This is phase II case-comparant diagnostic accuracy study. Its protocol is registered on ClinicalTrials.gov under identifier NCT04955197. The study was approved by the Research Ethics Committee (CREC), Faculty of Dentistry-Cairo University, with an approval number 19–9-22. The purpose of the study was described in details for the subjects before participation, followed by obtaining an informed consent.

The study was held at the Faculty of Dentistry, Cairo University and the National Cancer Institute from December 2020 to November 2021. The included study participants were enrolled in the study in a consecutive order of all patients evaluated for eligibility at the study location and satisfying the inclusion criteria. A total sample of 92 adult patients were divided into four groups: 23 participants with clinically normal oral mucosa and no evidence of any cancer risk factor (family history of cancer, cigarettes smoking and shisha, alcohol consumption), 23 participants with clinically normal oral mucosa and evidence of cancer risk factors, 23 patients with OPMLs, and 23 patients recently diagnosed with OSCC and not receiving yet any treatment. The exclusion criteria in the control groups were as follows: systemic disease and recent viral infection.

Questionnaire administration

Face-to-face interviews were performed with the study participants to obtain detailed information regarding their demographics, medical status, and cancer risk factors.

Clinical examination

All patients were subjected to a detailed intraoral and extraoral examination according to the National Institute of Dental and Craniofacial Research (2013) [11]. Subjects with oral lesions were assessed as follows:

◦ OSCC group: The tumor site was recorded, and grade was obtained after histopathologic examination.

◦ OPML group: The site and size of the lesion were recorded. The type of oral lichen planus was mentioned as; atrophic, erosive or papular. While, the leukoplakia cases were described as homogenous, non-homogenous, or proliferative verrucous leukoplakia.

◦ For control groups, no oral lesions were detected on conventional tactile and visual examination.

Index testOral exfoliative cytology procedure

All the cytological samples were taken at the same day of tissue biopsy in both OPMLs and malignant groups to avoid disease progression bias. The mechanical scraping of the oral mucosa was done by the use of cyto-brush. The sample was obtained from the mucosal lesion of the compared groups. While the buccal mucosa was the target of the cytological sampling in the risk factors and the control groups as it is an easily accessible tissue for sampling in a minimally invasive manner and does not cause any stress on study subjects, additionally, most of the encountered lesions were located at the buccal mucosa. For lesions that affect multiple sites — seen in the OPML group — the most suspicious and representative site was selected.

The sites of cytological smear were cleaned with a sterile cotton before sampling to remove viscous saliva covering the lesion and any debris from the oral cavity. Area was anesthetized with topical lignocaine spray.

The head of the cyto-brush was twisted to be perpendicular to the handle to facilitate the scrapping action and to direct pressure application. The brush was firmly applied with force against the mucosal lesion, and then pressure was applied while rolling the brush over the lesion area for 10 full turns until the bristles curled or pin point bleeding spot was evident [12]. The collected cells were smeared onto a labelled glass slide by rolling the cyto-brush in a continued motion from one end of the slide to the other (technique is shown in supplementary file 1). The slide was immediately immersed in 95% ethanol for fixation and stained with Papanicolaou stain (Pap stain) [13]. The Pap stain is the widely used cytological staining technique and is the preferred modality for demonstrating MN. It provides a distinct nuclear and cellular staining with highly defined nuclear details and cytoplasmic transparency [14].

Interpretation

The samples were evaluated and interpreted in reference to the alterations at the cytological level defined by the cytological features of malignancy and MN. The slides were completely screened by the cytopathologist for identification of any cytological abnormalities (negative or positive).

The cytological criteria of dysplasia or malignancy

The general features of malignancy in the cytological slides were reported as follows: high cellularity, increased nuclear/cytoplasmic ratio, nuclear hyperchromasia, discohesiveness of cells, nuclear membrane abnormalities, cellular pleomorphism, anisonucleosis, prominent nucleoli, and irregular mitoses.

Assessment of cytological features was performed on the basis of individual cells along with comparison between different cells. The sample must have a sufficient number of well-preserved cells (at least 30 well-preserved cells) [15]. The uniformity of the features was recorded as benign conditions Fig. 1, whereas pleomorphism was described as dysplasia/malignancy Figs. 2, 3,  4 and 5 [16].

Micronuclei

For a sample to be suitable for analysis, it has to meet the following criteria specified by Tolbert et al. (1991) which are as follows [17]:

1.

Intact cytoplasm and relatively flat cell position

2.

Little or no overlap with adjacent cells

3.

Little or no debris

4.

Nucleus normal and intact, nuclear perimeter smooth and distinct

5.

Fields having at least 1000 epithelial cells

The recommended criteria for the identification of a micronucleus were described by Tolbert et al. (1991) as follows [17]:

1.

Rounded smooth perimeter

2.

Less than one-third the diameter of the associated nucleus

3.

Staining intensity similar to that of the nucleus

4.

Texture similar to that of nucleus

5.

Same focal plane as nucleus

6.

The absence of overlap with, or bridge to, the nucleus

Cells with structures completely fulfilling the abovementioned MN criteria are counted high certainty. Those with objects slightly deficient in MN criteria 3, 4, and 5 and satisfying all of the other criteria are regarded medium certainty. Those assigned medium or high certainty are recorded as MN [17].

The cellular evaluation was performed using optic microscope with magnification (10 × 10) and (40 × 10), and the presence of micronuclei in all subjects was recorded (yes/no). Slides with at least 30 well-preserved cells (i.e., not obscured by blood or exudate or necrosis) are considered adequate for cytological evaluation.

Overlapping or clumped cells were excluded from the analysis. Also, hypocellularity in the slides was removed from analysis. Only cells with intact cytoplasmic border were included in the MN assessment.

The present study recorded MN as frequency (yes/no) rather than count (mean ± SD) in other studies. In reference to the systematic review and meta-analysis by de Geus et al. (2018), wide variations in MN score were observed among different studies. The baseline MN frequency was one of the most significant variable; the calibration of the upper limit of the baseline was not accurately estimated for a given population [18].

The histopathological assessment

Scalpel biopsy either incisional or excisional was taken from the highly suspicious areas in the OPMLs group. The histopathological picture is still the gold standard utilized for confirmation of the diagnosis and evaluation of the dysplastic changes [14].

OPML included leukoplakia diagnosed clinically and histologically or oral lichen planus diagnosed according to the modified WHO diagnostic criteria [19].

The dysplasia in the OPMLs was graded in reference to the squamous intraepithelial neoplasia/dysplasia (SIN/dysplasia) classification (2005) where mild dysplasia (SIN1) was recorded as low-grade dysplasia, while moderate dysplasia (SIN2), as well as severe dysplasia (SIN3), was categorized as high-grade dysplasia [20]. The OSCC was categorized based on Border’s classification into well differentiated (grade I), moderately differentiated (grade II), and poorly differentiated (grade III) [21].

Statistical methods

The sample size was calculated based on the previous work by Upadhyay et al. (2019) [22]. The statistical test used was the independent t-test with the power of the study 0.80 and level of significance of 0.05. Numerical data were presented as mean and standard deviation values. Categorical data were presented as frequency and percentage values and were analyzed using Fisher’s exact test followed by multiple pairwise comparisons using z-tests with Bonferroni correction. Diagnostic accuracy was assessed using ROC curve analysis. The significance level was set at p ≤ 0.05 within all tests. Statistical analysis was performed with R statistical analysis software version 4.1.2 for Windows [23].