlapping metabolites, and the connected dots below indicate the cluster combinations contributing to each overlap set. E Boxplots showing the relative abundance of representative overlapping metabolites across the control, low-titer ANA, and high-titer ANA groups. Statistical significance is indicated as follows: * P <  0.05; ** P <  0.01; *** P <  0.001. F SNF-based unsupervised clustering integrating proteomic and metabolomic data. Samples were stratified into three molecular subgroups (SNF-1 to SNF-3). Most negANA samples clustered together, whereas low- and high-titer ANA samples showed partial intermixing, suggesting that integrated molecular profiles distinguish ANA-positive from ANA-negative individuals more clearly than they distinguish ANA titer levels Expression trends of metabolites across different clinical ANA titers and their correlation with embryological outcomes. A PCA of metabolomic profiles in the control group (negANA), low-titer ANA group (1:80), and high-titer ANA group (≥ 1:160). Each point represents one sample, and the shaded ellipses indicate the group distribution. B Selection of the optimal number of Mfuzz clusters based on the within-cluster sum of squares (WSS) method. The elbow plot shows WSS values for cluster numbers ranging from 1 to 10. The vertical dashed line indicates the selected cluster number (k = 4), corresponding to the point at which the WSS curve begins to plateau after a marked decline. C Mfuzz clustering of 281 metabolites into four distinct expression clusters. For each cluster, the left panel shows the membership profiles of metabolites, and the right panel shows the corresponding expression heatmap across the control, low-titer ANA, and high-titer ANA groups. D UpSet plot showing the overlap of embryological outcome-related metabolites identified from different Mfuzz clusters. The upper bar plot indicates the number of overlapping metabolites, and the connected dots below indicate the cluster combinations contributing to each overlap set. E Boxplots showing the relative abundance of representative overlapping metabolites across the control, low-titer ANA, and high-titer ANA groups. Statistical significance is indicated as follows: * P <  0.05; ** P <  0.01; *** P <  0.001. F SNF-based unsupervised clustering integrating proteomic and metabolomic data. Samples were stratified into three molecular subgroups (SNF-1 to SNF-3). Most negANA samples clustered together, whereas low- and high-titer ANA samples showed partial intermixing, suggesting that integrated molecular profiles distinguish ANA-positive from ANA-negative individuals more clearly than they distinguish ANA titer levels To further explore expression trends, Mfuzz soft clustering was performed. Differences among the three groups were evaluated using ANOVA with Benjamini–Hochberg correction to obtain FDR-adjusted P values. DEMs were defined as those with VIP ≥ 1 and FDR < 0.05 (Additional file 7). The optimal number of clusters was determined to be four. Cluster assignments and their associations with ANA titers are presented in Fig.  6 B and C. Relevant metabolites associated with embryonic laboratory outcomes were summarized using an Upset plot, and their relative expression levels across ANA titer groups were visualized using boxplots. Importantly, the six metabolites previously identified as strongly associated with oocyte and embryo developmental competence were distributed across three distinct clusters rather than concentrated in a single pattern. Moreover, their expression levels did not differ significantly between low- and high-titer ANA groups but differed markedly from those of the negANA group ( P <  0.05). This strongly supports that the detrimental effects of ANA are dependent on ANA presence rather than titer magnitude. To validate this hypothesis, we applied SNF to integrate proteomic and metabolomic datasets in an unsupervised manner. SNF classified the samples into three clusters. Notably, 78.95% (30/38) of negANA patients clustered together and clearly separated from ANA-positive patients, indicating substantial differences in FF molecular composition between ANA-negative and ANA-positive individuals. However, no clear separation was observed between low- and high-titer ANA groups (Fig.  6 F). Correspondingly, SNF cluster 1 (comprising exclusively negANA patients) exhibited significantly better embryonic laboratory outcomes compared with SNF clusters 2 and 3, whereas outcomes did not differ between clusters 2 and 3 (Additional file: Table S3). Collectively, these findings demonstrate that:(1) ANA-negative patients possess a FF molecular signature favorable for oocyte and embryo development, whereas ANA-positive patients exhibit disrupted proteomic and metabolomic profiles; and(2) the adverse effects of ANA on follicular microenvironment and developmental outcomes are driven by ANA positivity itself rather than ANA titer levels. Through integrative analysis of follicular-fluid proteomic and metabolomic profiles, we found that the impact of serum ANA on oocyte and embryo outcomes was not dependent on ANA titers but was primarily determined by the presence or absence of ANA. To further identify key molecular determinants, we performed Venn-diagram–based integration of three analytical strategies: (1) SNF, which identified molecules most strongly associated with oocyte and embryo development in an unsupervised manner; (2) Mfuzz clustering, which captured metabolites exhibiting distinct expression patterns across ANA titer groups; and (3) correlation-based heatmap analysis of embryological outcomes. By combining these approaches, four metabolites were consistently identified as being strongly associated with embryological outcomes—L-phenylalanine, SM(d18:1/16:0), 3-methylindole, and hydrocortisone (Fig.  7 A). In contrast, due to the absence of significant protein differences between ANA titer subgroups, no protein candidates were identified that could explain titer-independent embryological outcomes (Additional file 3B). Fig. 7 Screening of candidate follicular fluid molecular features associated with oocyte and embryo quality independent of ANA titer in infertile patients with positive serum ANA. A Venn diagram showing the overlap among metabolites associated with embryological outcomes, metabolites identified from Mfuzz clustering analysis, and differential features identified from the SNF-based integrative analysis. B Pairwise correlation plot of the four selected metabolites. The lower triangle shows the Spearman correlation coefficients, and the upper triangle displays the corresponding correlation visualization. Asterisks indicate statistical significance ( * P <  0.05; ** P <  0.01). C Coefficients of variation (CVs) of the four key metabolites in the negANA and posANA groups, used to assess the relative expression stability of these metabolites within each group Screening of candidate follicular fluid molecular features associated with oocyte and embryo quality independent of ANA titer in infertile patients with positive serum ANA. A Venn diagram showing the overlap among metabolites associated with embryological outcomes, metabolites identified from Mfuzz clustering analysis, and differential features identified from the SNF-based integrative analysis. B Pairwise correlation plot of the four selected metabolites. The lower triangle shows the Spearman correlation coefficients, and the upper triangle displays the corresponding correlation visualization. Asterisks indicate statistical significance ( * P <  0.05; ** P <  0.01). C Coefficients of variation (CVs) of the four key metabolites in the negANA and posANA groups, used to assess the relative expression stability of these metabolites within each group A correlation heatmap further illustrated the inter-relationships among these four metabolites (Fig.  7 B). In addition, CV analysis demonstrated that three of them—L-phenylalanine, SM(d18:1/16:0), and hydrocortisone—exhibited stable expression patterns across samples (CV ≤ 40%) (Fig.  7 C). Taken together, these findings suggest that L-phenylalanine, SM(d18:1/16:0), and hydrocortisone in FF represent robust and biologically relevant metabolites associated with oocyte and embryo quality in ANA-related infertility, and their effects are independent of ANA titers but closely related to the presence of ANA itself. To evaluate the discriminatory performance of the three key metabolites (L-phenylalanine, SM(d18:1/16:0), and hydrocortisone) in ANA-related infertility, we constructed a logistic regression model based on these metabolites. In the discovery cohort, the model achieved an AUC of 0.968 (95% CI: 0.925–1.000), indicating excellent discrimination between posANA and negANA infertile patients (Fig.  8 A). In addition, the calibration curve showed good agreement with the ideal reference line. After 1,000 bootstrap resampling iterations, the Hosmer–Lemeshow test suggested acceptable calibration ( P =  0.816) (Fig.  8 B), indicating consistency between the predicted and observed outcomes within the study cohort. Fig. 8 Exploratory discriminatory performance of the combined metabolites L-phenylalanine, SM(d18:1/16:0), and hydrocortisone for differentiating negANA from posANA patients and for classifying dichotomized oocyte- and early embryo-related outcomes. A ROC curve assessing the ability of the three-metabolite panel to discriminate between posANA and negANA groups. The AUC (95% CI) is indicated in the legend. B Calibration curve analysis for the posANA and negANA groups. The ideal curve represents perfect calibration, the apparent curve was generated from the original dataset, and the bias-corrected curve was generated after 1,000 bootstrap resampling iterations. The Hosmer–Lemeshow test statistic and its corresponding P value are also shown. C ROC curve analysis of the three-metabolite panel for classifying dichotomized oocyte- and early embryo-related outcomes. For each embryological parameter, patients were dichotomized according to the median value of the corresponding outcome within the study population. AUC values with 95% confidence intervals are displayed for each outcome Exploratory discriminatory performance of the combined metabolites L-phenylalanine, SM(d18:1/16:0), and hydrocortisone for differentiating negANA from posANA patients and for classifying dichotomized oocyte- and early embryo-related outcomes. A ROC curve assessing the ability of the three-metabolite panel to discriminate between posANA and negANA groups. The AUC (95% CI) is indicated in the legend. B Calibration curve analysis for the posANA and negANA groups. The ideal curve represents perfect calibration, the apparent curve was generated from the original dataset, and the bias-corrected curve was generated after 1,000 bootstrap resampling iterations. The Hosmer–Lemeshow test statistic and its corresponding P value are also shown. C ROC curve analysis of the three-metabolite panel for classifying dichotomized oocyte- and early embryo-related outcomes. For each embryological parameter, patients were dichotomized according to the median value of the corresponding outcome within the study population. AUC values with 95% confidence intervals are displayed for each outcome We further explored whether the three-metabolite signature was associated with oocyte and early embryo developmental outcomes. Each embryological variable was dichotomized according to the median value within the study population, and separate logistic regression and ROC analyses were performed. The resulting AUC values ranged from 0.719 to 0.771 (Fig.  8 C), suggesting that this metabolite combination showed moderate discriminatory ability for these embryological outcomes in the present exploratory cohort.

From January 2024 to August 2025, we prospectively collected clinical data from 70 infertile women undergoing in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI) at the Reproductive Medicine Department of Nanjing Women and Children’s Healthcare Hospital. The cohort included women with both primary and secondary infertility. For each participant, FF samples obtained from the same patient were used for both proteomic and metabolomic analyses. Prior to IVF/ICSI treatment, baseline serum ANA titers were measured in the hospital laboratory. Patients with positive ANA titers underwent further testing for specific anti-ENA antibodies, and ANA titers were reassessed to exclude false positivity. Inclusion criteria were as follows: (1) age 20–40 years; (2) baseline follicle-stimulating hormone (FSH) < 10 IU/L, anti-Müllerian hormone (AMH) ≥ 1.1 ng/mL, and antral follicle count (AFC) ≥ 6; (3) first IVF/ICSI treatment cycle; (4) ACA and anti-β2-glycoprotein I antibody (A-β2-GPI) < 20 IgG or IgM phospholipid units, and lupus anticoagulant (LA) negativity [ 35 ]; and (5) no use of immunomodulatory agents, such as hydroxychloroquine or prednisone, within one month before ovulation induction. Exclusion criteria included: history of ovarian surgery, chemotherapy, HIV infection, PCOS, chromosomal abnormalities, severe male factor infertility, recurrent pregnancy loss (RPL), recurrent implantation failure (RIF), preimplantation genetic testing (PGT), and oocyte/embryo donation cycles. Finally, 32 patients with repeated ANA positivity and negative anti-ENA antibodies were assigned to the isolated ANA positivity group (posANA), and 38 patients with negative ANA titers were assigned to the ANA-negative control group (negANA). The study was approved by the Medical Ethics Committee of Nanjing Women and Children’s Healthcare Hospital (2024KY-116), and written informed consent was obtained from all participants before sample collection. Follicular fluid was aspirated from the leading follicle (> 18 mm in diameter) in each patient. A schematic overview of the study design is shown in Fig.  1 . Fig. 1 Flowchart of study design and patient allocation. ANA: antinuclear antibodies; ANA17s: a panel of 17 specific anti–extractable nuclear antigen (anti-ENA) antibodies; PCA: principal component analysis; SNF: similarity network fusion Flowchart of study design and patient allocation. ANA: antinuclear antibodies; ANA17s: a panel of 17 specific anti–extractable nuclear antigen (anti-ENA) antibodies; PCA: principal component analysis; SNF: similarity network fusion Before initiating IVF/ICSI treatment, all enrolled infertile patients underwent ANA screening in the Department of Laboratory Medicine at Nanjing Women and Children’s Healthcare Hospital. Peripheral venous blood was collected, and serum was separated for testing. The serum samples were incubated with antigen slides coated with HEp-2 cell substrates (Shenzhen YHLO Company). ANA in the serum binds to antigens present in the HEp-2 cells, forming antigen–antibody complexes. Fluorescence signals were subsequently detected using the iSlide 240 fully automated immunofluorescence analyzer (Shenzhen YHLO Company), following the manufacturer’s protocol. ANA titers were quantified by serial dilution, and a titer ≥ 1:80 was defined as positive [ 2 ]. Patients with positive ANA screening results underwent additional testing for specific anti-ENA IgG antibodies. Anti-ENA detection was performed using the highly sensitive line immunoassay (LIA) [ 36 ] on the Tenfly Auto automated immunoblot analyzer (Shenzhen Yaxilong Company). The assay panel included 17 autoantigens: nucleosome, dsDNA, histone, SmD1, PCNA, PO, SS-A/Ro60 kD, Ro52/TRIM21, SS-B/La, CENP-B, Scl-70, U1-snRNP, AMA-M2, Jo-1, PM-Scl, Mi-2, and Ku. Results were interpreted by comparing signal intensities with three internal control bands and classified as negative, weakly positive, moderately positive, or strongly positive [ 37 ]. All personnel responsible for IFA-ANA interpretation and reporting were rigorously trained, competency-assessed, and formally certified, each with a minimum of five years of professional laboratory experience. Although current evidence does not support a standardized or fixed interval for ANA retesting in ANA-positive individuals—and most patients remain ANA-positive even years after repeat testing [ 38 – 40 ]—to minimize the likelihood of potential false-positive ANA results [ 41 , 42 ], ANA titers were re-tested in all initially ANA-positive patients to ensure analytical accuracy and reporting reliability. The two ANA titer assessments were conducted at least one month apart, and the final confirmatory measurement was performed at the initiation of ovarian stimulation, ensuring that baseline ANA status was accurately established prior to COS. All patients in this study underwent controlled ovarian stimulation using a flexible gonadotropin-releasing hormone antagonist (GnRH-ant) protocol, as previously described by our group [ 22 ]. Final oocyte maturation was triggered when at least two follicles reached a diameter of > 18 mm, using either a single dose of human chorionic gonadotropin (hCG; Lizhu Pharmaceutical Factory, China) or a dual-trigger protocol consisting of 2000 IU hCG combined with a GnRH agonist (0.2 mg Decapeptyl; Ferring International Center SA). For patients at high risk of ovarian hyperstimulation syndrome (OHSS), a GnRH-ant protocol with a GnRH-agonist trigger alone was applied. The definition and grading of OHSS followed the recommendations of the American Society for Reproductive Medicine (ASRM) [ 43 ]. Transvaginal ultrasound–guided oocyte retrieval was performed 36 h after hCG administration. During oocyte retrieval, follicles with a mean diameter > 14 mm were routinely aspirated as part of the clinical assisted reproductive technology procedure. For research purposes, however, only one FF sample per patient was included in the multi-omics analyses. Specifically, the first aspirate of blood-free FF was collected from a single dominant follicle with a diameter > 18 mm. FF samples were centrifuged at 1500 g for 15 min at 4 °C, and the supernatant was aliquoted and stored at − 80 °C until further proteomic and metabolomic analyses. Oocytes were fertilized 4–6 h after retrieval, using either IVF or ICSI depending on semen quality. Resulting embryos were cultured in vitro for 3–6 days, followed by fresh embryo transfer (ET) or cryopreservation, based on clinical indications, embryo development, and patient preference. Assessment of fertilization status and the embryo culture system followed the procedures described in our previous study [ 44 ]. Briefly, on Day 3, the embryos were scored according to previously published criteria [ 45 ]. A high-quality cleavage-stage embryo was defined as having at least six blastomeres, less than 20% fragmentation, and no significant morphological abnormalities. Morphological observation and grading were performed on blastocysts formed on Day 5 and Day 6 post-fertilization. According to the Gardner blastocyst grading system [ 46 ], blastocysts were staged and rated based on three factors: the degree of blastocyst expansion, the number and morphology of the inner cell mass (ICM) cells, and the number and morphology of the trophoblast cells (TE). The degree of blastocoel expansion was graded on a scale from 1 to 6. When the blastocyst reached an expansion level of 3 or higher, both ICM and TE were evaluated based on cell size and density (A/B/C). High-quality blastocysts were defined as those at stage 3 or above, with an expanded blastocoel and both ICM and TE graded as B or higher (i.e., ≥ 3BB). Frozen FF samples were thawed slowly on ice prior to extraction. A total of 400 μL of methanol/acetonitrile (MeOH/ACN, 1:1, v/v) was added to 100 μL of FF to maintain a 2:2:1 extraction ratio (no water added). The mixture was vortexed for 30 s and subjected to sonication on ice for 10 min using a high-intensity ultrasonic processor. Samples were then incubated at − 20 °C for 1 h to facilitate protein precipitation. Following incubation, samples were centrifuged at 12,000 × g for 15 min at 4 °C, and the resulting supernatant was collected and evaporated to dryness at room temperature using a vacuum concentrator. The dried extracts were reconstituted in 100 μL of acetonitrile/water (ACN/H₂O, 1:1, v/v), vortexed for 30 s, and sonicated again on ice for 10 min. After a second centrifugation at 12,000 × g for 15 min at 4 °C, the clarified supernatant was transferred to clean tubes and stored at − 80 °C until LC–MS analysis. Metabolites were separated using a Waters UPLC system equipped with an ACQUITY UPLC BEH C18 column (1.7 µm, 2.1 × 100 mm). A sample volume of 10 µL was injected, and chromatographic separation was performed at a flow rate of 400 µL/min with the column temperature maintained at 45 °C. The mobile phase consisted of water containing 0.1% formic acid (solvent A) and acetonitrile containing 0.1% formic acid (solvent B). The elution gradient was set as follows: 0–11 min, 2%–98% B; 11.0–12.0 min, 98% B; 12.0–12.1 min, 98%–2% B; and 12.1–15.0 min, 2% B. After chromatographic separation, metabolites were ionized using an electrospray ionization (ESI) source and analyzed on a timsTOF Pro mass spectrometer (Bruker Daltonics, Bremen, Germany). The capillary voltage was set to 4.5 kV, and the mass scan range was m/z 50–1300. Data acquisition was performed using parallel accumulation–serial fragmentation (PASEF) mode. Following each full MS scan, two PASEF MS/MS scans were acquired for precursor ions with charge states ranging from 0 to 1. Dynamic exclusion was set to 6 s to avoid repeated fragmentation of the same precursor ions. Details of materials and reagents are provided in Additional file 1: Table S1. Raw LC–MS data were processed in MetaboScape 2022 for peak extraction, alignment, and retention time correction. The mass errors for both MS 1 and MS 2 features were controlled within ± 20 ppm to ensure high-confidence metabolite identification. Peaks with a missing rate > 50% within any group were removed prior to statistical analysis. Missing values were imputed using a combined strategy: features with > 50% zero values were imputed with one-fifth of the minimum detected value, whereas those with < 50% zeros were imputed using the k-nearest neighbor (KNN) algorithm. Peak intensities were further normalized using the support vector regression (SVR) method. Metabolite identification and structural annotation were performed by spectral matching against multiple databases, including NIST, HMDB, an in-house library, and integrated public spectral databases. Data normalization was performed in MetaboAnalyst 5.0( https://www.metaboanalyst.ca/MetaboAnalyst/docs/RTutorial.xhtml ) following the standard workflow. All metabolite intensities were normalized using the sample median. The normalization formula was as follows: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Rij=\frac{Iij}{Median\left(Ii\right)}$$\end{document} where R ij represents the normalized intensity, I ij is the original peak intensity, i is the sample index, and j is the metabolite index. To ensure analytical reproducibility and data quality, pooled quality control (QC) samples were prepared by mixing equal aliquots of all study samples. QC samples were injected at regular intervals (every 10 study samples) throughout the LC–MS run to monitor instrument stability and analytical performance. Frozen samples stored at –80 °C were thawed on ice, and total protein concentrations were determined using a BCA protein assay kit according to the manufacturer’s instructions. Equal amounts of protein from each sample were subjected to tryptic digestion. Protein volumes were first adjusted to the same level using lysis buffer. Dithiothreitol (DTT) was added to a final concentration of 5 mM, and samples were reduced at 56 °C for 30 min. Iodoacetamide (IAA) was then added to a final concentration of 11 mM and incubated for 15 min at room temperature in the dark. The urea concentration of each sample was diluted to below 2 M using 200 mM TEAB. Sequencing-grade trypsin was added at an enzyme-to-protein ratio of 1:50 (w/w) for overnight digestion, followed by a second digestion at a ratio of 1:100 (w/w) for an additional 4 h. Details of materials and reagents are provided in Additional file 1: Table S1. The tryptic peptides were reconstituted in solvent A and directly loaded onto a home-packed reversed-phase analytical column (25 cm × 100 μm i.d.). The mobile phases consisted of solvent A (0.1% formic acid, 2% acetonitrile in water) and solvent B (0.1% formic acid, 90% acetonitrile in water). Peptides were separated using an EASY-nLC 1200 system (Thermo Fisher Scientific) at a constant flow rate of 500 nL/min with the following gradient: 0–16 min, 6–20% B; 16–24 min, 20–32% B; 24–27 min, 32–80% B; and 27–30 min, 80% B. Eluted peptides were analyzed on an Orbitrap Exploris 480 mass spectrometer equipped with a nano-electrospray ion source. The spray voltage was set to 2100 V. Full MS scans were acquired in the Orbitrap at a resolution of 30,000 over an m/z range of 350–1050. MS/MS scans were acquired with a fixed first mass of 200 m/z at a resolution of 45,000. Higher-energy collisional dissociation (HCD) was performed with normalized collision energies (NCE) of 25%, 30%, and 35%. The automatic gain control (AGC) target was set to 3 × 10⁶ with a maximum injection time set to “Auto.” Data-independent acquisition (DIA) raw files were processed using the DIA-NN software package (version 1.8). MS/MS spectra were searched against the Homo_sapiens_9606_SP_20231220.fasta database (20,429 entries) concatenated with a reverse-decoy database. Trypsin/P was specified as the proteolytic enzyme, allowing up to one missed cleavage. N-terminal methionine excision and carbamidomethylation of cysteine residues were set as fixed modifications. Protein and peptide identifications were filtered using a stringent false discovery rate (FDR) threshold of < 1% at both the precursor and protein group levels. Based on the DIA-MS output, normalized intensity values were obtained for each protein across all samples. For quantitative comparison, the normalized intensity (I) of each protein was subjected to mean-centering to generate relative abundance values (R) for downstream analyses. The transformation was performed according to the following formula: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Rij=\frac{Iij}{Mean\left(Ij\right)}$$\end{document} where i denotes the sample index and j denotes the protein. Quality assessment of the proteomics data was performed by examining multiple metrics, including peptide length distribution, the number of peptides identified per protein, and protein sequence coverage distribution, ensuring the reliability and depth of mass spectrometry acquisition. Statistical analyses of clinical data were performed using R software (version 4.5.1). Categorical variables were compared between groups using Pearson’s chi-square test or Fisher’s exact test, as appropriate. To maintain a uniform data format, continuous variables were analyzed using the Mann–Whitney U test and presented as median with interquartile range (IQR) [ 21 ]. For comparisons across three or more groups, the Kruskal–Wallis rank-sum test was applied, followed by Bonferroni-adjusted post hoc tests for multiple comparisons. To evaluate the independent association between ANA status and reproductive outcomes, multivariable linear regression models were performed in internal validation using bootstrap methods with adjustment for potential confounders. Unstandardized coefficients (β) and 95% confidence intervals (CIs) were reported. All statistical tests were two-sided, and a P value < 0.05 was considered statistically significant. Adjusted variables were: bFSH, TSH, AFC, and P on trigger day. Multivariate analyses were performed on the annotated metabolites. All data were standardized by Z-score transformation (mean = 0; standard deviation = 1) to homogenise scale and variance among variables. Principal component analysis (PCA) was used to visualize overall metabolic variation and to assess clustering or separation patterns within and between groups. Variable importance in the projection (VIP) scores were calculated using orthogonal partial least squares–discriminant analysis (OPLS-DA) to identify metabolites contributing most significantly to group discrimination. In addition, permutation testing was performed to validate the OPLS-DA model and to assess the contribution and explanatory power of metabolite expression patterns in distinguishing between groups. The predictive performance of the OPLS-DA model was evaluated using the parameters R 2 X, R 2 Y, and Q 2 . R 2 X and R 2 Y represent the proportions of variance explained by the model in the X (predictor) and Y (response) matrices, respectively, whereas Q 2 reflects the predictive ability of the model. Values of these parameters closer to 1 indicate greater model robustness and reliability. Generally, a model with Q 2  > 0.5 is considered to have acceptable predictive performance, while Q 2  > 0.9 indicates excellent predictive capability. For univariate analysis, Student’s t-tests were conducted, and results were visualized using volcano plots. In this initial exploratory screening step, nominal, unadjusted P values were used. Metabolites with VIP ≥ 1 and a nominal, unadjusted P value < 0.05 were designated as candidate differentially expressed metabolites (DEMs) for subsequent exploratory analyses. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed and visualized using the ggplot2 package, with results displayed in bubble-plot format. To explore metabolic trends across different ANA titers, Mfuzz clustering was applied to group metabolites with similar expression patterns into four distinct clusters. Key metabolites associated with embryological outcomes within each cluster were visualized using UpSet plots, and their differential expression across ANA-titer subgroups was further presented using boxplots. To assess the differences between ANA-titer subgroups, ANOVA was performed with Benjamini–Hochberg correction to obtain FDR-adjusted P values. A threshold of FDR < 0.05 was used to identify differentially expressed metabolites. Differential expression analysis was performed using Student’s t-tests on the relative quantitative values of proteins across samples, and corresponding P values were calculated. Proteins with P <  0.05 were defined as potential differentially expressed proteins (DEPs) and visualized using volcano plots [ 47 , 48 ]. Hierarchical clustering analysis was conducted to assess changes in the relative abundance patterns of DEPs across samples. Gene Ontology (GO) functional annotation and KEGG pathway enrichment analyses for FF proteins were performed using Metascape ( https://metascape.org/gp/index.html#/main/step1 ) [ 49 ] and the ClueGO plugin implemented in Cytoscape. Two-sided Fisher’s exact tests were used for enrichment statistics, and a P value < 0.05 was considered significant. KEGG pathway mapping was conducted based on the KEGG database ( https://www.genome.jp/kegg/ ). Protein–protein interaction (PPI) information was retrieved from the STRING database ( https://string-db.org/ ), and interactions with a confidence score > 0.7 (high confidence) were selected for visualization and network analysis. To integrate FF proteomic and metabolomic datasets, multivariate dimension-reduction analysis was performed using DIABLO (Data Integration Analysis for Biomarker discovery using a Latent component method for Omics), implemented in the mixOmics R package. Briefly, DIABLO identifies linear combinations of variables from each omics layer that maximize correlations across datasets through stochastic optimization of the full design matrix. A tuning procedure was applied to determine the optimal number of selected features for each dataset by minimizing the overall classification error rate. Model performance and feature stability were evaluated using tenfold cross-validation. Joint KEGG pathway enrichment of DEPs and DEMs was conducted using the MetaboAnalyst online platform ( https://www.metaboanalyst.ca/ ). To further evaluate proteome–metabolome associations, a differential protein–metabolite interaction (PMI) network was constructed. The network incorporated DEPs and DEMs that were jointly mapped to KEGG pathways, and only protein–metabolite pairs with an absolute Spearman correlation coefficient |r|≥ 0.4 and P <  0.05 were retained. Network topology parameters, including node degree and betweenness centrality (computed using maximal-score weighting), were calculated using the R package igraph to characterize the importance of proteins and metabolites in the network. Node degree reflects the number of edges connected to a node, while betweenness centrality quantifies the number of shortest paths between all node pairs that pass through a given node, serving as an indicator of network centrality. To further identify key proteins and metabolites strongly associated with embryological outcomes in ANA-related infertility, a linear regression model with least absolute shrinkage and selection operator (LASSO) penalty was applied. LASSO analyses were performed using the glmnet R package. The optimal regularization parameter lambda (λ) was determined by tenfold cross-validation. As λ increased, coefficients of weaker predictors were shrunk toward zero, enabling the model to retain only the most influential proteins or metabolites. Variables corresponding to the minimum λ value (λ.min) were selected as the most predictive features. For each embryological endpoint, 100 repeated LASSO models were fitted, and features that appeared in ≥ 80 models were designated as key protein or metabolite predictors. To determine whether differential metabolites across ANA titers exhibited distinct effects on embryological outcomes, Similarity Network Fusion (SNF) was implemented using the R packages SNFtool and proxy . SNF integrates multi-omics datasets by constructing and iteratively fusing sample similarity networks from each data modality, thereby reducing the influence of measurement scale differences, sampling bias, and stochastic noise. Its nonlinear data-integration framework leverages shared and complementary information across omics layers to improve robustness and biological interpretability [ 50 ]. Data dispersion and reproducibility were assessed using the coefficient of variation (CV), calculated as CV = 100 × σ/μ, where σ represents the standard deviation and μ the mean. In the discovery cohort, Venn diagram analysis was used to identify overlapping proteins or metabolites jointly detected by SNF and Mfuzz as associated with embryological outcomes. Only molecules with CV ≤ 40% were considered reproducible and were retained for subsequent model development [ 51 ]. A multivariate logistic regression model was then established based on the selected factors, and its performance was evaluated using receiver operating characteristic (ROC) analysis, including the area under the ROC curve (AUC) and the calibration curve. Internal validation was performed using the Bootstrap method, and model goodness-of-fit was assessed with the Hosmer–Lemeshow test by comparing predicted probabilities with observed outcomes. For ROC and logistic regression analyses, continuous embryological outcomes (e.g., number of high-quality embryos, blastocysts) were dichotomized according to the median value of the study population to define binary endpoints. All statistical tests were two-tailed, and P values < 0.05 were considered statistically significant.

This prospective study investigated the relationship between positive serum ANA, in the absence of anti-ENA antibodies, and embryological outcomes in infertile patients. Using untargeted metabolomic and proteomic analyses of FF samples, we explored the metabolic and proteomic alterations associated with oocyte and early embryo quality in ANA-positive infertile women. Additionally, by integrating metabolomic and proteomic datasets derived from the same biological specimens, we conducted a multi-omics analysis to build a systems biology network linking proteins to metabolic products. This integrated approach identified molecular changes that were closely associated with embryological outcomes in ANA-positive patients. ROC analyses further characterized the exploratory discriminatory performance of these key metabolites for differentiating posANA from negANA patients and for classifying dichotomized oocyte- and embryo-related outcomes within the present cohort. These findings support their consideration as candidate molecular indicators. The uniformly reduced embryological outcomes observed in the posANA group, including oocyte yield, fertilization, cleavage, blastocyst formation, and blastocyst quality, may indicate that the primary defect occurs at the level of the oocyte, rather than being due to multiple independent disturbances at different embryonic stages. From this perspective, impairments at the embryo stage could be a downstream consequence of compromised oocyte competence. A plausible biological explanation is that ANA-associated immune dysregulation may interfere with oocyte maturation before fertilization, affecting both meiotic integrity and cytoplasmic maturation [ 10 , 18 ]. In particular, our findings revealed significant dysregulation of steroid hormone biosynthesis, amino acid metabolism, and glycerophospholipid metabolism in the follicular microenvironment, which could compromise developmental competence and contribute to early developmental arrest and poorer blastocyst outcomes. However, because this study did not assess the chromosomal status of oocytes or embryos, we cannot directly determine whether oocyte-derived aneuploidy may have contributed to the observed reductions in embryological outcomes. Amino acids are not only essential substrates for protein synthesis but also play central roles in energy metabolism, osmotic regulation, and intracellular signaling, thereby influencing oocyte maturation and embryo development [ 54 , 55 ]. In the present study, we observed significant alterations in amino acid-related pathways in the FF of ANA-associated infertile patients, including the biosynthesis of phenylalanine, tyrosine, and tryptophan, as well as phenylalanine and tryptophan metabolism. Notably, L-phenylalanine was markedly reduced in ANA-positive patients and showed positive correlations with several embryological outcomes ( r =  0.259–0.475, P <  0.05). Previous studies have reported divergent changes in L-phenylalanine across reproductive disorders: L-Phe is significantly elevated in patients with PCOS [ 56 , 57 ], whereas it is decreased in women with DOR [ 30 ], suggesting that L-phenylalanine levels may be associated with ovarian reserve and oocyte developmental competence. Moreover, L-phenylalanine has been shown to support the successful progression of 8-cell embryos to the blastocyst stage, particularly in embryos with high developmental potential [ 58 , 59 ]. Consistent with these findings, our study demonstrated positive correlations between L-phenylalanine levels and both the number and quality of oocytes and embryos, including blastocyst quality. Nevertheless, the mechanisms underlying reduced L-phenylalanine levels in ANA-positive patients remain unclear. In murine models of rheumatoid arthritis, progressive disruption of phenylalanine-related metabolic pathways has been documented with increasing disease activity [ 60 ]. Clinically, L-phenylalanine and its downstream metabolites have been proposed as serum biomarkers in rheumatoid arthritis and other immune-mediated inflammatory diseases (IMIDs) [ 61 , 62 ], suggesting that dysregulation of L-phenylalanine may reflect altered immune-metabolic status. In the setting of ANA-associated immune dysfunction, T cells may represent one possible contributor, given their established role in ANA-related immune responses and their capacity for L-phenylalanine uptake [ 33 , 63 – 65 ]. Therefore, we speculate that abnormal immune activation in ANA-positive infertile patients may be associated with altered consumption or redistribution of L-phenylalanine, which could contribute to reduced FF levels and impaired developmental competence. However, this interpretation remains hypothesis-generating, as the present study did not include direct immune profiling or functional validation. In addition, although early studies suggest that excess L-phenylalanine may inhibit antibody production [ 66 ], there is currently limited evidence regarding how reduced L-phenylalanine may participate in ANA-associated infertility. Further large-scale clinical studies and mechanistic investigations are needed to clarify its biological significance in this context. Previous studies have demonstrated that bioactive steroids, particularly androgens (e.g., testosterone, dihydrotestosterone), estrogens (e.g., estradiol, estrone), and progesterone, play important roles in regulating folliculogenesis, oocyte maturation, and early embryo development [ 67 , 68 ]. Interestingly, in our discovery cohort, although no statistically significant differences were observed between the negANA and posANA groups in these conventional steroid hormones, the steroid hormone biosynthesis pathway was markedly altered in ANA-positive infertile patients. This finding suggests potential involvement of non-classical or alternative steroid-related metabolic changes in the pathophysiology of ANA-associated infertility. Our correlation analysis further showed that hydrocortisone (cortisol) was significantly reduced in ANA-positive patients, and its concentration was positively associated with both embryo quantity and quality ( r =  0.333–0.402, P <  0.05). This reduction is broadly consistent with observations in systemic lupus erythematosus (SLE) and other autoimmune diseases characterized by ANA positivity, in which decreased cortisol and cortisone levels have been reported, possibly in relation to altered adrenal production or downstream metabolism [ 69 ]. These previous findings may provide a biologically relevant context for the reduced hydrocortisone observed in ANA-positive infertile patients in our study. Experimental evidence also suggests that glucocorticoids may influence reproductive processes. For example, supplementation with cortisol during bovine oocyte in vitro maturation has been reported to enhance blastocyst formation and upregulate GLUT1, FASN, and HSP70 expression [ 70 ]. In humans, FF cortisol levels have been positively correlated with total and mature oocyte counts [ 71 ], which is consistent with our findings. However, hydrocortisone may exert dose-dependent biphasic effects, as chronic or excessive corticosterone exposure has been shown to induce mitochondrial DNA mutations and impair oocyte maturation, fertilization, and embryo development [ 72 , 73 ]. Taken together, our findings suggest that altered hydrocortisone levels may be associated with the reproductive phenotype observed in ANA-positive infertile patients. However, given the observational design of this study, these results should be interpreted cautiously and should not be taken as direct evidence supporting glucocorticoid-based intervention. The potential clinical relevance of glucocorticoid-related metabolic alterations in this population requires further validation in well-designed prospective and mechanistic studies. Phospholipids play essential and highly complex roles in oocyte maturation and early embryogenesis, functioning in membrane structure, intracellular transport, signal transduction, and protein activity regulation [ 74 ]. In the present study, we observed a pronounced disruption in the glycerophospholipid metabolism pathway in ANA-positive infertile patients, accompanied by widespread lipid abnormalities in FF. Notably, LPA(18:2), PC(18:0/22:6), PC(20:4/20:4), and SM(d18:1/16:0) were significantly elevated in ANA-positive patients and were strongly associated with reduced oocyte and embryo quality. Consistent with our findings, previous studies in PCOS and DOR populations have demonstrated that lysophospholipids and glycerophospholipids, such as lysoPE and lysoPC, are strongly associated with oocyte competence and embryo development [ 75 , 76 ]. In our cohort, SM(d18:1/16:0) was persistently elevated in ANA-positive patients. Sphingolipids are key components of the cell membrane, involved in cellular signaling and apoptosis regulation. Their elevated levels may reflect adaptive responses to oxidative stress or impaired catabolism [ 77 ]. Prior studies have shown that sphingomyelin is enriched in small subordinate follicles [ 78 ], and its downstream metabolite ceramide is a well-established inducer of oocyte apoptosis [ 79 ]. These findings suggest that SM(d18:1/16:0) may impair oocyte developmental competence through apoptosis-related pathways. Although no direct evidence links serum ANA to lipid metabolic dysregulation, lipid metabolic rewiring is a recognized hallmark of T- and B-cell dysfunction in autoimmune diseases characterized by ANA positivity, such as SLE [ 80 , 81 ]. Patients with such disorders typically exhibit aberrantly increased lipid uptake and synthesis in immune cells, leading to lower circulating lipid levels [ 82 – 84 ]. As our study excluded patients with diagnosed autoimmune diseases, the absence of significant differences in baseline serum lipid profiles between groups is expected. Importantly, FF contains not only hormones, enzymes, electrolytes, and lipids but also immune cells including T and B lymphocytes [ 85 ]. Considering that lipid metabolic abnormalities are central to immune cell dysfunction in autoimmune diseases, we speculate that lipid dysregulation in the FF of ANA-positive infertile patients may reflect altered local immune cell activity, potentially reshaping the follicular microenvironment and impairing oocyte and embryo development. This hypothesis warrants further investigation through integrated immuno-lipidomic approaches. Previous studies have suggested that ANA titers may correlate with the activity of certain autoimmune diseases [ 86 ]. In ART cycles, high-titer ANA has been proposed as a potential predictor of poor ovarian stimulation response and adverse embryological outcomes, including fewer oocytes retrieved and fewer available embryos [ 17 , 87 ]. Nevertheless, other evidence indicates that ANA titers do not reliably reflect autoimmune disease activity and should not be used to monitor treatment response [ 52 ]. A retrospective cohort study demonstrated comparable oocyte maturation rates, fertilization rates, and implantation rates between patients with ANA titers ≤ 1:320 and > 1:320 undergoing IVF/ICSI, suggesting that the influence of ANA on ART outcomes may be independent of titer level [ 53 ]. Consistent with these findings, our study revealed no significant differences in metabolite or protein profiles within FF between the high-titer (≥ 1:160) and low-titer (1:80) ANA groups. In contrast, both ANA-positive subgroups exhibited marked molecular differences compared with the ANA-negative group. These observations indicate that the “presence” of ANA, rather than its “titer,” is the primary determinant of impaired oocyte and embryo quality, thereby providing molecular evidence for building ANA-titer–independent prognostic models. Specific anti-ENA antibodies—such as anti-dsDNA, anti-Scl-70, and ACA—have been strongly linked to impaired oocyte and embryo development. Their potential mechanisms include DNA topoisomerase disruption [ 16 , 17 ], and induction of apoptosis [ 18 , 19 ]. These mechanisms may explain why ANA titer alone does not predict IVF/ICSI outcomes, whereas specific ENA antibodies exert more direct biological effects. However, only a minority of ANA subtypes carry well-established clinical relevance [ 88 , 89 ], and many remain poorly characterized. Thus, studying ANA as an integrated immunological signal rather than parsing out individual subsets may more accurately reflect its biological impact [ 5 ]. Supporting this concept, previous studies have shown that:Patients who are ANA-positive but anti-dsDNA–negative have significantly fewer retrieved oocytes, lower fertilization rates, and fewer usable and high-quality embryos than ANA-negative women [ 20 ]; ANA-positive but ACA-negative patients also exhibit reduced oocyte retrieved and lower proportions of 2PN cleavage embryos [ 21 ]. In our study, even after excluding all clinically relevant anti-ENA antibodies, ANA positivity alone remained significantly associated with impaired oocyte competence and embryo development. This suggests that multiple unidentified autoantibodies or other immune mechanisms may act synergistically in ANA-positive infertile patients, ultimately affecting follicular physiology and embryogenesis. These findings highlight the complexity of ANA-related reproductive dysfunction and underscore the need for future mechanistic studies dissecting ANA subtypes and their biological effects. Metabolomic alterations in FF alone may not fully capture the metabolic activity of specific cell populations, thereby providing limited insights into the pathological mechanisms underlying ANA-related infertility [ 90 ]. To obtain a more comprehensive understanding, we performed an integrated multi-omics analysis combining proteomics and metabolomics derived from the same biological samples. This approach facilitates the identification of molecular changes that could be associated with disease processes, spanning from coding genes to end metabolites [ 91 ]. It should be noted that the integrative multi-omics strategy applied in this study was exploratory and hypothesis-generating. While concordant findings across multiple analytical approaches support the internal consistency of the results within this cohort, independent validation is still needed before broader biological or clinical inferences can be drawn. Previous FF proteomic studies in well-defined reproductive disorders—such as PCOS [ 92 ], DOR [ 30 ], and endometriosis [ 93 ]—have consistently reported large numbers of significantly altered proteins ( P  1.2 or < 0.67). In contrast, in our study, relatively few DEPs were identified between posANA and negANA groups under the criterion P <  0.05. This difference may be attributable to the fact that our participants were grouped solely based on serum ANA status, rather than on a clinically established disease phenotype. ANA-positive patients in this study did not exhibit overt pathological manifestations or pronounced embryological abnormalities, which likely resulted in more subtle protein expression changes compared with disease-specific cohorts. As the first multi-omics study profiling FF in ANA-related infertility, we applied a relatively inclusive threshold ( P <  0.05) to retain potentially relevant biomarkers. Under this integrative analytical framework, platelet degranulation emerged as the most significantly enriched pathway. Platelet degranulation has been implicated in folliculogenesis and oocyte maturation [ 94 – 96 ], and also plays a central role in the pathophysiology of various autoimmune diseases as part of the innate immune response [ 97 , 98 ]. Platelet degranulation is increasingly recognized as an integral component of innate immune activation and immunothrombosis, thereby linking coagulation-related pathways with immune dysregulation. Our subsequent network and correlation analyses revealed that several proteins involved in platelet degranulation—such as SERPINE1, CLU, PROC, KNG1, and SERPINF2—were strongly associated with embryonic developmental outcomes. Collectively, these findings suggest that ANA-associated perturbation of platelet degranulation may represent a key immune-coagulation mechanism through which the follicular microenvironment is altered, ultimately impairing oocyte maturation and early embryogenesis. This study has several limitations. The relatively small sample size may have introduced selection and statistical bias, potentially limiting the robustness and generalizability of the findings. Although patients with recurrent pregnancy loss and recurrent implantation failure were excluded to ensure population homogeneity, this exclusion may have reduced potential confounding while also introducing selection bias. As a result, the findings may not be generalizable to other infertile populations with different clinical profiles [ 99 , 100 ]. Further heterogeneity existed within the cohort, including variations in ovarian stimulation protocols, trigger strategies, fertilization methods (IVF vs. ICSI), and male factor characteristics. Despite efforts to minimize confounding, inter-individual variation in ovarian response may have contributed to differences in oocyte quality and follicular fluid metabolic profiles. The study focused on laboratory embryological outcomes in the first controlled ovarian stimulation cycle, without follow-up on clinical pregnancy or live birth outcomes. Future research should incorporate clinical follow-up data to assess whether the metabolic and proteomic signatures identified here correlate with pregnancy and live birth rates. Subgroup analyses were based solely on ANA titers, with ANA fluorescence patterns not further explored due to limitations in stratification based on sample size. Additionally, ANA levels in follicular fluid were not assessed, although previous studies suggest a closer association between follicular fluid ANA and oocyte developmental competence [ 11 , 12 ]. The exclusion of 17 common anti-ENA antibodies and technical challenges in subclassifying follicular fluid ANA prevented an in-depth investigation of the direct mechanisms by which ANA may affect oocytes or embryos. The chromosomal status of oocytes and embryos was also not evaluated. Given that the cohort consisted of non-advanced-age women and excluded infertility due to known chromosomal abnormalities, routine PGT-A was not clinically indicated. However, in the absence of chromosomal assessment, it remains unclear whether the impaired embryological outcomes in ANA-positive patients were partly due to increased oocyte-derived aneuploidy. Finally, the metabolite panel identified in this study was derived from an exploratory cohort without external validation. While the three-metabolite model demonstrated excellent discriminatory performance within this dataset, ROC analysis cannot establish clinical predictive utility or mechanistic causality. Additionally, the lack of appropriate animal models limited functional validation of the selected biomarkers. Larger multicenter studies with independent validation are necessary to confirm the reproducibility, generalizability, and clinical relevance of these findings. Despite these limitations, this study is the first to integrate a prospective cohort with multi-omics profiling to systematically evaluate oocyte and embryo quality in serum ANA-positive but anti-ENA-negative infertile patients undergoing their first IVF/ICSI cycle. Our findings suggest that metabolic alterations within the follicular microenvironment are associated with impaired oocyte and embryo developmental competence in ANA-positive patients. Notably, L-phenylalanine, SM(d18:1/16:0), and hydrocortisone emerged as candidate molecular indicators associated with ANA status and early embryological outcomes. Their relative stability and consistent expression patterns across the ANA-titer subgroups examined suggest that these features may inform the future development of titer-independent prognostic models. However, their reproducibility and clinical relevance require confirmation in larger, independent cohorts. This study advances our understanding of ANA-related reproductive pathology and highlights potential molecular targets for future mechanistic and therapeutic investigations.

Antinuclear antibodies (ANAs) are a group of autoantibodies that target intracellular structures and organelles, including the cell nucleus, cytoplasmic components, and mitotic structures [ 1 , 2 ]. According to the 2019 EULAR/ACR criteria, ANA positivity is defined by indirect immunofluorescence on HEp-2 cells with a titer ≥ 1:80 or by solid-phase assays with comparable sensitivity [ 2 ]. As one of the most common biomarkers of autoimmune diseases, studies have shown that approximately 15% of the healthy population test positive for ANA, with a higher prevalence among females; however, its exact clinical significance remains unclear [ 3 ]. In ANA-positive cases, further testing for extractable nuclear antigens (ENAs) can aid in diagnosing systemic autoimmune diseases, such as systemic lupus erythematosus (SLE) and Sjögren's syndrome (SS) [ 1 , 4 ]. However, in clinical practice, discrepancies between ANA screening results and specific autoantibody profiles are commonly observed. Specifically, approximately 40% of ANA-positive individuals have negative anti-ENA antibody tests [ 5 , 6 ]. This discrepancy is partly attributed to differences in target antigen epitopes, sensitivity, and specificity among different detection platforms. Nonetheless, ANA positivity is relatively prevalent in otherwise healthy populations, and the clinical management and diagnostic guidance for individuals with ANA positivity but negative anti-ENA antibodies remain insufficient. This discrepancy underscores an unmet clinical need to better understand the biological and clinical implications of ANA positivity in the absence of disease-specific autoantibodies. As of 2021, global infertility prevalence is estimated to range from 12.6% to 17.5% [ 7 ]. Among infertile patients undergoing assisted reproductive technology (ART), approximately 20% test positive for ANA, and fluctuations in hormone levels during controlled ovarian stimulation (COS) may further exacerbate ANA titers [ 2 , 8 ]. Accumulating evidence indicates that ANA positivity is closely associated with adverse ART outcomes, including impaired oocyte and embryo quality, reduced implantation rates, and poorer pregnancy outcomes [ 8 – 14 ]. Mechanistic insights into ANA-associated reproductive dysfunction have primarily been derived from studies focusing on specific anti-ENA antibodies. For instance, disease-specific anti-ENA antibodies, including anti-dsDNA, anti-Scl-70, and anti-centromere antibodies, have been shown to induce apoptosis [ 15 ], disrupt DNA replication [ 16 , 17 ], or impair meiotic spindle–centromere interactions [ 18 , 19 ], collectively leading to compromised oocyte maturation and embryonic development. However, clinical observations consistently demonstrate that even infertile women who are ANA-positive but anti-ENA antibody–negative still exhibit significantly lower oocyte retrieval rates, fertilization rates, numbers of two-pronucleus (2PN) embryos, available embryos, and high-quality embryos compared with ANA-negative counterparts [ 20 – 22 ]. This suggests that, in ANA-positive infertile women, in addition to the direct effects of individual antibodies, there may be synergistic interactions among multiple antibodies or other indirect mechanisms affecting oocyte and embryo development. Supporting this notion, in a mouse model, induced ACA was detected only in follicular fluid (FF) and early embryos, but not in oocytes, further indicating that ANA may impact meiosis without directly entering oocytes [ 18 , 19 ]. Given that ANAs comprise a heterogeneous group of more than one hundred autoantibodies, systematically dissecting the contribution of each antibody is impractical [ 23 , 24 ]. Consequently, ANA positivity has been proposed to function as an integrated immune marker that may better capture the overall autoimmune status relevant to reproductive outcomes [ 5 ]. Nevertheless, this concept has not yet been systematically investigated in infertile women undergoing ART. FF represents the immediate microenvironment surrounding the oocyte and integrates signals derived from local somatic cells and systemic physiological states [ 25 , 26 ]. As such, alterations in FF composition can reflect disturbances in follicular homeostasis that are not readily captured by circulating biomarkers. Accumulating evidence indicates that FF molecular profiling provides valuable insight into immune, endocrine, and metabolic perturbations associated with impaired oocyte competence and adverse ART outcomes [ 27 – 32 ]. ANA positivity, which reflects an underlying autoimmune abnormality, may similarly be regarded as an exposure-related immune state rather than the effect of a single pathogenic antibody [ 33 , 34 ]. Although clinical studies consistently report poorer oocyte and embryo quality in ANA-positive infertile women, the biological basis of these associations remains unclear, particularly in patients without detectable anti-ENA antibodies. This raises the possibility that isolated ANA positivity exerts its effects indirectly by reshaping the follicular microenvironment, rather than through direct antibody-mediated damage to oocytes or embryos. To test this hypothesis, the present study employed integrated proteomic and metabolomic analyses of FF from infertile women with isolated ANA positivity (serum ANA ≥ 1:80 and anti-ENA negative) and ANA-negative controls. By characterizing FF molecular alterations and examining their associations with oocyte and embryo developmental competence across different ANA titer levels, this study aims to elucidate immune–metabolic features underlying ANA-associated infertility and to identify candidate molecular biomarkers relevant to ART outcomes.

Additional file 1. Tables and figures selected for online publication. Additional file 2. Differentially expressed metabolites of the comparison between posANA group and negANA group. A Detailed information on differentially expressed metabolites between the posANA and negANA groups. Candidate DEMs were identified at the initial exploratory screening stage using VIP ≥ 1 and a nominal, unadjusted P value < 0.05. B Detailed KEGG pathway enrichment of differentially expressed metabolites. Additional file 3. Differentially expressed proteins of the comparison between posANA group and negANA group. A Detailed information on differentially expressed proteins between the posANA and negANA groups. Differential proteins were identified based on the criteria of P < 0.05. B Protein expression profiles across different ANA titers. H indicates an ANA titer ≥ 1:160, whereas L indicates an ANA titer of 1:80. Additional file 4. Functional and protein–protein interaction (PPI) analysis of differentially expressed proteins. A GO functional analysis of differentially expressed proteins using the ClueGO plugin in Cytoscape. B Topological properties of key proteins (nodes) within the functional network. C Functional enrichment analysis of proteins based on the Metascape platform. D Protein–protein interaction (PPI) analysis based on the STRING database. Additional file 5. Integrated analysis of 38 differentially expressed metabolites with annotated KEGG pathways and 50 differentially expressed proteins. A Detailed information on 38 differentially expressed metabolites. B Co-enriched pathways of related metabolites and proteins based on the MetaboAnalyst platform. C Detailed statistical information of the co-enriched pathways. Additional file 6. Correlation analysis of LASSO-selected proteins and metabolites with baseline characteristics and embryological parameters. A Spearman correlation coefficients between key molecules and clinical as well as embryology laboratory parameters. B P values of Spearman correlation coefficients between key molecules and clinical as well as embryology laboratory parameters. A key molecule was considered to be closely associated with embryology laboratory parameters if at least one Spearman correlation coefficient had an absolute value ≥ 0.4 with a corresponding P value < 0.05. Additional file 7. Proteomic analysis across three ANA-titer groups. Additional file 1. Tables and figures selected for online publication. Additional file 2. Differentially expressed metabolites of the comparison between posANA group and negANA group. A Detailed information on differentially expressed metabolites between the posANA and negANA groups. Candidate DEMs were identified at the initial exploratory screening stage using VIP ≥ 1 and a nominal, unadjusted P value < 0.05. B Detailed KEGG pathway enrichment of differentially expressed metabolites. Additional file 3. Differentially expressed proteins of the comparison between posANA group and negANA group. A Detailed information on differentially expressed proteins between the posANA and negANA groups. Differential proteins were identified based on the criteria of P < 0.05. B Protein expression profiles across different ANA titers. H indicates an ANA titer ≥ 1:160, whereas L indicates an ANA titer of 1:80. Additional file 4. Functional and protein–protein interaction (PPI) analysis of differentially expressed proteins. A GO functional analysis of differentially expressed proteins using the ClueGO plugin in Cytoscape. B Topological properties of key proteins (nodes) within the functional network. C Functional enrichment analysis of proteins based on the Metascape platform. D Protein–protein interaction (PPI) analysis based on the STRING database. Additional file 5. Integrated analysis of 38 differentially expressed metabolites with annotated KEGG pathways and 50 differentially expressed proteins. A Detailed information on 38 differentially expressed metabolites. B Co-enriched pathways of related metabolites and proteins based on the MetaboAnalyst platform. C Detailed statistical information of the co-enriched pathways. Additional file 6. Correlation analysis of LASSO-selected proteins and metabolites with baseline characteristics and embryological parameters. A Spearman correlation coefficients between key molecules and clinical as well as embryology laboratory parameters. B P values of Spearman correlation coefficients between key molecules and clinical as well as embryology laboratory parameters. A key molecule was considered to be closely associated with embryology laboratory parameters if at least one Spearman correlation coefficient had an absolute value ≥ 0.4 with a corresponding P value < 0.05. Additional file 7. Proteomic analysis across three ANA-titer groups.