JBRA Assist. Reprod. 2025;29(Suppl 1):7-7
ORAL PRESENTATION
doi: 10.5935/1518-0557.20250065
1Universidad Simón Bolívar-Department Life Sciences Research Center - Barraquilha - Colombia
2Institute Human Reproduction Procrear Esperanza de Vida - Barraquilha - Colombia
3University of Puerto Rico - Mayagüez - Department of Chemistry
Objective: Infertility affects approximately one in six people globally, with the male factor involved in up to 70% of cases. Conventional sperm evaluation methods assessing concentration, motility, and morphology frequently fail to elucidate idiopathic infertility. Raman spectroscopy (RS), enhanced through Surface-Enhanced Raman Spectroscopy (SERS), offers a non-invasive, highly specific molecular "fingerprint" analysis. The objective of this study was to develop a predictive Partial Least Squares Discriminant Analysis (PLS-DA) model using SERS for classifying fertile and infertile human semen samples by identifying key molecular differences.
Methods: Semen samples were collected from 93 male patients (45 infertile, 48 fertile controls) following 2-5 days of sexual abstinence. Gold nanoparticles (AuNPs) were synthesized and characterized (32±15nm) using dynamic light scattering. Sperm samples were processed with AuNPs to enhance Raman signals. SERS spectra were obtained using a 785 nm laser. Spectral preprocessing included Savitzky-Golay smoothing, baseline correction, and Standard Normal Variate normalization. The PLS-DA model was developed and validated through internal cross-validation and external validation, evaluating performance via confusion matrix parameters, sensitivity (TPR), specificity (TNR), accuracy (ACC), and receiver operating characteristic (ROC) analysis.
Results: SERS spectral analysis revealed distinct biochemical differences between fertile and infertile groups. Significant Raman peaks at 391.1cm-1 (S-S stretch), 730.34cm-1 (DNA methylation), 1310.77cm-1 (guanine nucleic acids), and 1560.04 m-1 (proteins from the equatorial segment and apical acrosomal vesicle) were identified. Univariate analysis indicated significant differences (p<0.0001) between fertile and infertile groups in intensities at these spectral regions. DNA methylation at 730 cm-1 demonstrated high specificity as a potential biomarker for assessing genetic integrity in spermatozoa, indicating a possible association with impaired sperm DNA packaging and increased susceptibility to fragmentation and oxidative damage. The S-S stretch at 391.1 cm-1, associated with protein structures and lipid residues primarily found in sperm membranes, highlighted differences in the biochemical composition of sperm membranes between groups, potentially influencing sperm function and motility. The Raman band at 1310.77 cm-1, corresponding to nucleic acids and guanine residues, suggested compromised genetic content integrity in infertile samples. Moreover, the intensity peak at 1560.04 cm-1 reflected alterations in protein integrity specifically within the equatorial segment and apical acrosomal vesicle, structures crucial for sperm-oocyte interaction and successful fertilization. The PLS-DA predictive model classified semen samples with high accuracy, achieving sensitivity and specificity values of 85.29% and 83.33% in the training set, and 90.91% and 83.33% in the test set, respectively. ROC curve analysis indicated robust model performance, with an Area Under Curve (AUC) of 0.9395 (cross-validation) and 0.9470 (external validation). Bivariate analysis combining DNA methylation and nucleic acid wagging correctly classified 82.8% of cases (sensitivity 80%, specificity 85.42%). Trivariate analysis, including wagging nucleic acid, further improved classification accuracy to 87.10% (sensitivity 82.22%, specificity 91.67%). The inclusion of additional variables beyond these three did not significantly enhance accuracy.
Conclusion: The application of SERS combined with PLS-DA effectively discriminated between fertile and infertile sperm samples based on specific molecular signatures, predominantly DNA methylation, guanine nucleic acids, and structural protein integrity. This non-invasive, label-free approach achieved high diagnostic accuracy, sensitivity, and specificity, highlighting its potential for clinical adoption to identify male infertility factors, particularly in idiopathic cases. The method's ability to detect molecular-level sperm damage, such as compromised genetic integrity and protein structure, could greatly enhance diagnostic accuracy and improve outcomes in assisted reproductive technologies (ART).