Increased Human Buccal Cell Autofluorescence Is a Candidate Biomarker of Tobacco Smoking

Oral cancer affects as many as 274,000 people annually (1, 2) and there exists today a need to identify biomarkers of cancer of the mouth (3-5). The frequency of oral cancer around the world is often indicative of the use of tobacco products (1, 6). Further, it has been established that there is a dose-response relationship between the amount of tobacco product used and the development of oral cancer (1, 2).

We have recently published a structured review of the literature that addresses smoking and smokeless tobacco-associated human buccal cell mutations and their association with oral cancer (7). The results of this literature review showed that diverse buccal cell changes have been associated with both smoking and smokeless tobacco. A partial listing of the tobacco-associated buccal cell changes that have been reported include (a) micronuclei formation, (b) bacterial adherence, (c) genetic mutations, (d) DNA polymorphisms, (e) carcinogen-DNA adducts, and (f) chromosome abnormalities (7).

Clinical studies have correlated buccal cell changes with malignant tumors, and some oral oncologists have reported that the buccal cell changes are useful biomarkers (7). Thus, some scientists and clinical practitioners have provided the rationale for using buccal cell changes as a useful surrogate biomarker of oral cancer; this subject has been reviewed elsewhere (7).

A shortcoming identified in this literature review is that the buccal cell biomarker assays that are currently prescribed are technically complex, expensive, often poorly reproducible, multiple-step methodologies that are not applicable for large-scale screening regimens (7). Accordingly, there exists a need to establish a buccal cell–based biomarker assay that is technically simple, fast, cheap, sensitive, and objective.

Schemes that use buccal cells are envisioned as offering a number of advantages in that the buccal cells can be collected in a noninvasive manner from most all sites of the mouth (e.g., mucosa, tongue, gum, and palate). Accordingly, buccal cells can be analyzed in longitudinal studies of healthy and diseased sites of the mouth of a given patient before, during, and following different therapeutic modalities.

Flow cytometry [e.g., fluorescence activated cell sorter (FACS)] has been used successfully for more than two decades for the simultaneous, real-time, multiparametric analysis of the physical (e.g., cell size), chemical (e.g., DNA), immunologic (e.g., membrane antigens), and many other attributes of single cells flowing in a fluid sheath through an optical/electronic detection apparatus (8). Approval of flow cytometry for diverse clinical studies by regulatory and government authorities (e.g., U.S. Food and Drug Administration) illustrates the potential of clinical cytometry for high-throughput analysis of epithelial cells.

The investigative protocol for these studies had been reviewed and approved by the Internal Review Board of the Roswell Park Cancer Institute. Healthy subjects and their friends were recruited with an internal review board–approved e-mail solicitation broadcasted to all Institute employees. All subjects were between 18 and 70 years old. No patients or subjects with known smoke-associated heart, lung, or oral disease were enrolled. Subject enrollment was determined by the chronological order that they responded to the e-mail. All subjects signed a written consent statement. The study group, collected without bias, was balanced across the demographic factors of age and sex, as well as the primary cofactor of smoking status (see Results). In addition, the cross-tabulations of smoking status by sex were balanced.

Pilot studies were conducted to define the optimal method for collecting buccal cells and preparing the cells for FACS analysis. By way of example, we tested different buccal cell harvesting methods (e.g., wood spatula, sponge, cytobrush, and the “swish and spit” method; ref. 7).

The collection procedure that proved to be most advantageous with respect to buccal cell yield and recovered cell homogeneity was as follows. In the morning, the subjects were asked to brush their teeth and rinse their mouth with water. They were instructed to refrain from eating or drinking for 1 h before donating buccal cells. The buccal cells were collected by a physician who softly scraped the left mucosal cheek with a new nylon-bristle toothbrush (Donovan Industries, Inc.). Before using, the bristles and head of the toothbrush were rinsed thoroughly under running tap water to remove any particles or debris from manufacturing and packaging.

The buccal cells were harvested using a gentle 120° rotation of the brush over the oral mucosa. This procedure was repeated to harvest cells from three adjacent fields of the mucosa. A new brush was used for each cheek, and was discarded after use. Thereafter, buccal cells were harvested, using the same method, from the right mucosal cheek of the subject.

A touch imprint was made of each brush to assess the harvested population of cells. Accordingly, after collecting the buccal cells, touch imprints were made by applying softly, and with an up and down stroke, the toothbrush to a precleaned 3 × 1–in. glass microscope slide. After allowing the slide to air dry, the slide was stained with a Quik-Dip (Mercedes Medical). A cover glass was mounted onto each of the two slides using Cytoseal 60 (Stephens Scientific).

The stained buccal cells were examined with a conventional white light microscope to determine the relative frequency of buccal cells (e.g., differential cell analysis), buccal cell quality, presence of epithelial cell clusters, and oral debris that may interfere with FACS analysis.

After preparing the touch imprints, the toothbrush was inserted immediately into a 50 mL polypropylene tube (Falcon, Becton Dickinson) containing 15 mL of Dulbecco's PBS without Ca²⁺ or Mg²⁺ (Life Technologies, Invitrogen Corp.). Thereafter, Parafilm (American National Can) was wrapped around the upper shaft of the brush to seal the brush into the tube. Cells were released from the brush by touching the tube to a vortex-mixer briefly three times, and then placing the tube onto an orbital shaker for 30 min.

Following a final vigorous vortexing of the tube for 3 s, 30 μL of cell suspension were collected for determining the number of buccal cells recovered. Cell counts were done of the samples collected from the mucosa surface using a hemocytometer. A differential analysis was done to assess the relative homogeneity of the cell population (e.g., buccal cells, leukocytes, and erythrocytes) and to define the number of buccal cells obtained.

Thereafter, the 50 mL tubes were centrifuged (1,400 × g, 5 min) to deposit the cells. Dulbecco's PBS was removed carefully with a pipette so as to leave 0.5 to 1.0 mL containing the cell pellet. The cells were then resuspended in 1.0 mL of 2% ultrapure formaldehyde in PBS (Polysciences, Inc.). The resuspended cells were passed through a 1-in. square nylon screen (90 μm mesh; Small Parts, Inc.) that had been positioned on top of a FACS tube (Becton, Dickinson and Company). Thereafter, in no instance were the unstained buccal cells further processed or treated.

Multivariable analyses of the buccal cells were done by cytometry using a conventional cytometer (FACScan) and an advanced multispectral cytometer (FACSAria); both FACS machines were purchased from Becton, Dickinson and Company. The cytometers were equilibrated daily with the use of freshly collected human peripheral blood leukocytes. The FACScan was configured with a 200 μm nozzle, and a FACSAria had a 100 μm orifice. Assay variables included forward scatter (relative cell size), side scatter (relative cell complexity/granularity), and the relative cell fluorescence as detected in each of nine channels (see below). For most (>85%; n = 120/138) of the samples, 1.0. × 10⁴ cells were analyzed; in no case was <2.0 × 10³ cells assayed.

Buccal cell autofluorescence was quantified using a set of fluorescence intensity standards consisting of five populations of calibrated fluorescent standards. These standards are a graded series of dye-impregnated latex microbeads having different levels of fluorescein (FITC) intensity (Quantum FITC Kit, Bangs Laboratories, Inc.). The stated fluorescence of the beads [mean equivalent soluble fluorescence (MESF)] was used to calculate a standard curve relating mean fluorescence intensity values to assigned MESF (9). Buccal cell mean fluorescence intensity values from the FITC channel were interpolated against this curve to determine MESF. In addition to the latex calibration beads, all experiments included, for comparison, freshly collected peripheral blood leukocytes of a healthy donor. Analysis of the data was done with WinList (version 5.0; Verity Software House, Inc.).

The primary end points that we examined were log₁₀-transformed total cells and MESF as a function of cell viability, age, sex, and smoking history. Both univariate ANOVA and regression models were fit depending on the nature of the covariate (nominal or continuous). In addition, multivariate analysis of covariance (ANCOVA) models were fit, modeling the main effects listed above and possible two-way interactions. Within the ANCOVA modeling, we blocked on a subject to account for measures from both the left and right sides. We also examined the linear correlation between left and right sides for log₁₀-transformed total cells, cell viability, and MESF using standard Pearson correlations.

Examination of the harvested population showed that it consisted almost exclusively (>98%) of buccal cells (n = 138 samples). The cells were present as single cells, and only rarely were clusters of cells, usually consisting of less than a dozen stratified epithelial cells, observed (Fig. 1).

Erythrocytes were seldom seen; each sample was examined for these cells as their appearance was interpretative of a trauma and use of the toothbrush that was too aggressive. Leukocytes, almost exclusively polymorphonuclear leukocytes, were present in “swish and spit” mouth wash procedures used for collecting buccal cells. Leukocytes, however, were seldom present (<1%) in buccal cell samples (n = 138) that had been collected with a toothbrush.

An initial study of 12 healthy subjects revealed that the buccal cells displayed a high level of autofluorescence. The high autofluorescence of the buccal cells was observed with both paraformaldehyde-fixed cells and freshly collected cells that were assayed within 3 h of collection. This and related studies excluded fixation as a cause of the high degree of cellular fluorescence. The results of these pilot studies also showed that it was unnecessary to remove the erythrocytes and leukocytes because both cell types could be discriminated from the large buccal cells and could be excluded from analysis by electronic gating.

In no instance was their clumping of the buccal cells that prevented analysis by FACS. Further, in no case was there a sample that had to be rejected due to a technical or processing error.

In one study, a total of 69 subjects were studied (32 men and 37 women; age, 43.5 ± 9.9 years old; mean ± SD). The number of buccal cells collected from 138 cheek specimens was 2.1 ± 1.4 × 10⁵ (mean ± SD). It is to be noted that a larger number of cells would have been recorded had the toothbrush not been used for making touch imprints before counting the cells. For all analyses, a minimum of 2.0 × 10³ cells were analyzed. For most samples (n = 120/138; 86.9%), 10 × 10³ cells were assayed; this was equivalent to ∼1/20th of the total buccal cells in a given sample. Thus, for all subjects and for all samples, a sufficient number of buccal cells were obtained for analysis by flow cytometry.

No technical problems were encountered in assaying the large flat buccal cells using a conventional flow cytometer and a standard protocol. The cells were assayed successfully with a FACScan (flow tube orifice size, 200 μm) and a FACSAria (size, 100 μm). Despite the large size of the buccal cells, there was never a case in which the orifice or other components of the flow cell became occluded.

To characterize the intensity and emitted fluorescence spectrum, the buccal cells were evaluated using a multispectral FACSAria flow cytometer. Presented in Fig. 2 are the histograms obtained from the analyses of buccal cells from three donors. The three subjects were selected to illustrate a representative profile of individuals who showed a low, intermediate, and high level of buccal cell autofluorescence.

The relative autofluorescence of the buccal cells from the three donors (subjects 6, 59, and 69) is shown in Fig. 2A to I. The fluorescence recorded for all (A-I) was high. By way of example, for C (FITC channel), the autofluorescence MESF values (×10³) of the buccal cells for the three subjects were 480, 2,760, and 16,640, respectively (35-fold). For normal human blood lymphocytes, run daily as controls in the Flow Cytometry Facility, the autofluorescence is ∼400 MESF.

The width of the buccal cell fluorescence spectrum from each of the three subjects is illustrated in A, 450 nm, to I, 780 nm of Fig. 2. For all three subjects, fluorescence profiles of the buccal cells were recorded in each of the nine different panels (A-I).

Particularly noteworthy is that the autofluorescence of buccal cells varied from subject to subject (Fig. 3). The results shown compare the autofluorescence intensity of the buccal cells collected from the left and right cheek of 69 different subjects (32 males and 37 females; age, 43.5 ± 9.9). In this population, there were 31 current smokers, and 38 former and never smokers.

Shown in Fig. 4 is a graph of nonlinear predicted fluorescence values as a function of smoking status, sex, and age. The values were first derived from the linear model of log fluorescence and then exponentiated. The actual errors/significance corresponding to the graph are summarized via the model statistics contained herein (see below) and summarized via the estimated variable and corresponding P values. Our univariate analysis indicated that there was an association between buccal cell fluorescence and smoking history (P = 0.018) and, to a lesser extent, by age (P = 0.036).

With a multivariate analysis using ANCOVA for smoking history, and sex by smoking history interaction, differences were significant at the 0.05 level. Smokers have higher levels of buccal cell fluorescence than nonsmokers. Buccal cell fluorescence levels increase faster for males compared with females as a function of age (P = 0.037). Further, younger females have high buccal cell fluorescence compared with males; however, the levels converge at the older ages.

The results of studies reported herein show that (a) numerous human buccal cells can be collected easily, reproducibly, and in a noninvasive manner with a toothbrush; (b) the buccal cells can be processed to provide a relatively homogeneous population of buccal cells that are perceived as being suitable for different assays; (c) ample numbers of buccal cells can be generated from scraping a single site of the mouth to generate samples for both morphologic and cytometric analysis; (d) despite their relatively large size, buccal cells can be analyzed with a common single-laser flow cytometer or with an advanced multilaser flow cytometer; (e) buccal cells consistently display a high level of autofluorescence; (f) buccal cells collected from the left and right cheek of a given donor displayed consistency for autofluorescence and cell number, thus illustrating the fidelity of the sampling technique; (g) the fluorescence spectrum of the buccal cells from the different subjects was displayed over a broad range of fluorescent emissions and all samples occupied all nine emission channels (range 450-780 nm) of the FACSAria; and (h) there was a significant correlation between autofluorescence of the buccal cells by smoking history (P = 0.018).

For all subjects, an ample number of cells were collected to permit FACS analysis. Further, in no instance was a sample rejected because of cell damage, cell clumping, or debris from food particles. The well-defined morphology and even distribution of individual buccal cells over the surface of the glass slide would suggest that the preparations would also be suitable for slide-based cytometry [reviewed recently by Tárnok (10)].

The origin of the high buccal cell autofluorescence has not been delineated. The broad emission spectrum would suggest that the fluorescence cannot be ascribed to any one agent, and that, most probably, it originates from diverse agents (e.g., NADP, Fad2, and flavoproteins; reviewed in ref. 11).

The fluorescence spectra and intensity observed for the single-cell, cytometry-based analyses of buccal cells of a subject that are reported herein differ from observations made of mouth or lung with autofluorescence bronchoscopy; reviewed in refs. (12-14). First, the buccal cell autofluorescence assayed by flow cytometry is a highly amplified signal of a single cell (8). In contrast, autofluorescence bronchoscopy and white light bronchoscopy is a view of a complex and heterogeneous tissue area that is observed with the naked eye (12-14).

Second, the origin of the autofluorescence is very different. Normal tissue fluoresces green (peak, 520 nm; ref. 13) when exposed to light in the violet blue spectrum (excitation, 400-450 nm; refs. 12, 13). With disease progression in the mucosa or submucosa, there is a progressive and highly significant (10-fold; ref. 13) diminution of the green autofluorescence (12-14).

Different explanations have been presented for the decreased tissue fluorescence (12-14). Most of the observed reduction in autofluorescence is not due to the tumor. The bulk of the fluorescence signal comes from the submucosa; the decrease observed by autofluorescence bronchoscopy has been attributed to changes in the submucosa (12-14), more specifically, to the blood content (e.g., microvasculature and angiogenesis; refs. 12-14), thickening of the epithelial cell layers, and other factors (14).

Additional contributors to tissue fluorescence include stroma and diverse types of cells as well as apoptotic cells and necrotic tissue (12-14). The major chromophores in normal and neoplastic tissue of the mucosa are elastin, fibrin, collagen, flavins, and intrinsic components (11, 12, 15). Thus, the autofluorescence profiles of single buccal cells characterized by FACS cannot be equated with observations of the autofluorescence of oral and pulmonary mucosa made by autofluorescence bronchoscopy.

Noteworthy is that quantitative autofluorescence imaging of excised tissue and paraffin block sections has been shown recently to be of considerable promise for biomarker analysis (16).

We and others have shown that the bright fluorescence observed in lung macrophages and lung tissue of the respiratory tract is due to tobacco “tar” (e.g., smoke particulates). Studies of laser-generated optical sections have shown that the greatest contributors are polycyclic aromatic hydrocarbons (see Discussion in ref. 17).

Diverse buccal cell changes have been associated with the use of smoking and smokeless tobacco using diverse assays (7). Buccal cell autofluorescence, however, has not been addressed previously.

Summarily, we present here the first report documenting the ability to analyze human buccal cells using flow cytometry. This technology is cheap, quantitative, versatile, reproducible, and amenable to statistical analyses. Notwithstanding, there is a noticeable absence of publications describing the application of new generation flow cytometers for the analysis of epithelial cells. Our experience in studies of buccal cells suggests that the advantages afforded by the new generation of multispectral flow cytometers will prove useful in crafting biomarker assay schemes to identify epithelial cells of tumors and premalignant lesions of diverse organs (e.g., mouth, lung, uterine cervix, and urinary bladder).

We thank Patti Coppola, R.N., Cancer Screening Clinic; Ed Podnieski, Department of Flow Cytometry; Doug Nixon, Creative Services; Mary M. Vaughan, Histology Research Specialist; and Puja Verma, Clinical Research Services.