Abstract Trichophyton rubrum and Candida species comprise the majority of onychomycosis pathogens. The aim of this study was to evaluate Raman spectroscopy for the differentiation between healthy and either T. rubrum or Candida infected nails. Raman measurements were performed on clippings (N = 52) infected either by T. rubrum (N = 12) or Candida species (N = 14; C. parapsilosis (sensu lato): N = 11, C. glabrata: N = 1, C. albicans: N = 2) with healthy nails (N = 26) used as controls. Systematic spectral differences were observed in the 500–520 cm−1 band region, attributable to a diverting imprint of the disulfide stretching of cystine and cysteine residues among samples. Particularly, Candida infected nails demonstrated a shoulder at 519 cm−1, corresponding to the signal of the less stable gauche-gauche-trans conformation of the disulfide bond. Two additional bands at 619 and 648 cm−1, corresponding to the C-S stretching vibration, were more evident in the T. rubrum infected nails. Finally, a Raman band at 1550 cm−1, attributable to amide II and tryptophan (Trp) content, was undetectable in Candida infected nails. Using principal component analysis (PCA), efficient differentiation of healthy, T. rubrum and Candida species infected nails was achieved. Soft independent modeling of class analogy (SIMCA) and partial least squares-discriminant analysis (PLS-DA) were further applied to generate diagnostic algorithms for the classification of Raman spectra. Both techniques succeeded in modeling clinical nail samples in three groups according to their mycological categories. Raman spectroscopy is a promising method for the differentiation of healthy vs. diseased nails, including efficient differentiation between onychomycosis caused by T. rubrum and Candida species. onychomycosis, Raman, spectroscopy, Candida, Trichophyton rubrum Introduction Onychomycosis is a common nail condition, predominantly affecting the toenails, with a prevalence of 4.3% across Europe and North America.1 In the aforementioned population, the principal single causative agent was the dermatophyte Trichophyton rubrum (44.9% of the cases; 95% confidence interval [CI]: 33.8–56.0), followed by yeasts (predominantly Candida species) in 21.1% (95% CI: 11.0–31.3) of infected nails. The remaining pathogens include other dermatophytes and different moulds. The gold standard of diagnosis remains the direct microscopic evaluation of nail clippings coupled with the cultivation of affected nail samples in specialized media. Direct microscopy is inexpensive and easy to perform,2 yet its sensitivity is strongly operator dependent. Culture is the mainstay of clinical mycology and holds high predictive value when a dermatophyte is grown after 3–4 weeks of incubation; yet it has lower sensitivity compared to microscopy.3 Newer techniques that have been developed to increase the sensitivity of direct microscopy include the staining of fungal chitin with calcofluor white and observation under a fluorescence microscope, or the histological evaluation of the infected nail after PAS staining.2 Both latter approaches provide only morphological information frequently insufficient to identify the culprit fungal pathogen at species level. However, the precise microbiological identification of the pathogen is important as this information guides appropriate treatment selection in order to reduce relapse rates.4 More sophisticated (and more costly) alternatives, currently available for onychomycosis diagnosis confirmation, species identification and suitable in the clinical setting, include the detection of dermatophyte DNA by polymerase chain reaction (PCR)-based amplification5,6 and an antibody-based immunochromatography strip method.7 It should be stressed that the diagnostic targets of these approaches are the principal pathogens of onychomycosis, that is, the dermatophytes. Raman spectroscopy is able to detect subtle biochemical changes in biological samples providing chemical and compositional information. In principle, it is suitable for biological specimens because it does not interfere with water molecules, it is noninvasive, and it requires minimal sample preparation; however, it is typically limited by considerable tissue auto-fluorescence interference and an overall low signal-to-noise ratio.8–10 Considering Raman spectroscopy applications to nail diseases, an ex vivo model of dermatophyte infection of nails has been evaluated and showed promising results in the differentiation of the dermatophyte as the causative agent of onychomycosis from other nondermatophytic pathogens, like Candida and Scopulariopsis species.11 Furthermore, successful identification of clinical Candida species with Raman was achieved in culture material.12,13 The purpose of this study was to evaluate the discriminative power of Raman spectroscopy in the ex vivo differentiation of nail clinical samples with confirmed infection by the dermatophyte T. rubrum or Candida species from healthy (normal) nails. To our knowledge, this is the first onychomycosis study that investigated the use of predictive classification modeling as a means for the diagnosis and pathogen differentiation in clinical nail samples, highlighting the potential of Raman spectroscopy in the clinical practice. Methods Samples Institutional Ethical Review Committee permission was granted prior to experiments confirming the protocol used and the anonymity of the donors. Raman measurements were performed on nail clippings (N = 52, nails of 52 volunteers). A randomly selected part of the acquired nail material of all cases was subjected to both direct KOH examination and mycological culture. Only macromorphologically diseased nails with a confirmed Trichophyton rubrum (N = 12) or Candida species infection (N = 14) were included in the spectroscopic analysis. Different Candida species were evaluated, including samples of Candida parapsilosis complex (N = 11), Candida glabrata (N = 1), and Candida albicans (N = 2). The negative controls (N = 26) were clippings of clinically healthy nails from volunteers with no evidence of any skin disease and in addition a negative mycological laboratory evaluation. The clippings used for Raman spectroscopy were not subjected to any chemical pretreatment, and care was taken to only include samples with no evident environmental contaminants. Spectra were collected at 30 mW output power, with 785 nm diode laser at ∼4.5 cm−1 resolution and 5 s accumulation time, in the 200–3200 cm−1 spectral range. Stokes Raman photons were detected with a 2048 pixel thermoelectrically cooled back-thinned CCD. For each nail clipping, four Raman spectra were measured on different spots to ensure statistical variability (208 spectra in total). Data analysis Signal processing and multivariate statistical analysis were performed using the Unscrambler X (CAMO Software AS, Oslo, Norway) software package. Data processing of the Raman spectra was carried out in the fingerprint region (400–2000 cm−1). Spectra were smoothed using Savitzky-Golay transform (11-point moving window width and second degree polynomial), baseline corrected and normalized by applying the standard normal variate (SNV) method. Data reduction of the 827 variables (wavenumbers) was accomplished by principal component analysis (PCA) with full cross validation. PCA reduces the dimensionality of the data set and also retains the original variability of the data by constructing orthogonal uncorrelated principal components (PCs). The first PC is a linear combination of the starting predictor variables and covers the maximum variance possible in the data set. Similarly, the second PC is uncorrelated (i.e., perpendicular) to the first and accounts for the highest possible of the remaining residual variance and so on for the remaining PCs. Clustering was visualized by PCA score plots while loading plots identified the spectral regions, which predominantly contribute to the variance in the data set. In order to be able to classify an unknown, future clinical sample, discrimination enhancement and class modeling were implemented using the soft independent modeling of class analogy (SIMCA) classification technique. In this supervised, pattern recognition scheme, construction of separate PCA models for each class are built at the training stage followed by a validation procedure to effectively map the spectral variables to the 3-classes (normal, T. rubrum, Candida species) subspace. The data set was split randomly into training and test sets, consisting of 38 and 14 samples, respectively. The latter consisted of eight normal samples and six samples of T. rubrum and Candida species equally divided (three of each species). Data were mean centered and PCA was performed with full cross-validation. The first two PCs were used in SIMCA classification and the significance limit was set to 5% (95% confidence). Partial least squares discriminant analysis (PLS-DA) employed the same sets with SIMCA. In this approach, the Raman spectra were grouped into the three predefined classes and a discriminant model was constructed and tested on the validation dataset. The classification error margin was evaluated by the root mean square error of calibration and validation (RMSEC and RMSEV, respectively). Prediction diagnostics were assessed by RMSE of prediction (RMSEP) of the test set. PLS-DA and SIMCA are two powerful classification algorithms that apply different criteria to construct predictive models. SIMCA computes an assembly of PCA submodels and captures within-class variability, while PLS-DA identifies latent variables that optimize the separation among classes. Therefore, PLS-DA typically provides better classification results because class-separation is maximized (given that within-class variance remains low). However, with a high number of classes, discrimination becomes increasingly more complex and results difficult to overview. On the other hand, SIMCA has some attractive features over other classification methods. First, it handles samples described as spectra in a dimensionally much lower subspace due to the prior principal component mapping and there is no restriction on the number of measurement variables. Second, with this method, new samples presenting spectral features not yet included in the predictive model can be efficiently assigned as either outliers or as members of some additionally recognized new class. In our case, this explorative model feature is particularly useful in expanding the clinical application of Raman spectroscopy to onychomycosis studies since it anticipates the identification of additional pathogens, not already belonging in model's feeding data. Consequently, for optimizing diagnostic output both analytical instruments PLS-DA (main strength: optimizing the separation among classes) and SIMCA (main strength: monitoring within-class variability) are utilized in parallel in this study. Results Figure 1 shows the mean Raman spectra recorded from 52 nail clippings of normal (N = 26) and infected nails: T. rubrum (N = 12), Candida spp. (N = 14). The acquired spectra are consistent with data in the literature, which provided the basis for the interpretation of the vibrational modes detailed in Table 1. The spectral features can be clearly resolved in all samples although intrinsic dispersion of data is apparent particularly in the Candida species group. Minor differences in peak positions and intensities could be mainly attributed to the biophysical variability of the α-keratin structure, the predominant structural macromolecule of the nail plate.14 Figure 1. View largeDownload slide Ex vivo, mean Raman spectra with standard deviations (SDev, cyan area) of normal nails (N = 26) and infected nails: T. rubrum (N = 12), Candida spp. (N = 14). The spectra (solid lines) are artificially offset for clarity. Dotted lines represent zero-offset spectra. This Figure is reproduced in color in the online version of Medical Mycology. Figure 1. View largeDownload slide Ex vivo, mean Raman spectra with standard deviations (SDev, cyan area) of normal nails (N = 26) and infected nails: T. rubrum (N = 12), Candida spp. (N = 14). The spectra (solid lines) are artificially offset for clarity. Dotted lines represent zero-offset spectra. This Figure is reproduced in color in the online version of Medical Mycology. Table 1. Tentative assignment of the most prominent Raman bands for normal and infected nail samples.15–18 Raman shift (cm−1) Band assignment Normal Trichophyton rubrum Candida spp. v(S-S) (gauche-gauche-gauche) 513 515 500/519 v(C-S) gauche 602 602/619 602 v(C-S) gauche (amide I) – 648 – ρ(CH2) in-phase 723 721 710 δ (CCH) aliphatic, Tyr 826 828 817 δ (CCH) aromatic, Tyr 855 855 866 v(C-C) skeletal, α-helix 936 938 942 v(C-C) aromatic, Phe 1004 1006 1010 v(C-H) Phe 1030 1032 1026 v(C-C) skeletal, trans 1104/1024 1104/1024 1106/1022 v(C-C), Tyr 1172 1176 1174 v(C-C6H5), Tyr and Phe 1207 1209 1222 δ(CH2) wagging; v(C-N) amide III 1250 1251 1253 δ(CH2) deformation 1319 1317 1307 δ(CH2) deformation 1338 1339 1343 δ(CH3) deformation 1418 1418 1415 δ(CH2) scissoring 1450 1450 1446 v(C = C), carotenoid 1514 1514 1514 δ(NH), v(C-N), Trp 1550 1550 – v(C-C) olefinic, 1610 1613 1605 ν(C = O) amide I α–helix 1654 1656 1661 ν(C = O) amide I β–sheet 1670 1674 1677 Raman shift (cm−1) Band assignment Normal Trichophyton rubrum Candida spp. v(S-S) (gauche-gauche-gauche) 513 515 500/519 v(C-S) gauche 602 602/619 602 v(C-S) gauche (amide I) – 648 – ρ(CH2) in-phase 723 721 710 δ (CCH) aliphatic, Tyr 826 828 817 δ (CCH) aromatic, Tyr 855 855 866 v(C-C) skeletal, α-helix 936 938 942 v(C-C) aromatic, Phe 1004 1006 1010 v(C-H) Phe 1030 1032 1026 v(C-C) skeletal, trans 1104/1024 1104/1024 1106/1022 v(C-C), Tyr 1172 1176 1174 v(C-C6H5), Tyr and Phe 1207 1209 1222 δ(CH2) wagging; v(C-N) amide III 1250 1251 1253 δ(CH2) deformation 1319 1317 1307 δ(CH2) deformation 1338 1339 1343 δ(CH3) deformation 1418 1418 1415 δ(CH2) scissoring 1450 1450 1446 v(C = C), carotenoid 1514 1514 1514 δ(NH), v(C-N), Trp 1550 1550 – v(C-C) olefinic, 1610 1613 1605 ν(C = O) amide I α–helix 1654 1656 1661 ν(C = O) amide I β–sheet 1670 1674 1677 View Large Table 1. Tentative assignment of the most prominent Raman bands for normal and infected nail samples.15–18 Raman shift (cm−1) Band assignment Normal Trichophyton rubrum Candida spp. v(S-S) (gauche-gauche-gauche) 513 515 500/519 v(C-S) gauche 602 602/619 602 v(C-S) gauche (amide I) – 648 – ρ(CH2) in-phase 723 721 710 δ (CCH) aliphatic, Tyr 826 828 817 δ (CCH) aromatic, Tyr 855 855 866 v(C-C) skeletal, α-helix 936 938 942 v(C-C) aromatic, Phe 1004 1006 1010 v(C-H) Phe 1030 1032 1026 v(C-C) skeletal, trans 1104/1024 1104/1024 1106/1022 v(C-C), Tyr 1172 1176 1174 v(C-C6H5), Tyr and Phe 1207 1209 1222 δ(CH2) wagging; v(C-N) amide III 1250 1251 1253 δ(CH2) deformation 1319 1317 1307 δ(CH2) deformation 1338 1339 1343 δ(CH3) deformation 1418 1418 1415 δ(CH2) scissoring 1450 1450 1446 v(C = C), carotenoid 1514 1514 1514 δ(NH), v(C-N), Trp 1550 1550 – v(C-C) olefinic, 1610 1613 1605 ν(C = O) amide I α–helix 1654 1656 1661 ν(C = O) amide I β–sheet 1670 1674 1677 Raman shift (cm−1) Band assignment Normal Trichophyton rubrum Candida spp. v(S-S) (gauche-gauche-gauche) 513 515 500/519 v(C-S) gauche 602 602/619 602 v(C-S) gauche (amide I) – 648 – ρ(CH2) in-phase 723 721 710 δ (CCH) aliphatic, Tyr 826 828 817 δ (CCH) aromatic, Tyr 855 855 866 v(C-C) skeletal, α-helix 936 938 942 v(C-C) aromatic, Phe 1004 1006 1010 v(C-H) Phe 1030 1032 1026 v(C-C) skeletal, trans 1104/1024 1104/1024 1106/1022 v(C-C), Tyr 1172 1176 1174 v(C-C6H5), Tyr and Phe 1207 1209 1222 δ(CH2) wagging; v(C-N) amide III 1250 1251 1253 δ(CH2) deformation 1319 1317 1307 δ(CH2) deformation 1338 1339 1343 δ(CH3) deformation 1418 1418 1415 δ(CH2) scissoring 1450 1450 1446 v(C = C), carotenoid 1514 1514 1514 δ(NH), v(C-N), Trp 1550 1550 – v(C-C) olefinic, 1610 1613 1605 ν(C = O) amide I α–helix 1654 1656 1661 ν(C = O) amide I β–sheet 1670 1674 1677 View Large The strong Raman band at 500–520 cm−1 is attributed to the disulfide stretching bond of cystine and cysteine residues.19 Disulfide cross linking is central for the physical and mechanical properties of keratin. The intensities and the position of the Cα-Cβ-S-S’-C’β-C’α band, infer the sulfur amount and the conformation of the disulfide bridges. The most energetically favorable gauche-gauche-gauche conformation is observed at 513–515 cm−1. However, samples infected with Candida species feature an additional shoulder at 519 cm−1, which is typical of the less stable gauche-gauche-trans conformation indicating heterogeneity among the S-S bonds conformation.20 In addition, the nails affected by T. rubrum are characterized by two weak, but distinct Raman bands at 619 cm−1 and 648 cm−1, which correspond to the C-S stretching vibration. A striking difference in Raman scattering also occurs at 1550 cm−1; this peak, which is attributed to amide II (60% N–H bend and 40% C–N stretch) and tryptophan (Trp), is absent from the samples of nails infected by Candida species. Strong bands found in common across all three groups of samples include the symmetric ring breathing mode and the C–H in-plane bending mode of phenylalanine (Phe) as well as the amide I, II, and III bands. Of these later bands, amide I (near 1650 cm−1) integrates contributions from the C = C in-plane bending mode of Phe and tyrosine (Tyr), the α-helix C = O stretching, and the β-sheet C = O stretching. Amide II (near 1450 cm−1) involves the deformation of –CH, –CH2, and –CH3 moieties and amide III (near 1350 cm−1) constitutes of broad bands attributed to C–N stretching and –CH2 deformation and scissoring. PCA has been employed in order to quantitatively discriminate between normal and infected nails. The spectra are represented in groups of similar variability; the construction of PCs permits the discrimination of different spectral groups in a dataset. In general, the first three PCs account for the highest variance present and give the best representative differentiation of the distinct clusters.21 In our case, the first three PCs combined, accounted for 89% of the total variability in the dataset (PC1: 49%, PC2: 21%, PC3: 16%). Visualizations of PCs in two-dimensional score plots often reveal which components must be taken into account, noting that PCs explain less variance in decreasing order, and consequently the first two contain most of the information. Figure 2 shows that distinct clustering among the three sample groups is achieved through PC1 and PC2. Figure 2. View largeDownload slide PCA Scores plot with Hotelling T2 (0.95) ellipse, partitioning 52 nail samples within the area defined by the two first PCs; (sl, sensu lato). This Figure is reproduced in color in the online version of Medical Mycology. Figure 2. View largeDownload slide PCA Scores plot with Hotelling T2 (0.95) ellipse, partitioning 52 nail samples within the area defined by the two first PCs; (sl, sensu lato). This Figure is reproduced in color in the online version of Medical Mycology. The potential of the PCA approach lies in the information obtained by the loading plots, which visualize the coefficients between the original variables (wavenumbers) and the PCs. The effect of the original variables upon a PC is the numerical value of the loading. Thus, the significant molecular vibrations that shape the clustering can be identified by inspecting the loading plot. The loadings of the first two PCs are shown in Figure 3. Figure 3. View largeDownload slide PCA loading plot for the first two PCs. Specific regions of interest are indicated in gray color. This Figure is reproduced in color in the online version of Medical Mycology. Figure 3. View largeDownload slide PCA loading plot for the first two PCs. Specific regions of interest are indicated in gray color. This Figure is reproduced in color in the online version of Medical Mycology. The PC1 loadings pattern is negatively correlated with the bands assigned to disulfide bridges, skeletal vibrational bands (α-helix and Phe) and amides I, II, and III (gray stripes in Fig. 3). These bands are responsible for the clear clustering of the normal nails as illustrated in the scores plot (Fig. 2). In contrast, PC2 loadings mostly contribute to the differentiation of T. rubrum and Candida species. This component is positively correlated with the -S-S- bands (490–530 cm−1) and the aromatic breathing mode of Phe, while it is negatively correlated with the amide bands. In an effort to predict the class membership (normal, T. rubrum, Candida species) of new observations, we first modeled the defined groups with SIMCA. Results showed 100% accuracy (calculated by the sum of true positives (TP) and true negatives (TN) divided by the total number of samples) for the classification into healthy and onychomycosis nails. When instead of the former two, three classes were used, false positives (FP) (i.e., samples that were classified as belonging to two classes) reduced the accuracy to the still satisfactory 92.8%, mainly as discrimination unambiguity between T. rubrum and Candida species affected nails (Supplementary information, Table S1 displays the confusion matrix obtained by cross-validation for training and test sets). Finally, discriminant PLS was employed to test the classification of new samples, not included in the construction of the model. PLS regression decomposes both X and Y variables as a product of orthogonal components similarly to PCA, albeit referred to as factors. By using the first two factors, 73% and 94% of the variance was explained by independent and dependent variables, respectively. As expected, a similar loading plot was obtained with amide stretches from the proteins backbone, peaks from side chain amino acids and the characteristic disulfide vibration. The low RMSEC (0.20) and RMSEV (0.23) indicate adequate stability in the classification of new samples. Indeed, the classification for the test set yielded 100% accuracy, with low RMSEP: 0.24 (Supplementary information, Fig. S1), albeit with significant uncertainty of the mean value (±31%) for the Candida species infected nail samples (Fig. 4). Figure 4. View largeDownload slide Predicted values with estimated uncertainties for the test set (14 samples) calibrated by PLS-DA (C, Candida spp., T, T. rubrum, N, normal). This Figure is reproduced in color in the online version of Medical Mycology. Figure 4. View largeDownload slide Predicted values with estimated uncertainties for the test set (14 samples) calibrated by PLS-DA (C, Candida spp., T, T. rubrum, N, normal). This Figure is reproduced in color in the online version of Medical Mycology. Discussion Three clinically important groups of human nails, that is, healthy and either T. rubrum or Candida infected ones, were studied by means of Raman spectroscopy. The main differences in the Raman spectra between healthy and onychomycosis nails were observed in the regions attributed to disulfide bond imprints, Phe vibration and amide I, II, and III bands. These observations are in accordance with previous studies in the literature that reported increased proportion of the less stable disulfide conformations in nail clippings from patients with onychomycosis undefined at species level.22 The two characteristic Raman bands at 619 cm−1 and 648 cm−1, which correspond to the C-S stretching vibration, they have been also described in nails with unspecified onychomycosis before and after Nd:YAG laser treatment.15 In addition, these findings are supported by the current knowledge of the molecular structure of healthy and infected nails. Normal nail plate is composed of hard α-keratins.23 These keratins contain large amounts of cysteine, which is cross-linked with disulfide bonds that make nails a biological material extremely resistant to biodegradation in nature.24,25 However, dermatophytes, including T. rubrum, employ a large array of specialized extracellular proteases, keratinases, that in cooperation with the segregation of sulfite,26 can digest these bonds and subsequently utilize degraded keratin as a protein source.27,28 A modified Raman spectrum could result from further invasion of the nail plate with development of hyphae that would increase the interaction surface25 and trigger the described chemical alterations. Also, Candida species possess extracellular enzymes with keratinolytic activity29–31 including the ability to resist large concentrations of cysteine by the excretion of sulfite32 which, as already mentioned, catalyses the disruption of the -S-S- bonds in nail keratin. Furthermore, the Raman band corresponding to the vibration of Trp is absent in the spectra of nails infected with Candida. Trp is a hydrophobic amino acid, which is essential for the structural stability of the lipid bilayers of the yeast cell membranes.33 Studies have shown that peptides with Trp residues yield enhanced anticandidal activity.34 The role of Trp in onychomycosis is still obscure and further studies are needed to decipher its implication to candidiasis, possibly in relation to immunosuppression.35 Raman spectroscopy can be easily integrated in the clinical setting to provide real time evaluation of diseased nails. It should be emphasized that in conjunction with SIMCA and PLA-DA approaches, Raman spectroscopy flawlessly classified normal nails. The effective discrimination between healthy versus fungal infected nails class model is especially important when on-site evaluation of onychomycosis is needed. The main limitation of this work is the relatively low number of evaluated samples, contrary to high prevalence of onychomycosis in the general population. The next challenge is to evaluate Raman spectroscopy in prospective studies and in comparison with standard fungal identification techniques (KOH microscopy and culture) and in association with patient characteristics (e.g., underlying diabetes), including also healthy and altered nails in patients with other dermatoses (e.g., psoriasis or lichen planus)36 as well as nails affected by multiple pathogens and possibly also bacterial species.37 Furthermore, inclusion in the dataset of other fungal pathogens and patient disease data (including clinical subforms of onychomycosis), could significantly improve the diagnostic potential of the proposed approach. Supplementary material Supplementary data are available at MMYCOL online. Acknowledgements The authors wish to express their appreciation to A. Milioni and S. Kritikou for expert technical assistance. This work was partially supported by the European Union (European Regional Development Fund-ERDF) and Greek national funds through the Operational Program “THESSALY-MAINLAND GREECE AND EPIRUS-2007–2013” of the National Strategic Reference Framework (NSRF 2007–2013). Declaration of interest The authors report no conflicts of interest. The authors alone are responsible for the content and the writing of the paper. References 1. Sigurgeirsson B , Baran R . The prevalence of onychomycosis in the global population: a literature study . J Eur Acad Dermatol Venereol . 2014 ; 28 : 1480 – 1491 . Google Scholar CrossRef Search ADS PubMed 2. Weinberg JM , Koestenblatt EK , Tutrone WD , Tishler HR , Najarian L . Comparison of diagnostic methods in the evaluation of onychomycosis . J Am Acad Dermatol . 2003 ; 49 : 193 – 197 . Google Scholar CrossRef Search ADS PubMed 3. Jung MY , Shim JH , Lee JH et al. Comparison of diagnostic methods for onychomycosis, and proposal of a diagnostic algorithm . Clin Exp Dermatol . 2015 ; 40 : 479 – 484 . Google Scholar CrossRef Search ADS PubMed 4. Ameen M , Lear JT , Madan V , Mohd Mustapa MF , Richardson M . British Association of Dermatologists' guidelines for the management of onychomycosis 2014 . Br J Dermatol . 2014 ; 171 : 937 – 958 . Google Scholar CrossRef Search ADS PubMed 5. Arabatzis M , Bruijnesteijn van Coppenraet LES , Kuijper EJ et al. Diagnosis of common dermatophyte infections by a novel multiplex real-time polymerase chain reaction detection/identification scheme . Br J Dermatol 2007 ; 157 : 681 – 689 . Google Scholar CrossRef Search ADS PubMed 6. Brillowska-Dąbrowska A , Saunte DM , Arendrup MC . Five-hour diagnosis of dermatophyte nail infections with specific detection of Trichophyton rubrum . J Clin Microbiol . 2007 ; 45 : 1200 – 1204 . Google Scholar CrossRef Search ADS PubMed 7. Tsunemi Y , Hiruma M . Clinical study of Dermatophyte Test Strip, an immunochromatographic method, to detect tinea unguium dermatophytes . J Dermatol . 2016 ; 43 : 1417 – 1423 . Google Scholar CrossRef Search ADS PubMed 8. Elka A , Moulia V , Spyridonos P , Kourkoumelis N . The effect of irradiance and integration time in in vivo normal skin Raman measurements assessed by multivariate statistical analysis . Biomed Spectrosc Imag . 2016 ; 5 : 217 – 223 . Google Scholar CrossRef Search ADS 9. Kourkoumelis N , Polymeros A , Tzaphlidou M . Background estimation of biomedical Raman spectra using a geometric approach . Spectroscopy . 2012 ; 27 : 441 – 447 . Google Scholar CrossRef Search ADS 10. Kourkoumelis N , Balatsoukas I , Moulia V , Elka A , Gaitanis G , Bassukas ID . Advances in the in vivo Raman spectroscopy of malignant skin tumors using portable instrumentation . Int J Mol Sci . 2015 ; 16 : 14554 – 14570 . Google Scholar CrossRef Search ADS PubMed 11. Smijs TG , Jachtenberg JW , Pavel S et al. Detection and differentiation of causative organisms of onychomycosis in an ex vivo nail model by means of Raman spectroscopy . J Eur Acad Dermatol Venereol . 2014 ; 28 : 1492 – 1499 . Google Scholar CrossRef Search ADS PubMed 12. Ibelings MS , Maquelin K , Endtz HP , Bruining HA , Puppels GJ . Rapid identification of Candida spp. in peritonitis patients by Raman spectroscopy . Clin Microbiol Infect . 2004 ; 11 : 353 – 358 . Google Scholar CrossRef Search ADS 13. Maquelin K , Choo-Smith L-P , Endtz HP , Bruining HA , Puppels GJ . Rapid identification of Candida species by confocal Raman microspectroscopy . J Clin Microbiol . 2002 ; 40 : 594 – 600 . Google Scholar CrossRef Search ADS PubMed 14. Moll R , Divo M , Langbein L . The human keratins: biology and pathology . Histochem Cell Biol . 2008 ; 129 : 705 – 733 . Google Scholar CrossRef Search ADS PubMed 15. Shin MK , Kim TI , Kim WS , Park H-K , Kim KS . Changes in nail keratin observed by Raman spectroscopy after Nd:YAG laser treatment . Microsc Res Tech . 2017 ; 80 : 338 – 343 . Google Scholar CrossRef Search ADS PubMed 16. Widjaja E , Seah RK . Use of Raman spectroscopy and multivariate classification techniques for the differentiation of fingernails and toenails . Appl Spectrosc . 2006 ; 60 : 343 – 345 . Google Scholar CrossRef Search ADS PubMed 17. Williams AC , Edwards HGM , Barry BW . Raman spectra of human keratotic biopolymers: skin, callus, hair and nail . J Raman Spectrosc . 1994 ; 25 : 95 – 98 . Google Scholar CrossRef Search ADS 18. Widjaja E , Lim GH , An A . A novel method for human gender classification using Raman spectroscopy of fingernail clippings . Analyst 2008 ; 133 : 493 – 498 . Google Scholar CrossRef Search ADS PubMed 19. Widjaja E , Garland M . Detection of bio-constituents in complex biological tissue using Raman microscopy . Application to human nail clippings. Talanta . 2010 ; 80 : 1665 – 1671 . 20. Gniadecka M , Nielsen OF , Christensen DH , Wulf HC . Structure of water, proteins, and lipids in intact human skin, hair, and nail . J Invest Dermatol . 1998 ; 110 : 393 – 398 . Google Scholar CrossRef Search ADS PubMed 21. Bonnier F , Byrne HJ . Understanding the molecular information contained in principal component analysis of vibrational spectra of biological systems . Analyst . 2012 ; 137 : 322 – 332 . Google Scholar CrossRef Search ADS PubMed 22. Wen W , Meng Y , Xiao J , Zhang P , Zhang H . Comparative study on keratin structural changes in onychomycosis and normal human finger nail specimens by Raman spectroscopy . J Mol Struct . 2013 ; 1038 : 35 – 39 . Google Scholar CrossRef Search ADS 23. de Berker DA , Andre J , Baran R . Nail biology and nail science . Int J Cosmet Sci . 2007 ; 29 : 241 – 275 . Google Scholar CrossRef Search ADS PubMed 24. Kornillowicz-Kowalska T , Bohacz J . Biodegradation of keratin waste: Theory and practical aspects . Waste Manag . 2011 ; 31 : 1689 – 1701 . Google Scholar CrossRef Search ADS PubMed 25. Filipello Marchisio V. Keratinophilic fungi: their role in nature and degradation of keratinic substrates . In: Kushawaha RKS , Guarro J , eds. Biology of Dermatophytes and Other Keratinophilic Fungi . Bilbao : Revista Iberoamericana Micologia , 2000 : 86 – 92 . 26. Grumbt M , Monod M , Yamada T et al. Keratin degradation by dermatophytes relies on cysteine dioxygenase and a sulfite efflux pump . J Invest Dermatol . 2013 ; 133 : 1550 – 1555 . Google Scholar CrossRef Search ADS PubMed 27. Monod M. Secreted proteases from dermatophytes . Mycopathologia . 2008 ; 166 : 285 – 294 . Google Scholar CrossRef Search ADS PubMed 28. Léchenne B , Reichard U , Zaugg C et al. Sulphite efflux pumps in Aspergillus fumigatus and dermatophytes . Microbiology . 2007 ; 153 : 905 – 913 . Google Scholar CrossRef Search ADS PubMed 29. Ray TL , Payne CD . Comparative production and rapid purification of Candida acid proteinase from protein-supplemented cultures . Infect Immun . 1990 ; 58 : 508 – 514 . Google Scholar PubMed 30. Negi M , Tsuboi R , Matsui T , Ogawa H . Isolation and characterization of proteinase from Candida albicans: substrate specificity . J Invest Dermatol . 1984 ; 83 : 32 – 36 . Google Scholar CrossRef Search ADS PubMed 31. Vermelho AB , Mazotto AM , de Melo AC et al. Identification of a Candida parapsilosis strain producing extracellular serine peptidase with keratinolytic activity . Mycopathologia . 2010 ; 169 : 57 – 65 . Google Scholar CrossRef Search ADS PubMed 32. Hennicke F , Grumbt M , Lermann U et al. Factors supporting cysteine tolerance and sulfite production in Candida albicans . Eukaryot Cell . 2013 ; 12 : 604 – 613 . Google Scholar CrossRef Search ADS PubMed 33. Chan DI , Prenner EJ , Vogel HJ . Tryptophan- and arginine-rich antimicrobial peptides: structures and mechanisms of action . Biochim Biophys Acta . 2006 ; 1758 : 1184 – 1202 . Google Scholar CrossRef Search ADS PubMed 34. Lum KY , Tay ST , Le CF et al. Activity of novel synthetic peptides against Candida albicans . Sci Rep . 2015 ; 5 : 9657 . Google Scholar CrossRef Search ADS PubMed 35. Belladonna ML , Puccetti P , Orabona C et al. Immunosuppression via tryptophan catabolism: the role of kynurenine pathway enzymes . Transplantation . 2007 ; 84 : S17 – S20 . Google Scholar CrossRef Search ADS PubMed 36. Hay RJ , Baran R . Onychomycosis: a proposed revision of the clinical classification . J Am Acad Dermatol . 2011 ; 65 : 1219 – 1227 . Google Scholar CrossRef Search ADS PubMed 37. Baran R , de Berker DA , Holzberg M , Thomas L , (Eds). Baran and Dawber's Diseases of the Nails and Their Management , 4th edn . London : John Wiley & Sons , 2012 . Google Scholar CrossRef Search ADS © The Author(s) 2017. Published by Oxford University Press on behalf of The International Society for Human and Animal Mycology. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
Medical Mycology – Oxford University Press
Published: Oct 9, 2017
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera