Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Sep 1.
Published in final edited form as: Laryngoscope. 2017 Jul 28;127(Suppl 4):S1–S9. doi: 10.1002/lary.26763

The Color of Cancer: Margin guidance for oral cancer resection using elastic scattering spectroscopy

Gregory A Grillone 1,2, Zhi Wang 2, Gintas P Krisciunas 1, Angela C Tsai 3, Vishnu R Kannabiran 4, Robert W Pistey 5, Qing Zhao 5, Eladio Rodriguez-Diaz 6, Ousama M A’Amar 6, Irving J Bigio 6
PMCID: PMC5568964  NIHMSID: NIHMS882483  PMID: 28752518

Abstract

Objectives/Hypothesis

To evaluate the usefulness of Elastic Scattering Spectroscopy (ESS) as a diagnostic adjunct to frozen section analysis in patients with diagnosed squamous cell carcinoma of the oral cavity.

Study Type /Design

Prospective, analytic study

Methods

Subjects for this single institution, IRB-approved study were recruited from among patients undergoing surgical resection for squamous cell cancer of the oral cavity. A portable, Elastic Scattering Spectroscopy (ESS) device with a contact fiberoptic probe was used to obtain spectral signals. Four to ten spectral readings were obtained on each subject from various sites including gross tumor and normal appearing mucosa in the surgical margin. Each reading was correlated with the histopathologic findings of biopsies taken from the exact location of the spectral readings. A diagnostic algorithm based on multidimensional pattern-recognition/machine-learning was developed. Sensitivity and specificity, error rate, and area under the curve (AUC) were used as performance metrics for tests involving classification between disease and non-disease classes.

Results

Thirty-four (34) subjects were enrolled in the study. One hundred seventy six (176) spectral data-point/biopsy-specimen pairs were available for analysis. ESS distinguished normal from abnormal tissue with a sensitivity ranging from 84% to 100% and specificity ranging from 71% to 89%., depending on how the cutoff between normal and abnormal tissue was defined (i.e., mild, moderate or severe dysplasia). There were statistically significant differences in malignancy scores between histologically normal tissue and invasive cancer and between non-inflamed tissue and inflamed tissue.

Conclusions

This is the first study to evaluate the effectiveness of ESS in guiding mucosal resection margins in oral cavity cancer. ESS provides fast, real-time assessment of tissue without the need for pathology expertise. ESS appears to be effective in distinguishing between normal mucosa and invasive cancer and between “normal” tissue (histologically normal and mild dysplasia) and “abnormal” tissue (severe dysplasia and carcinoma-in-situ) that might require further margin resection. Further studies, however, are needed with a larger sample size to validate these findings and to determine the effectiveness of ESS in distinguishing visibly and histologically normal tissue from visibly normal but histologically abnormal tissue.

Keywords: Oral cancer, Scattering Spectroscopy, Optical Biopsy, Non-Invasive Diagnosis

INTRODUCTION

Oral cancer constitutes a significant global health problem with over 300,000 new cases and over 145,000 deaths worldwide in 2012 1. It is estimated that 39,500 to 45,780 new cases of oral cavity and oropharyngeal cancer will be diagnosed in the United States and an estimated 7,500 to 8,650 people will die from these types of cancer in 20152, 3. Globally, oral cancer remains a lethal disease for nearly half of all patients diagnosed annually, in part because most cases are already in advanced stages at the time of detection and because local recurrence following treatment is common.

Management of oral cavity cancer may include surgical excision, radiation therapy, or both. Early stage, superficial cancers (Stage I and II) can be treated with surgery alone or radiation therapy alone, but surgery is often preferred4. Advanced cancers (Stage III and IV) are typically treated with surgery and radiation therapy, with or without chemotherapy, depending on the stage of the disease at the time of diagnosis and the histological features of the tumor 4. The main goal of surgery is to completely remove the tumor, leaving no residual malignant cells, while preserving as much function as possible. In cases where there are positive or close surgical margins (tumor within 5 mm of surgical margin), surgical re-resection is recommended if the original margins are identifiable and if further surgical resection does not introduce significant functional impairment4.

Various studies have demonstrated that the status of the surgical margin may be the single most important factor in the outcome of oral cavity cancer excision, as positive margins are associated with increased complications, recurrence and morbidity5, 6. Also, positive margins are more associated with oral carcinoma than with other head and neck tumors7, 8. The current standard of care is to perform intraoperative frozen section analysis. However, such analysis is typically based on randomly selected samples for frozen section biopsies, thus leaving much of the surgical margin unexamined. Furthermore, a number of studies have shown that despite the fact that the specificity of this technique in predicting the status of margins is very high, the sensitivity is unacceptably low9. Additionally, even in cases where the frozen section sample has been reported to be negative, the final histologic evaluation in the immediately adjacent final surgical margin can be positive10. Currently there is no standard method of surgical margin analysis when frozen section technique is employed. Other potential disadvantages to the use of frozen section analysis are miscommunication between the surgeon and the pathologist and disorientation of the samples received for frozen section. Finally, frozen section technique can be time consuming and highly expensive with a cost-benefit ratio of 20:19, 10.

The goal of this study was to evaluate the usefulness of optical technology, specifically Elastic Scattering Spectroscopy (ESS), as a diagnostic adjunct to frozen section analysis in patients undergoing surgical resection for squamous cell carcinoma of the oral cavity.

A number of studies have evaluated the usefulness of optical technology for evaluation of normal and abnormal tissue in the oral cavity. McGee et al. demonstrated the significance of anatomy in the production of distinctive spectra from various subsites in the oral cavity. Specifically, a Fast Excitation Emission Matrix (FastEEM) spectroscopy device was used to obtain diffuse reflectance, intrinsic fluorescence and light scattering spectroscopic measurements (Trimodal Spectroscopy) from different sites in the oral cavity of healthy volunteers and revealed that intrinsic factors in the anatomy of the oral cavity account for the significant differences observed in the extracted spectral measurements, reflecting the need for the development of subsite-specific spectral algorithms in the oral cavity11. A subsequent study by McGee et al. investigated the potential use of FastEEM to distinguish benign from dysplastic/malignant lesions in the oral cavity. Data were collected from patients undergoing biopsy for clinically suspicious lesions and patients undergoing surgical resection of diagnosed malignancy. The results suggested that distinction between benign and dysplastic/malignant lesions was most successful and consistent when anatomy-based algorithms for single subsites or groups of spectrally similar subsites were used rather than a single algorithm for all subsites in the oral cavity12. Several studies have evaluated the usefulness of optical technologies to detect dysplastic or early malignant lesions in normal appearing oral mucosa. Tsui et al. used direct fluorescence during the excision of a clinically apparent lateral tongue tumor. Application of this technique resulted in the discovery of visually undetectable but histologically present dysplasia and squamous cell carcinoma up to 25 mm from the clinically detected lesion, even though the surgical margins at 10 mm were histologically normal13. A multimodality approach employing Optical Coherence Tomography (OCT) and polarimetry has also been proposed as a screening tool for mapping areas of field cancerization in order to detect premalignant or malignant lesions in the oral cavity of an animal model14.

ESS technology has shown promising results to support its use in settings outside of the head and neck, such as breast cancer15–17, Barrett’s disease18–21, and during colonoscopy22,23. It has to date never been investigated as tool to guide surgical resection in the oral cavity.

Optical tissue diagnosis using ESS, mediated by fiber-optic probes, has been shown to be useful to perform noninvasive, or minimally invasive, real-time assessment of tissue pathology in-situ24.

ESS is a point spectroscopic measurement technique, which is sensitive to cellular and subcellular morphological features25–32. Normal and abnormal tissues can generate different spectral signatures as a result of changes in nuclear size, chromatin granularity or density, organelle sizes and densities, and other sub-cellular features, the optical-spectroscopy equivalent of histopathological readings33,34. An important advantage of ESS is that it provides an objective and quantitative assessment of tissue pathology, which does not require on-site special expertise, and may avoid the need for subjective image interpretation as in conventional histopathology.

This study evaluated the potential application of ESS patterns to distinguish benign, dysplastic and malignant tissue in the surgical margin, which ultimately could improve the prognosis of patients treated for squamous cell carcinoma of the oral cavity.

METHODS

Study subjects and research design

This study was conducted at a Boston University Medical Center (BUMC) from 2012 to 2015 and was approved by the BUMC Institutional Review Board. Subjects were recruited from among patients already diagnosed with squamous cell cancer of the oral cavity who were to undergo surgical resection under general anesthesia as part of their treatment. Written informed consent was obtained after thoroughly explaining the study to each subject. In the operating room the surgeon of record determined the resection margin to be used. Spectral readings and biopsies were obtained by the principle investigator or the operating surgeon. Care was taken to clear the mucosa of any surface blood or other debris prior to obtaining spectral readings. Depending on the dimension of the tumor and the margin, four to ten spectral readings and biopsies were taken from each subject from sites that included visible tumor, normal appearing mucosa immediately adjacent to visible tumor, and mucosa at various distances from the edge of the visible tumor up to, but not extending beyond, the planned resection margin. The Elastic Scattering Spectroscopy (ESS) device was operated by a technician trained in its use. Histological analysis was performed independently by two pathologists blinded to the spectral results. If the two histological analyses differed for any given sample, the two pathologists reviewed that sample together to agree on a single analysis. The spectra obtained intraoperatively were then correlated with the corresponding histological findings.

For each subject the following information was recorded: Age, gender, race/ethnicity, TNM stage, tumor subsite (dorsal tongue, lateral tongue, ventral tongue, floor of mouth, buccal mucosa, soft palate, gingiva, hard palate, retromolar trigone), HPV (P16) status, history of tobacco and alcohol consumption, and history and location of prior head neck cancer surgery and/or radiation.

Elastic Scattering Spectroscopy (ESS) System-Description

The method of elastic-scattering spectroscopy (ESS) was developed by Bigio et al 27,28,33, and has been demonstrated to have potential clinical use in tissue diagnosis based on the scattering properties of cellular micromorphology24,35. ESS is a point spectroscopic measurement technique, over a broad wavelength range (320–900 nm in our current system), which, when performed using appropriate fiberoptic geometry (Figure 1), has a depth of penetration of 0.5 millimeters and is sensitive to the microscopic morphological changes occurring in tissues at the cellular and sub-cellular level. These include nuclear size and density, hyperchromaticity, DNA condensation and chromatin granularity, nuclear crowding, and changes in the size/density of mitochondria and other cellular organelles, or structural proteins. ESS spectra are derived from the wavelength-dependent optical scattering efficiency and the effects of changes in the scattering phase function (the angular probability distribution) due to the optical index gradients of cellular and subcellular structures) (Figure 1). As a result of such ultrastructural changes, normal and abnormal tissues generate different scattering spectral signatures, which represent the optical-spectroscopy equivalent of histological appearances. The ESS method senses those morphology changes in a quantitative manner, without actually imaging the microscopic structure27,34.

Figure 1.

Figure 1

Open in a new tab

Schematic diagram of the optical geometry for the ESS fiber-optic tissue measurements.

ESS Device Description, Spectra Acquisition and Biopsy Technique

The ESS device is small, portable (Figure 2) and can be easily carried into the operating room and placed on a standard small utility table near the surgical field. The fiberoptic probes (core diameter = 200 microns, center-to center separation = 250 microns) used to obtain the spectral readings were integrated into hand-pieces custom designed for use in the oral cavity. The diameter of the outer tubing of the fiberoptic probe is 2 millimeters. Two hand-piece designs were made, one straight and one angled. The integrated fiberoptic probe/hand-piece units were designed to be sterilized and reused. Prior to recording spectra from the subjects, a calibration spectrum from a spectrally-flat diffuse reflection standard (Spectralon, Labsphere, Inc., North Sutton, NH) was taken. The diffuse reflectance of the standard ranges between 250 and 1000 nm, and can be used to account for spectral variations in the light source, spectrometer, fiber transmission and fiber coupling. All spectra in this study were recorded as ratios to the Spectralon-reference spectrum, providing data independency from the spectral characteristics of the system. The optical probe was calibrated at the time of the measurement, and the tip of the probe was placed in gentle contact with the tissue site to be investigated. Spectral measurements were obtained by the use of either a foot pedal or the keyboard on the integrated computer. The instrumentation compensates for the background light by taking a first measurement without firing the xenon lamp and a subsequent ESS measurement with the pulsed lamp triggered. The ESS spectra used in this study are the result of the subtraction of the background spectrum from the ESS spectrum. Thus, output spectra were calculated according to:

I(λ)=I(λ)tissue−I(λ)tissue backgroundI(λ)ref−I(λ)ref background

Figure 2.

Figure 2

Figure 2

Open in a new tab

(A) Schematic diagram of the ESS system (B) Photo of portable ESS optical biopsy system

Reproducible and precise measurements were assured by the use of appropriate software that provides graphical illustration of the spectra obtained. Spectrum acquisition for each site measurement and appropriate analysis from the software takes approximately one second. After the spectral data are taken, the fiber-probe tip is pressed firmly into the tissue to mark the exact site in the tissue where the spectrum was obtained. A standard 2 millimeter diameter biopsy punch (Seamless Premier Uni-Punch, Premier Medical Products Company, Plymouth Meeting, PA, USA) is then used to biopsy the exact tissue site where the spectral reading was taken. Tissue specimens were then fixed in formalin and sent to the pathology laboratory for hematoxylin and eosin staining and histopathological analysis.

Analysis of Spectra

All ESS spectra were pre-processed prior to analysis. Spectral measurements taken at each site before biopsy were averaged, smoothed, and then cropped, resulting in spectra encompassing 216 wavelength bands, from 330 to 760 nm. These spectra were then normalized to the intensity at 650nm to enhance spectral shape, not relative intensities. To classify measured spectra, a diagnostic algorithm based on multidimensional pattern-recognition/machine learning was developed. Given the high-dimensional nature of the data, a framework consisting of dimensionality-reduction followed by classification was used. Sequential floating forward selection (SFFS) was used for dimensionality-reduction36, followed by multidimensional classification using linear support vector machines (SVM)30,37–39. The diagnostic algorithm was designed to have as an output a score that relates to whether or not the input spectrum is found to be malignant or benign by the algorithm. This output is referred to as the malignancy score, which has a value in the range of 0 to 1. The closer the score is to 1, the higher the chance that the spectrum was obtained from a malignant sample, and, conversely, the closer the score is to 0 the higher the chance it is from a benign sample. Leave-one-out cross-validation was used to optimize classifier parameters and obtain classification performance estimates. Sensitivity and specificity, error rate, and area under the curve (AUC) were used as performance metrics for tests involving classification between disease and non-disease classes40. Statistical analysis was performed using Fisher's exact test and the Bonferroni corrected multiple comparison test where appropriate.

RESULTS

Subject data

Thirty-four subjects were enrolled in the study, 15 males and 19 females, with a mean age of 59 years (range 25–81). Twenty seven subjects (79%) were Caucasian, 1 (3%) was African American, 4 (12%) were Hispanic and 2(6%) had no race/ethnicity recorded. Almost half (47%) had a history of ethanol use and 71% had a history of tobacco use. Of the 15 subjects who were tested for HPV status 5 were positive. The majority of subjects (75%) had advanced stage (III, IV) disease. The most common subsite involved in the oral cavity was the lateral tongue (15 subjects), followed by the floor of mouth (12 subjects) and gingiva (7 subjects). Only one patient had prior surgery and no patients had radiation to the oral cavity. Subject data is summarized in table I.

Table I.

Subject Data

Subject Data (N=34)
Gender
  Male 15
  Female 19
Race/Ethnicity
  Black /African American 1
  White/Caucasian 27
  Hispanic /Latino 4
  Unknown/Not Reported 2
Mean Age (years) 59 (range 25–81)
HPV Status (P16)
  Positive 5
  Negative 10
  Unknown 19
Cancer Stage
  1 5
  2 3
  3 2
  4 22
  Stage unknown 2
Ethanol use history
  Yes 16 (12 current)
  No 17
  Unknown 1
Mean Drinks per day (n=16) 4 (range 0.5–12)
Tobacco use history
  Yes 24 (9 current)
  No 8
  Unknown 2
Mean Pack Years (n=24) 26.7 (range 2–80)
Oral Cavity Subsite Involved By Tumor (Some tumors involved more than 1 subsite)
  Buccal Mucosa 3
  Gingiva 7
  Dorsal tongue 2
  Floor of mouth 12
  Hard palate 1
  Lateral tongue 15
  Retromolar trigone 3
  Soft palate 0
  Ventral tongue 4
Prior radiotherapy to oral cavity 0
Open in a new tab

Correlation of spectral data with pathology

Twenty-seven subjects had spectroscopic data that could be analyzed. Seven subjects were excluded from this part of the study either because biopsies were not obtained (e.g. case aborted) or spectral data was not available (e.g. technical difficulty with obtaining or processing the data).

One hundred seventy six (176) spectral data point/biopsy specimen pairs were available for analysis. Table II shows the distribution of pathological findings in these 176 biopsies.

Table II.

Pathologic Diagnoses

Pathologic diagnoses Number of biopsy specimens
Normal Tissue (N) 91
Mild Dysplasia (MiD) 7
Moderate Dysplasia (MoD) 2
Severe Dysplasia/ Carcinoma in situ (SD/CIS) 4
Invasive Cancer (IC) 72
TOTAL 176
Open in a new tab

Several comparisons were made to determine the effectiveness of elastic scattering spectroscopy (ESS) to discriminate between the various pathologic categories outlined in table II. ESS distinguished normal tissue and mild dysplasia from severe dysplasia, CIS (carcinoma-in-situ) and invasive cancer with sensitivity 84% and specificity of 71%. The ROC (Receiver Operating Characteristic) curve for this comparison is shown in figure 3. ESS distinguished normal tissue and mild dysplasia from moderate dysplasia, severe dysplasia and CIS with a sensitivity of 100% and specificity 89%. ESS also distinguished normal tissue from mild, moderate and severe dysplasia and CIS with a sensitivity 85% and specificity 78%. This data is summarized in table III.

Figure 3.

Figure 3

Open in a new tab

Sample ROC Curve for Comparison Group 1

Table III.

Pathology/ESS
Comparison Groups		 Sensitivity Specificity Total
Error		 PPV NPV AUC
1 N/MiD vs SD/CIS/IC 0.8421 0.7143 0.2299 0.6957 0.8537 0.8161
2 N/MiD vs MoD/SD/CIS 1.0000 0.8878 0.1058 0.3529 1.0000 0.9354
3 N vs MiD/MoD/SD/CIS 0.8462 0.7802 0.2115 0.3548 0.9726 0.8453
4 N(NI) vs IC(I) 0.8592 0.7273 0.1720 0.9104 0.6154 0.8025
5 N(I) vs IC(I) 0.8310 0.7391 0.2143 0.7662 0.8095 0.8048
Open in a new tab

N=normal tissue; MiD=Mild Dysplasia; MoD=Moderate Dysplasia; SD=Severe Dysplasia; CIS=Carcinoma-in-Situ; IC=Invasive Cancer; PPV=Positive Predictive Value; NPV=Negative Predictive Value; AUC=Area Under Curve; N(NI)=Normal tissue with No Inflammation); N(I)=Normal tissue with Inflammation (chronic or mixed chronic/acute); IC(I)=Invasive Cancer with Inflammation (chronic or mixed chronic/acute).

For normal tissue and mild dysplasia versus moderate dysplasia, severe dysplasia and CIS, and for normal tissue versus mild, moderate and severe dysplasia and CIS the differences in the accuracies of the diagnostic algorithms were not found to be statistically significant (p=0.06, Fisher’s exact test).

Impact of inflammation

Several comparisons were also made to determine if the presence of inflammation affected the ability of ESS to discriminate between the various pathological categories. ESS distinguished histologically normal tissue without inflammation from invasive cancer with inflammation (chronic or mixed chronic/acute) with a sensitivity of 86% and specificity of 73%; and distinguished histologically normal tissue with inflammation from invasive cancer with inflammation with a sensitivity of 83% and specificity of 74%. This data is also summarized in table III. The differences in the accuracies of these diagnostic algorithms were not found to be statistically significant (p=.50, Fisher’s exact test).

Malignancy score

Malignancy scores were developed from the classification generated by the analysis in table III. The higher the value of the score, the higher the likelihood of malignancy. Figures 4 and 5 show the malignancy scores for the various pathologic and inflammation grades, respectively. For the pathology grades, there was a statistically significant difference between normal tissue and invasive cancer. For the inflammation grades, there was a statistically significant difference between non-inflamed tissue and chronically inflamed tissue, and between non-inflamed tissue and tissue with mixed (chronic and acute) inflammation.

Figure 4.

Figure 4

Open in a new tab

Malignancy Score-Pathology

Figure 5.

Figure 5

Open in a new tab

Malignancy Score-Inflammation

DISCUSSION

Patients diagnosed with squamous cell carcinoma of the oral cavity typically undergo surgical excision with 10 mm margins. The goal is to achieve surgical margins with no residual malignant cells since positive surgical margins are associated with increased risk of morbidity5, 6. Sutton et al. reported that patients with close surgical margins (tumor <5 mm from surgical margin) after resection of oral squamous cell carcinoma had significantly higher rates of perineural invasion and vascular invasion compared to patients with negative surgical margins. Additionally, in those patients with close surgical margins the tumor’s characteristics demonstrated greater diameter and a more aggressive invasion pattern with nodal metastatic disease41. Another study also demonstrated that a higher incidence of recurrence at the primary site and significantly lower 5-year survival rate were observed when tumor was present or close (<2 mm) to the surgical margins as compared to clear surgical margins6. Moreover, patients with oral cavity cancer are at increased risk for local recurrences and the development of second primary tumors, even after surgical excision of the primary malignancy. This observation could be explained by a phenomenon known as field cancerization, which was first described by Slaughter et al. almost six decades ago. In this study of 783 patients with oral cavity carcinoma, 88 instances of independent multiple tumors were detected, which was far beyond the statistical possibility of chance occurrence. The group suggested that field cancerization is a significant factor in local recurrences and multifocality (“satellite” or “skip” lesions) of oral cavity tumors42. Recent studies also strongly support the dynamics of the field cancerization theory in the oral cavity43,44. According to these studies, local recurrence of excised aggressive oral cavity carcinoma can be observed even in cases where clear surgical margins, as determined by histopathology, were achieved43. Additionally, Tabor et al. reported that squamous cell carcinoma in the oral cavity may show common clonal origin with second primary oral cancer, even if the distance between them was >7 cm and the mucosa separating them appeared visually and histologically normal. These results indicate that large fields of healthy mucosa are replaced by a population of genetically altered cells of monoclonal origin44. Modern molecular techniques, such as measuring the loss of heterozygosity with microsatellite analysis, demonstrate genetic alterations associated with malignant conversion, despite the fact that biopsies from surgical margins of excised oral cavity carcinoma were determined to be histopathologically normal45. Nevertheless, frozen section analysis remains the current standard of care for the intraoperative detection of safe surgical margins during excision of squamous cell carcinoma in the oral cavity. However, this technique can be unreliable in distinguishing benign from malignant tissue since it can have a high false negative rate9, 10. Furthermore it typically samples only a small area of the entire mucosal margin.

Achieving negative margins in oral cavity cancer resection may be the only factor related to recurrence that is within the control of the surgeon. Discovery of newer, faster and more reliable methods to monitor surgical margins are therefore critical in aiding surgeons. The ideal system for intraoperative assessment of surgical margins is one that can be integrated clinically without disrupting the flow of the surgical procedure, that can rapidly provide real-time, accurate assessment of margins, and that does not require on-site special expertise other than the surgeon. To this end, various optical technologies have been studied. Early work by Backman et al. showed that Light Scattering Spectroscopy (LSS) has the potential to detect epithelial precancerous lesions and preinvasive cancers in various organs throughout the body including the oral cavity46. Subsequently, work by Muller et al and McGee et al using a FastEEM device with Trimodal Spectroscopy (TMS), which utilizes diffuse reflectance and intrinsic fluorescence measurements in addition to LSS, showed promising results in distinguishing histologically normal tissue from dysplastic/malignant tissue and benign lesions from dysplastic/malignant lesions in the oral cavity but is dependent on anatomic subsite-based algorithms11,12,47. Furthermore the device is large and not easily transportable to the operating room, and spectral readings can take more than a few seconds to record and can thus be distorted if the contact probe moves during the recording process. Direct fluorescence has also shown promise in detecting occult premalignant and malignant disease in the oral cavity in a handful of patients, but more studies are needed to determine its effectiveness13,48. Optical coherence tomography (OCT) has also been shown to be useful in detecting premalignant malignant lesions in the oral cavity of animals and humans14,49. However, the depth of penetration can be limited by other factors such as absorption and it requires special training and expertise to interpret the images. Furthermore, the speed of data acquisition is slow and thus OCT is sensitive to any movement of the tissue or probe.

There are several important differences between ESS and other optical methods for assessing oral cancer. The information contained in the ESS spectrum most closely correlates to the micro-structural and architectural changes that a histopathologist looks for but without the need for image interpretation. In contrast, fluorescence spectroscopy, and in particular autofluorescence, assesses biochemical changes and can be affected by variations in the local pH and electronegativity as well as by blood oxygenation50. Autofluorescence spectra can also be influenced by factors such as gender, pigmentation, tobacco use, alcohol consumption, and the wearing of dentures51. Moreover, while ESS can be performed in the presence of full room light, autofluorescence spectroscopy and Raman spectroscopy generally require temporary reduction or complete elimination of ambient light which may be impractical in clinical settings.

Diffuse reflectance spectroscopy (DRS) has also been investigated as an optical method to assess oral cancer. Einstein et al used DRS to study large, excised tissue samples52. However, the analysis was based on assumptions of the diffusion approximation to transport theory. Diffuse reflectance requires measurement of optical reflectance from large tissue volumes, whereas the sensing volume of ESS is much smaller, such that ESS can detect changes in tissue micromorphology on a millimeter range spatial scale, and thus it is better suited to assessing the precise location of changes at margins.

Confocal Reflectance Microscopy (CRM) has also been studied as a non-spectroscopic optical approach to in-situ assessment of oral mucosa. The large size of early CRM devices limited study to the lip and tongue. More recently, Maitland et al used a miniature microscope to apply CRM technology to oral tissue53. This was demonstrated in vivo to provide moderate-resolution microscopic images of oral mucosa, aided by an applied acetowhitening agent to increase scattering from nuclei. This method images features that can be correlated with conventional histopathology findings and shows promise for assessing resection margins. However, similar to OCT, real-time interpretation of the microscopic images would require the surgeon to develop the skills to read the images and render subjective interpretation. In contrast, the ESS method provides an objective, quantitative determination of the tissue microarchitecture, rendered instantaneously and not requiring subjective interpretation.

ESS has shown promising data in areas of the body other than the oral cavity15–23. To our knowledge, this study is the first to evaluate ESS technology to guide surgical resection in the oral cavity. The main goal was to evaluate its effectiveness in distinguishing normal, dysplastic and malignant tissue in the oral cavity.

The ESS system used was lightweight, portable and could be easily transported to the operating room. The custom made hand-pieces facilitated easy access of the fiberoptic probe to all areas of the oral cavity. Spectral readings were obtained typically in under one second. With a depth of penetration of 0.5 mm, the system is well suited to evaluate oral cavity mucosa, which varies in thickness from about 100 to 300 um. The ESS system is not commercially available at the present time. However, if produced in large numbers, the ESS system would cost approximately $10,000 and the disposable fiberoptic probes approximately $20 each.

ESS effectiveness was calculated by grouping several different pathologic categories rather than comparing single pathologic categories. This was done for two reasons. First, such groupings often reflect real clinical situations. For example, during resection a surgeon may be most interested in distinguishing severe dysplasia from mild dysplasia but not severe dysplasia from carcinoma-in-situ (CIS), since in the latter case both would need to be resected in order to safely achieve a negative margin, whereas leaving mild dysplasia at the margin may be acceptable. Second, there was a paucity of specimens with dysplasia and CIS, making it difficult to provide meaningful comparisons without combining some groups.

ESS was effective in distinguishing normal and mildly dysplastic tissue from severe dysplasia, CIS and invasive carcinoma and from moderate dysplasia, severe dysplasia and CIS with a sensitivity/specificity of 84%/71% and 100%/89% respectively. These are clinically relevant comparisons, since most surgeons will resect additional tissue if there is severe dysplasia or CIS at the margin, but there are instances where it is not feasible to “chase” mild dysplasia.

ESS was also effective in distinguishing histologically normal tissue from all grades of dysplasia and CIS. This is a clinically relevant comparison since dysplasia and CIS often occur in visibly normal mucosa, and the major challenge for surgeons in determining resection margins is in distinguishing visibly and histologically normal tissue from visibly normal but histologically abnormal tissue. While these results are encouraging, it is important to note that the number spectral data point/biopsy specimen pairs was small, particularly for the various grades of dysplasia and CIS. For normal tissue and mild dysplasia versus moderate and severe dysplasia and CIS, and for normal tissue versus mild, moderate and severe dysplasia and CIS the differences in the accuracies of the diagnostic algorithms, while not statistically significant, had a p-value close to .05, suggesting that more significant differences might be noted with more data points. Furthermore we were unable to determine from this study whether the accuracy of ESS would have been improved further by analyzing the above comparisons for each anatomic subsite separately, since the number of data points/biopsy specimen pairs available for each subsite was not sufficient.

Inflammation is often a confounding factor in optical diagnosis, particularly with those technologies that include assessment parameters such as vessel size or hemoglobin saturation, since these can change significantly in inflammation. Inflammation did not appear to have a significant effect on the ability of ESS to distinguish between normal tissue and invasive cancer, regardless of whether the normal tissue was inflamed or not. This may be due to the fact that ESS technology relies primarily on parameters such as nuclear size and density, hyperchromaticity and chromatin granularity, which may be less affected by inflammation.

The malignancy scores for the various pathology grades further reflect the effectiveness of ESS in distinguishing benign, dysplastic and malignant tissue in the oral cavity. While the only statistically significant difference seen was between normal mucosa and invasive cancer, it is possible that significant differences between normal mucosa and moderate/severe dysplasia/CIS would be seen with a larger sample size for the latter group. The difference in malignancy scores between non-inflamed tissue and tissue with chronic or mixed inflammation is an interesting finding but the significance of this is unclear. It may reflect the fact that more malignant specimens showed inflammation compared to normal tissue specimens.

There are several limitations to this study and to the ESS system itself. First, the presence of blood on the surface mucosa may have affected spectral readings. Even with careful irrigation and sponging, it was sometimes difficult to keep the surface completely free of blood. In several cases the margins were incised by the surgeon prior to the taking biopsies, and this added to the problem of surface blood. Second, deep margins were not assessed. It has been shown that most positive margins occur at the deep margin in oral cavity cancer8. It is not known how effective ESS would be in evaluating deep margins, and this will require further study. Third, there were only a few biopsy specimens showing dysplasia, and it is not known if these came from visibly normal or abnormal tissue. Clinically, the most important application of optical technologies like ESS is their potential to detect severe dysplasia, carcinoma-in-situ or even microinvasive cancer in visibly normal mucosa. To determine this, a much larger sample size of dysplastic and CIS biopsies from normal appearing mucosa would be necessary. Finally, the current ESS technology utilizes a fiber based contact probe, which can only assess a 2mm diameter area at a time. Even though spectral readings can be obtained in less than one second, the small probe size and need for tissue contact make it impractical to assess the entire mucosal margin in most cases. A device (for example, a non-contact imaging device) that utilizes EES technology and is capable of rapidly assessing many millimeter range surface areas simultaneously would allow assessment of a larger surface area would be more practical in assessing margins.

CONCLUSIONS

This is the first study to evaluate the effectiveness of ESS in guiding mucosal resection margins in oral cavity cancer. ESS provides fast, real-time assessment of tissue without the need for pathology expertise. ESS appears to be effective in distinguishing between normal mucosa and invasive cancer and between “normal” tissue (histologically normal and mild dysplasia) and “abnormal” tissue (severe dysplasia and carcinoma-in-situ) that might require further margin resection. Further studies, however, are needed with a larger sample size to validate these findings and to determine the effectiveness of ESS in distinguishing visibly and histologically normal tissue from visibly normal but histologically abnormal tissue.

Acknowledgments

This study was carried out at Boston University Medical Center, Boston, MA, USA

This study was supported by a grant from the National Institutes of Health NIDCR (National Institute of Dental and Craniofacial Research) - 1R21DE023192

The authors wish to acknowledge the following individuals for their significant contributions to this study:

Scharukh Jalisi, MD

Andrew Salama, MD, DDS

Waleed Ezzat, MD

Alix Rubio, BS

Footnotes

Presented at the Triological Society Combined Sections Meeting New Orleans, LA, January 19, 2017.

The authors have no financial disclosures or conflicts of interest to declare.

Level of Evidence: N/A (Basic science research)

References

  • 1.Ferlay J, Soerjomataram I, Ervik M, et al. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11 [Internet] Lyon, France: International Agency for Research on Cancer; 2013. [accessed on September 8, 2015]. Available from: http://globocan.iarc.fr. [Google Scholar]
  • 2. [Accessed September 8, 2015];Oral Cavity Cancer [American Cancer Society Website] Last revised March 2, 2015. Available at http://www.cancer.org/cancer/oralcavityandoropharyngealcancer/detailedguide/oral-cavity-and-oropharyngeal-cancer-key-statistics.
  • 3. [Accessed on September 8, 2015];SEER Cancer Statistics Factsheets: Oral Cavity and Pharynx Cancer [National Cancer Institute Website] 2015 Apr; Available at http://seer.cancer.gov/statfacts/html/oralcav.html.
  • 4.Genden E, Ferlito A, Silver CE, et al. Contemporary management of cancer of the oral cavity. Eur Arch Otorhinolaryngol. 2010;267(7):1001–1017. doi: 10.1007/s00405-010-1206-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Jones AS, Bin Hanafi Z, Nadapalan V, Roland NJ, Kinsella A, Helliwell TR. Do positive resection margins after ablative surgery for head and neck cancer adversely affect prognosis? A study of 352 patients with recurrent carcinoma following radiotherapy treated by salvage surgery. Br J Cancer. 1996;74(1):128–132. doi: 10.1038/bjc.1996.327. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Binahmed A, Nason RW, Abdoh AA. The clinical significance of the positive surgical margin in oral cancer. Oral Oncol. 2007;43(8):780–784. doi: 10.1016/j.oraloncology.2006.10.001. [DOI] [PubMed] [Google Scholar]
  • 7.Jones AS. Prognosis in mouth cancer: tumour factors. Eur J Cancer B Oral Oncol. 1994;30B(1):8–15. doi: 10.1016/0964-1955(94)90044-2. [DOI] [PubMed] [Google Scholar]
  • 8.Woolgar JA, Triantafyllou A. A histopathological appraisal of surgical margins in oral and oropharyngeal cancer resection specimens. Oral Oncol. 2005;41(10):1034–1043. doi: 10.1016/j.oraloncology.2005.06.008. [DOI] [PubMed] [Google Scholar]
  • 9.DiNardo LJ, Lin J, Karageorge LS, Powers CN. Accuracy, utility, and cost of frozen section margins in head and neck cancer surgery. Laryngoscope. 2000;110:1773–1776. doi: 10.1097/00005537-200010000-00039. [DOI] [PubMed] [Google Scholar]
  • 10.Gerber S, Gengler C, Gratz KW, Kruse AL. The impact of frozen sections on final surgical margins in squamous cell carcinoma of the oral cavity and lips: A retrospective analysis over an 11 years period. Head Neck Oncology. 2011;3(1):56–63. doi: 10.1186/1758-3284-3-56. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.McGee S, Mirkovic J, Mardirossian V, Elackattu A, et al. Model-based spectroscopic analysis of the oral cavity: impact of anatomy. J Biomed Opt. 2008;13(6):064034. doi: 10.1117/1.2992139. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.McGee S, Mardirossian V, Elackattu A, et al. Anatomy-based algorithms for detecting oral cancer using reflectance and fluorescence spectroscopy. Ann Otol Rhinol Laryngol. 2009;118(11):817–826. [PMC free article] [PubMed] [Google Scholar]
  • 13.Tsui IF, Garnis C, Poh CF. A dynamic oral cancer field: unraveling the underlying biology and its clinical implication. Am J Surg Pathol. 2009;33(11):1732–1738. doi: 10.1097/PAS.0b013e3181b669c2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Ahn YC, Chung J, Wilder-Smith P, Chen Z. Mutimodality approach to optical early detection and mapping of oral neoplasia. J Biomed Opt. 2011;16(7):076007. doi: 10.1117/1.3595850. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Keshtgar MRS, Chicken DW, Autwick MR. Optical scanning for rapid intraoperative diagnosis of sentinel node metastases in breast cancer. Br J Surg. 2010;97(8):1232–1239. doi: 10.1002/bjs.7095. [DOI] [PubMed] [Google Scholar]
  • 16.Bigio IJ, Bown SG, Briggs G, et al. Diagnosis of breast cancer using elastic-scattering spectroscopy: preliminary clinical results. J Biomed Opt. 2000;5(2):221–228. doi: 10.1117/1.429990. [DOI] [PubMed] [Google Scholar]
  • 17.Johnson KS, Chicken DW, Pickard DC, et al. Elastic scattering spectroscopy for intraoperative determination of sentinel lymph node status in the breast. J Biomed Opt. 2004;9(6):1122–1128. doi: 10.1117/1.1802191. [DOI] [PubMed] [Google Scholar]
  • 18.Lovat L, Bown S. Elastic scattering spectroscopy for detection of dysplasia in Barrett’s esophagus. Gastrointest Endosc Clin N Am. 2004;14(3):507–517. doi: 10.1016/j.giec.2004.03.006. [DOI] [PubMed] [Google Scholar]
  • 19.Lovat LB, Johnson K, MacKenzie GD, et al. Elastic scattering spectroscopy accurately detects high grade dysplasia and cancer in Barrett’s oesophagus. Gut. 2006;55(8):1078–1083. doi: 10.1136/gut.2005.081497. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Wong Kee Song LM. Optical spectroscopy for the detection of dysplasia in Barrett’s esophagus. Clin Gastroenterol Hepatol. 2005;3(7 Suppl 1):S2–7. doi: 10.1016/s1542-3565(05)00250-8. [DOI] [PubMed] [Google Scholar]
  • 21.Zhu Y, Fearn T, MacKenzie G, et al. Elastic scattering spectroscopy for detection of cancer risk in Barrett’s esophagus: experimental and clinical validation of error removal by orthogonal subtraction for increasing accuracy. J Biomed Opt. 2009;14(4):044022. doi: 10.1117/1.3194291. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Dhar A, Johnson KS, Novelli MR, et al. Elastic scattering spectroscopy for the diagnosis of colonic lesions: initial results of a novel optical biopsy technique. Gastrointest Endosc. 2006;63(2):257–261. doi: 10.1016/j.gie.2005.07.026. [DOI] [PubMed] [Google Scholar]
  • 23.Bigio IJ, Singh SK. Gastrointestinal cancer surveillance by optical sensing. Clin Cancer Res. 2007;13(15 Pt 1):4315–4316. doi: 10.1158/1078-0432.CCR-07-0367. [DOI] [PubMed] [Google Scholar]
  • 24.Mourant JR, Bigio IJ, Boyer J, Johnson T, Conn R. Spectroscopic diagnosis of bladder cancer with elastic light scattering. Lasers Surg Med. 1995;17(4):350–357. doi: 10.1002/lsm.1900170403. [DOI] [PubMed] [Google Scholar]
  • 25.Boyer J, Mourant JR, Bigio IJ. Theoretical and experimental demonstrations of elastic scattering spectroscopy as a diagnostic for tissue pathologies. In: Alfano RR, editor. Optical Imaging, and Photon Migration. Washington DC: Optical Society of America; 1994. [Google Scholar]
  • 26.Upile T, Jerjes W, Sterenborg H, et al. Head & neck optical diagnostics: vision of the future of surgery. Head Neck Oncol. 2009;1(1):25. doi: 10.1186/1758-3284-1-25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Mourant JR, Fuselier T, Boyer J, Johnson T, Bigio IJ. Predictions and measurements of scattering and absorption over broad wavelength ranges in tissue phantoms. Appl Opt. 1997;36(4):949–957. doi: 10.1364/ao.36.000949. [DOI] [PubMed] [Google Scholar]
  • 28.Hielscher AH, Mourant JR, Bigio IJ. Influence of particle size and concentration on the diffuse backscattering of polarized light from tissue phantoms and biological cell suspensions. Appl Opt. 1997;36(1):125–135. doi: 10.1364/ao.36.000125. [DOI] [PubMed] [Google Scholar]
  • 29.Mourant JR, Jack D, Johnson TM, Miller HD, Bigio IJ. Measuring absorption coefficients in small volumes of highly scattering media: source-detector separations for which path lengths do not depend on scattering properties. Appl Opt. 1997;36(22):5655–5661. doi: 10.1364/ao.36.005655. [DOI] [PubMed] [Google Scholar]
  • 30.Hielscher AH, Eick AA, Mourant JR, Shen D, Freyer JP, Bigio IJ. Diffuse backscattering Mueller matrices of highly scattering media. Opt Express. 1997;1(13):441–453. doi: 10.1364/oe.1.000441. [DOI] [PubMed] [Google Scholar]
  • 31.Reif RO, A’Amar, Bigio IJ. Analytical model of light reflectance for extraction of the optical properties in small volumes of turbid media. Appl Opt. 2007;46(29):7317–7328. doi: 10.1364/ao.46.007317. [DOI] [PubMed] [Google Scholar]
  • 32.Reif R, Amorosino MS, Calabro KW, Aamar O, Singh SK, Bigio IJ. Analysis of changes in reflectance measurements on biological tissues subjected to different probe pressures. J Biomed Opt. 2008;13(1):010502. doi: 10.1117/1.2870115. [DOI] [PubMed] [Google Scholar]
  • 33.Mourant JR, Boyer J, Hielscher A, Bigio IJ. Influence of the scattering phase function on light transport measurements in turbid media performed with small source-detector separations. Opt Lett. 1996;21(7):546–548. doi: 10.1364/ol.21.000546. [DOI] [PubMed] [Google Scholar]
  • 34.Fang H, Ollero M, Vitkin E, et al. Noninvasive sizing of subcellular organelles with light scattering spectroscopy. IEEE Journal of Selected Topics in Quantum Electronics. 2003;9:267–276. [Google Scholar]
  • 35.Mourant JR, Bigio IJ, Boyer J, et al. Elastic scattering spectroscopy as a diagnostic tool for differentiating pathologies in the gastrointestinal tract: preliminary testing. J Biomed Opt. 1996;1:192–199. doi: 10.1117/12.231372. [DOI] [PubMed] [Google Scholar]
  • 36.Pudil P, Novovicová J, Kittler J. Floating search methods in feature selection. Pattern Recognition Letters. 1994;15(11):1119–1125. [Google Scholar]
  • 37.Cortes C, Vapnik V. Support-vector networks. Machine Learning. 1995;20(3):273–297. [Google Scholar]
  • 38.Vapnik VN. Statistical learning theory. New York, NY: John Wiley & Sons; 1998. [Google Scholar]
  • 39.Rodriguez-Diaz E, Castanon DA. Support vector machine classifiers for sequential decision problems; Proc. of 48th IEEE Conf. on Decision and Control; 2009. pp. 2558–2563. [Google Scholar]
  • 40.DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–845. [PubMed] [Google Scholar]
  • 41.Sutton DN, Brown JS, Rogers SN, Vaughan ED, Woolgar JA. The prognostic implications of the surgical margin in oral squamous cell carcinoma. Int J Oral Maxillofac Surg. 2003;32:30–34. doi: 10.1054/ijom.2002.0313. [DOI] [PubMed] [Google Scholar]
  • 42.Slaughter DP, Southwick HW, Smejkal W. Field cancerization in oral stratified squamous epithelium: clinical implications of multicentric origin. Cancer. 1953;6(5):963–968. doi: 10.1002/1097-0142(195309)6:5<963::aid-cncr2820060515>3.0.co;2-q. [DOI] [PubMed] [Google Scholar]
  • 43.Leemans CR, Tiwari R, Nauta JJ, van der Waal IE, Snow GB. Recurrence at the primary site in head and neck cancer and the significance of neck lymph node metastases as prognostic factor. Cancer. 1994;73:187–190. doi: 10.1002/1097-0142(19940101)73:1<187::aid-cncr2820730132>3.0.co;2-j. [DOI] [PubMed] [Google Scholar]
  • 44.Tabor MP, Brakenhoff RH, Ruijter-Schippers HJ, van der Wal JE, Snow GB, Leemans CR, et al. Multiple head and neck tumors frequently originate from a single paraneoplastic lesion. Am J Pathol. 2002;161:1051–1060. doi: 10.1016/S0002-9440(10)64266-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Tabor MP, Brakenhoff RH, van Houten VM, et al. Persistence of genetically altered fields in head and neck cancer patients: biological and clinical implications. Clin Cancer Res. 2001;7:1523–1532. [PubMed] [Google Scholar]
  • 46.Backman V, Wallace MB, Perelman LT, et al. Detection of preinvasive cancer cells. Nature. 2000;406:35–36. doi: 10.1038/35017638. [DOI] [PubMed] [Google Scholar]
  • 47.Muller MG, Valdez TA, Georgakoudi I, et al. Spectroscopic detection and evaluation of morphological and biochemical changes in early human oral carcinoma. Cancer. 2003;97:1681–1692. doi: 10.1002/cncr.11255. [DOI] [PubMed] [Google Scholar]
  • 48.Poh CF, Ng SP, Williams PM, et al. Direct fluorescence visualization of clinically occult high-risk oral premalignant disease using a simple hand-held device. Head Neck. 2007;29:71–76. doi: 10.1002/hed.20468. [DOI] [PubMed] [Google Scholar]
  • 49.Ridgway JM, Armstrong WB, Guo S, et al. In vivo optical coherence tomography of the human oral cavity and oropharynx. Arch Otolaryngol Head Neck Surg. 2006;132(10):1074–1081. doi: 10.1001/archotol.132.10.1074. [DOI] [PubMed] [Google Scholar]
  • 50.Schwarz RA, Richards-Kortum RR, Gillenwater AM. Fluorescence and Reflectance Spectroscopy for Detection of Oral Dysplasia and Cancer. In: Wong BJF, Ilgner J, editors. Biomedical Optics in Otorhinolaryngology: Head and Neck Surgery. New York, NY: Springer; 2016. pp. 431–449. ©SpringerScience+Business Media New York. [DOI] [Google Scholar]
  • 51.De Veld DCG, Witjes MJH, Sterenborg HJCM, Roodenburg JLN. The status of in vivo autofluorescence spectroscopy and imaging for oral oncology. Oral Oncology. 2005;41:117–131. doi: 10.1016/j.oraloncology.2004.07.007. [DOI] [PubMed] [Google Scholar]
  • 52.Einstein G, Udayakumar K, Aruna PR, Koteeswaran D, Ganesan S. Diffuse reflectance spectroscopy for monitoring physiological and morphological changes in oral cancer. Optik. 2016;127:1479–1485. [Google Scholar]
  • 53.Maitland KC, Gillenwater AM, Williams MD, El-Naggar AK, Descour MR. Richards-Kortum RR. In vivo imaging of oral neoplasia using a miniaturized fiber optic confocal reflectance microscope. Oral Oncology. 2008;44:1059–1066. doi: 10.1016/j.oraloncology.2008.02.002. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES