Oral EC specimens were obtained from Nihon University Hospital at Matsudo between January 2017 and December 2018. Twenty patients were divided into three subgroups: 10 patients classified as negative for intraepithelial lesion or malignancy (NILM) (comprising 5 patients with inflammation, 5 patients with hyperkeratosis); 5 patients with low-grade epithelial dysplasia (dysplasia); and 5 patients with malignant lesions (OSCC, well-differentiated type). Details of cases are shown in Table 1. The 20 patients comprised 10 men (mean age, 55.1 ± 18.9 years) and 10 women (mean age, 63.9 ± 16.5 years). Biopsy was performed after smear collection to confirm the histological diagnosis.

All pathological lesions were identified on the edge of the tongue. In control subjects (normal, n = 5, mean age 36.0 ± 13.7 years), smears were obtained by brushing from the edge of the tongue from healthy volunteers. Exclusion criteria included: presence of any systemic illness or use of any medication 2 years prior to the study; presence of habits that were considered likely to affect to the oral mucosa (drinking, smoking, or heavy use, more than 10 times a day, of mouthwash); Candida infection; or use of dentures/orthodontic appliances.

Informed consent was obtained from all individuals prior to inclusion in the study. All procedures in studies involving human participants were performed in accordance with the ethical standards of the Committee on Studies Involving Human Beings of Nihon University School of Dentistry at Matsudo (EC-15-14-033-1) and with the 1964 Declaration of Helsinki and its later amendments.

1) Cell collection

All cell collection was performed by the same skilled oral surgeon. Specimens were collected using a cell collecting brush (Orcellex^Ò brush RT; Becton, Dickinson and Company, Tokyo, Japan) [21]. Each brush was rotated at the lesion 10 times. The split sample method was used for specimen preparation of all cytological samples.

¹Edge of the tongue; ²Negative for intraepithelial neoplasia or malignancy; ³Oral squamous intraepithelial lesion; ⁴Oral squamous cell carcinoma; ⁵No biopsy was performed.

2) Specimen preparation

Conventional smears were prepared from the brush, fixed in 95% ethanol for 3 hours and subsequently stained by routine Papanicolaou (PAP) staining. PAP staining dyes the cytoplasm into light green/blue, orange, and red by several substances with different molecular weights. After conventional smear preparation, the brush containing the remaining cells was immersed in a preservative vial (Cytorich^TM red; BD). Fixation was performed for over 30 minutes. Specimen preparation for LBC (BD Cytorich^TM as SurePath^TM manual method; BD) followed the instructions from the manufacturer. In short, the preservative vial was centrifuged at 2000 rpm for 5 minutes, then the supernatant was discarded. After adding 500 ml of purified water, the sediment was stirred. About 250 mL of sediment was dispensed onto a pre-coated slide with a settling chamber and left to stand for 10 minutes. After inverting the slide tray and discarding the solution, the slide was washed twice with 100% ethanol. The slide tray was then turned upside-down for no more than 1 min to remove excess alcohol. Thereafter, PAP staining was performed.

3) Cytological and pathological diagnosis

Cytological diagnosis was performed under microscopy (Olympus BX51; Olympus, Tokyo, Japan) by at least one cyto-screener for each specimen. After that, two or more dentists board certified trainer of oral cytology by the Japan Society of Clinical Cytology. Each cytological specimen was then observed for cytological findings including thickness of the smear, cellular distribution, resolution/clarity of cells, cellular staining characteristics and the presence of unsatisfactory background/artifacts and cytoplasm/nuclear form. Definitive pathological diagnosis was made by two or more oral pathologists certified by the Japan Society of Pathology. The criteria for definitive diagnoses followed the criteria published by the WHO [4].

4) Cytomorphological methods

Using a marking pen, a line was drawn across the center of each specimen. Ten discontinuous fields across the middle diameter of each preparation were evaluated using a ×40 objective lens. Only well-visualized squamous cells were counted. The average cell count from the 10 fields (×40) was measured. An average of ≥50 cells/case was required for cytomorphological analysis. Selection of atypical squamous cells was performed by two cyto-screeners and two dentists specializing in oral cytology. Dysplastic cells of OSCC were divided into deep and superficial layers. After selecting about 50 cells/case, pictures were taken using a digital camera (Olympus DP74; Olympus), and cytomorphological analyses were performed using imaging software for microscopy (cellSens^Ò software; Olympus). Cell number and nuclear findings (area, hyperchromasia, irregularity, and circularity) were quantified.

All statistical analyses were performed using SPSS for Windows version 14. OJ (IBM, Tokyo, Japan). Statistical analyses were performed using the Wilcoxon signed-rank test, the Kruskal-Wallis test and Steel-Dwass test. Values of p < 0.05 were considered statistically significant.

Representative cytological pictures are shown in Figure 1 and Figure 2. A significantly larger number of superficial cells was observed with LBC (Figures 1(a2)-(d2)) than with CVC (Figures 1(a1)-(d1)) in all cases. Squamous cells in LBC formed clusters in the superficial layer, middle layer and deep layer.

Slight intussusception was apparent in LBC but did not impact cell observations. In the background, inflammatory cells, necrotic debris and erythrocytes were decreased in LBC. In cases of hyperkeratosis, keratin debris and orthokeratinized squamous cells were mainly observed (Figure 1(a1), Figure 1(a2)). With CVC, keratinized superficial cells with keratohyaline granules and keratin debris formed

a clustered appearance. Slight thickening of nuclear contours, pyknosis and no increase in nuclear/cytoplasm ratio were observed (Figure 2(a1)). With LBC, nuclei showed slightly more irregular contours, clear nuclear structures and fine chromatin distribution (Figure 2(a2)). In inflammation cases, non- or para-keratinized cells with mild nuclear swelling and keratin debris were observed intermingled in both methods (Figure 1(b1), Figure 1(b2)). With LBC, each cell was isolated and showed increased transparency of cytoplasm and clear nuclear structures (Figure 2(b2)). In cases of dysplasia, the difference in the number of cells collected between LBC and CVC is clear (Figure 1(c1), Figure 1(c2)). Keratinized dysplastic cells showed cytoplasmic thickening with staining by deep orange G, nuclei with irregular contours, irregular chromatin distribution, pyknosis and an increased N/C ratio with CVC (Figure 1(c1)). With LBC, keratinized dysplastic cells appeared solitary with a torn appearance of cytoplasmic margins, but brightness was higher than in CVC. Nuclear structure was clearer with LBC than with CVC (Figure 1(c2)). In OSCC, diversity of cytoplasmic staining was observed in both methods (Figure 1(d1), Figure 1(d2)). Deep atypical cells and superficial atypical cells were appeared intermingled in CVC (Figure 1(d1)), and the two types of atypical cells each formed small clusters in LBC (Figure 1(d2)). Thickened cytoplasm with deep blue coloration, deep dysplastic cells with dense chromatin, thickening of nuclear contours and pyknosis were observed with CVC (Figure 1(d1)). Conversely, moderate thickening of nuclear contours and slight nuclear swelling was shown with LBC, but nuclear transparency was increased, and circularity was maintained (Figure 1(d2)).

1) Comparisons between lesions

Results of nuclear cytomorphological analysis with CVC are shown in Table 2. Nuclear area was significantly lower in normal (35.8 um²) and hyperkeratosis (35.7 um²) cases than in other lesions. Inflammation (175.8) cases showed higher hyperchromasia than normal (91.3), hyperkeratosis (87.1) or dysplasia cases (80.7), and OSCC (superficial cells, 102.8) showed higher hyperchromasia than those of all other lesions. Irregularity was significantly lower in normal cases (11.0) than in all lesions except OSCC (superficial cells, 11.7). No significant difference in irregularity was seen between dysplasia (12.8) and either layer of OSCC. Regarding circularity, OSCC (deep cells, 0.72) displayed greater irregularity than hyperkeratosis (0.65) or dysplasia (0.66).

Results of LBC are shown in Table 3. Nuclear area was larger for inflammation (69.6 um²) than for all other lesions. Inflammation (156.3) cases showed higher

a: vs normal; b: vs hyperkeratosis; c: vs inflammation; d: vs dysplasia; e: vs OSCC (deep cells). *p < 0.05; **p < 0.01 (the Kruskal-Wallis test and Steel-Dwass test).

a: vs normal; b: vs hyperkeratosis; c: vs inflammation; d: vs dysplasia; e: vs OSCC (deep cells). *p < 0.05; **p < 0.01 (the Kruskal-Wallis test and Steel-Dwass test).

hyperchromasia than normal (122.2) and hyperkeratosis (71.5), dysplasia (76.9) and both layers of OSCC. No significant difference in irregularity was seen between dysplasia (14.4) and OSCC (superficial cells, 13.3). No significant difference in circularity was seen between OSCC (superficial cells, 0.76) and all other lesions.

2) Comparison between sample preparation methods

Results of cytomorphological analysis are shown in Table 4. In all 5 groups, LBC yielded significantly higher cell numbers than CVC. Comparing among cases, hyperkeratosis showed the lowest number of cells among all 5 groups with both CVC (21.27) and LBC (202.40). The LBC-to-CVC ratio for cell numbers ranged from 9.52 for hyperkeratosis to 1.87 for deep cells in OSCC. For nuclear area, the most prominent nuclear swelling was shown in inflammation with both CVC (57.99) and LBC (74.95), and cells were significantly larger with LBC than with CVC for normal, hyperkeratosis, and inflammation cases. As for hyperchromasia, denser, darkish changes were seen for inflammation with both CVC (173.59) and LBC (157.36), next to deep cells in CVC, and hyperkeratosis, inflammation, dysplasia and OSCC cases showed significantly bigger cells with LBC than with CVC. Irregularity was prominent in inflammation (23.64) and OSCC (deep cells, 18.18) with LBC, and cells of normal mucosa and OSCC (deep cells) were significantly bigger with LBC than with CVC. No significant difference in circularity was seen between CVC and LBC among all 5 groups.