1.

Introduction

Early detection of neoplastic changes in the oral cavity and initiation of treatment is important to improve the quality of life of patients and their survival rates. Eighty-five percent of malignancies of the oral cavity are squamous cell carcinomas¹ (SCCs). Despite tremendous advancements made in ablative and reconstructive techniques for treating SCC of the oral cavity, survival rates over the past $30\mathrm{yr}$ are not very significant. Patients usually present themselves for checkup when the cancer of their oral cavity is already in an advanced stage. The $5\text{‐}\mathrm{yr}$ overall survival rates remain around 50% and are even less in an advanced stage of the disease.² The treatment modalities of the oral cancer in an advanced stage is expensive and often disfiguring and debilitating. In addition, there is the additional risk of metastasis, which again requires timely detection and follow up treatment.

Discrete lesions of the oral cavity such as oral leukoplakia (white plaques) and erythroplakia (reddish, velvety mucosal lesions) have the potential for malignant transformation. Many of the currently available screening and detection techniques for oral cancer do not provide adequate sensitivity and specificity for detecting precancerous tissues. Although the most common screening method of staining with agents such as toludine blue and Lugol’s iodine has a sensitivity of around 90%, its specificity is lower and requires experienced clinicians for diagnosis. Sometimes, the whole tissue lining of the oral cavity of patients is at risk or has premalignant changes and it becomes difficult to determine biopsy location for diagnosis. Thus, more sensitive, noninvasive, cost-effective, and real-time screening and diagnostic methods are essential to detect cancer in its early stages.

The optical properties of tissues are important and informative, and the spectroscopic aspects are preeminent in the field of lesion localization and determination. Therefore, diagnostic techniques based on optical spectroscopy are fast, noninvasive, and quantitative and utilize the biochemical and morphological properties of tissues. This technique can also be used to elucidate key tissue features, such as the cellular metabolic rate, vascularity, intravascular oxygenation, and alterations in tissue morphology, which helps in precisely locating the neoplastic mucosal tissues of the oral cavity.

A number of new quantitative optical techniques, including diffuse reflectance spectroscopy, fluorescence spectroscopy, and Raman spectroscopy, have been investigated for detecting epithelial neoplasms.^{3, 4, 5, 6, 7, 8} Many of the instruments developed and tested, both in vitro and in vivo, show promise to increase both sensitivity and specificity. Among them, diffuse reflectance spectroscopy is one of the simplest and most cost effective methods for understanding biological tissue characteristics. This technique involves detection and analysis of a portion of the incident light that undergoes multiple elastic scattering owing to inhomogeneities in the refractive index of the tissue. Recent results have shown that tissue back scattering is altered as the size of the nucleus increases and the nuclear texture becomes coarser.⁹ Tissue absorption in the UV and visible region is dominated by hemoglobin, with the oxygenated and deoxygenated forms having different absorption spectral features.¹⁰ The potential for using diffuse reflectance spectroscopic techniques is greater because elastic interactions are much stronger than inelastic interactions.

By illuminating the tissue with continuous-wave white light from an optical fiber, and collecting the emanating light at various distances using several other fibers, the optical properties of the tissue (absorption and scattering coefficients) can be calculated.^{4, 11, 12} Many researchers have explored the use of reflectance to correct the tissue fluorescence spectral data and for understanding the effect of changes in oxygenation and tissue perfusion.^{13, 14} Mourant ⁶ used reflectance between 330 and $370\mathrm{nm}$ , affected by nuclear-to-cytoplasmic ratio, to distinguish malignant from nonmalignant sites in the bladder and found a sensitivity of 100% and a specificity of 97%. Ge ¹⁵ used an algorithm that utilized intensities at the absorption peaks of oxy- and deoxyhemoglobin to distinguish neoplastic from nonneoplastic sites in colonic tissue and reported a sensitivity of 86% and specificity of 84%. Reflectance spectroscopy was used to detect Barrett’s oesophagus with an agreement of 71% in comparison to the diagnosis by four different pathologists.¹⁶ Zonios ⁷ used diffuse reflectance data from human adenomatous colon polyps to fit an analytical model and found that differences can be attributed to hemoglobin concentration and effective scatterer size. Elastic light scattering spectroscopy with polarized illumination∕detection, which reduces the contribution from hemoglobin absorption, was used by Sokolov ¹⁷ to characterize in vitro cervical biopsies and in vivo oral cavity mucosa. Utizinger ¹⁸ have investigated the slope of reflectance spectral features between 530 and $585\mathrm{nm}$ at different source-detector separations for in vivo characterization of ovarian cancer and found that reflectance spectral slopes between 510 and $530\mathrm{nm}$ , strongly affected by oxy- and deoxyhemoglobin absorption, provided useful discriminatory information. These bands, seen in both normal and malignant tissues, have been utilized to extract intrinsic tissue fluorescence, which is free from artifacts induced by tissue scattering and absorption.¹⁹

The diffuse reflectance spectrum shows characteristic hemoglobin absorption valleys at 430, 542, and $577\mathrm{nm}$ on excitation with white light and it is reported that hemoglobin content is more pronounced in malignant and premalignant lesions owing to increased microvasculature.⁷ In this study, we explore the potential of using the diffuse reflectance spectral ratio (R540∕R575) of the oxygenated hemoglobin absorption bands at 540 and $575\mathrm{nm}$ for ex vivo detection of oral cavity SCC. Surgically excised tissues from various parts of the oral cavity have been examined and the results presented show the potential of diffuse reflectance spectral ratio R540∕R575 for in vivo quantitative analysis of mucosal tissues.

2.

Materials and Methods

2.2.

Instrument

The portable spectroscopic system for measurement of tissue reflectance is shown in Fig. 1 . The system consists of a compact tungsten halogen lamp (Ocean Optics, USA, model: LS-1-LL) whose output is sent through the central fiber ( $400\mu \mathrm{m}$ diameter) of a $3\text{‐}\mathrm{m}$ long reflection probe that has six surrounding fibers ( $400\mu \mathrm{m}$ diameter, each) for collection of diffuse reflectance from the lesion. The reflection probe tip is terminated in a stainless steel ferrule ( $15\mathrm{cm}$ long and $6\mathrm{mm}$ in diameter) for easy access to interior areas and to facilitate sterilization, before and after use. The distance from the fiber tip to a lesion was optimized to obtain maximum overlap between the excitation and collection areas. A spacer was used to maintain this distance during measurements. The optical fiber probe tip is kept in close contact with the tissue surface to avoid ambient light from entering the detection system and to avoid potential light loss through the specimen edges. The light emerging from the collection fiber is delivered to a miniature fiber-optic spectrometer (Ocean Optics, USA, model: USB 2000FL) driven via the universal serial bus (USB) port of a laptop computer. This spectrometer is equipped with a 2048-element linear silicon CCD array that gives a resolution of $8\mathrm{nm}$ in conjunction with the $400\text{‐}\mu \mathrm{m}$ -diam. optical fiber light guide. The diffuse reflectance spectrum is recorded in the $400\text{to}600\text{‐}\mathrm{nm}$ region using the OOI software supplied with the spectrometer.

Fig. 1

Experimental setup for diffuse reflectance spectral measurements from intact tissues.

3.

Results

The diffuse reflectance spectra depend on tissue morphology, constitution, and surface features. Figure 2 shows the mean of 10 measurements of the ex vivo (immediately after surgery) diffuse reflectance spectra of the excised lesion from the buccal mucosa of a patient, pathologically confirmed as moderate to well-differentiated SCC, and of an adjoining uninvolved area, considered as normal. Average in vivo tissue reflectance spectrum from the buccal mucosa of three healthy volunteers (10 measurements each) and the diffuse reflectance spectrum of the tungsten halogen lamp, with a ground glass plate as the scatterer, are also shown in Fig. 2 for comparison. In all our measurements, the reflectance spectral intensity from malignant lesion was always greater than that of normal mucosa. Notable differences were observed in the short-wavelength region where the broad hemoglobin (Hb) absorption valley around $420\mathrm{nm}$ is prominent. Secondary dips in reflectance spectra at 542 and $577\mathrm{nm}$ owing to absorption from oxygenated hemoglobin $\left(\mathrm{Hb}{\mathrm{O}}_2\right)$ bands were evident in both malignant lesions and adjoining uninvolved areas. As compared to normal tissues, absorption at the 542- and $577\text{‐}\mathrm{nm}$ bands were more prominent in malignant lesions, signifying increased Hb presence and microvascular volume. However, the hemoglobin bands are not seen in the in vivo diffuse reflectance spectra of healthy volunteers.

Fig. 2

Mean reflectance spectra of malignant and normal buccal mucosal tissues superimposed on the diffusely scattered light from a tungsten-halogen lamp. Mean in vivo reflectance spectra from the normal buccal mucosa of three healthy volunteers is also shown for comparison.

Mean R540∕R575 ratios, with their standard deviations, are shown in Table 1 , for normal and malignant tissue samples from various sites of the oral cavity of the patients studied. Irrespective of tissue location, the mean R540∕R575 ratio was always found to be lower for malignant lesions as compared to the adjoining uninvolved normal areas. Significant changes were also noticed in the diffuse reflectance ratio according to the stage of malignancy. In the case of moderate to poorly defined SCC of the tongue, which is the highest grade of SCC studied, the percentage decrease in the diffuse reflectance ratio from 0.94 to 0.82 was maximum (12.77%). The patient with buccal mucosa, pathologically diagnosed as moderate to well-defined SCC, reported minimum variation (8.54%) from 0.82 to 0.75. Thus, the R540∕R575 ratio derived from the absorption due to oxygenated hemoglobin bands appears to be a useful parameter in the grading of oral cavity SCC.

The temporal profile of ex vivo diffuse reflectance spectra is shown in Fig. 3 . The overall ex vivo diffuse reflectance spectral intensity and absorption band intensities at 540 and $575\mathrm{nm}$ in both normal and malignant tissues were found to decrease gradually during the $3\text{‐}\mathrm{h}$ measurement period owing to tissue degradation. During this period, the percentage variation in R540∕R575 ratio between the normal and malignant lesions also decreased from 12.5 to 1.5% (Fig. 4 ), thereby demonstrating the necessity to carry out ex vivo diffuse reflectance spectra measurements within the shortest possible time for maximum tissue discrimination. This also points to the significance of using oxygenated hemoglobin band ratios (R540∕R570) as indicators of malignancy.

Fig. 3

Mean diffuse reflectance spectra of malignant and adjoining normal lesion from the tongue diagnosed as moderate to poorly differentiated SCC. These spectra were recorded after 1, 2, and $3\mathrm{h}$ of surgical excision.

Fig. 4

Mean R540∕R575 ratio of surgically excised malignant lesion of the tongue and its adjoning normal lesion. The percentage decrease in ratio with respect to the normal lesion and standard deviation for each ratio are shown over the histogram bar.

Table 1

The mean reflectance spectral intensity ratio R540∕R575 of surgically excised oral lesions from patients with different stages of oral cavity squamous cell carcinoma (SCC), within 1h of tissue removal.

4.

Discussion

Tumor or cancerous tissues are known to exhibit increased microvasculature and hence, increased blood content.²⁰ They often have abnormally low hemoglobin (Hb) oxygenation owing to the disturbed metabolism.²¹ Zonios ⁷ observed in the case of adenomatous colon polyps an increase in Hb concentration, a larger average effective scatter size, and a smaller average scatter density, with no significant change in Hb oxygenation. Using morphometry and vascular casting, they observed that precancerous tissues are characterized by increased microvascular volume. Our results are in agreement with these observations as the malignant tissues exhibit increased absorption at the oxygenated hemoglobin bands and overall enhancement in spectral intensities with concomitant decrease in the R540∕R575 reflectance ratio.

Palmer ⁵ observed that the reflectance intensities at 415, 540, and $570\mathrm{nm}$ remained relatively constant over a $90\text{‐}\min$ time period. However, in the case of malignant tissues, the decrease of ex vivo reflectance intensity at 540 and $575\mathrm{nm}$ in $120\min$ , was 46.4 and 46.3%, respectively. They also reported that the diffuse reflectance of both ex vivo and in vivo tissues are similar at $500\mathrm{nm}$ and decreases at longer wavelengths due to tissue deoxygenation and changes in tissue scattering. Our measurements (Fig. 2) also show similarity in reflectance intensity profile between in vivo normal tissues with ex vivo malignant tissues up to $500\mathrm{nm}$ and around $600\mathrm{nm}$ . However, the temporal variations in reflectance intensity (Fig. 3) are prominent in the $500\text{to}580\text{‐}\mathrm{nm}$ region where the oxygenated Hb bands are absorbing, whereas at around $600\mathrm{nm}$ the changes are minimal between normal and malignant tissues. This increase in tissue absorbance is due to increase in hemoglobin absorption and hemoglobin deoxygenation brought about by cell lysing that increases the amount of blood content in tissues.⁵

Studies carried out on surgically excised tissue samples show that decrease in the diffuse reflectance intensity ratio R540∕R575 of $\mathrm{Hb}{\mathrm{O}}_2$ bands at 542 and $577\mathrm{nm}$ can be used to discriminate between malignant lesions and normal tissues and in determining the grade of malignancy. Since the normal tissue samples used for comparison were from areas adjoining the tumor site, there is every possibility that some parts of these tissues could also be malignant. Note that histology was not performed on the same tissue samples, but on another portion of the tissue removed during surgery. To check the validity of results, pathological analysis was conducted in a tissue sample from upper alveolus, fixed immediately after surgery, and also on a lesion fixed in the laboratory after optical sampling. In both of these cases, biopsy reports showed the lesions as moderately differentiated SCC. However, tissues from uninvolved areas considered as normal, also showed up as squamous epithelial lining with focal mild dysplasia. Hence, under in vivo conditions where one has the opportunity to sample healthy and suspicious tissues from the oral cavity of the same patient, the changes in the diffuse reflectance ratio between normal tissues and malignant lesions could be much more pronounced.

In comparison, the in vivo diffuse reflectance spectra from the buccal mucosa of healthy volunteers showed the absence of the oxygenated hemoglobin bands. The mean reflectance spectra shown in Fig. 2 relate to sampling from both sides of the buccal cavity of the three volunteers who had no visible symptoms of any oral cavity disease or inflammatory condition. Since the surgical procedures are usually not carried out in the presence of overt inflammatory conditions, the changes seen in diffuse reflectance spectral could be attributed to the vascularity associated with increased blood content in malignant lesions. Further, the percentage of decrease in the reflectance ratio R540∕R575 due to increase in vascularity of the lesions was found to depend on the grade of malignancy (Table 1).

The precancerous condition is usually characterized by cellular and microvascular proliferation and, hence, increased volume occupied by cells in tissue. Since visible light penetration is limited to the top one-half millimeter of the mucosal layer, where precancerous changes occur, the diffuse reflectance ratio of oxygenated Hb bands could help in early detection of changes to the mucosal layer of the oral cavity and cervix. Since the ex vivo and in vivo tissues have similar diffuse reflectance spectral features between $400\text{to}600\mathrm{nm}$ , the ex vivo measurements, within a reasonable time frame, have the advantage of reproducing clinically relevant in vivo conditions.

On the basis of the results obtained in this study, the diffuse reflectance ratio technique appears to have potential for the in vivo detection of oral and cervical cancer and superficial tumors of internal organs with the help of endoscopes. The optical fiber probe could further aid in demarcation of malignant lesions during surgical interventions, thereby limiting tissue removal to the barest minimum. Moreover, the diffuse reflectance instrument due to its compactness, low cost, and portability proves to be an advantageous and suitable alternative for in vivo cancer screening through community extension centers.