Cancers of the oral cavity (lip, oral tongue, gingivae, floor of mouth, hard palate, and other mouth tissues including buccal mucosa) are amongst the most common worldwide, with an estimated annual burden of over 300,000 incident cases. Most oral cancers (>75%) are attributable to cigarette smoking, alcohol drinking, and chewing of areca nut/betel quid with or without tobacco, and very few are related to human papillomavirus infection. Oral cancer incidence geographically tracks with the prevalence of these risk factors and is notably high in the Indian subcontinent (due to tobacco chewing and smoking) and southeast Asia (due to betel quid chewing without tobacco and smoking).

The current standard for screening for oral precancer/cancer is visual and tactile examination by an expert for the presence of clinical/visual lesions (leukoplakia, erythroplakia, and oral submucous fibrosis). Such visually identified lesions are further triaged based on clinical impression for a biopsy to determine histopathologic presence/grade of dysplasia. Several observations point to key limitations of oral cancer screening based on clinical impression-based biopsy of visually identified lesions, including the decision to biopsy a lesion, which lesion to biopsy, and where within the lesion to direct a biopsy. Thus, there is a need for tools for improved triage of visual precancers for biopsy and targeting areas for biopsy within a lesion for more effective risk stratification and better provision of care.

Two non-invasive methods hold promise for triage of lesions for biopsy-oral liquid-based cytology and high-resolution microendoscopy (HRME). Oral cytology provides a method to non-invasively sample visible oral lesions and holds promise to enable triage of lesions for biopsy. HRME utilizes optical fiber-based imaging in combination with the fluorescent contrast agent proflavine to image sub-cellular features in vivo in lesions/epithelial tissues, functionally an in situ biopsy.

The investigators propose to conduct a cross-sectional study to evaluate the clinical utility of these two non-invasive methods for detecting oral precancer and early oral cavity cancer- the performance of oral cytology and HRME as an adjunct for triage of visible lesions for biopsy and the performance of HRME as an adjunct to enable better within-lesion targeting of areas for biopsy. If successful, this study would facilitate the development of a non-invasive, 3-step algorithm for oral cancer screening: identification of lesions through visual inspection, triage for biopsy through cytology or HRME, and targeted within-lesion biopsy (if needed) through HRME.

The investigators propose to conduct a cross-sectional randomized study to evaluate the performance of oral cytology and HRME (separately) versus clinical impression for triage of lesions for biopsy and the performance of HRME-directed biopsies versus clinician-directed biopsies. Performance will be evaluated based on yield of histopathologic disease- defined as diagnosis of any dysplasia (encompassing mild dysplasia, moderate dysplasia, severe dysplasia, or cancer) and high-grade disease (encompassing moderate dysplasia or above).

Upon identification of visible lesions, study investigators will ask the clinicians to provide a clinical impression regarding whether or not they would biopsy each lesion. Patients with visual oral precancer will then be randomized to one of two arms: 1) a standard-of-care clinical-directed biopsy arm or 2) HRME-directed biopsy arm. The biopsy decision in the clinical-biopsy arm will be guided by clinical impression/ examination under white light and the biopsy decision in the HRME-biopsy arm will be guided by HRME features. However, clinical impression and HRME features will be collected in both arms. Such availability of data will enhance the sample sizes for comparisons of clinical site of biopsy versus HRME site of biopsy.

The study specific aims are:

Aim 1: To evaluate the performance of oral cytology or HRME for triage of visually-identified lesions for biopsy. For cytology evaluation, this aim will combine participants from both arms and thus be non-randomized. For HRME evaluation, this aim will utilize participants from the HRME-directed biopsy arm. The yield of histopathologic disease will be compared between clinicians' response as to whether or not they would biopsy a lesion versus oral cytology or versus HRME.

Aim 2: To evaluate the performance of HRME for within-lesion targeting of regions for biopsy. This aim will be a randomized comparison of diagnostic yields of histopathologic disease between the clinician-directed biopsy arm versus the HRME-directed biopsy arm.

Combining both arms, different cut points (HRME scores range from 0-100) will be evaluated for the identification of any dysplasia or high-grade disease.

Aims 1 and 2 together will validate the performance of two non-invasive methods of diagnosing oral precancer. Cytology samples the entire lesion and could be a better representation of underlying disease status of the lesion when compared to visually-guided triage, while HRME allows better targeting of within-lesion areas for biopsy than visual impression.

Aim 3: To evaluate the within-lesion agreement between the area of clinician-directed biopsy and HRME-directed biopsy. This aim will combine participants from both arms and thus be non-randomized. This aim is anticipated to provide a potential explanation for why HRME does or does not improve the yield of disease when compared to clinical impression of within-lesion regions for biopsy.

Study design and procedures: This is a cross-sectional randomized study of 400 patients with visual oral precancer (leukoplakia, erythroplakia). Patients will be randomized (1:1 ratio) into 2 arms: 1) clinician-directed biopsy based on clinical examination/ impression or 2) HRME-directed biopsy based on HRME features.

Standard-of-care screening for visual oral precancer (through their national screening program) will be conducted by ENT clinicians through standard visual and tactile examination. Those eligible (defined as individuals ages 30-80 years with visible lesions with at least 1 cm in greatest diameter) will be approached and offered participation with informed consent.

Pre-randomization, all consenting participants will provide detailed demographic and behavioral information and will receive (in order) digital images of visible lesions, recording of whether or not the clinician would biopsy each lesion, recording of clinically-guided area of biopsy (regardless of clinician's decision to biopsy a lesion), and recording of HRME and HRME-guided area of biopsy. Participants will be randomized individually to one of the two study arms.

1. In the clinician-directed biopsy arm, the clinician will take biopsy(s) from the predetermined location from the initial clinical assessment.
2. In the HRME-directed biopsy arm, the clinician will take biopsies from lesion areas based on the prerecorded HRME reading.

Oral cytology will be collected after the HRME procedure to avoid tissue changes through cytology.

Study sites: The study would be conducted at the National Taiwan University Hospital (NTUH), Taipei.

Inclusion Criteria:

• Individuals aged 30-80 years
• Individuals with visible oral precancerous lesions of at 1 cm in greatest diameter

Exclusion Criteria:

• Individuals who are undergoing current cancer treatment or had a cancer within the last 12 months
• Individuals who are unwilling or unable to provide informed consent to participate

Deidentified individual-level patient data will be shared by study investigators upon written request and review

Upon publication of study results

Deidentified data can be accessed by contacting the study team