Metabolomic Analysis Reveals Association between Decreased Ovarian Reserve and In Vitro Fertilization Outcomes

Abstract

In vitro fertilization (IVF) is a highly effective treatment for infertility; however, it poses challenges for women with decreased ovarian reserve (DOR). Despite the importance of understanding the impact of DOR on IVF outcomes, limited research has explored this relationship, particularly using omics approaches. Hence, we conducted a study to investigate the association between DOR and IVF outcomes, employing a metabolomic approach. We analyzed serum samples from 207 women undergoing IVF treatment, including 89 with DOR and 118 with normal ovarian reserve (NOR). Our findings revealed that DOR was significantly associated with unfavorable IVF outcomes, characterized by a reduced oocyte count, lower embryo quality, and decreased rates of pregnancy and live births. Furthermore, we identified 82 metabolites that displayed significant alterations in DOR patients, impacting diverse metabolic pathways. Notably, a distinct panel of metabolites, including palmitic acid, stearic acid, LysoPC(9:0(CHO)/0:0), PC(18:0/9:0(CHO)), and PC(16:0/9:0(CHO)), exhibited discriminatory power between the DOR and NOR groups, showcasing a strong correlation with IVF outcomes. These findings emphasize the crucial role of metabolomic disruptions in influencing IVF outcomes among women with DOR.

Keywords: diminished ovarian reserve; human serum; in vitro fertilization; metabolomics.

Figure 1

Overview of the study design. (A) Sample collection. (B) Metabolomic workflow. (C) Association workflow. DOR, diminished ovarian reserve; NOR, normal ovarian reserve.

Figure 2

Identification of differential metabolomics profiles in serum between DOR and NOR. Score plots of PCA (A) and OPLS-DA (B) based on the combinational data of RPLC-ESI(+) TOF-MS, RPLC-ESI(−) TOF-MS, HILIC-ESI(+) TOF-MS, and HILIC-ESI(−) TOF-MS from the discovery set. The pink circles represent DOR; the blue circles represent NOR; the green circles represent QC samples. (C) Volcano plot, down-regulated, up-regulated, and not significantly changed metabolites in DOR compared to NOR are marked in blue, red, and grey, respectively. (D) Distribution of metabolites across super/sub-classes in the discovery set.

Figure 3

Associations between differential metabolites. (A) Chord diagram displaying the Pearson correlation of the superclasses for the differential metabolites between DOR and NOR. (B) Debiased sparse partial correlation network analysis illustrating the differential correlation between individual significantly different metabolites. The node size of each metabolite is reflected by its betweenness centrality (how frequently a metabolite occurs on the shortest paths between other metabolites). The thickness of the lines connecting metabolites is scaled in relation to the -lg (adjust p-values). Metabolite names are listed in the legend.

Figure 4

(A) Pathway analysis of significantly different metabolites in DOR according to the KEGG pathway. (B) Human disease states that correlated with DOR-related metabolites on the basis of published metabolomics data.

Figure 5

Performance of the biomarker signature for the diagnosis of DOR. (A) Boxplots of the five DOR-associated metabolites in the discovery set and validation set: (A1) palmitic acid, (A2) stearic acid, (A3) LysoPC(9:0(CHO)/0:0), (A4) PC(16:0/9:0(CHO)), and (A5) PC(18:0/9:0(CHO)). **, 0.001 < p < 0.01; ***, p < 0.001. (B) Receiver operating characteristics curves and corresponding area under the curve (AUC), confidence interval, and the sensitivity and specificity of the biomarker signature for differentiating DOR from NOR. (B1) Discovery set; (B2) Validation set. (C) Heatmap of the Spearman correlation coefficients between five DOR-associated metabolites and clinical parameters. The colors in the heatmap represent the positive (represented by red) or negative correlation (represented by blue). *, p ≤ 0.05. E2, Estradiol; P, Progesterone; LH, Luteinizing hormone; T, Testosterone; PRL, Prolactin; FT3, Free triiodothyronine; FT4, Free thyroxine; TSH, Thyroid-stimulating hormone; ALT, Alanine aminotransferase; AST, Aspertate aminotransferase; ALP, Alkaline phosphatase; STB, Serum total bilirubin; TP, Total protein; GGT, Glutamyl transpeptidase; TC, Total cholesterol; Hb, Hemoglobin; LDH, Lactate dehydrogenase; Cr, Creatinine; UA, Uric acid.

Figure 6

The associations between serum metabolites and IVF outcomes based on GLM models. The models were adjusted by age (continuous), body mass index (BMI, <25.0 kg/m² vs. ≥25.0 kg/m²), passive smoking status (yes vs. no), alcohol status (never vs. ever/current), educational level (less than high school vs. high school and above), income (≤5000 vs. >5000 yuan/month) and infertility diagnosis (female factor, male factor, mixed factor vs. unexplained). Data for count and proportional outcomes are presented as adjusted β (95% CI) and for binary outcomes as adjusted RR (95% CI).