The Specific Changes of Urine Raman Spectra Can Serve as Novel Diagnostic Tools for Disease Characteristics in Patients with Crohn’s Disease

Abstract

Purpose: Crohn's disease (CD) is a chronic recurrent intestinal inflammatory disease that requires repeated invasive examinations. Convenient and noninvasive diagnostic tools for CD are lacking. Surface-enhanced Raman spectroscopy (SERS) can rapidly provide specific metabolite information in various samples. Our previous study has showed urine Raman spectrum can distinguish CD patients from healthy controls noninvasively. In this study, we further investigated the value of urine Raman spectra on identifying the disease characterizations in patients with CD. Patients and Methods: Urine samples were analyzed by SERS to acquire specific changes of the spectra from 100 active CD (aCD) patients and 88 inactive CD (iCD) patients. The accuracy of classifier models yielded by SERS was assessed by principal component analysis and support vector machine (PCA-SVM) to investigate spectral differences and disease characterizations. Results: Given a panel of 16 specific Raman spectra, the classifier model was established to predict disease activity between patients with aCD and iCD and achieved higher efficacy than fecal calprotectin (AUC value, 0.864 vs 0.596, P=0.02). After leave-one-patientout cross-validation, the classifier model still obtained 75.5% of accuracy. The correlation analysis showed it had negative correlation with endoscopic results (r=−0.616, P<0.0001). We further established the classifier model in identifying disease location to discriminate colonic-type from ileal-type CD with 63.6% of accuracy with the significantly increased intensity of 1643 cm −1 band, and the model to predict the spectra changes of before and after treatment in tumor necrosis factor inhibitor responders with 91.2% of accuracy with a panel of 11 specific spectra. The metabolic changes of amino acids, proteins, lipids, and other compounds in urine levels were noted by SERS in patients with CD. Conclusion:The specific changes of urine Raman spectra can reflect changes in urine metabolism. It has the potential value on being the promising diagnostic tool for disease characterizations in CD patients by a convenient and noninvasive way.

Review–Vibrational spectroscopy‐aided diagnosis, prognosis and treatment of inflammatory bowel disease

“…and Kirchberger‐Tolstik et al 36,64 . Other studies harnessed dimensionality reduction and feature extraction methods, such as principal component analysis‐linear discriminant analysis (PCA‐LDA), 9,45,67 partial least squares‐discriminant analysis (PLS‐DA) 67 or principal component analysis‐support vector machine (PCA‐SVM) 68,71 . In each instance, the goal was to provide a diagnostic tool which could rapidly and accurately identify the presence or absence of IBD based on the obtained spectral signature.…”

“…This approach was employed by Ding et al., Addis et al., Chernavskaia et al., Bi et al. and Wu et al., 12,61,62,66,69,71 where the authors applied statistical methods including the parametric ( t‐test ). Shapiro W ilk and Kolmogorov‐Smirnov statistical analysis were employed by Wu et al., while Addis et al.…”

“…When identifying disease location, either colonic or ileal-type tissue, the authors observed a significant increase in the 1643 cm −1 Raman band assigned to the C-C stretch in lipids and the C-O stretch of protein amide I band. 71 Interestingly, spectral changes associated with treatment of tumour necrosis factor inhibitor were also investigated in this study, where the classification accuracy of before and after treatment groups achieved 91.2%. Again, most of the metabolic changes in urine samples were attributed to proteins, lipids and amino acids.…”

“…Several further researchers have developed and applied further forms of supervised machine learning approaches including the artificial neural networks (ANNs) by Smith et al, Buchan et al and Kirchberger-Tolstik et al 36,64 Other studies harnessed dimensionality reduction and feature extraction methods, such as principal component analysis-linear discriminant analysis (PCA-LDA), 9,45,67 partial least squaresdiscriminant analysis (PLS-DA) 67 or principal component analysis-support vector machine (PCA-SVM). 68,71 In each instance, the goal was to provide a diagnostic tool which could rapidly and accurately identify the presence or absence of IBD based on the obtained spectral signature. Overall, both approaches complement one another and are vital to gain a comprehensive understanding as to the disease development and diagnosis using RS.…”

Identification of circulating biomarkers of Crohn's disease and spondyloarthritis using Fourier transform infrared spectroscopy

“…In the specific case of CD and SpA, previous work was done by Wu et al to diagnose CD on urine samples [29]. But the contribution of infrared spectroscopy has never been tested before in the diagnosis of these two pathologies using plasma.…”