Open Access
Add to BibSonomy
Abstract
Early detection is critical for improving the survival rate and quality of life of oral cancer patients; unfortunately, dysplastic and early-stage cancerous oral lesions are often difficult to distinguish from oral benign lesions during standard clinical oral examination. Therefore, there is a critical need for novel clinical technologies that would enable reliable oral cancer screening. The autofluorescence properties of the oral epithelial tissue provide quantitative information about morphological, biochemical, and metabolic tissue and cellular alterations accompanying carcinogenesis. This study aimed to identify novel biochemical and metabolic autofluorescence biomarkers of oral dysplasia and cancer that could be clinically imaged using novel multispectral autofluorescence lifetime imaging (maFLIM) endoscopy technologies. In vivo maFLIM clinical endoscopic images of benign, precancerous, and cancerous lesions from 67 patients were acquired using a novel maFLIM endoscope. Widefield maFLIM feature maps were generated, and statistical analyses were applied to identify maFLIM features providing contrast between dysplastic/cancerous vs. benign oral lesions. A total of 14 spectral and time-resolved maFLIM features were found to provide contrast between dysplastic/cancerous vs. benign oral lesions, representing novel biochemical and metabolic autofluorescence biomarkers of oral epithelial dysplasia and cancer. To the best of our knowledge, this is the first demonstration of clinical widefield maFLIM endoscopic imaging of novel biochemical and metabolic autofluorescence biomarkers of oral dysplasia and cancer, supporting the potential of maFLIM endoscopy for early detection of oral cancer.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
It is estimated that 11,230 people will die from oral cancer in 2022 and around 54,000 new cases will be diagnosed in the United States during the same year [1]. Patients who are diagnosed at advanced stages commonly require complex and highly invasive surgery and have a five-year survival rate of only 40%, while patients who are identified at early stages usually require minor surgery and have an 85% chance of survival [1]. Unfortunately, benign oral lesions are often difficult to distinguish from dysplastic lesions and oral squamous cell carcinoma (SCC) [2,3]. As a result, early-stage oral cancer can only be identified in three out ten patients [1]. The current gold standard for the diagnosis of SCC and dysplasia is conventional oral examination followed by tissue biopsy and histopathological analysis of clinically suspicious oral lesions [4]. This procedure, however, has several limitations, including underdiagnosis or misdiagnosis resulting from sampling bias, lengthy time to diagnosis, subjective grading of dysplastic lesions, and patient morbidity and discomfort due to invasive biopsy tissue surgical removal [4]. Therefore, fast and accurate tools for oral cancer and precancer screening are urgently needed to improve the survival rate and quality of life of these patients, while reducing the number of unnecessary tissue biopsies of benign oral lesions.
Several commercially available diagnostic adjuncts have been developed to assist with the clinical evaluation of potentially malignant and premalignant oral lesions, including Toluidine blue [5], brush cytology [6], acetowhitening with chemiluminescence (ViziLite) [7], and autofluorescence imaging (VELscope, Identafi, and OralID) [8–10]. Nevertheless, these diagnostic adjuncts have low specificity and are not generally recommended for the assessment of clinically suspicious oral lesions [4,11]. Autofluorescence-based optical imaging technologies recently emerged as novel non-invasive diagnostic adjuncts of oral cancer. These tools have been widely explored for imaging and quantifying the autofluorescence of the endogenous fluorophores reduced-form nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD) in the oral epithelium, and collagen in the underlying lamina propria or connective tissue. Oral carcinogenesis process has been shown to cause changes in the levels of epithelial NADH and FAD [12–14], and collagen density in the connective tissue [15], which can modulate the autofluorescence properties of oral epithelial tissue. In addition, the optical redox ratio, an autofluorescence biomarker of cell metabolism typically defined as the ratio of fluorescence intensity of NADH to FAD, decreases as oral cancer develops [16,17].
Several studies have explored the use of autofluorescence spectroscopy (AFS), time-resolved fluorescence spectroscopy (TRFS), confocal fluorescence microscopy (CFM), and fluorescence lifetime imaging microscopy (FLIM) to identify optical biomarkers of oral SCC and dysplasia. In an in vivo study in 38 patients, Wang et al. used TRFS at the 633 nm emission wavelength under 410 nm excitation to differentiate dysplastic vs. benign lesions with a sensitivity of 93% and specificity of 75% [18]. In an ex vivo study using CFM in human oral biopsies, Gillenwater et al. reported increased epithelial and decreased connective tissue autofluorescence in dysplasia relative to benign inflammation upon ultraviolet excitation at 351 nm and 364 nm, collected from 380 to 500 nm [19]. Krishnakumar et al. performed ex vivo AFS in hamsters and reported decreased autofluorescence intensity at the 385 nm emission band upon 320 nm excitation, and decreased redox ratio (450 and 520 nm bands) in well-differentiated squamous cell carcinoma (WDSCC) relative to benign [20]. Our group previously performed in vivo multispectral FLIM in hamsters using 355 nm excitation and reported lower intensity at the 390 ± 20 nm band, shorter lifetimes at both the 390 ± 20 nm and 450 ± 20 nm bands, and higher intensity at the >500 nm band in high-grade dysplastic/carcinoma vs. benign oral lesions [21]. Even though these studies have provided promising results, none of these imaging tools have been translated yet to the clinic for aiding in the discrimination of cancerous/precancerous vs. benign oral lesions.
In this study, we used a multispectral autofluorescence imaging (maFLIM) endoscopy system [22] to image the human oral mucosa in vivo. We demonstrate the capabilities of endogenous widefield maFLIM endoscopy to clinically assess a plurality of metabolic and biochemical autofluorescence biomarkers of oral epithelial dysplasia and cancer (SCC). The outcomes of this study provide the basis for the development of maFLIM based clinical tools for the differentiation of precancerous/cancerous vs. benign oral lesions in a clinical setting.
2. Materials and methods
The workflow and corresponding methodology of this study are summarized in Fig. 1, and a detailed description is presented in the following sub-sections.
Fig. 1. Schematic summarizing the methodology of this study. 1) Clinical maFLIM images of both the oral lesion and a healthy-appearing tissue area in the contralateral side were acquired from the patient. 2) The raw maFLIM data was preprocessed to improve its quality. 3) maFLIM features were computed for every pixel and spectral channel. 4) Statistical analyses were performed on the image-pixel median value of each maFLIM feature to detect difference in their distribution between cancerous/precancerous vs. benign oral lesions.
2.1 maFLIM system and clinical imaging of oral lesions
Endogenous clinical maFLIM endoscopic images of benign, dysplastic, and cancerous oral lesions from 67 patients were acquired in vivo using a maFLIM endoscope system previously developed by Cheng et al. [22] and depicted in Fig. 2. Tissue autofluorescence was excited using a pulsed laser (355 nm, 1 ns pulse width, ∼1µJ/pulse at the tissue, Advanced Optical Technology), and collected at the emission spectral bands of 390 ± 20 nm, 452 ± 22.5 nm, and >500 nm, which were selected to preferentially acquire the tissue autofluorescence at the emission peaks of collagen, NADH, and FAD, respectively. The total energy delivered by the system did not exceed 2.8 mJ, which is an order of magnitude lower than the maximum permissible exposure (MPE = 29.8 mJ) provided by the American National Standards Institute (ANSI) [23]. The maFLIM images were acquired with a circular field-of-view (FOV) of 10 mm in diameter, lateral resolution of ∼100 µm, and acquisition time of less than 3 seconds per image. The time-resolved multispectral autofluorescence signal at each pixel of the maFLIM images were acquired by a multichannel plate photomultiplier tube (MCP-PMT, 25ps transient time spread, R3809U-50, Hamamatsu) followed by a preamplifier (AMP) and a high-speed digitizer (PXIe-5185, National Instruments) with a temporal resolution of 250 ps (sampling rate of 4 GS/s).
Fig. 2. Schematic of the maFLIM endoscope system. L: Lens, DM: Dichroic mirror, F: Filter. The purple light path represents the 355 nm excitation; and the pink, blue, and green light paths represent the autofluorescence at the 390 ± 20 nm, 452 ± 22.5 nm, and >500 nm emission bands, respectively. Note: Modified from “Handheld multispectral fluorescence lifetime imaging system for in vivo applications,” by S. Cheng, R. M. Cuenca, B. Liu, B. H. Malik, J. M. Jabbour, K. C. Maitland, J. Wright, Y.-S. L. Cheng, and J. A. Jo, 2014, Biomedical optics express 5, 921-931 [22].
After clinical examination of the patient’s oral cavity by an experienced head and neck surgeon (M.M., M.A.K., H.A.E), endoscopic maFLIM images were acquired from both the suspicious oral lesion and a clinically healthy appearing area in the corresponding contralateral anatomical side following an image acquisition protocol approved in September 2017, by the Institutional Review Board at Hamad Medical Corporation (Doha, Qatar, study 16332/16). Once the maFLIM images from the patient’s oral cavity were acquired, the tissue biopsy examination procedure was performed following standard clinical protocols. Each imaged oral lesion was annotated based on its tissue biopsy histopathological diagnosis (gold standard), which was blinded to the maFLIM endoscopy imaging data acquisition and processing. The anatomical location and histopathological diagnosis of the 67 imaged oral lesions is summarized in Table 1.
Table 1. Distribution of the 67 imaged oral lesions based in both anatomical location and histopathological diagnosis (MiD: mild dysplasia; MoD: moderate dysplasia; HiD: high-grade dysplasia; SCC: squamous cell carcinoma)
2.2 maFLIM data preprocessing
The maFLIM data is composed of fluorescence intensity temporal decay signals ${{\boldsymbol y}_{\boldsymbol \lambda }}({{\boldsymbol x},{\boldsymbol y},{\boldsymbol \; t}} ),$ measured at each emission spectral band $({\boldsymbol \lambda } )$ and at each spatial location or image pixel $({{\boldsymbol x},{\boldsymbol y}} ).$ Each of the acquired maFLIM images were preprocessed as follows: 1) offset and background subtraction was performed on the temporal signal at each pixel of the maFLIM image, 2) a threshold on the maximum signal amplitude was applied to mask pixels with temporal signal saturation, 3) the temporal signal-to-noise ratio (SNR) at every pixel was increased through the application of a 5 × 5 spatial averaging filter, 4) pixel masking based on SNR was performed using an SNR threshold value of 15 decibels, 5) additional pixels were manually masked from images containing teeth regions characterized by strong autofluorescence emission. Representative preprocessed maFLIM temporal decays from benign, cancerous (SCC), and precancerous (HiD) oral lesions are presented in Fig. 3.
Fig. 3. Representative preprocessed decays from cancerous (SCC), precancerous (HiD), and benign oral lesions for each emission spectral band.
2.3 Spectral feature computation
After all the acquired maFLIM images were preprocessed, maFLIM-derived features were computed per pixel as described here. The multispectral absolute fluorescence intensity ${{\boldsymbol I}_{\boldsymbol \lambda }}({{\boldsymbol x},{\boldsymbol y}} )$ was computed by numerically integrating the fluorescence intensity temporal decay signal as shown in Eq. (1).
Each of the multispectral absolute fluorescence intensities ${{\boldsymbol I}_{\boldsymbol \lambda }}({{\boldsymbol x},{\boldsymbol y}} )$ were normalized (${{\boldsymbol I}_{{\boldsymbol \lambda },{\boldsymbol n}}}({{\boldsymbol x},{\boldsymbol y}} )$) using Eq. (2). .
From the multispectral absolute fluorescence intensities, three spectral ratios were also computed at each spatial location to quantify the relative autofluorescence intensities between individual spectral channels: ${{\boldsymbol I}_{\mathbf{390}}}({{\boldsymbol x},{\boldsymbol y}} )$/${{\boldsymbol I}_{\mathbf{452}}}({{\boldsymbol x},{\boldsymbol y}} )$, ${{\boldsymbol I}_{\mathbf{390}}}({{\boldsymbol x},{\boldsymbol y}} )$/${{\boldsymbol I}_{\mathbf{500}}}({{\boldsymbol x},{\boldsymbol y}} )$, and ${{\boldsymbol I}_{\mathbf{452}}}({{\boldsymbol x},{\boldsymbol y}} )$/${{\boldsymbol I}_{500}}({{\boldsymbol x},{\boldsymbol y}} )$. The latter represents an estimation of the tissue metabolic redox ratio originally reported by Chance et al. [17].
2.4 Time-resolved feature computation
The fluorescence decay ${{\boldsymbol y}_{\boldsymbol \lambda }}({{\boldsymbol x},{\boldsymbol y},{\boldsymbol t}} )$ measured at every pixel $({{\boldsymbol x},{\boldsymbol y}} )$ can be modeled as the convolution of the fluorescence impulse response (FIR) ${{\boldsymbol h}_{\boldsymbol \lambda }}({{\boldsymbol x},{\boldsymbol y},{\boldsymbol t}} )$ of the sample and the measured instrument response function (IRF) ${{\boldsymbol u}_{\boldsymbol \lambda }}({\boldsymbol t} )$ as shown in Eq. (3) [24].
The sample FIR ${{\boldsymbol h}_{\boldsymbol \lambda }}({{\boldsymbol x},{\boldsymbol y},{\boldsymbol t}} )$ can be estimated by temporally deconvolving the IRF ${{\boldsymbol u}_{\boldsymbol \lambda }}({\boldsymbol t} )$ from the measured fluorescence decay ${{\boldsymbol y}_{\boldsymbol \lambda }}({{\boldsymbol x},{\boldsymbol y},{\boldsymbol t}} )$. Temporal deconvolution was performed using a nonlinear least squares iterative reconvolution algorithm [24], in which the FIR was modeled as the bi-exponential decay presented in Eq. (4).
In Eq. (4), ${{\boldsymbol \tau }_{{\boldsymbol {fast}},{\boldsymbol \lambda }}}({{\boldsymbol x},{\boldsymbol y}} )$ and ${{\boldsymbol \tau }_{{\boldsymbol {slow}},{\boldsymbol \lambda }}}({{\boldsymbol x},{\boldsymbol y}} )$ represent the lifetime of the fast and slow decay components, respectively, while ${{\boldsymbol \alpha }_{{\boldsymbol {fast}},{\boldsymbol \lambda }}}({{\boldsymbol x},{\boldsymbol y}} )$ and ${{\boldsymbol \alpha }_{{\boldsymbol {slow}},{\boldsymbol \lambda }}}({{\boldsymbol x},{\boldsymbol y}} )$ represent their corresponding relative contributions, which are complementary to each other (${{\boldsymbol \alpha }_{{\boldsymbol {fast}}}} + {{\boldsymbol \alpha }_{{\boldsymbol {slow}}}} = 1$). The optimal number of exponential components was selected by analyzing the model-fitting mean squared error (MSE) as a function of the number of exponential components. In this study, two exponential components were selected, since the addition of a third component did not reduce the MSE. After temporal deconvolution was performed, the average fluorescence lifetime (${{\boldsymbol \tau }_{{\boldsymbol {avg}},{\boldsymbol \lambda }}}({{\boldsymbol x},{\boldsymbol y}} )$) for each pixel and emission spectral band were estimated from the FIR ${{\boldsymbol h}_{\boldsymbol \lambda }}({{\boldsymbol x},{\boldsymbol y},{\boldsymbol t}} )$ using Eq. (5) [24].
2.5 Computation of relative values
Finally, relative values ${\mathbf{\Delta}}{\boldsymbol f}({{\boldsymbol x},{\boldsymbol y}} )$ for each maFLIM feature were computed as follows. First, maFLIM feature maps were generated for both the lesion and paired healthy tissue images. Second, the difference between each pixel value in the lesion feature map ${\boldsymbol f}({{\boldsymbol x},{\boldsymbol y}} )$ and the median value of the corresponding healthy feature map ${{\boldsymbol \mu }_{{\boldsymbol f},\,{\boldsymbol {Healthy}}}}$ was computed for each of the maFLIM features previously described as shown in Eq. (6). The computed relative values are thus the result of normalizing the lesion image feature distribution with respect to the median (${{\boldsymbol \mu }_{{\boldsymbol f},\,{\boldsymbol {Healthy}}}}$) of the healthy image feature distribution.
In summary, a total of 36 maFLIM-derived autofluorescence features were computed per pixel, as summarized in Table 2.
Table 2. Summary of maFLIM-derived features computed per pixel.
2.6 Statistical analysis of maFLIM features
As summarized in Table 1, the 67 imaged oral lesions corresponded to 33 benign, 5 precancerous (1 MiD, 1 MoD, and 3 HiD), and 29 cancerous (SCC) lesions. To identify statistically significant differences in the distribution of the extracted maFLIM features (Table 2) from precancerous/cancerous vs. benign oral lesions, the following statistical analysis was performed. The maFLIM features computed at the pixel level enabled the generation of 36 feature maps. For each maFLIM feature map, the median value from all pixels was computed. Hence, each maFLIM image was represented by a feature vector composed of the (image pixel) median values of the 36 maFLIM features. A non-parametric two-tailed Wilcoxon rank-sum test comparing the median value distributions of precancerous/cancerous (n = 34) vs. benign (n = 33) lesions was performed on each of the 36 maFLIM features. Finally, the most relevant features for the discrimination of precancerous/cancerous from benign oral lesions were identified based on the p-value (P) resulting from the Wilcoxon rank-sum test. All tests were performed with a type-1 error probability of P < 0.05.
3. Results
The statistical analysis performed on the maFLIM feature median values revealed that the distributions of 14 maFLIM features were statistically significantly different (P < 0.05) in cancerous/precancerous (n = 34) vs. benign (n = 33) oral lesions. Violin plots of these feature median value distributions are presented in Fig. 4 (absolute feature values) and Fig. 5 (relative feature values). The dashed line shown on each plot indicates the trend of the maFLIM feature distribution in cancerous/precancerous relative to benign lesions.
Fig. 4. Violin plots of absolute maFLIM feature median value distributions of benign (n = 33) and cancerous/precancerous (n = 34) oral lesions. P-values resulting from two-tailed Wilcoxon rank-sum tests are shown on top of each plot.
Fig. 5. Violin plots of relative maFLIM feature median value distributions of benign (n = 33) and cancerous/precancerous (n = 34) oral lesions. P-values resulting from two-tailed Wilcoxon rank-sum tests are shown on top of each plot.
Representative maFLIM feature maps comparing cancerous (SCC) and benign tongue lesions from two patients, and precancerous (HiD) and benign buccal mucosal lesions from two other patients are shown in Fig. 6 and Fig. 7, respectively. In both representative cases, the pixel distributions of each maFLIM feature map were consistent with the observed trends in the distribution of median values of each of the 14 statistically different maFLIM features from precancerous/cancerous versus benign oral lesions (Fig. 4 and Fig. 5).
Fig. 6. Representative absolute maFLIM feature maps and corresponding pixel value histograms comparing cancerous (SCC) vs. benign tongue lesions (left panels), and high-grade dysplastic (HiD) vs. benign buccal mucosal lesions (right panels).
Fig. 7. Representative relative maFLIM feature maps and corresponding pixel value histograms comparing cancerous (SCC) vs. benign tongue lesions (left panels), and high-grade dysplastic (HiD) vs. benign buccal mucosal lesions (right panels). Δ: Relative values.
4. Discussion
In this study, in vivo maFLIM endoscopic imaging of benign, precancerous, and cancerous oral lesions from 67 patients was performed in a clinical setting. The acquired maFLIM endoscopic images were processed to generate maps of autofluorescence spectral and time-resolved features of benign and precancerous/cancerous oral lesions. The statistical analysis applied to each of the computed maFLIM features enabled the identification of 14 biomarkers that displayed contrast between precancerous/cancerous vs. benign oral lesions Fig. 4 and Fig. 5). We have previously demonstrated that a plurality of biochemical and metabolic autofluorescence biomarkers of oral cancer and dysplasia can be clinically imaged using widefield maFLIM technologies [25], and that several of these biomarkers were relevant to automatically discriminate cancerous and precancerous oral lesions from healthy oral epithelial tissue [26]. Hence, further studies will determine the potentials of maFLIM endoscopy for automated image-guided discrimination of benign vs. precancerous and cancerous oral lesions. The trends of the most relevant autofluorescence spectral and time-resolved maFLIM features observed in this study are summarized in Table 3 and compared with previously reported observations [19–21].
Table 3. Summary of trends in statistically different maFLIM-derived features in cancer/precancer vs. benign oral lesions. FLIM: fluorescence lifetime imaging; AFS: autofluorescence spectroscopy; CFM: confocal fluorescence microscopy.
Decreased normalized autofluorescence intensity at the 390 ± 20 nm spectral band (${I_{\mathbf{390},n}}$) was observed in cancerous/precancerous vs. benign oral lesions. Collagen in lamina propria is the main contributor to the oral tissue autofluorescence acquired at this spectral band upon 355 nm excitation; thus, this observation is in agreement with previously reported findings [19–21] and likely reflects a breakdown of collagen crosslinks [15,27] and increased epithelial thickness and tissue optical scattering accompanying carcinogenesis [28]. We also report what is to the best of our knowledge the first observation of a lower relative normalized autofluorescence intensity at this band ($\Delta {I_{\mathbf{390},n}}$), which further indicates that collagen autofluorescence signal with respect to healthy oral tissue from the same patient decreases more in cancerous/precancerous than in benign oral lesions.
Shorter average lifetimes (${\tau _{avg,390}}$, $\Delta {\tau _{avg,390}}$), fast (${\tau _{fast,390}}$) and slow (${\tau _{slow,390}}$) component lifetimes at the 390 ± 20 nm spectral band were observed for the first time in cancerous/precancerous vs. benign oral lesions. Due to the spectral overlap in the autofluorescence signals of collagen and NADH at this band, the observed shorter autofluorescence lifetimes might be the result of a decrease in the slower-decaying collagen signal, leading to a faster-decaying (shorter) autofluorescence temporal response mainly associated to NADH, which increases in precancerous and cancerous lesions due to increased use of glycolysis as a complementary metabolic pathway to oxidative phosphorylation [29].
Increased absolute and relative normalized autofluorescence intensities (${I_{\mathbf{500},n}}$, $\Delta {I_{\mathbf{500},n}}$) measured at the >500 nm band were observed in cancerous/precancerous vs. benign oral lesions. The main contributor to the oral tissue autofluorescence acquired at this spectral band upon 355 nm excitation is FAD in the mitochondria. Hence, the higher ${I_{\mathbf{500},n}}$ is likely associated to increased metabolic rate of cancerous cells due to oxidative phosphorylation, which requires the oxidation of FADH2 into FAD [30], resulting in higher levels of mitochondrial FAD [31], and is consistent with previous observations [19,21]. The higher $\Delta {I_{\mathbf{500},n}}$, which has not been previously reported, further indicates that the FAD autofluorescence signal relative to healthy oral tissue from the same patient increases more in cancerous/precancerous than in benign oral lesions.
Decreased absolute and relative optical redox ratio (${I_{{452}}}$/${I_{{500}}}$, $\Delta {I_{{452}}}$/${I_{{500}}}$) were observed in cancerous/precancerous with respect to benign oral lesions. The process of oxidative phosphorylation requires the oxidation of both NADH and FADH2 molecules, resulting in lower NADH/FAD ratio [31]. Therefore, the observed decrease in ${I_{{452}}}$/${I_{{500}}}$, consistent with previous findings [20], reflects an increase in cellular metabolic activity, which is characteristic of carcinogenesis [16]. The lower $\Delta {I_{{452}}}$/${I_{{500}}}$, which has not been previously observed, further indicates that the redox ratio relative to healthy oral tissue from the same patient decreases more in precancerous/cancerous compared to benign oral lesions.
Finally, our findings also displayed four novel autofluorescence biomarkers of oral cancer and dysplasia derived from the autofluorescence intensity ratios. Decreased absolute (${I_{390}}$/${I_{{452}}}$) and relative ($\Delta {I_{390}}$/${I_{{452}}}$) intensity ratios were observed in cancerous/precancerous vs. benign oral lesions, likely reflecting a higher loss of collagen autofluorescence signal measured at the 390 ± 20 nm band relative to NADH signal measured at the 452 ± 22.5 nm band. Similarly, the intensity ratios ${I_{390}}$/${I_{{500}}}$ and $\Delta {I_{390}}$/${I_{{500}}}$ decreased in cancerous/precancerous lesions, reflecting lower collagen signal measured at the 390 ± 20 nm band relative to higher FAD signal measured at the >500 nm band.
4.1 Study limitations
Even though the maFLIM-derived biochemical and metabolic autofluorescence biomarkers of oral cancer and dysplasia presented in this study demonstrate the potential of maFLIM endoscopy for early detection of oral cancer, some important limitations were identified. The relatively small sample sizes of dysplastic lesions included in this study (Table 1) prevented the application of statistical analyses comparing them to benign lesions independently. Ongoing research efforts to overcome this limitation include the generation of a larger and more balanced maFLIM endoscopic image database of dysplastic and cancerous oral lesions. The maFLIM endoscope used in this study did not enable specific interrogation of collagen, NADH, and FAD autofluorescence due to the use of a single excitation wavelength and broad emission spectral bands. Because of the spectral overlap of collagen and NADH at both the 390 ± 20 nm and 452 ± 22.5 nm bands, and of FAD and NADH at the >500 nm band, the autofluorescence acquired at these spectral bands could not be associated to a single fluorophore. Nevertheless, the analysis of the fluorescence emission as a biexponential decay can help separating different fluorophore signals from a single band. For example, since collagen has a longer fluorescence lifetime than NADH, the long and short time-constants of biexponential fluorescence decays measured at the 390 ± 20 nm and 452 ± 22.5 nm bands can be attributed to collagen and NADH, respectively. Similar interpretation could be applied to the longer lifetime of FAD relative to the NADH fluorescence lifetime expected at the >500 nm band. In addition, the use of multiple excitation wavelengths can also help separating fluorescence signals with overlapping emission spectra. Our maFLIM imaging system is currently undergoing several improvements, including the implementation of dual wavelength excitation (375 nm/445 nm) with narrower emission bands to enable more specific interrogation of collagen, NADH, and FAD autofluorescence [32], and optimized fluorescence emission detection for maFLIM imaging with cellular spatial resolution and improved sensitivity [33].
5. Conclusion
We performed label-free clinical biochemical and metabolic imaging of benign, precancerous, and cancerous oral lesions by means of widefield maFLIM endoscopy. Several spectral and time-resolved maFLIM features were found to provide contrast between dysplastic/cancerous vs. benign oral lesions. To the best of our knowledge, this is the first demonstration of clinical widefield maFLIM endoscopic imaging of novel biochemical and metabolic autofluorescence biomarkers of benign, precancerous and cancerous oral lesions. Future studies will assess the capabilities of maFLIM endoscopy as a label-free imaging tool for in situ, objective, and accurate discrimination of precancerous and cancerous from benign oral lesions during conventional oral examination.
Funding
Qatar National Research Fund (NPRP8-1606-3-322); Cancer Prevention and Research Institute of Texas (RP180588); National Institutes of Health (R01CA218739, R21CA267236).
Acknowledgments
The statements made herein are solely the responsibility of the authors. Research reported in this publication was also supported in part by the Oklahoma Tobacco Settlement Endowment Trust awarded to the University of Oklahoma, Stephenson Cancer Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Oklahoma Tobacco Settlement Endowment Trust. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board at Hamad Medical Corporation (study 16332/16, approved on 09/13/2017).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. A. C. Society, “American Cancer Society: Cancer Facts and Figures 2022,” (American Cancer SocietyAtlanta, GA, 2022).
2. J. B. Epstein, P. Güneri, H. Boyacioglu, and E. Abt, “The limitations of the clinical oral examination in detecting dysplastic oral lesions and oral squamous cell carcinoma,” The Journal of the American Dental Association 143(12), 1332–1342 (2012). [CrossRef]
3. J. L. Cleveland and V. A. Robison, “Clinical oral examinations may not be predictive of dysplasia or oral squamous cell carcinoma,” J Evid Based Dent Pract 13(4), 151–154 (2013). [CrossRef]
4. A. M. Gillenwater, B. Nadarajah Vigneswaran, and D. DrMedDent, “Noninvasive diagnostic adjuncts for the evaluation of potentially premalignant oral epithelial lesions: current limitations and future directions,” (2018).
5. D. H. Kim, E. A. Song, S. W. Kim, and S. H. Hwang, “Efficacy of toluidine blue in the diagnosis and screening of oral cancer and pre-cancer: A systematic review and meta-analysis,” Clinical Otolaryngology 46(1), 23–30 (2021). [CrossRef]
6. A. H. Alsarraf, O. Kujan, and C. S. Farah, “The utility of oral brush cytology in the early detection of oral cancer and oral potentially malignant disorders: a systematic review,” Journal of Oral Pathology & Medicine 47(2), 104–116 (2018). [CrossRef]
7. V. Khanna, R. Patil, A. Tripathi, and V. Singh, “Comparative Study to Determine the Efficacy of Chemi-Luminescent Method (Vizilite) and Toludine Blue Staining as a Screening Test in Oral Pre-Malignant Lesions,” (2019).
8. M. Cicciù, G. Cervino, L. Fiorillo, C. D’Amico, G. Oteri, G. Troiano, K. Zhurakivska, L. Lo Muzio, A. S. Herford, and S. Crimi, “Early diagnosis on oral and potentially oral malignant lesions: a systematic review on the VELscope® fluorescence method,” Dentistry journal 7(3), 93 (2019). [CrossRef]
9. I. F. P. Lima, L. M. Brand, J. A. P. de Figueiredo, L. Steier, and M. L. Lamers, “Use of autofluorescence and fluorescent probes as a potential diagnostic tool for oral cancer: A systematic review,” Photodiagn. Photodyn. Ther. 33, 102073 (2021). [CrossRef]
10. A. E. Sadakah, K. Saad, E. Megahed, and O. El-Bahgy, “The Role of OralID Guided Biopsy in Diagnosis of Oral Epithelial Dysplastic Lesions,” Egyptian Journal of Oral and Maxillofacial Surgery 12(1), 26–34 (2021). [CrossRef]
11. M. W. Lingen, E. Abt, N. Agrawal, A. K. Chaturvedi, E. Cohen, G. D’Souza, J. Gurenlian, J. R. Kalmar, A. R. Kerr, and P. M. Lambert, “Evidence-based clinical practice guideline for the evaluation of potentially malignant disorders in the oral cavity: a report of the American Dental Association,” The Journal of the American Dental Association 148(10), 712–727.e10 (2017). [CrossRef]
12. M. G. Müller, T. A. Valdez, I. Georgakoudi, V. Backman, C. Fuentes, S. Kabani, N. Laver, Z. Wang, C. W. Boone, and R. R. Dasari, “Spectroscopic detection and evaluation of morphologic and biochemical changes in early human oral carcinoma,” Cancer: Interdisciplinary International Journal of the American Cancer Society 97, 1681–1692 (2003). [CrossRef]
13. A. T. Shah, M. D. Beckler, A. J. Walsh, W. P. Jones, P. R. Pohlmann, and M. C. Skala, “Optical metabolic imaging of treatment response in human head and neck squamous cell carcinoma,” PloS one 9(3), e90746 (2014). [CrossRef]
14. J. A. Jo, B. E. Applegate, J. Park, S. Shrestha, P. Pande, I. B. Gimenez-Conti, and J. L. Brandon, “In vivo simultaneous morphological and biochemical optical imaging of oral epithelial cancer,” IEEE Trans. Biomed. Eng. 57(10), 2596–2599 (2010). [CrossRef]
15. I. Pavlova, K. Sokolov, R. Drezek, A. Malpica, M. Follen, and R. Richards-Kortum, “Microanatomical and Biochemical Origins of Normal and Precancerous Cervical Autofluorescence Using Laser-scanning Fluorescence Confocal Microscopy,” Photochem. Photobiol. 77(5), 550–555 (2003). [CrossRef]
16. Z. Zhang, D. Blessington, H. Li, T. M. Busch, J. D. Glickson, Q. Luo, B. Chance, and G. Zheng, “Redox ratio of mitochondria as an indicator for the response of photodynamic therapy,” J. Biomed. Opt. 9(4), 772–779 (2004). [CrossRef]
17. B. Chance, B. Schoener, R. Oshino, F. Itshak, and Y. Nakase, “Oxidation-reduction ratio studies of mitochondria in freeze-trapped samples. NADH and flavoprotein fluorescence signals,” J. Biol. Chem. 254(11), 4764–4771 (1979). [CrossRef]
18. H. M. Chen, C. P. Chiang, C. You, T. C. Hsiao, and C. Y. Wang, “Time-resolved autofluorescence spectroscopy for classifying normal and premalignant oral tissues,” Lasers Surg Med 37(1), 37–45 (2005). [CrossRef]
19. I. Pavlova, M. Williams, A. El-Naggar, R. Richards-Kortum, and A. Gillenwater, “Understanding the biological basis of autofluorescence imaging for oral cancer detection: high-resolution fluorescence microscopy in viable tissue,” Clin Cancer Res 14(8), 2396–2404 (2008). [CrossRef]
20. R. Sethupathi, K. Gurushankar, and N. Krishnakumar, “Optical redox ratio differentiates early tissue transformations in DMBA-induced hamster oral carcinogenesis based on autofluorescence spectroscopy coupled with multivariate analysis,” Laser Phys. 26(11), 116202 (2016). [CrossRef]
21. B. H. Malik, J. Lee, S. Cheng, R. Cuenca, J. M. Jabbour, Y. S. L. Cheng, J. M. Wright, B. Ahmed, K. C. Maitland, and J. A. Jo, “Objective detection of oral carcinoma with multispectral fluorescence lifetime imaging in vivo,” Photochem. Photobiol. 92(5), 694–701 (2016). [CrossRef]
22. S. Cheng, R. M. Cuenca, B. Liu, B. H. Malik, J. M. Jabbour, K. C. Maitland, J. Wright, Y.-S. L. Cheng, and J. A. Jo, “Handheld multispectral fluorescence lifetime imaging system for in vivo applications,” Biomed. Opt. Express 5(3), 921–931 (2014). [CrossRef]
23. “Safe Use of Lasers, ANSI Z136.1–2007,” L. I. O. America, ed. (American National Standards Institute, 2007).
24. J. R. Lakowicz, Principles of Fluorescence Spectroscopy (Springer US, 2007).
25. E. Duran-Sierra, S. Cheng, R. Cuenca-Martinez, B. Malik, K. C. Maitland, Y. L. Cheng, J. Wright, B. Ahmed, J. Ji, and M. Martinez, “Clinical label-free biochemical and metabolic fluorescence lifetime endoscopic imaging of precancerous and cancerous oral lesions,” Oral Oncology 105, 104635 (2020). [CrossRef]
26. E. Duran-Sierra, S. Cheng, R. Cuenca, B. Ahmed, J. Ji, V. V. Yakovlev, M. Martinez, M. Al-Khalil, H. Al-Enazi, and Y.-S. L. Cheng, “Machine-Learning Assisted Discrimination of Precancerous and Cancerous from Healthy Oral Tissue Based on Multispectral Autofluorescence Lifetime Imaging Endoscopy,” Cancers 13(19), 4751 (2021). [CrossRef]
27. P. Lu, V. M. Weaver, and Z. Werb, “The extracellular matrix: a dynamic niche in cancer progression,” J Cell Biol 196(4), 395–406 (2012). [CrossRef]
28. R. A. Drezek, K. V. Sokolov, U. Utzinger, I. Boiko, A. Malpica, M. Follen, and R. R. Richards-Kortum, “Understanding the contributions of NADH and collagen to cervical tissue fluorescence spectra: modeling, measurements, and implications,” J. Biomed. Opt. 6(4), 385–397 (2001). [CrossRef]
29. M. G. Vander Heiden, L. C. Cantley, and C. B. Thompson, “Understanding the Warburg effect: the metabolic requirements of cell proliferation,” science 324(5930), 1029–1033 (2009). [CrossRef]
30. O. A. Deshpande and S. S. Mohiuddin, “Biochemistry, Oxidative Phosphorylation,” StatPearls [Internet] (2021).
31. O. I. Kolenc and K. P. Quinn, “Evaluating cell metabolism through autofluorescence imaging of NAD (P) H and FAD,” Antioxid. Redox Signaling 30(6), 875–889 (2019). [CrossRef]
32. G. P. Tortorelli, R. C. Martinez, M. J. Serafino, W. Kim, B. Applegate, and J. A. Jo, “Dual-wavelength excitation multispectral autofluorescence lifetime imaging handheld probe for label-free metabolic clinical imaging of the oral mucosa,” in Imaging, Therapeutics, and Advanced Technology in Head and Neck Surgery and Otolaryngology 2022, (SPIE, 2022), PC1193508.
33. E. A. Shirshin, M. V. Shirmanova, A. V. Gayer, M. M. Lukina, E. E. Nikonova, B. P. Yakimov, G. S. Budylin, V. V. Dudenkova, N. I. Ignatova, and D. V. Komarov, “Label-free sensing of cells with fluorescence lifetime imaging: the quest for metabolic heterogeneity,” bioRxiv (2022).
