Next Article in Journal
Metabolomics-Based Machine Learning Models Accurately Predict Breast Cancer Estrogen Receptor Status
Previous Article in Journal
Rapid LC-MS/MS Evaluation of Collagen and Elastin Crosslinks in Human and Mouse Lung Tissue with a Novel Bioanalytical Surrogate Matrix Approach
Previous Article in Special Issue
Investigating the Link between STAT4 Genetic Variants, STAT4 Protein Concentrations, and Laryngeal Squamous Cell Carcinoma: A Comprehensive Analysis of Clinical Manifestations
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Comment

Comment on Borșa et al. Developing New Diagnostic Tools Based on SERS Analysis of Filtered Salivary Samples for Oral Cancer Detection. Int. J. Mol. Sci. 2023, 24, 12125

by
Ivan Bratchenko
* and
Lyudmila Bratchenko
Laser and Biotechnical Systems Department, Samara National Research University, Moskovskoe Shosse 34, 443086 Samara, Russia
*
Author to whom correspondence should be addressed.
Int. J. Mol. Sci. 2024, 25(23), 13030; https://doi.org/10.3390/ijms252313030
Submission received: 6 March 2024 / Revised: 20 September 2024 / Accepted: 17 October 2024 / Published: 4 December 2024
(This article belongs to the Special Issue Pathogenesis and Therapy of Oral Carcinogenesis, 2nd Edition)

Abstract

This comment discusses a recent research paper on the classification of saliva samples with SERS by Borsa et al. The authors suggested utilizing PCA-LDA to detect oral cancer and claimed to achieve an accuracy of up to 77%. Despite the high prediction capability of the proposed approach, the demonstrated findings could be treated as unclear due to possible overestimation of the proposed classification models. Data should be provided for both the training and the validation sets to make sure that there were no repeated data from the same sample in either set. Moreover, the authors proposed to measure opiorphin in saliva with SERS as a potential biomarker of oral cancer. However, opiorphin in saliva is contained in ng/mL concentrations, and the proposed technique is most likely not capable of recording the real concentration of opiorphin.

Today, liquid biopsies based on spectral techniques [1,2] and especially on Raman spectroscopy [3,4] provide a way for the fast and precise analysis of body fluids [5,6,7]. The analysis of Raman spectra of tissues and biofluids can be utilized in a variety of applications such as cancer screening, non-communicable disease detection, virus tracking, etc. [8]. At the same time, the complexity and variability of spectral data and the even higher complexity of modern statistical approaches can be misleading and provide incorrect results [9,10,11,12].
Borsa et al. [13] examined saliva samples with SERS (surface-enhanced Raman spectroscopy) [14] to determine the possibility of oral cancer detection. To find the most important spectral areas and classify normal samples vs. cancerous ones, the authors utilized PCA-LDA (principal component analysis with linear discriminant analysis) [15]. The obtained accuracy in cancer detection was stated to be up to 77%, and opiorphin was proposed as a possible oral cancer biomarker. Basically, these methods are quite common in liquid biopsy; however, some details of the study indicate that the proposed classification models may have been overestimated and that opiorphin cannot be detected with the proposed approach in saliva in natural concentrations.
First, Borsa et al. [13] suggested using Leave-One-Out Cross-Validation (LOOCV) and compared the LOOCV accuracy with the training accuracy. This looks like an appropriate approach to validating classification models but, in conclusion to their study, the authors stated: “However, PCA-LDA yielded an overfit model (97% training accuracy vs. 83% Leave-One-Out Cross-Validation (LOOCV) accuracy). For that reason, we used five PCA components which account for 90% of the total variance and are less overfit in the PCA-LDA analysis (85% training accuracy vs. 77% LOOCV accuracy).” The authors’ choice of the principal components (PCA components) was based on the proximity of the results for training and LOOCV. However, this approach may have led to an inadequate estimation of the classification model results. For reference, we have provided Figure 1 with the data from our previous studies [16,17], where we compared the data used for training (blue line) and cross-validation (red line). As can be seen from the figure, a higher number of principal components (e.g., eight principal components) can yield slightly overfitted data, but the number of principal components should be limited in the first local minimum on a cross-validation curve. Otherwise, the utilized classification model will utilize junk components that may even decrease the performance of the model. Note here that high performance of eight and nine principal components is just a random result. Thus, the authors should provide the same data to prove the correct number of principal components. In addition, Borsa et al. could have analyzed a scree plot to choose the proper number of principal components [15]. Overall, for groups of about 17 samples, only 2–3 first principal components can be effectively utilized for constructing classification models [9].
Another important issue in the commented paper is the ability to detect opiorphin. Their conclusion stated the following: “The presence of opiorphin in salivary samples was experimentally proven for the first time”. The SERS spectra of opiorphin (Figure 2) were collected with a concentration of 0.1 mg/mL. However, in real saliva samples, the concentration of opiorphin can be in the range of about 3–30 ng/mL [18]. Therefore, to estimate the contribution of opiorphin nanograms (per ml) in real saliva samples, the authors should have reduced the opiorphin spectra in Figure 2 by a factor of at least 10,000 (about 104 to 106). In doing so, the opiorphin contribution to SERS spectra of saliva would decrease and may become as low as measurement noise. Note here that Borsa et al. associated the 1002 cm−1 band with the opiorphin presence; however, this spectral band may be associated with tens of substances that are present in saliva with higher concentrations (e.g., uric acid, exosomes [19,20,21], etc.).
In addition, it should be noted that certain molecules have higher affinities for metal surfaces, and therefore the SERS spectra of biofluids can be dominated, for example, by the contributions of purine (uric acid, hypoxathine, adenine) and sulfur-containing (ergothioneine, glutathione) metabolites [22,23]. Thus, to demonstrate the real concentration of an exact analyte, SERS measurements should be accompanied with other well-established techniques (such as liquid chromatography or other laboratory techniques) [4]. Without determining the exact concentration of the target analyte and the ability to detect this analyte in real concentrations with SERS, numerous biological molecules may be hypothetically treated as biomarkers due to their potential contribution to the recorded SERS spectra.
In summary, to prove the correctness of the classification models, it is advisable to provide additional data regarding the utilized cross-validation procedure. It is necessary to compare the performance of training and cross-validation and to estimate a scree plot. The authors should demonstrate that their technique is capable of detecting opiorphin in saliva samples at ng/mL concentrations (or to establish the limit of opiorphin detection).

Author Contributions

Conceptualization, I.B. and L.B.; methodology, I.B. and L.B.; software, I.B. and L.B.; validation, I.B. and L.B. and I.B.; formal analysis, I.B. and L.B.; investigation, I.B. and L.B.; writing—original draft preparation, I.B. and L.B.; writing—review and editing, I.B. and L.B.; visualization, I.B. and L.B. All authors have read and agreed to the published version of the manuscript.

Funding

The findings of this paper were supported by the Russian Science Foundation Project No. 21-75-10097.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Fornasaro, S.; Alsamad, F.; Baia, M.; Batista de Carvalho, L.A.; Beleites, C.; Byrne, H.J.; Chiadò, A.; Chis, M.; Chisanga, M.; Daniel, A.; et al. Surface enhanced Raman spectroscopy for quantitative analysis: Results of a large-scale European multi-instrument interlaboratory study. Anal. Chem. 2020, 92, 4053–4064. [Google Scholar] [CrossRef] [PubMed]
  2. Bonifacio, A.; Cervo, S.; Sergo, V. Label-free surface-enhanced Raman spectroscopy of biofluids: Fundamental aspects and diagnostic applications. Anal. Bioanal. Chem. 2015, 407, 8265–8277. [Google Scholar] [CrossRef] [PubMed]
  3. Chen, S.; Zhang, H.; Yang, X.; Shao, X.; Li, T.; Chen, N.; Chen, Z.; Xue, W.; Pan, J.; Liu, S. Raman Spectroscopy Reveals Abnormal Changes in the Urine Composition of Prostate Cancer: An Application of an Intelligent Diagnostic Model with a Deep Learning Algorithm. Adv. Intell. Syst. 2021, 3, 2000090. [Google Scholar] [CrossRef]
  4. Bratchenko, L.A.; Al-Sammarraie, S.Z.; Tupikova, E.N.; Konovalova, D.Y.; Lebedev, P.A.; Zakharov, V.P.; Bratchenko, I.A. Analyzing the serum of hemodialysis patients with end-stage chronic kidney disease by means of the combination of SERS and machine learning. Biomed. Opt. Express 2022, 13, 4926–4938. [Google Scholar] [CrossRef]
  5. Moisoiu, T.; Iancu, S.D.; Burghelea, D.; Dragomir, M.P.; Iacob, G.; Stefancu, A.; Cozan, R.G.; Antal, O.; Bálint, Z.; Muntean, V.; et al. SERS Liquid Biopsy Profiling of Serum for the Diagnosis of Kidney Cancer. Biomedicines 2022, 10, 233. [Google Scholar] [CrossRef]
  6. Moisoiu, V.; Iancu, S.D.; Stefancu, A.; Moisoiu, T.; Pardini, B.; Dragomir, M.P.; Crisan, N.; Avram, L.; Crisan, D.; Andras, I.; et al. SERS liquid biopsy: An emerging tool for medical diagnosis. Colloids Surf. B Biointerfaces 2021, 208, 112064. [Google Scholar] [CrossRef]
  7. Esposito, A.; Bonifacio, A.; Sergo, V.; Fornasaro, S. Label-free Surface Enhanced Raman Scattering (SERS) on Centrifugal Silver Plasmonic Paper (CSPP): A Novel Methodology for Unprocessed Biofluids Sampling and Analysis. Biosensors 2021, 11, 467. [Google Scholar] [CrossRef]
  8. Khristoforova, Y.; Bratchenko, L.; Bratchenko, I. Raman-Based Techniques in Medical Applications for Diagnostic Tasks: A Review. Int. J. Mol. Sci. 2023, 24, 15605. [Google Scholar] [CrossRef]
  9. Khristoforova, Y.; Bratchenko, L.; Bratchenko, I. Combination of Raman spectroscopy and chemometrics: A review of recent studies published in the Spectrochimica Acta, Part A: Molecular and Biomolecular Spectroscopy Journal. arXiv 2022, arXiv:2210.10051. [Google Scholar] [CrossRef]
  10. Bratchenko, I.A.; Bratchenko, L.A. Comment on “Serum Raman spectroscopy combined with multiple classification models for rapid diagnosis of breast cancer”. Photodiagn. Photodyn. Ther. 2023, 41, 103215. [Google Scholar] [CrossRef]
  11. Queral-Beltran, A.; Marín-García, M.; Lacorte, S.; Tauler, R. UV-Vis absorption spectrophotometry and LC-DAD-MS-ESI(+)-ESI(−) coupled to chemometrics analysis of the monitoring of sulfamethoxazole degradation by chlorination, photodegradation, and chlorination/photodegradation. Anal. Chim. Acta 2023, 1276, 341563. [Google Scholar] [CrossRef] [PubMed]
  12. Kucheryavskiy, S. mdatools—R package for chemometrics. Chemom. Intell. Lab. Syst. 2020, 198, 103937. [Google Scholar] [CrossRef]
  13. Borșa, R.-M.; Toma, V.; Onaciu, A.; Moldovan, C.-S.; Mărginean, R.; Cenariu, D.; Știufiuc, G.-F.; Dinu, C.-M.; Bran, S.; Opriș, H.-O.; et al. Developing New Diagnostic Tools Based on SERS Analysis of Filtered Salivary Samples for Oral Cancer Detection. Int. J. Mol. Sci. 2023, 24, 12125. [Google Scholar] [CrossRef] [PubMed]
  14. Al-Sammarraie, S.Z.; Bratchenko, L.A.; Typikova, E.N.; Zakharov, V.P.; Bratchenko, I.A.; Lebedev, P.A. Silver nanoparticles-based substrate for blood serum analysis under 785 nm laser excitation. J. Biomed. Photonics Eng. 2022, 8, 010301. [Google Scholar] [CrossRef]
  15. Jolliffe, I.T. Principal Component Analysis, 2nd ed.; Springer Series in Statistics: New York, NY, USA, 2002. [Google Scholar]
  16. Bratchenko, L.A.; Bratchenko, I.A.; Khristoforova, Y.A.; Artemyev, D.N.; Konovalova, D.Y.; Lebedev, P.A.; Zakharov, V.P. Raman spectroscopy of human skin for kidney failure detection. J. Biophotonics 2021, 14, e202000360. [Google Scholar] [CrossRef]
  17. Bratchenko, I.A.; Bratchenko, L.A. Comment on “Combining derivative Raman with autofluorescence to improve the diagnosis performance of echinococcosis”. Spectrochim. Acta Part A Mol. Biomol. Spectrosc. 2021, 252, 119514. [Google Scholar] [CrossRef]
  18. Brkljačić, L.; Sabalić, M.; Salarić, I.; Jerić, I.; Alajbeg, I.; Nemet, I. Development and validation of a liquid chromatography-tandem mass spectrometry method for the quantification of opiorphin in human saliva. J. Chromatogr. B 2011, 879, 3920–3926. [Google Scholar] [CrossRef]
  19. Jaiswal, A.; Madaan, S.; Acharya, N.; Kumar, S.; Talwar, D.; Dewani, D. Salivary Uric Acid: A Noninvasive Wonder for Clinicians? Cureus 2021, 13, e19649. [Google Scholar] [CrossRef]
  20. Zlotogorski-Hurvitz, A.; Dayan, D.; Chaushu, G.; Korvala, J.; Salo, T.; Sormunen, R.; Vered, M. Human Saliva-Derived Exosomes. J. Histochem. Cytochem. 2015, 63, 181–189. [Google Scholar] [CrossRef]
  21. Parlatan, U.; Ozen, M.O.; Kecoglu, I.; Koyuncu, B.; Torun, H.; Khalafkhany, D.; Loc, I.; Ogut, M.G.; Inci, F.; Akin, D.; et al. Label-Free Identification of Exosomes using Raman Spectroscopy and Machine Learning. Small 2023, 19, 2205519. [Google Scholar] [CrossRef]
  22. Bonifacio, A. Label free SERS for clinical applications. In Principles and Clinical Applications of SERS; Wang, Y., Ed.; Elsevier: Amsterdam, The Netherlands, 2021; ISBN 9780128231982. [Google Scholar]
  23. Gurian, E.; Di Silvestre, A.; Mitri, E.; Pascut, D.; Tiribelli, C.; Giuffrè, M.; Crocè, L.S.; Sergo, V.; Bonifacio, A. Repeated double cross-validation applied to the PCA-LDA classification of SERS spectra: A case study with serum samples from hepatocellular carcinoma patients. Anal. Bioanal. Chem. 2021, 413, 1303–1312. [Google Scholar] [CrossRef] [PubMed]
Figure 1. Example of utilizing RMSE for determining the number of PCs in the PLS-DA model for discriminating between a group of patients with kidney failure and a group of healthy volunteers in the analysis of human skin spectra (obtained with permission from the authors [17]).
Figure 1. Example of utilizing RMSE for determining the number of PCs in the PLS-DA model for discriminating between a group of patients with kidney failure and a group of healthy volunteers in the analysis of human skin spectra (obtained with permission from the authors [17]).
Ijms 25 13030 g001
Figure 2. SERS spectrum of opiorphin, uric acid, hypoxanthine, and salivary control probes using the excitation wavelength of 785 nm (from Figure S5 in the commented paper [13]).
Figure 2. SERS spectrum of opiorphin, uric acid, hypoxanthine, and salivary control probes using the excitation wavelength of 785 nm (from Figure S5 in the commented paper [13]).
Ijms 25 13030 g002
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Bratchenko, I.; Bratchenko, L. Comment on Borșa et al. Developing New Diagnostic Tools Based on SERS Analysis of Filtered Salivary Samples for Oral Cancer Detection. Int. J. Mol. Sci. 2023, 24, 12125. Int. J. Mol. Sci. 2024, 25, 13030. https://doi.org/10.3390/ijms252313030

AMA Style

Bratchenko I, Bratchenko L. Comment on Borșa et al. Developing New Diagnostic Tools Based on SERS Analysis of Filtered Salivary Samples for Oral Cancer Detection. Int. J. Mol. Sci. 2023, 24, 12125. International Journal of Molecular Sciences. 2024; 25(23):13030. https://doi.org/10.3390/ijms252313030

Chicago/Turabian Style

Bratchenko, Ivan, and Lyudmila Bratchenko. 2024. "Comment on Borșa et al. Developing New Diagnostic Tools Based on SERS Analysis of Filtered Salivary Samples for Oral Cancer Detection. Int. J. Mol. Sci. 2023, 24, 12125" International Journal of Molecular Sciences 25, no. 23: 13030. https://doi.org/10.3390/ijms252313030

APA Style

Bratchenko, I., & Bratchenko, L. (2024). Comment on Borșa et al. Developing New Diagnostic Tools Based on SERS Analysis of Filtered Salivary Samples for Oral Cancer Detection. Int. J. Mol. Sci. 2023, 24, 12125. International Journal of Molecular Sciences, 25(23), 13030. https://doi.org/10.3390/ijms252313030

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop