Abstract
Human seminal fluid is a complex mixture of secretions originated from epididymis and the male accessory sex glands. It contains a variety of both inorganic and organic components, among which proteins are a major part of the high molecular-mass substances. In this study, 83 human seminal plasma samples were analyzed using a combined Nuclear Magnetic Resonance (NMR) Spectroscopy and Principal Component Analysis (PCA) approach to discriminate patients in relation to semen characteristics and/or conditions affecting the fertility status. Results showed a discrimination between patients with leukocytospermia and with the concomitant presence of varicocele/ex varicocele and leukocytospermia. Patients with testicular cancer, necrozoospermia, and azoospermia were separated from the other patient clusters. In addition, a differentiation of semen quality was also possible. This study represents to first use of sperm parameters together with NMR data as variables in the PCA analysis. Furthermore, this methodology allows the identification of the metabolites which play the most important role in identifying differences among human seminal plasma samples.
Introduction
Nuclear Magnetic Resonance (NMR) spectroscopy provides a useful approach for the simultaneous measurement of many metabolites in biofluids. In particular, NMR is a simple, reliable tool for exploring alterations in the patterns of metabolites present in biological materials and can give diagnostic information and mechanistic insights into the biochemistry of disease processes [Nicholson and Wilson, Citation1989].
More than 70 different metabolic disorders have been studied by the analysis of biofluid by NMR techniques [Brown et al. Citation1989; Holmes et al. Citation1997; Lindon et al. Citation1999; Wevers et al. Citation1999]. Some studies have shown their potential utility in the analysis of male infertility and for understanding the human testicular function [Bahl et al. Citation1988; van der Grond, Citation1995]. More recently, 1H NMR studies in boar spermatozoa have found glycolytic pathways as important in maintaining a supply of lactate for mitochondrial ATP production [Jones and Bubb, Citation2000]. In other studies using ornidazole, a glycolysis inhibitor, 13C labeled substrates coupled with 13C NMR detection have been used to estimate glycolytic turnover suggesting that glycolysis supports hyperactivation as well as zona binding and penetration into spermatozoa [Bone et al. Citation2000]. Gupta et al., [Citation2011a, Citationb] have used NMR spectroscopy to measure quantitatively the concentration of some metabolites and linear discriminant function analysis to determine the signature biomarkers and descriptors of infertility. Recently, Mc Rae et al., [Citation2012] utilized 1H NMR to investigate changes in follicular fluid and plasma composition in patients undergoing assisted reproduction. The ability to interpret the metabolic responses of different diseases by NMR studies of altered composition of human seminal plasma [Hamamah et al. Citation1998; Tomlins et al. Citation1998], depends upon the availability of databases of 1H NMR spectra of a large number of samples. Their analysis can be managed through data reduction and pattern recognition techniques such as Principal Component Analysis (PCA) [El-Deredy, Citation1997].
PCA is a multivariate data analysis technique that transforms the variables of the experimental data into principal components (PCs), which are linear combinations of the original input descriptors. The resulting PC maps can be used to visualize any clustering patterns associated with a biological response. The overall distribution of information (eigenvalue analysis) can be used to examine correlations between variables (loading plot), to identify objects (score plot), outliers, clusters, and eliminate noise and useless information, as well as to define the principal properties of the system under consideration [Massart et al. Citation1988].
In this study, 1H-NMR spectroscopy, sperm biological parameters, and multivariate analysis were combined to identify correlations among metabolites in human seminal plasma and semen parameters and between metabolites present in different pathologies/conditions. This allowed the discrimination of unselected patients with respect to semen quality.
Results and Discussion
On the basis of the methods reported in the Materials and Methods section semen collection was performed following WHO [Citation2010] guidelines. All semen samples were treated following the same protocol in order to assess the sperm parameters, including motility. The seminal plasma was liquefied at 37°C for 30 minutes, and then centrifuged (10 minutes), and finally frozen [Gupta et al. Citation2013, Jayaraman et al. Citation2014]. Before the NMR analysis, the samples were lyophilized and resuspended in deuterium oxide. shows the NMR proton spectrum of a human seminal plasma sample in D2O solution acquired at 600 MHz and 298 K. The assignment reported in detail in was obtained by bidimensional NMR spectra (data not shown) according to previous papers [Gupta et al., Citation2011a,Citationb; Averna et al. Citation2005; Gupta et al. Citation2013; Lynch et al. Citation1994; Patel et al. Citation1999; Seaglen et al. Citation1995]. The enlarged signals at 7.18, 6.92, 6.89, and 6.75 ppm, marked with an asterisk in , in the region of aromatic protons of phenylalanine, tyrosine, and histidine, show a characteristic line width broadening of proton resonances. In fact, 1H spectrum of seminal plasma shows resonances having different line width and resolution. These characteristics were much more evident at low field, between 6.5 and 8.0 ppm, where a number of peaks having a wider line width were observed. We hypothesized that the signal spectral line width broadening may be due to the presence of macromolecular systems which determines a slow reorientational dynamics [Wuthrich, Citation1986].
Figure 1. 1H NMR spectrum of human seminal plasma in D2O. This proton spectrum was recorded at 600 MHz and 298 K without TMS as internal standard. The asterisks show signal line width broadening at 7.18, 6.92, 6.89, and 6.75 ppm, assigned to phenylalanine and tyrosine residues linked to macromolecules.
Table 1. Assignments of resonances in proton Nuclear Magnetic Resonance spectrum of human seminal plasma.
All human seminal fluid samples were analyzed using the same procedures. NMR spectra were recorded for 83 human seminal plasma samples and all NMR integrals (assigned and non-assigned NMR signals), obtained as reported in the Experimental section, were used to construct the matrix of experimental data for the PCA approach. The objective of the study was to explore a PCA method to NMR integrals on semen samples, without investigating proteolytic and/or enzymatic processes.
Due to the high number of variables, a statistical approach was used with the aim of identifying correlations between metabolites, seminal parameters, and different conditions related to fertility (varicocele, ex varicocele, presence of leukocytes) detected in the patients under study. Multivariate analysis techniques allow the extraction of the most relevant information from the experimental data reducing the complexity of the data set. PCA was used to reduce the set of original variables and to extract a small number of latent factors (PCs) for analyzing relationships among them and understand the biological implications [Jolliffe, Citation2002].
Previous studies used only NMR parameters as variables in the statistical data analysis of biological samples. In the present study, NMR data have been coupled to seminal parameters, i.e., semen volume, sperm concentration, progressive motility, and morphology, with the aim to increase the understanding of the complexity of the system. The score plot PC1 vs. PC2 is reported in .
Figure 2. Statistical analysis results. Score plot of first principal component (PC1) vs. the second (PC2) illustrating the distribution of the 83 human seminal plasma samples. Samples are indicated by consecutive numbers. The first information arising from this analysis was that sample 54 had a higher value of residual variance with respect to the other samples.
The first information arising from this analysis was that sample 54 had a higher value of residual variance with respect to the other samples. Since the sample fell outside the normal range of the response data, it can be identified as the only outlier coming from the application of PCA to the entire set of samples. Its deletion constitutes a tool to extract information from the application of PCA to the rest of the data; these results are shown in a new two-dimensional PC space, where the abscissa and ordinate axis represent each PC (). The analysis of the eigenvalues as a function of the main components indicated that the first three components were able to explain 85% of the total variance. shows the variance explained by the first eight components. The first principal component (PC1) explained 60% of the overall variance while the second (PC2) explained 13% of the simple variance. Together they explained the 73% of the variance of the original dataset and the third represents only the 12% variance. PC1 and PC2 eigenvalues were the only values higher than one (EV1 = 4.75 and EV2 = 1.08) ().
Figure 3. Principal Component Analysis (PCA) results without sample 54. First principal component (PC1) and the second (PC2) scatter plot of all data collected without the sample no. 54 (patient was affected by necrozoospermia). The cluster ellipses were determined according to a cluster analysis. Cluster on the left has been indicated by number 1 while the other one with number 2.
Table 2. Principal components results: eigenvalues and individual variance percent.
A cluster analysis was used to identify samples that shared a common location in the PC1-PC2 space. The resulting clusters were obtained using an unweighted pair group clustering method which calculates Euclidean distance between samples and a standard deviation. Similarity of 60% was used as a cut-off value (). The cluster analysis resulted in two clusters of samples falling in the same area of the PC1-PC2 space and five outliers (). A description of demographic data, infertility-related pathologies, and fertility status of 83 consecutive patients included in the study was reported ().
Figure 4. Dendrogram representation of the samples data (X-axis: 83 human seminal samples and Y-axis: % of similarity). The resulting clusters were obtained using an unweighted pair group clustering method which calculates Euclidean distance between samples and a standard deviation. Sixty percent of similarity was used as a cut-off value.
Table 3. Number of patients, age (mean ± standard deviation), country, pathologies, and fertility status of the group of 83 consecutive patients included in the study.
The first result from PCA was the identification of six patients as outliers. Patient 54 was affected by necrozoospermia (a pathology characterized by 100% non-viable sperm detected in at least three ejaculates of the patient), patients 34 and 28 were azoospermic, patient 2 was affected by testicular cancer (he was carrier of an implant in left testis and the patient's right testis was operated on for the same cancer), patient 3 and 51 were fertile men (they fathered during the last two years). All the other 77 patients were included in cluster 1 or in cluster 2. In the two considered clusters four other azoospermic patients were present: patients 1, 29 and 69 in cluster 2; patient 30 in cluster 1. In seminal samples from patients 1, 29, and 69 a condition of leukocytospermia was also highlighted. Excluding these azoospermic patients in which the sperm parameters cannot be determined, the mean percentages of semen volume, sperm concentration, progressive motility, and normal morphology were evaluated according to WHO [Citation2010] guidelines in both clusters (). A better quality in seminal parameters was observed in patients included in cluster 1; particularly considering progressive motility and normal morphology (). All patients declared any drug or alcohol consumption. Nine patients were smokers (more than 10 cigarettes daily), three of which were present in the outlier group (patient 28, 54, 51), while the remaining six smokers were equally distributed in both clusters. As this factor (smoking) was present in both clusters as well as in the outlier group, it was not included as a discriminatory factor for our study.
Table 4. The number of patients not showing or showing pathologies such as varicocele, ex varicocele, presence of leukocytospermia, leukocytospermia associated with varicocele, or ex varicocele and the means and standard deviation of seminal parameters from patients included in clusters 1 and 2.
In cluster 1, 18 patients out of 39 showed a history of varicocele (8 varicocele, 10 ex varicocele) and 21 subjects did not have a history of varicole or leucocytospermia. In cluster 2, nine patients out of 38 showed leukocytospermia, five patients had the presence of varicocele associated with leukocytospermia, five had ex varicocele associated with leukocytospermia, and two had the presence of a varicocele only. Finally 17 individuals did not show any of these conditions.
After the characterization of the samples within the clusters, the identification of the variables affecting PCA results were carried out by the calculation of the loading factors for each variable within each PC. shows the loading plot of PC1 and PC2, which displays the metabolites playing the most important role in the clustering processes. The figure suggests that variables referred to the integrals of some NMR signals as well as concentration, motility, and morphology of spermatozoa, contributed the most to the discrimination among samples and identifies the metabolites playing the most important role in the detection of differences among samples.
Figure 5. Loading plot of first principal component (PC1) and the second (PC2) variables. The plot suggests which metabolites contribute to the discrimination among samples playing a role in the detection between seminal plasma of unselected patients.
Regarding PC1, which explains the largest variance, the variables which play the most significant role are glycerophosphorylcholine (GPC) and a group including choline, phenylalanine, citrate, lactate, and histidine. For PC2 only uridine was the most important variable. Moreover the plot () shows that the uridine and GPC metabolites were directly correlated to each other, while the other variables are indirectly related.
Direct correlations between pairs of variables have been found by PCA of the entire dataset (83 samples) with R-squared >0.995 and p < 0.001. The correlations that gave the best statistical fitting were histidine versus glycine, tyrosine, citrate, and valine; valine versus citrate, phenylalanine, and histidine; citrate versus tyrosine and spermine; glycine versus leucine. On the contrary, sperm concentration, motility, morphology, and volume did not show any linear correlation towards the other variables, as aminoacids, citrate, or spermine.
The next step was to repeat the statistical analysis using cluster 1 and 2 separately. Within cluster 1 we found linear correlations between histidine and several amino acids (phenylalanine, tyrosine, glycine, choline, leucine, isoleucine) and between tyrosine and lysine, glycine, isoleucine, and leucine. Citrate was the only variable which did not show any correlation. Cluster 2 gave different results about variables involved in linear correlations. In particular, ATP showed numerous linear dependences with several aminoacidic residues such as histidine, tyrosine, choline, lysine, glycine, isoleucine, and leucine. No correlations were found for uridine and citrate.
In this study 83 samples of human seminal plasma were analyzed using a combined NMR/PCA approach as a challenge to discriminate patients in relation to their semen characteristics and/or conditions that may influence the fertility status (the presence of varicocele, ex varicocele, and leukocytospermia). We developed this methodology by including important biological data (concentration, motility, and morphology of spermatozoa) in the set of the variables used in the PCA analysis that enabled the identification of two clusters of patients. The metabolites playing the most important role in the detection of differences among samples were GPC, a group of metabolites (choline, phenylalanine, citrate, lactate, and histidine), uridine together with seminal parameters (concentration, motility, and morphology). Seminal uridine is likely to be an essential precursor to metabolites required for capacitation and is a potential marker for the prognosis of post-spinal cord injury functional fertility recovery [Maher et al. Citation2008].
Within cluster 1, citrate was the only variable which did not show any linear correlation with the other variables. In fact, citrate is the unique prostatic metabolite in seminal plasma investigated by high resolution NMR analysis as a marker to diagnose prostatic cancer [Averna et al. Citation2005] or in men with spinal cord injury [Alexandrino et al. Citation2009]. None of these conditions were detected in our patient population. In cluster 2 no correlations were found for uridine and citrate.
NMR was able to discriminate different azoospermic conditions, such as maturation arrest or histological pattern of Sertoli-cell only [Aaronson et al. Citation2010] and obstructive from non-obstructive azoospermia [Roy et al. Citation2001]. The importance of the applied method resides in the ability to discriminate among seminal specimens of unselected patients between cluster 1 that showed a better semen quality than cluster 2. In our population cluster 2 was characterized by a decreased semen quality and by the presence of all the cases with leukocytospermia only or associated with the other conditions. It is known that leukocytospermia has a negative impact on semen quality due to the production of the reactive oxygen species that are able to decrease sperm motility [Cavagna et al. Citation2012]. We can hypothesize that in cluster 2 leukocytospermia may cause a reduction in sperm motility, altering the ability of the spermatozoa to utilize substrates involved in ATP production and subsequently fertility [Goodson et al. Citation2012]. In this cluster, ATP showed several linear dependences with aminoacidic residues such as histidine, tyrosine, choline, lysine, glycine, isoleucine, and leucine.
The presence of patients with varicocele or with a past history of varicocele was almost equally distributed in clusters 1 and 2, since this pathology may have different impact on the quality of seminal fluids related to considered metabolites. The method was able to allocate in cluster 2 all the patients with the concomitant presence of varicocele/ex varicocele and leukocytospermia. Moreover, data show that patients without the considered pathologies were distributed in the two clusters.
In conclusion, NMR spectroscopy coupled to statistical analysis (PCA) was applied to an unselected group of patients with the aim of developing a statistical method to investigate complex biological samples. The novelty of this work resides in the use of sperm parameters together with NMR data as variables in the PCA analysis. In the first step of the statistical analysis, patients with testicular cancer, necrozoospermia, azoospermia, and fertile men came out as outliers. After removing these samples from further analysis, the method enabled the identification of two populations of men with different semen quality. In addition all patients with leukocytospermia were located in a specific cluster.
Materials and Methods
Semen samples
All patient information was assigned with a progressive number. Patients signed an informed consent following Ethics Committee regulations (Ethics Committee of Azienda Ospedaliera Universitaria Senese, CEAOUS). Semen samples were obtained from 83 men (aged between 20-41 y) consecutively attending at the Interdepartmental Centre for Research and Therapy of Male Infertility, University of Siena, for infertility problems or from those requesting fertility testing. Each subject filled a form specifying the presence/absence of varicocele, indicating a resolved history of varicocele (ex varicocele). The volume, position, and consistency of the testes and epididymis were checked, and each spermatic cord was palpated in the standing position and during Valsalva maneuver. None of the patients received any therapy before semen analysis. Patients were questioned about their smoking habits, and alcohol and drugs.
Semen samples were collected by masturbation after 4 d of sexual abstinence and examined after liquefaction for 30 min at 37°C. Volume, pH, sperm concentration, and progressive motility were evaluated according to WHO [Citation2010] guidelines. Sperm morphology was assessed by the Papanicolaou (PAP) staining modified for spermatozoa following the WHO [Citation2010] guidelines. Leukocytes were identified by peroxidase stain; a concentration higher than 1x106 cell/ml was considered out of range and this condition was considered to be leukocytospermia. Five hundred µl of each whole semen sample were fractioned by centrifugation (1,500 rpm for 15 min) 1 h after collection. The supernatant composed of seminal plasma without spermatozoa was recovered and maintained at −80°C until analysis was performed.
NMR experiments
The collected human seminal plasma samples were lyophilized and resuspended in 500 µL of deuterium oxide in order to decrease the NMR water signal. 1H NMR measurements were made on a Bruker AMX-600 Avance spectrometer operating at 600 MHz and 298 K. 1H NMR spectra of the resulting solutions were acquired using the pulse sequence with presaturation of the solvent signal (zgpr pulse sequence). The 128 free induction decays (FIDs) were collected into 64 K data points using a spectral width of 6009 Hz, utilizing a 8.5 µs pulse width and a relaxation delay (D1) of 4.0 s [Jayaraman et al. Citation2014]. For each sample, the proton NMR signals were integrated after the baseline correction and using the internal standard (Tetramethylsilane, TMS, 0.0 ppm) in coaxial NMR tubes. All the spectra were processed using the Bruker Software XWINNMR, version 2.5 on Silicon Graphics O2 equipped with RISC R5000 processor, working under the IRIX 6.3 operating system.
Data analysis
NMR spectra were normalized to the total sum of the spectral integral to accommodate concentration variations between patient samples. These data were also subjected to a scaling procedure. In particular, in autoscaling, the variable mean was subtracted from each variable and then each variable was divided by its standard deviation. After autoscaling all integrals have a standard deviation of one and therefore the data is analyzed on the basis of correlations. Scaling enables small changes in small signals to have similar influence to small changes in big signals [van den Berg et al. Citation2006]. The biological data was normalized using median and range (minimum and maximum value) since the distribution of biological variables is non normal.
PCA
PCA was conducted using normalized proton NMR integrals and seminal variables (as concentration, motility, and morphology of spermatozoa) data, and the results are presented as the scores and loadings plots. This matrix was analyzed using the NCCS 2007 statistics package [Hintze, Citation2007] to obtain different clusters of dataset.
Acknowledgments
The authors thank the University of Siena.
Declarations of interest
The authors declare that there were no conflicts of interest. The authors report no declarations of interest.
Author contributions
Conceived and designed the experiments: CB, GC SM, EM; Performed the experiments: CB, GC, EM, AD; Analyzed the data: CB, AD, GC, EM; Contributed reagents/materials/analysis tools: CR; Wrote the manuscript: CB, GC, EM, CR.
Notes
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/iaan.
References
- Aaronson, D.S., Iman, R., Walsh, T.J., Kurhanewicz, J. and Turek, P.J. (2010) A novel application of 1H magnetic resonance spectroscopy: non-invasive identification of spermatogenesis in men with non-obstructive azoospermia. Hum Reprod 25(4):847–852
- Alexandrino, A.P., Rodrigues, M.A., Matsuo, T., Schuquel, I.T., Costa, W.F. and Santilli, J.C. (2009) Evaluation of seminal citrate level by 1H nuclear magnetic resonance spectroscopy in men with spinal cord injury. Spinal Cord 47(12):878–881
- Averna, T.A., Kline, E.E., Smith, A.Y. and Sillerud, L.O. (2005) A decrease in 1H nuclear magnetic resonance spectroscopically determined citrate in human seminal fluid accompanies the development of prostate adenocarcinoma. J Urol 173:433–438
- Bahl, S., Bawa, S.R. and Srivastava, S. (1988) Magnetic resonance studies of intact human spermatozoa. Physiol Chem Phys Med NMR 20:183–188
- Bone, W., Jones, N.G., Kamp, G., Yeung, C.H. and Cooper, T.G. (2000) Effect of ornidazole on fertility of male rats: inhibition of a glycolysis-related motility pattern and zona binding required for fertilization in vitro. J Reprod Fertil 118:127–135
- Brown, J.C., Mills, G.A., Sadler, P.J. and Walker, V. (1989) 1H NMR studies of urine from premature and sick babies. Magn Reson Med 11:193–201
- Cavagna, M., Oliveira, J.B., Petersen, C.G., Mauri, A.L. Silva, L.F., Massaro, F.C., et al. (2012) The influence of leukocytospermia on the outcomes of assisted reproductive technology. Reprod Biol Endocrinol 10:44
- El-Deredy, W. (1997) Pattern recognition approaches in biomedical and clinical magnetic resonance spectroscopy: a review. NMR Biomed 10:99–124
- Goodson, S.G., Qiu, Y., Sutton, K.A., Xie, G., Jia, W. and O’Brien, D.A. (2012) Metabolic substrates exhibit differential effects on functional parameters of mouse sperm capacitation. Biol Reprod 87(3):75
- Gupta, A., Mahdi, A.A., Ahmad, M.K., Shukla, K.K., Jaiswer, S.P. and Shankhwar, S.N. (2011a) 1H NMR spectroscopic studies on human seminal plasma: A probative discriminant function analysis classification model. J Pharm Biomed Anal 54:106–113
- Gupta, A., Mahdi, A.A., Ahmad, M.K., Shukla, K.K., Bansal, N., Jaiswer, S.P., et al. (2011b) A proton NMR study of the effect of Mucuna pruriens on seminal plasma metabolites of infertile males. J Pharm Biomed Anal 55(5):1060–1066
- Gupta, A., Mahdi, A.A., Shukla, K.K., Ahmad, M.K., Bansal, N., Sankhwar, P., et al. (2013) Efficacy of Withania somnifera on seminal plasma metabolites of infertile males: A proton NMR study at 800 MHz. J of Ethnopharmac 149:208–214
- Hamamah, S., Seguin, F., Bujan, L., Barthelemy, C., Mieusset, R. and Lansac, J. (1998) Quantification by magnetic resonance spectroscopy of metabolites in seminal plasma able to differentiate different forms of azoospermia. Hum Reprod 13:132–135
- Holmes, J.C.C., Foxall, P.J.D., Spraul, M., Farrant, R.D., Nicholson, J.K. and Lindon, J.C. (1997) 750 MHz 1H NMR spectroscopy characterization of the complex metabolic pattern of urine from patients with inborn errors of metabolism: 2-hydroxyglutaric aciduria and maple syrup urine disease. J Pharm Biomed Anal 15:1647–1659
- Hintze, J. (2007) NCSS 2007. NCSS, LLC. Kaysville, Utah, USA. www.ncss.com
- Jayaraman, V., Ghosh, S., Sengupta, A., Srivastava, S., Sonawat, H.M. and Narayan, P.K. (2014) Identification of biochemical differences between different forms of male infertility by nuclear magnetic resonance (NMR) spectroscopy. J Assist Reprod Genet 31(9):1195–1204
- Jolliffe, I.T. (2002) Principal Component Analysis. 2nd Edition. Springer -Verlag, New York, USA
- Jones, A. and Bubb, W. (2000) Substrates for endogenous metabolism by mature boar spermatozoa. J Reprod Fertil 119:129–135
- Lindon, J.C., Nicholson, J.K. and Everett, J.R. (1999) NMR spectroscopy of biofluids. Annul Rep. NMR Spectrosc 38:1–88
- Lynch, M.J., Masters, J., Pryor, J.P., Lindon, J.C., Spraul, M., Foxall, P.J., et al. (1994) Ultra high field NMR spectroscopic studies on human seminal fluid, seminal vesicle and prostatic secretions. J Pharm Biomed Anal 121:5–19
- Maher, A.D., Patki, P., Lindon, J.C., Want, E.J., Holmes, E., Craggs, M., et al. (2008) Seminal oligouridinosis: low uridine secretion as a biomarker for infertility in spinal neurotrauma. Clin Chem 54(12):2063–2066
- Massart, D.L., Vandeginste, B.G.M., Deming, S.N., Michotte, Y. and Kauffman, L. (1988) Chemometrics: A Textbook. Elsevier. New York, USA
- Mc Rae, C., Baskind, N.E., Orsi, N.M., Sharma, V. and Fisher, J. (2012) Metabolic profiling of follicular fluid and plasma from natural cycle in vitro fertilization patients- a pilot study. Fertil Steril 98(6):1449–1457
- Nicholson, J.K. and Wilson, I.D. (1989) High resolution proton magnetic resonance spectroscopy of biological fluids. Prog NMR Spectrosc 21:449–501
- Patel, A.B., Srivastava, S., Phadke, R.S. and Govil, G. (1999) Identification of low-molecular-weight compounds in goat epididymis using multinuclear nuclear magnetic resonance. Anal Biochem 266:205–215
- Roy, S., Banerjee, A., Pandey, H.C., Singh, G. and Kumari, G.L. (2001) Application of seminal germ cell morphology and semen biochemistry in the diagnosis and management of azoospermic subjects. Asian J Androl 3(2):55–62
- Seaglen, J., de Certains, JD., Le Calvè, M., Colleu, D., Bansard, J.Y. and Rio, M. (1995) 1H nuclear magnetic resonance of human seminal plasma in in vitro fertilization attempts: use of automatic spectrum analysis. J Reprod Fertil 103:181–187
- Tomlins, A.M., Foxall, P.J.D., Lynch, M.J., Parkinson, J., Everett, J.R. and Nicholson, J.K. (1998) High resolution 1H NMR spectroscopic studies on dynamic biochemical processes in incubated human seminal fluid samples. Biochimica et Biophysica Acta 1379:367–380
- van den Berg, R.A., Hoefsloot, H.C., Westerhuis, J.A., Smilda A.K. and van der Werf, M.J. (2006) Centering, scaling, and transformations: improving the biological information content of metabolomics data. BMC Genomics 7:142
- van der Grond, J. (1995) Diagnosing testicular function using 31P magnetic resonance spectroscopy: a current review. Hum Reprod 1(3):276–283
- Wevers, R.A., Engelke, U.F.H., Moolenaar, S.H., Bräutigam, C., Brautigam, C., De Jong, J.G.N., et al. (1999) 1H-NMR spectroscopy of body fluids: Inborn errors of purine and pyrimidine metabolism. Clin Chem 45(4):539–548
- WHO (2010) WHO laboratory manual for the examination and processing of human semen. Fifth edition. World Health Organization, WHO Press, Geneva, Switzerland
- Wuthrich, K. (1986) NMR of proteins and nucleic acids. Wiley & Sons, New York, USA