Fig 1. OPLS-DA model for two group comparison. The OPLS-DA models of two group comparison are shown in A (healthy control- Diabetic neuropathy) and B (healthy control-diabetes). The 95 % confidence ellipse of the group is depicted. The circle in the score plot represents the healthy control sample and the square represents the diabetic neuropathy (A) and diabetes (B), respectively. The values of R2Y, Q2 and p-value of CV-ANOVA were 0.983, 0.745 and 0.025 (A) and 0.984, 0.843 and 0.002 (B). The Mahalanobis p-values for two group comparison were 3.0583e-10 and 1.3211e-14 respectively (A and B).
Fig 2. Box and Whisker plots of metabolites with significant difference. Box and whisker plots of four metabolites are illustrated (HC, healthy control; DNP, Diabetic neuropathic patient; D, Diabetic patient). Lactate is not statistically significant but affecting group separation. The groups of which the comparison was identified as significant are linked with lines. The horizontal line in the middle portion of the box is median value. The bottom and top boundaries of boxes represent lower and upper quartile. The open circles represent outliers.
Fig 3. The S-plot from OPLS-DA model between two groups of healthy control-diabetic neuropathy group (A) and healthy control-diabetes group (B). The S-plot between two groups of healthy control-diabetic neuropathy group (A) and healthy control-diabetes group (B) from OPLS-DA are shown and metabolites that were highly contributed to the group separation are depicted on the plots. The important metabolites (p < 0.05, FDR <0.05) with the strongest association to disease are depicted on the S-plot. Lactate was also depicted because it contributed group separation, although it was not significant in univariate analysis.
Fig 4. The ROC curve analysis for two composite metabolites (lactate and citrate). The AUC values were obtained from OPLS-DA models of healthy control – diabetic group (diabetic neuropathy + diabetes) with combination of metabolites. All NMR signals of glucose were not included in the metabolite list to remove the effect of glucose signal on the discrimination. The combination of two metabolites in the group comparison of healthy control – diabetic group (diabetic neuropathy + diabetes) provided the AUC value, 0.952.
Fig 5. Three representative 1H NMR spectra from the serum samples of healthy control (A), diabetic neuropathy (B) and diabetic group (C). 1H CPMG spectra were processed using Mnova 10.0.20 Statistically significant metabolites (ascorbate, glucose, and citrate) were labeled on the spectrum. Lactate was also labeled because it contributed group separation. The enlarged spectra for ascorbate, citrate, glucose and lactate are depicted in the figure. The black line in the enlarged figure represent the spectra of healthy control group, the red line represents the spectra of diabetic neuropathy group and the blue line represent the spectra of diabetic group.
Table 1. Demographic and clinical characteristics of the patients
Table 2. Statistical analysis of the non-parametric Kruskal-Wallis test
References
- B. S. Galer and M. P. Jensen. Neurol. 48, 332 (1997) https://doi.org/10.1212/WNL.48.2.332
- E. L. Feldman, M. J. Stevens, P. K. Thomas, M. B. Brown, N. Canal, and D. A. Greene. Diabetes Care 17, 1281 (1994) https://doi.org/10.2337/diacare.17.11.1281
- C. A. Daykin, P. J. Foxall, S. C. Connor, J. C. Lindon, and J. K. Nicholson. Anal. Biochem. 304, 220 (2002) https://doi.org/10.1006/abio.2002.5637
- A. Guleria, A. Pratap, D. Dubey, A. Rawat, S. Chaurasia, E. Sukesh, S. Phatak, S. Ajmani, U. Kumar, C. L. Khetrapal, P. Bacon, R. Misra, and D. Kumar, Sci. Rep. 6, 35309 (2016) https://doi.org/10.1038/srep35309
- M. Bylesjo, M. Rantalainen, O. Cloarec, J. K. Nicholson, E. Holmes, and J. Trygg, J. Chemom. 20, 341 (2006) https://doi.org/10.1002/cem.1006
- A. Craig, O. Cloarec, E. Holmes, J. K. Nicholson, and J. C. Lindon, Anal. Chem. 78, 2262 (2006) https://doi.org/10.1021/ac0519312
- F. Dieterle, A. Ross, G. Schlotterbeck, and H. Senn, Anal. Chem. 78, 4281 (2006) https://doi.org/10.1021/ac051632c
- J. Trygg, E. Holmes, and T. Lundstedt, J. Proteome Res. 6, 469 (2007) https://doi.org/10.1021/pr060594q
- S. Wiklund, E. Johansson, L. Sjoestroem, E. J. Mellerowicz, U. Edlund, J. P. Shockcor, J. Gottfries, T. Moritz, and J. Trygg, Anal. Chem. 80, 115 (2008) https://doi.org/10.1021/ac0713510
- J. Boccard and D. N. Rutledge, Anal. Chim. Acta 769, 30 (2013) https://doi.org/10.1016/j.aca.2013.01.022
- F. Pesarin, "Multivariate permutation tests: with applications in biostatistics", Wiley Chichester, 2001
- J. A. Westerhuis, H. C. Hoefsloot, S. Smit, D. J. Vis, A. K. Smilde, E. J. van Velzen, J. P. M. van Duijnhoven, and F. A. van Dorsten., Metabolomics 4, 81 (2008) https://doi.org/10.1007/s11306-007-0099-6
- B. Worley, S. Halouska, and R. Powers, Anal. Biochem. 433, 102 (2013) https://doi.org/10.1016/j.ab.2012.10.011
- L. Eriksson, J. Trygg, and S. Wold, J. Chemom. 22, 594 (2008) https://doi.org/10.1002/cem.1187
- J. D. Spurrier, Nonparametr. Stat. 15, 685 (2003) https://doi.org/10.1080/10485250310001634719
- Y. Benjamini and Y. Hochberg, J. R. Stat. Soc. B 57, 289 (1995)
- K. Hajian-Tilaki, Caspian J. Intern. Med. 4, 627 (2013)
- S. Wold, E. Johansson, and M. Cocchi, 3D QSAR in drug design 1, 523 (1993)
- S. Hur, H. Lee, A. Shin, and S. J. Park, J. Kor. Mag. Reson. Soc. 18, 10 (2014) https://doi.org/10.6564/JKMRS.2014.18.1.010
- J. Hyun and S. J. Park, J. Kor. Mag. Reson. Soc. 20, 76 (2016) https://doi.org/10.6564/JKMRS.2016.20.3.076