INFLAMMATORY BOWEL DISEASE DIAGNOSIS METHOD

Abstract

A highly efficient and non-invasive automated diagnosis method of some inflammatory bowel diseases, such as Ulcerative Colitis (UC) and Crohn's Disease (CD), which always allows to avoid duodenal biopsy. Such a method is based on spectroscopic analysis by Raman technique, which operates directly on a protein extract of feces, and reduces the burden of the diagnosis, simultaneously being preferably useful for the automated distinction between Ulcerative Colitis (UC) and Crohn's Disease (CD).

Claims

1. A diagnosis method for detecting inflammatory bowel diseases comprising Ulcerative Colitis (UC) and Crohn's Disease (CD), the method comprising the following steps: a) providing, as input data, a Raman spectrum of a protein extract of a fecal sample; b) isolating, from said Raman spectrum, a band in a wavenumber range of about 1550-1750 cm.sup.−1, which defines amide-I vibration modes; c) calculating a second derivative profile of said band of Raman spectrum, and identifying the minima of said profile; d) performing a deconvolution of said band, setting a number of sub-bands equal to the number of the minima of said second derivative profile, obtaining a number of sub-bands equal to at least said number of minima, each sub-band having a respective center-frequency ω.sub.i and a respective area A.sub.i, with i=1, 2, . . . n; e) associating the sub-bands obtained in step d), based on the center-frequency ω.sub.i thereof, with corresponding vibration modes of protein secondary structures comprising vibration modes of disordered protein secondary structures of amide-I and vibration modes of the higher frequency anti-parallel β-sheet protein secondary structures; f) calculating a ratio R of the area of the sub-band, associated with the vibration modes of the disordered protein secondary structures of amide-I, with the area of the sub-band associated with the vibration modes of the higher frequency anti-parallel β-sheet protein secondary structures; g) verifying that the ratio R is greater than or equal to a first threshold value to confirm that the protein extract of the fecal sample belongs to a patient suffering from an inflammatory bowel disease.

2. A method according to claim 1, wherein after step g) if said ratio R is greater than or equal to a second threshold value, the protein extract of the fecal sample belongs to a patient suffering from Ulcerative Colitis (UC), while if said ratio R is lower than said second threshold value, the protein extract of the fecal sample belongs to a patient suffering from Crohn's disease (CD).

3. A method according to claim 1, wherein in step d) in order to perform the deconvolution, a minimum number of parameters characterizing each sub-band is left free, and an iterative best-fit procedure is started, which thus uses said minimum number of parameters for each sub-band, preferably using a Voigt profile for each sub-band, each Voigt profile being mathematically expressed by a convolution of a Lorentzian curve with a Gaussian curve, and the characteristic parameters of which defining said minimum number of parameters comprise: the center-frequency ω.sub.Vgt; the amplitude a.sub.Vgt; the half-width Γ.sub.VG related to the half-width at half-height of the Gaussian Γ.sub.G, being Γ.sub.G=Γ.sub.VG√{square root over (2 ln 2)}, and the half-width Γ.sub.VL depending on the ratio of the half-width at half-height of the Lorentzian Γ.sub.L with respect to said half-width Γ.sub.VG, being Γ.sub.L=Γ.sub.VG.Math.Γ.sub.VL.

4. A method according to claim 1, wherein between step a) and step b) a fit of a region of the Raman spectrum between 1300-1800 cm.sup.−1 is performed setting a predetermined profile, preferably a Voigt profile; preferably wherein, to obtain said fit, a deconvolution is performed, leaving a minimum number of parameters characterizing each sub-band free, and an iterative best-fit procedure is started, which thus uses said minimum number of parameters for each sub-band, preferably using a Voigt profile for each sub-band, each Voigt profile being mathematically expressed by a convolution of a Lorentzian curve with a Gaussian curve, and the characteristic parameters of which defining said minimum number of parameters comprise: the center-frequency ω.sub.Vgt; the amplitude a.sub.Vgt; the half-width Γ.sub.VG related to the half-width at half height of the Gaussian Γ.sub.G, being Γ.sub.G=Γ.sub.VG√{square root over (2 ln 2)}, and the half-width Γ.sub.VL depending on the ratio of the half-width at half-height of the Lorentzian Γ.sub.L with respect to said half-width Γ.sub.VG, being Γ.sub.L=Γ.sub.VG.Math.Γ.sub.VL.

5. A method according to claim 1, wherein before performing the Raman spectroscopy, the fecal sample protein extract is homogenized, preferably by centrifugation.

6. A method according to claim 2, wherein said first threshold value can be defined by performing the following steps: providing, as input data, a Raman spectrum of a fecal sample protein extract for each patient of a first known group of patients with Ulcerative Colitis (UC) or Crohn's Disease (CD) and for each patient of a second known group of healthy patients; for each patient, of both the first group and second group, isolating from the respective Raman spectrum the band in the wavenumber range of about 1550-1750 cm.sup.−1; performing steps c), d), e) and f) for each Raman spectrum; performing an analysis of a first ROC curve considering the ratios R of the first group and the second group as database and calculating the first threshold value by Youden's index.

7. A method according to claim 6, wherein said second threshold value can be defined by performing the following steps: providing, as input data, a Raman spectrum of a fecal sample protein extract for each patient of a first known subgroup of said first group consisting of patients with Ulcerative Colitis (UC) and for each patient of a second known subgroup of said first group consisting of patients with Crohn's Disease (CD); for each patient, of both the first subgroup and the second subgroup, isolating from the respective Raman spectrum the band in the wavenumber range of about 1550-1750 cm.sup.−1; performing steps c), d), e) and f) for each Raman spectrum; performing an analysis of a second ROC curve considering the ratios R of the first subgroup and the second subgroup as database and calculating the second threshold value by Youden's index.

8. A method according to claim 2, wherein said second threshold value is greater than said first threshold value.

9. A method according to claim 1, wherein the Raman spectrum is acquired in a wavenumber range between 400 cm.sup.−1 and 3300 cm.sup.−1, preferably between 500 cm.sup.−1 and 2000 cm.sup.−1.

10. A method according to claim 1, wherein a Raman spectrometer connected to a detector, preferably a CCD (charge-coupled device), is used to obtain the Raman spectrum of the fecal sample protein extract.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In the description of the invention, reference is made to the accompanying drawings, which are provided by way of non-limiting example, in which:

[0028] FIG. 1a shows an example of an average Raman spectrum in the region 1300-1800 cm.sup.−1 of the fecal protein extract of non-IBD subjects (black line), subjects with CD (red line) and subjects with UC (green line);

[0029] FIG. 1b shows an example of a profile obtained by calculating the second derivative of the band comprised in a wavenumber range of 1550-1750 cm.sup.−1;

[0030] FIG. 2a shows best-fit results for the region 1550-1750 cm.sup.−1 of a sample related to a non-IBD patient and the related sub-bands;

[0031] FIG. 2b shows best-fit results for the region 1550-1750 cm.sup.−1 of a sample related to a patient with CD and the related sub-bands;

[0032] FIG. 2c shows best-fit results for the region 1550-1750 cm.sup.−1 of a sample related to a patient with UC and the related sub-bands;

[0033] FIG. 3 shows the calculated values of the ratio R for non-IBD subjects (black circles), subjects with CD (red squares) and subjects with UC (green triangles);

[0034] FIG. 4 shows an ROC curve for non-IBD and IBD subjects from a sample of patients;

[0035] FIG. 5 shows the best cutoff-point to maximize sensitivity and specificity for said non-IBD and IBD subjects;

[0036] FIG. 6 shows the distribution graph for said non-IBD and IBD subjects;

[0037] FIG. 7 shows an ROC curve for the group of IBD subjects of said patient sample;

[0038] FIG. 8 shows the best cutoff-point to maximize sensitivity and specificity for said IBD subjects;

[0039] FIG. 9 shows the distribution graph for UC and CD subjects belonging to said group of IBD subjects.

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION

[0040] The method of the invention for the diagnosis of inflammatory bowel diseases, such as Ulcerative Colitis (UC) and Crohn's Disease (CD), comprises the following steps:

[0041] a) providing, as input data, a Raman spectrum of a protein extract of a fecal sample;

[0042] b) isolating, from said Raman spectrum, the band in a wavenumber range of about 1550-1750 cm.sup.−1, which defines the amide-I vibration modes;

[0043] c) calculating the second derivative profile of said Raman spectrum band, and identifying the minima of said profile;

[0044] d) performing a deconvolution of said band, setting a number of sub-bands equal to the number of minima of the profile of said second derivative, obtaining a number of sub-bands equal to at least said number of minima, each sub-band having a respective center-frequency ω.sub.i and a respective area A.sub.i, with i=1, 2, . . . n;

[0045] e) associating the sub-bands obtained in step d), based on the center-frequency ω.sub.i thereof, with the corresponding vibration modes of the protein secondary structures comprising at least the vibration modes of the disordered protein secondary structures of amide-I and the vibration modes of the higher frequency anti-parallel β-sheet protein secondary structures;

[0046] f) calculating the ratio R of the area of the sub-band, associated with the vibration modes of the disordered protein secondary structures of amide-I, with the area of the sub-band associated with the vibration modes of the higher frequency anti-parallel β-sheet protein secondary structures;

[0047] g) verifying that the ratio R is greater than or equal to a first threshold value to confirm that the protein extract of the fecal sample belongs to a patient suffering from an inflammatory bowel disease between Ulcerative Colitis (UC) and Crohn's Disease (CD).

[0048] For example, the vibration modes of the disordered protein secondary structures of amide-I have the respective center-frequency at a wavenumber in the range 1660-1689 cm.sup.−1, for example about 1662 cm.sup.−1, compatibly with the resolution of the Raman spectrometer; while the vibration modes of the higher frequency anti-parallel β-sheet protein secondary structures have the respective center-frequency at a wavenumber in the range 1690-1700 cm.sup.−1, for example about 1694 cm.sup.−1, compatibly with the resolution of the Raman spectrometer.

[0049] Other significant vibration modes which can fall within the band comprised in the wavenumber range of about 1550-1750 cm.sup.−1 are: [0050] the vibration modes of the parallel β-sheet protein secondary structures, which have the respective center-frequency at a wavenumber in the range 1610-1619 cm.sup.−1, compatibly with the resolution of the Raman spectrometer, [0051] the vibration modes of the lower frequency anti-parallel β-sheet protein secondary structures, which have the respective center-frequency at a wavenumber in the range 1620-1639 cm.sup.−1, always compatibly with the resolution of the Raman spectrometer, [0052] and the vibration modes of the α-helix protein secondary structures, which have the respective center-frequency at a wavenumber in the range 1640-1654 cm.sup.−1, always compatibly with the resolution of the Raman spectrometer.

[0053] Preferably, it is possible to automatically distinguish whether the ill patient is suffering from Ulcerative Colitis (UC) or Crohn's Disease (CD). In fact, if the above ratio R is greater than or equal to a second threshold value it can be confirmed that the protein extract of the fecal sample belongs to a patient suffering from Ulcerative Colitis (UC). Otherwise, if the ratio R is less than the second threshold value, it can be confirmed that the protein extract of the fecal sample belongs to a patient with Crohn's disease (CD).

[0054] The first threshold value can be defined by performing the following steps: [0055] providing, as input data, a Raman spectrum of a fecal sample protein extract for each patient of a first known group of patients with Ulcerative Colitis (UC) or Crohn's Disease (CD) and for each patient of a second known group of healthy patients; [0056] for each patient, of both the first group and second group, isolating from the respective Raman spectrum the band in the wavenumber range of about 1550-1750 cm.sup.−1; [0057] performing the steps c), d), e) and f) described above for each Raman spectrum; [0058] performing an analysis of a first ROC curve considering the ratios R of the first group and of the second group as database and calculating the first threshold value by Youden's index.

[0059] As is known, most diagnostic tests produce a quantitative result. To discriminate between healthy and ill people it is necessary to have a threshold or cut-off value.

[0060] In an ideal situation, healthy and ill people return different test values and the cut-off value is immediately determined. In real situations, there is always some overlap in the distribution of healthy and ill people. Sensitivity and specificity are inversely related in relation to the selection of the cut-off. The adoption of a threshold which offers high sensitivity leads to a loss of specificity and vice versa.

[0061] In the proposed method, although the distribution of healthy and ill people is distinct enough, to obtain the threshold value which minimizes the probability of finding false positives and false negatives, it is preferable to construct a ROC (Receiver Operating Characteristic) curve in a known manner, with sensitivity on the ordinates and (1-specificity) on the abscissas, considering the ratios R as the database, therefore obtaining the first optimal threshold value. Accuracy, sensitivity and specificity of the proposed diagnostic test were determined using said ROC curve and Youden's index, providing excellent results.

[0062] The second threshold value can instead be defined by performing the following steps: [0063] providing, as input data, a Raman spectrum of a fecal sample protein extract for each patient of a first known subgroup of said first group consisting of patients with Ulcerative Colitis (UC) and for each patient of a second known subgroup of said first group consisting of patients with Crohn's Disease (CD); [0064] for each patient, of both the first subgroup and the second subgroup, isolating from the respective Raman spectrum the band in the wavenumber range of about 1550-1750 cm.sup.−1; [0065] performing the steps c), d), e) and f) described above for each Raman spectrum; [0066] performing an analysis of a second ROC curve considering the ratios R of the first subgroup and of the second subgroup as database and calculating the second threshold value by Youden's index.

[0067] It was found that the second threshold value is greater than the first threshold value. The information obtained from a Raman scattering experiment is graphically depicted as a diagram (Raman spectrum) where the abscissas report the Raman shifts corresponding to the differences in energy between the vibrational levels involved in the transition with respect to the fundamental level, expressed in cm.sup.−1 (wavenumber, keeping in mind the direct proportionality between the energy and the inverse of the wavelength of an electromagnetic radiation). Raman intensities proportional to the number of Stokes photons collected by the instrument detector are shown on the ordinates. The range of energies reported in a normal Raman spectrum can extend from a few tens of cm.sup.−1 up to about 3500 cm.sup.−1, a region in which almost all the fundamental molecular vibrations fall.

[0068] In particular, the Raman spectrum of the protein extract of a fecal sample can be acquired in a wavenumber range between 3500 cm.sup.−1 and 300 cm.sup.−1, preferably between 3300 cm.sup.−1 and 400 cm.sup.−1, even more preferably between 2500 cm.sup.−1 and 500 cm.sup.−1.

[0069] Raman spectroscopy can be performed by means of any Raman spectrometer having a laser source, such as, for example, a diode laser source, with a wavelength preferably, but not necessarily, equal to 780-790 nm.

[0070] The spectrometer is connected to a detector, preferably a charge-coupled device (CCD), possibly integrated with a monochromator.

[0071] The Raman spectrum can, for example, be acquired by performing a number of scans of the protein extract of a fecal sample between 25 and 35, with an exposure time between 30 and 90 seconds for each scan.

[0072] Between step a) and step b) it is preferable to perform a first fit of a region of the Raman spectrum in the range 1300-1800 cm.sup.−1 by means of a predetermined profile for the sub-bands, preferably the Voigt profile, to isolate the band related to the amide-I in the range 1550-1750 cm.sup.−1. To obtain said first fit, a first deconvolution is performed, leaving a minimum number of parameters free which characterize each sub-band, and an iterative best-fit procedure is started which therefore uses a minimum number of parameters.

[0073] Similarly, in order to perform the deconvolution of step d) (second curve-fitting), a minimum number of parameters characterizing each sub-band are left free and an iterative best-fit procedure is started which therefore uses a minimum number of parameters.

[0074] Before performing the Raman spectroscopy, the protein extract of the fecal sample is homogenized, preferably by centrifugation. In particular, a liquid extract is collected to homogenize the sample for Raman analysis.

[0075] An experimental study is described below, which allowed developing the method of the invention.

Materials and Methods

[0076] A protein extract from fecal samples of 15 subjects with CD, 9 subjects with UC (both of these groups of subjects form the so-called IBD group) and 19 subjects for whom IBD was excluded (non-IBD group) was analyzed. In total, samples of 43 patients were analyzed, all of them in pediatric age. In particular, the IBD group had a median age of 15.6±3.15 years, while the non-IBD group had a median age of 12.6±5.17 years.

[0077] A fecal sample was collected for all subjects to assess the fecal levels of calprotectin. In the IBD group, the assessment was needed to assess disease activity (i.e., intensity of intestinal inflammation), in the non-IBD group the assessment was performed as part of the assessments to rule out the diagnosis of IBD. The patients in the IBD group, with CD or UC, had a different degree of disease activity based on calprotectin levels. Instead, the patients in the control group were those for whom intestinal inflammation was ruled out based on normal calprotectin levels (e.g., 100 mg/kg). From the same sample prepared for the ELISA test for fecal calprotectin, treated in a known manner with an extraction buffer solution in order to recover the protein content, a liquid extract of 100 μl was collected in order to homogenize the sample for Raman analysis.

[0078] The Raman measurements were performed using a DXR-Smart Raman spectrometer (Thermo Fisher Scientific) using a diode laser with an excitation wavelength of 785 nm. All the Raman spectra were acquired in the wavenumber range 400-3300 cm.sup.−1, with a resolution of 1.9285 cm.sup.−1 and irradiated with a laser power of 24 mW in output from a 50 μm pinhole aperture.

[0079] The detector connected to the spectrometer is a charge-coupled device (CCD) which, in addition to having very high sensitivity, allows simultaneously studying a broad spectral band, instead of a single wavelength at a time. In particular, the use of a CCD integrated with a monochromator allows obtaining, at once, the entire Raman spectrum with advantages in terms of experimental simplicity and speed of execution of the experiment.

[0080] The vials containing the fecal protein extract were placed in the sample holder thereof and the 180 degree sampling accessory of said DXR-Smart Raman spectrometer was used for the measurements. In order to obtain spectra with a high signal-to-noise ratio (S/R), each Raman spectrum was obtained after collecting 32 sample exposures and the duration of each exposure during the data collection was set equal to 60 s. The total acquisition time was therefore 32 minutes for each spectrum.

[0081] It proved preferable to fit a region of the Raman spectrum in the range 1300-1800 cm.sup.−1 by means of a predetermined profile, preferably Voigt profiles, to isolate the band related to the amide-I in the range 1550-1750 cm.sup.−1.

[0082] The experimental data, first in the range 1300-1800 cm.sup.−1 and then in the range 1550-1750 cm.sup.−1, were fitted using a multiple curve-fitting routine, for example provided in the PeakFit 4.0 software package (Systat Software, Inc., USA).

[0083] The fitting procedure adopted is a non-linear regression procedure, based on the least squares method, which adjusts the parameters in small steps in order to improve the goodness of the fit, minimizing the sum of squares of the vertical deviations r.sup.2=Σ[y.sub.i−ƒ(x.sub.i, a.sub.1, a.sub.2, . . . , a.sub.n)].sup.2 of a set of n data points from a function ƒ. In the previous relation, y.sub.i represents the ordinate, experimentally observed, corresponding to the abscissa x.sub.i, while ƒ(x.sub.i, a.sub.1, a.sub.2, . . . , a.sub.n) represents the ordinate, calculated using the function ƒ depending on the n parameters a.sub.1, a.sub.2, . . . , a.sub.n, at the same abscissa x.sub.i. The analysis of the second derivative of the selected band of the Raman spectrum was preliminarily applied to obtain, by evaluating the minima in the profiles obtained, a first indication of the minimum number of components of the band and the peak position thereof.

[0084] In the associated systems, the vibrational bands undergo both homogeneous and non-homogeneous broadening. Therefore, the Voigt profile, which is a convolution of a Lorentzian curve with a Gaussian curve, is particularly suitable for the analysis of the aforesaid bands. Considering that, the experimental data were deconvoluted into the minimum number of fit functions having a Voigt profile, defined as the following convolution of a Lorentzian curve with a Gaussian curve:

[00001]$V\left(\omega \right)=\frac{\left\{{\int }_{-\infty }^{+\infty }\frac{a_{\mathrm{Vgt}}\exp \left(-x^2\right)\mathrm{dx}}{{\Gamma }_{\mathrm{VL}}^2+{\left[\left(\frac{\omega -{\omega }_{\mathrm{Vgt}}}{{\Gamma }_{\mathrm{VG}}}\right)-x\right]}^2}\right\}}{{\int }_{-\infty }^{+\infty }\frac{\exp \left(-x^2\right)\mathrm{dx}}{{\Gamma }_{\mathrm{VL}}^2+x^2}},$

[0085] where

[0086] ω.sub.Vgt is the center-frequency;

[0087] a.sub.Vgt is the amplitude;

[0088] Γ.sub.VG is the half-width related to the half-width at half-height of the Gaussian Γ.sub.G, being Γ.sub.G=Γ.sub.VG√{square root over (2 ln 2)},

[0089] and Γ.sub.VL is the half-width related to the half-width at half-height of the Lorentzian Γ.sub.L, and is dependent on the ratio of the half-width at half-height of the Lorentzian Γ.sub.L with respect to said half-width Γ.sub.VG, being Γ.sub.L=Γ.sub.VG.Math.Γ.sub.VL. All these four parameters, a.sub.Vgt, ω.sub.Vgt, Γ.sub.VG, Γ.sub.VL, were left free to vary at the time of iteration. For each fit session, multiple iterations were performed until a convergent solution (best-fit) was reached, minimizing the value of r.sup.2. The well-known problem of finding a local false minimum was overcome by running each nonlinear regression several times (20 times on average), each time providing a different set of initial parameter values. All the fits generated the same parameter values, with the same sum of squares, regardless of the initial values, and this confirmed not having encountered a false minimum.

[0090] As an alternative to the fit functions having a Voigt profile, other suitable fit functions could be used.

[0091] The ROC curve, as mentioned above, is a graph with the sensitivity on the ordinates and (1-specificity) on the abscissas, where the sensitivity represents the percentage of true positives, while (1-specificity) represents the percentage of true negatives. In this study, the sensitivity and specificity of the diagnostic test were determined by considering the positive and negative test in the two different groups (IBD subjects vs. non-IBD subjects). Sensitivity refers to the ability of Raman spectroscopy to correctly identify those patients in inflammatory phase. Instead, specificity refers to the ability of Raman spectroscopy to correctly identify patients without inflammation. The area under the ROC curve (AUC=Area under curve) represents the probability Pr (Y≥X), i.e., the probability of the marker value Y of a subject randomly selected from the ill population to be greater than the marker value X of a subject randomly selected from the healthy population.

[0092] The obtained AUC result is an effective mode to summarize the general diagnostic accuracy of the test. Values from 0 to 1 are used, where a value of 0 indicates a perfectly inaccurate test and a value of 1 reflects a perfectly accurate test. In general, an AUC value of 0.5 suggests no discrimination (i.e., the ability to diagnose patients with and without the disease or condition based on the test), an AUC value of 0.7 to 0.8 is considered acceptable, an AUC value of 0.8 to 0.9 is considered excellent, an AUC value greater than 0.9 is considered perfect.

[0093] On the other hand, since AUC serves as a measure of the overall diagnostic precision of a biomarker over all possible cut-points, the AUC cannot directly provide an optimal threshold or cut-point or cut-off point necessary to clinicians for diagnoses, e.g., to classify a subject as ill or healthy. For this reason, the cut-point was determined using the Youden index J, defined as J=sensitivity y+specificity y−1. This index represents the maximum vertical distance between the ROC curve and the reference represented by the diagonal, and is taken at the cut-point which optimizes the biomarker differentiation capacity when sensitivity and specificity are given equal weight. The Youden index J offers the optimal cut-off point and is simultaneously a direct measure of the diagnostic accuracy at the optimal cut-off point, i.e., the maximum correct general classification rate which a marker can achieve.

Results and Discussion

[0094] The average Raman spectrum of the protein extract of the fecal samples of non-IBD patients was considered, obtained by averaging the spectra collected for all the subjects analyzed, previously normalized to the total area.

[0095] The spectrum highlighted some of the main vibrational modes typical of proteins, associated with the polypeptide backbone (amide bands) and the residual side of aromatic and non-aromatic amino acids.

[0096] The vibration modes of amide-I are clearly evident as a wide band centered at about 1650 cm.sup.−1, and comprised in the wavenumber range of about 1550-1750 cm.sup.−1. These vibration modes are mainly due to the stretching mode v(C=O) and a small amount of out-of-phase stretching v(C−N). The position of the amide-I band will depend on the conformations of the polypeptide backbone, e.g., on the type of secondary structure, and on the intramolecular and intermolecular hydrogen bond of the protein sample. Each type of secondary structure will have a typical stretching frequency v(C=O), and this justifies the broadening of the band.

[0097] FIG. 1a depicts the average Raman spectrum, in the range 1300-1800 cm.sup.−1, of the fecal protein extract of non-IBD subjects (black line), subjects with CD (red line) and subjects with UC (green line). In fact, with the onset of the inflammatory disease, an evident increase in the spectral intensity in the broad band detected in the range 1300-1800 cm.sup.−1 was observed.

[0098] The high quality of the spectra allowed us to quantitatively evaluate the spectral modifications possibly induced by the inflammatory state as a consequence of the changes in the protein secondary structures, focusing our attention in the amide-I region. In fact, this band is an excellent indicator for quantifying the secondary structure and conformational changes of the proteins, as a consequence of the role played by the amide part in the cross-linking.

[0099] To do this, the various protein secondary structures which contribute to the unresolved amide-I band must be distinguished, separating the corresponding components by means of second derivative, deconvolution and curve-fitting calculations.

[0100] In this sense, the main problem encountered in the spectra analysis is the overlapping of different components which contribute to the Raman vibrational profile observed in the range 1300-1800 cm.sup.−1.

[0101] Preferably, a first curve-fitting can be performed only for the region or band of the Raman spectrum in the range 1300-1800 cm.sup.−1 by means of, for example, Voigt profiles. The Voigt profile, described by a convolution of a Lorentzian curve with a Gaussian curve, is the most suitable in the case of associated systems, for which the vibrational bandwidth is affected by both homogeneous and non-homogeneous broadening.

[0102] After this step, all the sub-bands which fell outside the wavenumber range 1300-1800 cm.sup.−1 were subtracted from the experimental profile, and a second more detailed curve-fitting was then performed only for the amide-I band, in the range 1550-1750 cm.sup.−1. Also in this case it is preferable to perform the curve-fitting using Voigt profiles. The analysis of the second derivative of said amide-I band was advantageously applied to obtain, by evaluating the minima in the profiles obtained, a first indication of the positions of the peaks of the components of the amide-I band. FIG. 1b shows an example of a profile obtained by calculating the second derivative of the experimental Raman spectrum of a non-IBD subject in the wavenumber range 1550-1750 cm.sup.−1. Four main sub-bands were therefore recognized, centered respectively at about 1617 cm.sup.−1, 1634 cm.sup.−1, 1663 cm.sup.−1 and 1694 cm.sup.−1. Based on a comparison with the scientific literature these sub-bands were related to the [0103] vibration modes of the parallel β-sheet protein secondary structures, characterized by a strong hydrogen bond (indicative of protein aggregates) and which have the respective center-frequency at a wavenumber in the range 1610-1619 cm.sup.−1, compatibly with the resolution of the Raman spectrometer; [0104] vibration modes of the lower frequency anti-parallel β-sheet protein secondary structures, which have the respective center-frequency at a wavenumber in the range 1620-1639 cm.sup.−1, compatibly with the resolution of the Raman spectrometer; [0105] vibration modes of the disordered protein secondary structures of amide-I, which have the respective center-frequency at a wavenumber in the range 1660-1689 cm.sup.−1, compatibly with the resolution of the Raman spectrometer; [0106] vibration modes of the higher frequency anti-parallel β-sheet protein secondary structures, which have the respective center-frequency at a wavenumber in the range 1690-1700 cm.sup.−1, compatibly with the resolution of the Raman spectrometer. The results of a curve-fitting procedure obtained for a non-IBD subject, for a subject with CD and for a subject with UC are reported, as an example, in FIGS. 2a, 2b and 2c. The circles represent the experimental points; the line which follows the trend of the circles represents the best-fit; and curves 1, 2, 3 and 4 represent four sub-bands that are components of the amide-I band.

[0107] In particular, in the study case, the protocol used provided an optimal reproduction of the experimental data. In fact, for all the samples studied, the best-fit is characterized by r.sup.2˜0.9999.

[0108] Advantageously, the alterations of the protein network following the inflammatory state were monitored by evaluating the ratio R between the area of the sub-band A.sub.3, associated with the vibration modes of the disordered protein secondary structures of amide-I, and the area of the sub-band A.sub.4 associated with the vibration modes of the higher frequency anti-parallel β-sheet protein secondary structures. In particular, the ratio R is related to the non-reducible (trivalent)/reducible (bivalent) cross-linking ratio.

[0109] The R values obtained for all the subjects analyzed are reported in FIG. 3.

[0110] It can be noted that active inflammation gives rise to a significant increase in the cross-linking ratio R, which passes from an average value of 0.93±0.05 in the case of non-IBD subjects up to 1.26±0.06 and 1.45±0.07 in the case of subjects with CD and with UC disease, respectively. This increase is indicative of an increase, following inflammation, in the quantity of non-reducible trivalent cross-links at the expense of reducible bivalent cross-links, and/or following the reduced formation thereof. As a result, a more strongly interconnected protein network is conceivable, which clearly results in altered mechanical and functional properties. It could be argued that an increase in the ratio R in the amide-I band could also be attributed to conformational changes in proteins induced by factors other than inflammatory state, including aging, dehydration and radiological conditions. However, these factors do not influence the reported data, as the groups of subjects are very similar to each other, differing only with regard to intestinal inflammation.

[0111] FIG. 4 illustrates the ROC curve for IBD and non-IBD subjects (AUC=1; 95% confidence range). FIG. 5 shows the best cut-off point to maximize sensitivity and specificity (threshold value=1.2). These curves show that the method of the invention is capable of clearly discriminating IBD subjects from non-IBD subjects. To highlight the above results, FIG. 6 shows the distribution graph for IBD and non-IBD subjects.

[0112] Considering the values obtained by the ratio R, the automatic differentiation between CD and UC was also advantageously performed in IBD patients. FIG. 7 illustrates the ROC curve for IBD subjects (AUC=1; 95% confidence range). FIG. 8 shows the best cut-off point for maximizing sensitivity and specificity (threshold value=1.4). These curves show that the method of the invention is also capable of clearly discriminating subjects with UC from subjects with CD considering the entire group of IBD subjects. To highlight the above results, FIG. 9 shows the distribution graph for UC and CD subjects.

[0113] The results obtained with the method of the invention surprisingly demonstrated an optimal use of Raman spectroscopy as a diagnostic mode for the rapid detection of IBD and, among IBD subjects, for a clear and unambiguous distinction between CD and UC.

[0114] In addition to non-invasiveness, the proposed method offers significant advantages, including ease of use, experimental repeatability and speed of execution. Furthermore, the method is extremely low cost, with respect to the diagnostic techniques currently used which require special diagnostic kits, and is also non-destructive, since the sample can be stored and analyzed several times after some time, the only analysis limit being the degradation of the sample itself.