A brief review of cytology in dentistry

Abstract

Oral cytology is a non-invasive adjunctive diagnostic tool with a number of potential applications in the practice of dentistry. This brief review begins with a history of cytology in medicine and how cytology was initially applied in oral medicine. A description of the different technical aspects of oral cytology is provided, including the collection and processing of oral cytological samples, and the microscopic interpretation and reporting, along with their advantages and limitations. Applications for oral cytology are listed with a focus on the triage of patients presenting with oral potentially malignant disorders and oral mucosal infections. Furthermore, the utility of oral cytology roles across both expert (for example, secondary oral medicine or tertiary head and neck oncology services) and non-expert (for example, primary care general dental practice) clinical settings is explored. A detailed section covers the evidence-base for oral cytology as a diagnostic adjunctive technique in both the early detection and monitoring of patients with oral cancer and oral epithelial dysplasia. The review concludes with an exploration of future directions, including the integration of artificial intelligence for automated analysis and point of care ‘smart diagnostics’, thereby offering some insight into future opportunities for a wider application of oral cytology in dentistry.

Introduction and history of cytology

Cytology (or cytopathology) is a discipline of pathology examining individual cells collected using various methods (for example, brush, fine-needle aspiration, urine collection) to diagnose diseases. Cytology is currently used across medicine to help diagnose a number of diseases and continues to play a critical role in cancer prevention (that is, screening and early detection), particularly in resource-limited settings.¹

The history of cytology begins with the invention of the microscope and the study of human cells.^2,3,4,5,6 One of the earliest reports of cytology to detect malignant cells published by Lionel S. Beale (King’s College, London, 1860s),⁷ but it wasn’t until the 1930s and 40s when George Papanicolaou, considered the pioneer of modern cytology,⁸ coined the term ‘exfoliative cytology’ in the context of cervical neoplasia.^9,10,11 He described the characteristic features of cancer cells, including variability in nuclear size, increase in nuclear-to-cytoplasmic ratio and abnormal distribution of chromatin, and this led to the development of the pap-test (or pap-smear), revolutionising cervical cancer screening and leading to significant reduction in the burden of cervical cancer.

These results led to an exploration of cytology across other cancer types,^{12,13,14,15,16,17,18} including oral cancer. The evaluation of oral cytology began in 1951 with studies in the United States by Paul Montgomery in Ohio and Samuel Charles Miller in New York City that initially characterised exfoliative cytology of normal mucosa,^19,20 and subsequently to ‘leukoplacia’ and oral carcinoma in two case series.^21,22 This spurred a flurry of publications in the 1950s and 60s, largely reports of case series (reviewed by Tiecke),²³ and promoted the utility of cytology to facilitate the early detection of oral squamous cell carcinoma.²⁴ Improvements in oral cytology specimen collection and processing were made in the 1980s and 90s. The introduction of a cytological brush (that is, the cytobrush) provided a higher yield than the wooden spatula²⁵ and facilitated a more even spread of cells onto glass slides.²⁶ Automation in cytological image analysis improved the ability to quantify the morphological changes, such as nuclear to cytoplasmic area.²⁷ The introduction of quantitative deoxyribonucleic acid (DNA) cytometry using stoichiometric DNA staining (eg Feulgen stain) also assisted in distinguishing between normal and abnormal (dysplastic or malignant) smears based on the premise that malignant cells have abnormal DNA content.^28,29 The introduction of liquid-based cytology (LBC) to improve the accuracy of oral cytology demonstrated improvements in ‘sample preservation, specimen adequacy, visualisation of cell morphology and reproducibility’.³⁰ Further modification in brush technology led to ‘transepithelial’ sampling (that is, the ability to sample cells from all layers of the epithelium), although the ability to consistently collect basal and parabasal cells has been reported to be questionable.³¹ Research studies in the 1990s began to explore the utility of molecular biomarkers in oral cytology,³² particularly in the surveillance of patients with a history of oral squamous cell carcinoma (OSCC) (that is, detecting occult disease and predicting cancer recurrence).^33,34 Cytology has also been applied to nodal disease in regionally metastatic OSCC through samples procured by fine needle aspiration but this topic is beyond the scope of our review.

Beyond the cytology of OSCC, studies have described the cytology of other benign and malignant oral mucosal diseases or conditions, such as pemphigus vulgaris, viral infections (herpes simplex virus [HSV], Epstein-Barr virus, varicella zoster virus [VZV]), primary oral melanoma and radiation-related mucosal changes.^{35,36,37,38,39,40,41,42,43,44,45}

The aim of this concise narrative review is to provide an updated discussion on current applications for oral cytology, both in primary care settings, such as general dental practice, and in secondary (for example, oral medicine) and tertiary (for example, head and neck oncology) services, as well as to project future trends for clinical expansion of cytology in clinical practice catalysed by point of care data collection and artificial intelligence (AI)-enhanced analysis.

Oral cytology today

Oral cytology remains focused on the early diagnosis of OSCC, although there are other indications for the diagnosis of infectious and inflammatory oral mucosal diseases. The major advantage of oral cytology is that sample procurement is simple, fast and minimally invasive. This contrasts with tissue biopsy, which requires local anaesthesia and leads to post-surgical discomfort. Primary care ‘frontline’ clinicians, such as general dentists (and dental hygienists or other medical colleagues in primary care settings) with variable training and experience in the diagnosis of oral mucosal diseases encounter a myriad of benign and potentially malignant diseases with overlapping clinical features. This circumstance can create a diagnostic challenge and leads to uncertainty about whether the patient should be referred to a secondary care setting (for example, an oral medicine service). Cytology can be useful in the interrogation or ‘triage’ of abnormal mucosal examination findings that are detected by conventional oral examination and that arouse ‘suspicion’ for a possible early malignancy, or an infectious disease, such as candidosis (also known as candidiasis). Such non-expert clinicians can easily be trained to perform cytological sampling, thereby expanding the pool of diagnosticians and optimising referral. Table 1 lists some broad indications and contraindications for the use of oral cytology. In addition to oral potentially malignant disorders (OPMDs) and infectious diseases, cytology of vesiculobullous disorders, such as pemphigus vulgaris, may also be diagnosed by oral cytology in situations where a tissue sample cannot be easily procured, although the rarity of this disease in general dental practices suggests very limited applicability. Before expanding on the specific indications for oral cytology, it is important to appreciate some of the current methods for collecting, processing and interpreting oral cytological samples.

Table 1.

Indications and contraindications for oral cytology

Indications	Contraindications
• Triage of OPMDs for OSCC/OED (for lesions detected by non-expert, for example, general dentists) and/or for monitoring of patients with a history of OSCC or OED by experts (for example, oral medicine specialists) • Triage of oral lesion(s) that arouse suspicion for oral infectious diseases (for example, candidosis, herpes infections [HSV/VZV] and possibly others).	• Lesions arousing a high index of suspicion for OSCC should be promptly referred to an expert for tissue biopsy • Pigmented lesions and non-epithelial lesions (for example, submucosal lesions) are not indicated for cytology • Lesions with a clear-cut clinical diagnosis of pseudomembranous candidosis, denture stomatitis, herpes labialis/intra-oral herpes do not require cytology for confirmation.

Sample collection

Cytological samples of the mucosa may be collected either by scraping with a spatula or wooden tongue depressor or by rotation of a brush. A spatula or tongue depressor can collect exfoliative samples, akin to a ‘broom sweep’, and are ideally suited to diagnosing superficial mucosal infectious diseases. Brushes come in various forms and some, such as Oral CDx (CDx Diagnostics, New York, USA) and Orcellix (Rovers Medical Devices, Oss, Netherlands), have the advantage of capturing a transepithelial sample, which is important when a full thickness epithelial sampling, including basal and parabasal cells, is required to detect atypical cells in dysplastic epithelial lesions. This technique is commonly referred to as a ‘brush biopsy’. Figure 1 compares conventional exfoliative and transepithelial sampling methods.

Fig. 1.

Comparison of oral cytology methods. Created with BioRender.com

Processing

For conventional cytological preparations, the collected cells are immediately smeared onto a glass slide by the clinician and fixed using a commercially available cytology fixative (for example, 95% ethyl alcohol, 2.5% polyethylene glycol and 2.5% purified water in a single use packet) which preserves cellular morphology. For liquid-based cytological (LBC) platforms, such as ThinPrep (Hologic, Massachussets, USA) and SurePath (Becton, Dickinson and Company, New Jersey, USA), the clinician places the specimen (that is, the brush head) directly into a liquid fixative. Methanol-based buffered fixatives, such as CytoLyt (Hologic, Massachussets, USA) which is typically used in the ThinPrep platform, is both haemolytic and mucolytic. Ethanol-based fixatives often contain a small percentage of formaldehyde. In both cases, the specimen is typically shipped to a commercial laboratory for further processing and interpretation.

For LBC samples, the cells are gently dispersed from the brush head by vortexing followed by centrifugation of the cells to form a pellet. Cell pellets can be either resuspended and sprayed in a homogeneous monolayer onto a glass slide or formed into a cell block which can be paraffin-embedded and processed akin to standard tissue specimens for histopathology. The benefits of LBC processing outweigh those of conventional smears with consistent production of high-quality slide preparation with minimal overlapping of cells. A comparison of conventional and LBC methods is summarised in Table 2.

Table 2.

A comparison of conventional and liquid based cytology

Conventional cytology

Liquid-based cytology

• Samples are collected with a spatula or brush which is smeared onto a glass slide, fixed and sent to a pathology laboratory for staining, interpretation and reporting • No special instrumentation required. Glass slides and fixative packets can be procured from an oral pathology laboratory. Any oral pathology laboratory can process samples • Sampled cells are from the superficial epithelial layers. If the sample is from the keratinised mucosa (that is, gingivae) or highly keratinised lesions (that is, thick white lesions), the sample may contain excess keratin and an inadequate cell sample • Cells that are directly smeared to the slide can clump leading to impurities like blood cells, mucus and debris overlapping and deforming the cells • Fewer cells are available for interpretation and reporting • The slide can be used only for single testing.

• Samples are collected with a specialised brush, the brush head is placed into a vial of a buffered fixative and sent to a pathology laboratory, where the brush head is rinsed to remove the cells and the cells undergo automated processing for staining interpretation and reporting • Specialised brushes and sampling kits may be required depending on commercial platform (may not be available in every country). Only specific commercial laboratories are equipped with specific LBC processing • Samples from commercial brushes are typically transepithelial which allows the evaluation of basal and parabasal cells and optimises the identification of atypical cells in both OSCC and OED • Through the automated processing, the impurities are removed and the smear is uniformly spread on the slide with minimal overlap and mechanical deformation • Cells that are rinsed from the brush provides a higher proportion of representative epithelial cells for staining and interpretation • Multiple slides can be generated from a single sample and can be used for other analyses, such as molecular analysis (for example, genomics, epigenomics, proteomics). Some platforms also create cell blocks which can be processed similar to histopathology.

The cytological slides then undergo various staining protocols depending upon the disease being investigated. Examples of stains include modified Papanicolaou or Feulgen staining (for OSCC/oral epithelial dysplasia [OED]), periodic acid-Schiff or Gram stain (for candidosis), and Giemsa (for HSV/VZV). Some commercial laboratories also perform immunohistochemical staining for diagnostic biomarkers (for example, CDX Diagnostics, Sufferin, New York USA, which includes coverage for CDX2, MUC2, A-Methylacyl-CoA Racemase [AMACR], P16, P53, KI67, HSV and Helicobacter pylori).

Microscopic interpretation and reporting

The final step is the interpretation of cytological samples by a pathologist and the microscopic analysis is typically designated as one of four categories: negative (no epithelial abnormality/benign); atypical (abnormal epithelial changes of uncertain diagnostic significance; positive (definitive cellular evidence of epithelial dysplasia or malignancy); or inadequate (incomplete specimen).⁴⁶ To be clear, the brush biopsy cannot render a definitive diagnosis completed by tissue biopsy and histology, but rather provides the clinician with additional data to facilitate decision-making. Traditionally, the pathologist views the stained slide under the microscope; however, this can be subjective and time-consuming. The implementation of automated cytological assessment, including cytometry, image recognition software and machine learning algorithms are now used to facilitate interpretation.^47,48,49,50 However, there are challenges to automated assessment and these include the data management of high-magnification digital scanning of cytological slides, the risk of over-fitting the data used to train an algorithm, the inability to learn from new data without retraining the algorithm, and issues related to ethics, privacy and data security. There is also a lack of cost-effectiveness studies compared to traditional pathologist reporting and we hypothesise that automation may reduce costs.

Oral cytology in the diagnosis of OPMDs

OPMDs are defined as ‘any oral mucosal abnormality that is associated with a statistically increased risk of developing oral cancer’ and they include leukoplakia (OL), erythroplakia (OE), oral submucous fibrosis (OSF), proliferative verrucous leukoplakia, oral lichen planus (OLP) and oral lichenoid lesions (OLLs).⁵¹ The clinical diagnosis of OL/OE are made based upon clinical features which have a wide spectrum and overlap with those of a number of mucosal diseases with no malignant potential. Highly suspicious lesions, such as advanced stage OSCC aside, and for which prompt referral is indicated, it can be challenging for ‘frontline’ clinicians to perform chairside risk-stratification and referral decisions based on clinical features alone.⁵² The current gold standard definitive diagnosis and the basis for treatment decisions is tissue biopsy and histological evaluation. However, tissue biopsy is invasive and requires referral to an expert. Triage in primary care using an accurate adjunctive technique, a non-invasive surrogate to tissue biopsy and histopathology, yet without the associated morbidity, to optimise the correct referral of higher-risk patients (and reduce over-referral of patients with no to little risk), is needed from both a patient management perspective and health care costs. In secondary or tertiary care clinics (that is, oral medicine or head and neck oncology services), where patients with a history of OED or OSCC undergo surveillance, a non-invasive adjunct to monitor such higher-risk patients for new or recurrent OSCC or worsening OED also reduces morbidity. Management of patients with mild epithelial dysplasia, by far the largest subset of patients with OED, and with the lowest risk for cancer development (approximately 5% transform),⁵³ is controversial. If the clinician choses a ‘watch and wait’ strategy, performing serial invasive tissue biopsies is not an optimal strategy, particularly given the high rates of disagreement by pathologists of mild OED.⁵⁴ With the exception of a small Fanconi anaemia pilot study,⁵⁵ the current literature lacks longitudinal surveillance studies, making it more challenging to develop effective treatment protocols for these patients. To address this essential gap, a current multi-institution prospective longitudinal surveillance cytology study in a cohort of patients with a history of OED and OSCC is now in progress.⁵⁶

Numerous studies have explored the accuracy of cytology for the detection of OED or OSCC in OL/OE. Such ‘accuracy’ studies enroll subjects with OPMDs detected following a conventional oral examination. Cytological results, typically dischotomised into ‘positive’ versus ‘negative’, are then compared to the matched gold standard histologic diagnosis. Positive cytology is largely based on morphological criteria and ‘true positives’ are matched histologically demonstrating any grade of OED or OSCC. A recent Cochrane review⁵⁷ pooled the results from 24 datasets generated by accuracy studies and analysed a total of 2,892 OPMDs (that is, OL/OE), with 1,496 definitively diagnosed as OED/OSCC, and reported an overall sensitivity of 0.90 (95% confidence interval [CI]: 0.82–0.94) and specificity, 94 (95% CI: 0.88–0.97) for the detection of OED/OSCC by cytology. These results met a moderate-certainty level of evidence. The review was further stratified to explore any differences with respect to groupings based on collection methods (datasets based on brush collection [16 sets] versus scrapings [three sets]), and processing (conventional morphological analysis [19 sets] versus DNA cytometry [five sets]). The type of collection method showed no statistical difference and the DNA cytometry datasets showed a lower sensitivity, largely due to the false-negative test results in patients with mild epithelial dysplasia (the group with the lowest risk for malignant transformation). When the threshold for a ‘true positive’ was increased to moderate dysplasia or worse, the sensitivity of DNA cytometry increased to 89%. Overall, the authors of the Cochrane review acknowledged that none of the studies were performed by ‘frontline clinicians’, yet contended that cytology might be indicated for the triage of patients detected with low-risk/innocuous lesions in a general dental practice.

A 2017 systematic review and meta-analysis report by the American Dental Association expert consensus group⁵⁸ evaluated the diagnostic accuracy of the adjuncts available for the evaluation of OL/OE by frontline general dentists. The authors concluded that, compared to other adjuncts, cytologic testing offered the highest levels of accuracy, with pooled sensitivity of 0.92 (95% CI: 0.86–0.98) and specificity, 0.94 (95% CI: 0.88–0.99). Based on these promising results, the group recommended that in light of the high-test accuracy against the gold standard histological outcomes, that cytology has utility and is indicated in general dental practice when tissue biopsy is not feasible.

Neumann et al. recently reported the first study exploring the utility of oral cytology in 40 dentists in Germany (four of the dentists were specialised in oral surgery).⁵⁹ A total of 670 patients with 814 oral lesions, including lesions that were clinically suspicious for carcinoma, underwent ‘brush biopsies’ (cytological samples were collected by the Orcellex brush [Rovers Medical Devices B.V., Oss, The Netherlands] and processed using LBC). Compared to histological endpoints, cytology demonstrated 100% sensitivity and 85% specificity for detecting carcinoma (OED was not reported). Unfortunately, the authors did not stratify by clinician type to evaluate the effect on cytology performance by excluding the dentists who were specialised in oral surgery. Despite these methodological limitations, the study demonstrates the feasibility for general dentists to perform cytology in private dental practice.

The utility of cytology has also been explored for other OPMDs. For OSF, a disorder with progressive oral fibrosis and burning which affects habitual areca nut chewers (largely found in immigrants from Southeast Asia), there is conflicting support for cytology. One study recommended cytology for the monitoring of patients with OSF because, compared to controls, OSF samples were deemed atypical.⁶⁰ In contrast, a second study reported normal DNA cytometry in 27 cases of OSF, although the authors reported that abnormal cytometry was present in two of three cases of OSF with concomitant oral leukoplakia.⁶¹ Beyond a few selected cases, there have been no specific studies employing cytology for the diagnosis or monitoring of patients with proliferative verrucous leukoplakia.

OLP and OLLs are relatively common in the general population and classified as OPMDs because of their risk, albeit small, for cancer development.⁶² Patients with these disorders, particularly non-reticular and unilateral presentations of OLP or OLLs, can be challenging for clinicians to distinguish from those of OL/OED.⁶³ Even for expert clinicians who monitor these patient populations, it is difficult to predict which patients with clinical features of OLP/OLL harbour dysplasia or will evolve to malignancy. Using cytology to assist in such risk stratification has been explored and suggests utility in both primary and secondary/tertiary care settings. Maraki et al. demonstrated that samples from patients with OLP deemed positive by conventional cytology were aneuploid by DNA image cytometry.⁶⁴ Subsequently, Yarom et al. reported 16 of 57 (28.1%) OLP patients undergoing cytological sample collection were aneuploid (by fluorescent in situ hybridisation for chromosomes 2 and 8).⁶⁵ Subsequent serial brushing on 30 patients, with an average follow-up of almost four years, showed that the aneuploid fluorescence in situ hybridisation (FISH) signature remained consistent over time and heralded cancer development in three patients.⁶⁶ Idrees et al. recently assessed the efficacy of using LBC to differentiate dysplastic versus non-dysplastic patients presenting with a clinical diagnosis of OLP or OLL. Using LBC for both conventional cytology and immunohistochemical (IHC) assays of cell blocks (IHC for Ki67, BAX, NF-κB-p65 and AMACR), the authors reported high accuracy in detecting dysplasia in OLLs.⁶⁷

More recent research developments in oral cytology for OPMDs over the last decade include the exploration of various novel ‘omics’ biomarkers, such as gene expression profiling. Some examples include panels of genomic markers, such as ribonucleic acid (RNA) expression (both messenger RNA and micro RNA expression)^68,69 and epigenetic markers.^70,71,72 Such biomarkers, typically batched as panels, are generated to ‘classify’ OSCC from a cohort of cytological samples obtained from patients with histologically confirmed OSCC and normal tissue controls. Machine learning algorithms are used to validate and test classification models and AI utilising deep learning algorithms is currently being explored (see discussion below). Unfortunately, only a few studies included study cohorts also comprising a blend of patients with benign and dysplastic OPMDs, hindering the ability to develop models to accurately stratify higher- versus lower-risk patients in a general population. Along this theme, a large international prospective cytology study involving 999 patients^73,74,75,76 recorded over 200 cellular features related to biomarker expression, nuclear parameters and cellular morphology per cell, encompassing ~2,000 cells per patient, and representing nearly 13 million indexed cells. The study revealed 128 ‘features’ to allow better cellular discrimination across the diagnostic histological spectrum of disease.⁷⁴ The power of the database gathered by the effort is documented in a number of papers^{54,73,74,75,76,77} wherein detailed comparisons are made across a broad range of protein biomarkers, nuclear parameters, risk factors, patient demographics and lesion appearance categories. Further, machine learning classifiers (that is, k-nearest neighbor) were also trained to identify and count cytological cell phenotypes correlating to OED and OSCC and showed that cell size and F-actin distribution (cytoskeleton microfilaments) differentiated cell phenotypes. Importantly, changes to cellular actin have been implicated in cancer initiation and progression.^78,79 Models trained to discriminate histopathology across multiple dichotomous splits, including ‘early’ (where the model is geared to include mild and moderate OED in the ‘disease’ group, that is, benign | mild/moderate/severe OED and OSCC) and ‘late’ disease models (where the model is geared to include mild and moderate OED with benign diseases, that is, benign, mild/moderate OED | severe OED and OSCC). As expected, the late disease model was more accurate than the early disease model with area under the curve (AUC) of 0.93 and 0.82, respectively. Such late disease modeling has been implemented using Clearcyte, a Health Canada-approved cytological platform, using a quantitative DNA-cytometry test, now with an additional automated machine learning algorithm (iClearcyte); this platform also demonstrated and high accuracy in detecting ‘high-grade’ disease (AUC: 0.99).⁴⁷ These results suggest the adaptation of modeling to improve the triage and surveillance of the entire spectrum of OPMD depending on the care setting, although the automation to reduce the time spent by cytopathologists is confirm these diagnostic interpretations will be key to the success of these emerging platforms.

Among head and neck cancers, human papillomavirus (HPV) + oropharynx squamous cell carcinoma (OPSCC) incidence rates are steadily rising⁸⁰ and both the feasibility and utility of using cytology for diagnosis have been explored. Such cancers typically involve the palatine or lingual tonsil/base of tongue cancers and dysplastic precursor lesions, similar to those preceding with OSCC (eg dysplastic OPMDs), are not found in OPSCC.^81,82 More recent studies in patients with HPV + OPSCC accessible to brush cytological collection (that is, a palatine tonsil mass visualised during conventional examination, or a base of tongue mass visualised by endoscopic examination by an otolaryngologist) do demonstrate cytological findings and presence of HPV (or surrogate p16 immunohistochemistry), which accurately correlate to histopathology.^83,84 Yet, primary care clinicians, such as general dentists, should promptly refer a patient with a palatine tonsil mass rather than sampling it for cytology. Furthermore, such clinicians cannot reliably visualise the base of tongue, let alone sample a mass. Benign oral HPV-related lesions, such as a viral papilloma, verruca vulgaris, condyloma accuminatum, or florid oral papillomatosis could feasibly be sampled by cytology and tested for HPV, although there would be little advantage in doing so, since high-risk HPV-genotypes are rarely encountered in these lesions and would not change current treatment strategies. Table 3 summarises the advantages and disadvantages of oral cytology for OPMDs and Figure 2 provides examples of cytology used for the diagnosis of an OPMD.

Table 3.

Advantages and disadvantages of oral cytology

Advantages

Disadvantages

• Fast and simple procedure requiring minimal expertise and training of clinicians. Painless in most cases (brushing ulcerated lesions may require topical anaesthesia). Minimal (if any) bleeding. Well-tolerated by patients • Cytology has been demonstrated in meta-analysis of accuracy studies to have a very high sensitivity and specificity against matched histology for OSCC/OED. It is the most accurate of all diagnostic adjuncts • Allows triage of lesions with a larger area that may not be adequately sampled by incisional biopsy (or biopsies) • Facilitates repeated non-invasive sampling of patients with a history of OSCC/OED undergoing monitoring/surveillance • Cell samples are amenable to molecular testing • The cost of consumables and processing are less than for histopathology • Cytology could be applied in lower resource settings where patients are detected during routine screening with asymptomatic OPMDs and who are referred to secondary care but do not follow through • Cytology may be able to help differentiate OLP/OLLs from OED and OSCC

• Does not provide a definitive diagnosis and is not a replacement for the gold standard tissue biopsy and histological evaluation. A positive or atypical cytology result necessitates additional referral for further evaluation by an expert • Leukoplakias with thick white appearance can be challenging to obtain an adequate sample for interpretation • Limited availability and access in many countries to commercially available cytopathology platforms and laboratories. Even if available, insurance companies may not cover the clinician fee or laboratory processing • Cost-benefit analyses for oral cytology in terms of survival and quality of life have not been performed.

Fig. 2.

a, b) Papanicolaou and periodic acid–Schiff-stained (PAP/PAS) conventional cytologic smears from two patients with candidosis, most likely from candida albicans. The presence of candidal hyphae indicate candidosis. Panel A reveals numerous candidal yeast forms (Y) and a few hyphae (H). Yeast forms alone are not consistent with a diagnosis of candidosis. Panel b reveals numerous characteristic filamentous hyphae indicating candidosis. c) PAP-stained transepithelial brush sample from an OPMD (erythroleukoplakia) which is positive for cellular evidence of epithelial dysplasia or carcinoma and characterised by abnormal variations in nuclear size and shape, abnormal variations in cell size, dyskeratosis and increased number of large nuclei. d) Subsequent incisional biopsy of the OPMD confirmed the diagnosis of moderate to focally severe oral epithelial dysplasia (stained with haematoxylin and eosin)

Oral cytology in the diagnosis of candidosis

Oral candidosis (or candidiasis) is the most prevalent mucocutaneous fungal infection, caused by Candida species.⁸⁵ This condition typically arises as a result of immune suppression, which may be either local or systemic, including extremes of age (such as newborns and the older person), diseases like HIV/AIDS, as well as prolonged use of systemic steroids and antibiotics.⁸⁶ Acute pseudomembranous candidosis, also known as thrush, is the most common form of oral candidosis and its classic presentation of wipeable whitish plaques makes for a simple clinical diagnosis requiring no further confirmatory diagnostic steps.⁸⁷ Other forms of candidosis, such as the non-wipeable white plaques of chronic hyperplastic candidosis, or the red patches associated with chronic erythematous candidosis, are less easy to diagnose clinically. In such cases where candidosis is the working diagnosis on a differential, it may be definitively diagnosed either by a culture-based method, commonly employed in medical practices, or by oral exfoliative cytology. The principal disadvantage of culturing for candidosis is that Candida species are commensal organisms, and a positive culture cannot distinguish the presence of candidal pseudohyphae, a sign of active infection, from the non-infectious yeast forms. Oral cytology can easily detect candidal pseudohyphae (Fig. 2).^85,88

Future directions for oral cytology

As discussed above, important developments in oral cytology derived from biomarker discoveries suggest promise to aid in the diagnosis and monitoring of patients with OPMDs. While these efforts search for essential insights into early signs of malignant transformation, there remains a key gap between the scientific discoveries and broad-scale implementation/acceptance of these new oral cytology-based approaches. Such barriers become even more dominant when these potential predictive cytological biomarkers are derived only from expensive and time-consuming -omics-based diagnostic infrastructure that also require ultra-sophisticated interpretation. Likewise, stresses placed on health care delivery and shortage of expert providers further limits the scalability of these new cell-based diagnostic capabilities. Oral cytological sampling is suited to the area of ‘smart diagnostics’ in that it has the potential to integrate the following four capabilities: i) point of care non-invasive testing; ii) multivariate measurements; iii) machine learning/AI diagnostic modeling; and iv) cloud-based intuitive reporting of results.

One of the initial embodiments of this smart diagnostic concept is shown in Figure 3. Here, panel A illustrates the microfluidic engine that allows for the replacement of normal remote lab infrastructure, thereby enabling sophisticated cytomics-on-a-chip tests to be completed chairside. Panel B reveals the sequence of steps starting with a non-invasive brush sampling or an oral lesion, placement of the brush in a stablised solution, transfer to the cartridge and inserting of the labcard into the analyser that then completes the analysis. This platform allows for the simultaneous acquisition of data elements, including 200 cellular features related to biomarker expression, nuclear parameters, cellular morphology and cell phenotypes.^48,49,89,90 The same platform has been used to develop a numerical index that is key to the delivery of a simplified summary result.^75,76

Fig. 3.

a) Here, a microfluidic engine is used for the processing of the cell suspension that is collected following a non-invasive brush biopsy, transferred to a buffer containing tube, agitate,d and then the cell suspension is introduced into a credit card-sized structure, which integrates many of the lab functions that are more traditionally used for remote lab measurements. b) Basic sequence of steps that, in the future, may be completed chairside en route to completing the oral lesion risk profiling alongside the production on an intuitive report

These next generation diagnostics are now positioned to serve as highly scalable in vitro diagnostic devices which harness the power of AI to exceed the performance of remote lab-based diagnostics.⁸⁹ Regulatory bodies like the Food and Drug Administration recognise the value of AI in enhancing treatment decisions and have provided guidance for AI-driven clinical tools, underscoring the significance of this integration in the context of oral cancer diagnosis and treatment planning.⁹¹ Looking ahead, smart cytology could expand its applications to encompass various other oral mucosal diseases.

The advent of smart diagnostics and implementing cytology has the potential to improve the ability to diagnose and predict disease, although we must remain cognisant of the need for carefully designed clinical validation efforts as we negotiate this emerging field.

Conclusions

The review has explored current and emerging applications for oral cytology and offers some insight into opportunities for a wider application of oral cytology for use in clinical decision-making in dentistry.

Key points.

The major potential application for oral cytology is as a diagnostic adjunct in the triage of patients with oral potentially malignant disorders.

Oral cytology is recommended over fungal culture for the initial diagnosis of patients with possible oral candidosis.

There is potential for oral cytology to be incorporated into both general dental practice and across expert clinic settings.