Efficacy of spectroscopy in oral cancer detection: systematic review

Introduction

Timely identification of oral cancer is of utmost importance, given that the health status of an individual may be beyond repair once the clinical manifestations become apparent. However, the present screening and diagnostic methods with imaging techniques usually detect cancer in advanced stages with the tumor significantly grown and visible, and the existing screening tests lack the necessary sensitivity and specificity to detect early stages of cancer.
Histopathology continues to serve as the primary method for diagnosing solid cancers. Notwithstanding, histopathology report and clinical presentation always present undeniable association. In order to assist patho­logists in formulating a clinically significant assessment and diagnosis, clinicians should deliver a comprehensive clinical history and detailed physical examination findings when submitting a sample. In case of any inconsistency arising between the clinical and patholo­gical findings, a secondary histopathological review may be requested, subsequent to a discussion with the pathologist. All of these assessments can lead to intra- and inter-observer bias, since these are inherently subjective in nature. Consequently, when subjective judgement is employed, the occurrence of false-negative and false- positive rates becomes a common factor in histopathological assessment.
Utilization of alternative techniques to facilitate the complete process of evaluation, diagnosis, and reporting has been occasionally employed. In this fast-paced world, a dire need for a much simpler and objective- based technique is imperative. Unfortunately, no signi­ficant advancements in the diagnosis based on objective assessment have been achieved. The current review aimed to explore the implementation of spectroscopy in the realm of oral cancer diagnosis. Spectroscopy has a history of being highly specific, and has gathered widespread recognition in the industrial sector for identifying chemical compositions of various substances.
Spectroscopy is a technique, in which the interaction between electro-magnetic radiation and matter causes absorption of light and its splitting into desirable wavelengths based on functional group. This study involved the interpretation of electro-magnetic spectra produced during such interactions. For each functional group present in a substance, there is a specific and unique wavelength.

Material and methods

This study was recorded in the International Prospective Register of Systematic Reviews database (No.: CRD42023405597), and the preferred reporting items for systematic reviews and meta-analyses guidelines were followed.
The review question was: “Is spectroscopy a better tool for oral cancer screening comparing with histopathology and cytopathology diagnosis?”. PICO included the following: population – participants of all ages with a confirmed diagnosis of oral cancer, and no restrictions on gender or ethnicity; intervention – exposure is spectroscopy for early oral cancer detection. The absorption spectrum of a solid, liquid, or gas can be obtained using spectroscopy. It is one of the non-destructive characteri­zation techniques for molecules and materials. It analyses a molecule’s vibrational modes that are sensitive to its chemical bonds, providing a unique “fingerprint” for molecules to be identified. These compounds feature a fingerprint region that allows them to be specific to that chemical bond. The spectroscopy technique is a promising diagnostic tool, and provides molecular-specific methodology for developing a fundamental biochemical understanding of tissue physiology and pathology. It focuses on the differentiation and characterization of cells and tissues by recognizing individual bands or groups, to accurately determine molecular configurations, types of bonding, functional groups, and inter-molecular interactions within the specimen. Therefore, rather than traditional approaches, it can be employed as an early diagnostic tool for oral cancer detection.
The outcomes of the current review included: the vibration band assessment of normal, dysplastic, and cancer cells as well as the analysis of spectra for determining the principle molecular component of spectral profiles of normal and abnormal cells.
Original studies were investigated, and extensive literature search was performed without language restriction from electronic databases, such as Cochrane Library, PubMed, Scopus, NLM, and Web of Science from the year 2000 onwards, using the following key words: “Oral Cancer”, “Spectroscopy”, “Raman Spectroscopy”, “FT-IR Spectroscopy”, “Biopsy”, “Cytology”, “Diagnosis”.
Inclusion criteria were: 1) studies using spectroscopy for detection of oral cancer with biopsy, cytology, serum, and saliva samples; 2) randomized controlled trials, dia­gnostic studies, and other observational designs, such as cross-sectional, case-control, and cohort studies; 3) studies including diagnostic sensitivity and specificity values, or other relevant parameters, such as true- positive (TP), false-positive (FP), true-negative (TN), and false-negative (FN) values, from which sensitivity and specificity could be derived.
Exclusion criteria were: 1) grey literature, reviews, letters, individual viewpoints, books’ chapters, reports on specific cases, and summaries of conferences; 2) duplicate publications; 3) experimental research both in vivo and in vitro.
All articles were screened by two independent authors, and the following data were obtained: country of origin, author(s), year of publication, types of spectroscopy techniques used in medical and dental field, most effective among spectroscopy techniques, specificity and sensitivity of spectroscopy in diagnosis of solid and liquid samples, and efficacy of spectroscopy use in different types of samples (solid and liquid). The main outcome was the analysis of different spectroscopic techniques used in screening and diagnosis of oral cancer, and their efficacy when used with different sample types. Microsoft Excel was employed to tabulate and process the collected quantitative data.
To assess the risk of bias for each study, quality assessment of diagnostic accuracy studies tool 2 (QUADAS-2) was employed, and analysis was done by computation using Review Manager (RevMan v. 5.3, Nordic Cochrane Center, Cochrane Collaboration, Copenhagen, Denmark). Checklist items for risk bias and applicability concerns were patient selection, index test, reference standard, and flow and timing (Figures 1 and 2).

Results

The searching strategy resulted in 256 articles published in various scientific databases. After removal of duplicates, 230 articles remained. These were further probed by reading the titles and abstract, resulting in 66 articles. Further evaluation was done by reading the full text, and 18 articles were included for the final quantitative synthesis (Figure 3). The methodological characteristics of the study are presented in Table 1. Demographic and clinical details were availed and tabulated. Spectroscopic findings of the selected studies, with their sensitivity and specificity are showed in Table 2. Spectral data of the selected studies for normal tissue, pre-cancerous, and cancerous lesions are presented in Table 3.

Discussion

This research shows that the application of spectroscopy is a promising diagnostic tool for detecting oral diseases. The levels of sensitivity and specificity exhibited an increase in each of the 18 studies. Among various spectroscopic techniques, Raman spectroscopy was found to be the most widely used. It was also determined that biopsy samples were the most frequently used samples. Data shows that the use of tissue samples, especially biopsy samples, can help diagnose cancer. In addition to biopsy specimens, cytology specimens as well as saliva and serum samples were utilized, and demonstrated their effectiveness for detection of oral cancer. A study conducted by Behl et al. [1] revealed notable differences between healthy donors and patients with potentially malignant lesions. Specifically, when examining cytoplasm through spectroscopy, healthy donor samples displayed pronounced protein- (636, 845, 997, 1237, 1361, 1597, and 1645 cm−1) and lipid- (1437 cm−1) derived bands, while patient samples demonstrated prominent nucleic acid (721, 780, and 1180 cm−1) and lipid (1060, 1135, 1300, 1417, and 1745 cm−1) bands. These findings provide valuable insight into the variations between healthy individuals and those with potentially malignant lesions using spectroscopy. Additionally, results of another study indicate higher lipid content in healthy conditions and higher DNA, heme, and protein-related features in severe patholo­gical conditions [2]. According to Jang et al. [3], the appli­cation of Raman spectroscopy to oral cryopreserved freshly-excised tissue samples would yield several advan­tages, including its rapidity, lack of the need for labeling, and inexpensiveness. It has the potential to improve the efficiency of screening procedures for oral cancers and identify the boundary for tumor-free resection margin during surgery. Spectral analysis of normal tissues revealed lipid signatures, specified by ester bands, strong CH2 bend, two sharp features around amide III, and sharp peak around amide I. In contrast, the mean tumor spectrum showed dominating protein features, indicated by broad amide III, broadened CH2, and broad features in the amide I region [4]. Furthermore, significant differences were observed between normal and malignant oral pouch tissues in Raman spectra, specifically in the range between 800 and 1800 cm–1. Normal tissues displayed strong peaks in 858 and 94 cm–1 assigned to the molecular vibrational modes of proteins (mainly due to Try and C-C stretch), while malignant tissues showed higher peaks in the same range, indicating protein dominance [5].
The fingerprint region in biological tissues contains various biomolecules, with normal tissues being lipid- dominated and malignant tissues being protein-dominated. This is demonstrated by higher peaks in tumor samples compared with normal tissues, which differentiate between the two [6]. Specifically, malignant tissues have higher peaks than normal tissues, with proteins dominating in spectral peaks in malignant tissues and lipids dominating in normal tissues [3]. Normal tissues have dominant peaks at 1004, 1156, 1339, 1450, 1523, and 1656 cm–1, while malignant tissues have dominant peaks at 1064, 150, 1168, and 1220 cm–1. Additionally, carotenoid peaks at 1518 cm–1 and 1156 cm–1 were present in both normal and tumor tissues, but tumor tissues had more intense peaks [7].
A comparative study has been accomplished between diagnostic mediums, i.e., human oral tissue and saliva for oral cancer detection using Stokes shift (SS) spectroscopy (SSS) [8]. SS spectra obtained from tissue and saliva consist of major bands of collagen, tryptophan, NADH, and minor bands of flavin adenine dinucleotide (FAD) and porphyrin [7]. When examining oral cancer spectra, higher Tyr, Trp (doublet at 830, 850, and 1552 cm–1), amide III (1270 cm–1), and CH2 deformation (1450 cm–1) were observed, but slightly lower amide I (1660 cm−1) and sharper DNA bases (1342 cm–1) were present. These spectral variations confirm previous findings. Moreover, pre-malignant spectra showed higher Phe (1008 cm–1), lower amide III (1270 cm–1), higher DNA bases (1320 cm–1 and 1342 cm–1), higher CH2 deformation (1450 cm–1), and slightly lower amide I (1660 cm–1) compared with the normal group [7].
Research findings indicate that different sub-sites of the oral mucosa exhibit varying percentages of collagen and elastin. Notably, the buccal tumor sub-site displays more prominent amide I and amide III bands at 1655 cm–1 and 1250 cm–1, respectively, comparing with the gingiva and tongue tumor sub-sites. Furthermore, the protein/lipids bands at 1155 cm–1 and 1523 cm–1 are more intense at the tongue and gingival sub-sites, as opposed to the buccal sub-site [3]. Despite variations in prognosis, metastasis to lymph nodes, aggressiveness, and overall survival rate among the three sub-sites, it is crucial to recognize that these genetic and biological differences serve as the basis for classification of buccal, tongue, and gingiva cancers [3].
According to Francisco et al. [8], excitation under violet light demonstrates higher levels of sensitivity and specificity for discrimination of normal versus carcinoma samples, as compared with excitation at 532 nm. The J48 algorithm of violet excitation produced better analysis results than the 532 nm wavelength. Highly intense Raman bands assigned to β-carotene could be due to resonance Raman spectroscopy and were observed in all sera, with the highest relative intensity in normal samples. The average cancer spectra demonstrated an elevation in both DNA and protein levels [9]. Fluctuations in the intensity of fluorescence in normal and neoplastic oral tissues stem from variations in coenzyme concentrations and oxidation states as well as alterations in tissue characteristics, signifying structural and metabolic changes linked to cancer [10]. The spectral parameters of diagnostic models exhibit patterns that correspond to the progression of disease, including heightened scattering caused by small particles, increased cHb levels indicating vascular dilation and inflammation, and reduced collagen fluorescence [11].
Therefore, Raman spectroscopy has displayed signi­ficant potential in the identification of oral pre-cancers, cancers, and field cancerization in humans as well as mechanical irritation-induced tumors in animals. The early diagnosis of oral cancers, which have a poor survival rate, was extensively studied using Raman spectroscopy. The detection of biochemical changes during the onset of disease [19, 20] is possible with Raman spectroscopy, and it has been explored for the purpose of diagnosis, surgical margin assessment, and prediction of treatment response in oral cancers. Anatomical differences between various oral sub-sites were reported using Raman spectroscopy, and these differences can have an impact on the classification of healthy versus pathological conditions. In vivo Raman spectroscopy demonstrated promising results in the differential dia­gnosis of malignant and potentially malignant lesions of the oral cavity, with high accuracy and sensitivity. Before clinical applications, large-scale validation studies are necessary. All of the discoveries collectively mentioned in the current review confirm the opinion that spectroscopy is indeed the optimal method for oral cancer screening.

Conclusions

It can be assumed that spectroscopy may be deemed as a technique for oral cancer preventive scanning tool, with its potential to substitute histopathology. This precise, exceedingly sensitive, and specific instrument can be employed to identify the initial alterations and serve as a prognostic tool. The collection of spectra for normal tissue, pre-cancerous lesions, and cancerous lesions can be segregated and classified for the utmost accurate assessment, prompt intervention, treatment, and management. A wide variety of options exist for categorizing spectral data sets, such as creating separate data sets for cytology, biopsy specimens, saliva specimens, and serum samples. Literature suggests that data sets can be organized based on the location of the lesion. It is also possible to have separate data sets for leukoplakia, OSMF, and other OPMDS. Therefore, spectroscopy techniques allow for the creation of highly specific and objective data sets. In order to further validate the claim of spectroscopy, it is essential to conduct additional investigations to identify potential confounding variables, including age, systematic disorders, and habits, which might have an impact on the efficacy of spectroscopy as a screening tool. The transition in spectra that occurs from normal to dysplastic oral cancer has been observed to yield positive outcomes in terms of early detection and final diagnosis of the disease. Additionally, the effectiveness of spectroscopy in detecting positivity in resection margins has further validated its utility in the screening and diagnosis of oral cancer.
Given these outcomes, it is imperative that we integrate spectroscopy into our daily diagnostic and treatment routines in a timely manner. In that way, we can ensure the early detection and prompt treatment of oral cancer that is crucial for improving patients’ outcomes and enhancing overall survival rates.

Disclosures

1. Institutional review board statement: SRB/SDC/PhD/OPATH-2314/23/TH-070.
2. Assistance with the article: None.
3. Financial support and sponsorship: None.
4. Conflicts of interest: The authors declare no potential conflicts of interest concerning the research, authorship, and/or publication of this article.

References