Profiling cancer risk in oral potentially malignant disorders-A patient cohort study

Deep neural network uncertainty estimation for early oral cancer diagnosis

BACKGROUND

Early diagnosis in oral cancer is essential to reduce both morbidity and mortality. This study explores the use of uncertainty estimation in deep learning for early oral cancer diagnosis.

METHODS

We develop a Bayesian deep learning model termed 'Probabilistic HRNet', which utilizes the ensemble MC dropout method on HRNet. Additionally, two oral lesion datasets with distinct distributions are created. We conduct a retrospective study to assess the predictive performance and uncertainty of Probabilistic HRNet across these datasets.

RESULTS

Probabilistic HRNet performs optimally on the In-domain test set, achieving an F1 score of 95.3% and an AUC of 96.9% by excluding the top 30% high-uncertainty samples. For evaluations on the Domain-shift test set, the results show an F1 score of 64.9% and an AUC of 80.3%. After excluding 30% of the high-uncertainty samples, these metrics improve to an F1 score of 74.4% and an AUC of 85.6%.

CONCLUSION

Redirecting samples with high uncertainty to experts for subsequent diagnosis significantly decreases the rates of misdiagnosis, which highlights that uncertainty estimation is vital to ensure safe decision making for computer-aided early oral cancer diagnosis.

Oral cancer and oral potentially malignant disorders in patients with Fanconi anemia - A systematic review

The purpose of the present study was to perform a systematic review focusing on oral squamous cell carcinoma (OSCC) and oral potentially malignant disorders (OPMD) in Fanconi anemia (FA) individuals. Electronic searches were undertaken in five databases supplemented by manual scrutiny and gray literature. Case reports and/or cases series were included. The searches yielded 55 studies describing 112 cases of OSCC (n = 107) and/or OPMD (n = 5) in FA individuals. The mean age at diagnosis of OSCC/OPMD was 27.1 (±9.6) years, and females (51.8 %) were slightly more affected. Ulcer (n = 37) or mass (n = 25) were described as clinical presentations for OSCC and OPMD. White lesions (n = 4) were the most common manifestation in OPMD. Tongue (47.2 %) was the most frequent location. Sixty-one (54.5 %) individuals underwent HSCT. Surgical resection (n = 75) was the main treatment adopted. The estimated rate of OPMD malignant transformation was 1.8 % and recurrences following OSCC excision occurred in 26.8 % of individuals. Overall, at 60 months of follow-up, the probability of survival fell to 25.5 % and at 64 months the probability of recurrence increased to 63.2 %. The present data support the need for strict surveillance of patients with FA, even in the absence of OPMD, for early OSCC detection and reduction of mortality.

Clinical assessment for the detection of oral cavity cancer and potentially malignant disorders in apparently healthy adults

BACKGROUND

The early detection of oral cavity squamous cell carcinoma (OSCC) and oral potentially malignant disorders (OPMD), followed by appropriate treatment, may improve survival and reduce the risk for malignant transformation respectively. This is an update of a Cochrane Review first published in 2013.

OBJECTIVES

To estimate the diagnostic test accuracy of conventional oral examination, vital rinsing, light-based detection, mouth self-examination, remote screening, and biomarkers, used singly or in combination, for the early detection of OPMD or OSCC in apparently healthy adults.

SEARCH METHODS

Cochrane Oral Health's Information Specialist searched the following databases: Cochrane Oral Health's Trials Register (to 20 October 2020), MEDLINE Ovid (1946 to 20 October 2020), and Embase Ovid (1980 to 20 October 2020). The US National Institutes of Health Trials Registry (ClinicalTrials.gov) and the World Health Organization International Clinical Trials Registry Platform were searched for ongoing trials. No restrictions were placed on the language or date of publication when searching the electronic databases. We conducted citation searches, and screened reference lists of included studies for additional references.

SELECTION CRITERIA

We selected studies that reported the test accuracy of any of the aforementioned tests in detecting OPMD or OSCC during a screening procedure. Diagnosis of OPMD or OSCC was provided by specialist clinicians or pathologists, or alternatively through follow-up.

DATA COLLECTION AND ANALYSIS

Two review authors independently screened titles and abstracts for relevance. Eligibility, data extraction, and quality assessment were carried out by at least two authors independently and in duplicate. Studies were assessed for methodological quality using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2). We reported the sensitivity and specificity of the included studies. We provided judgement of the certainty of the evidence using a GRADE assessment.

MAIN RESULTS

We included 18 studies, recruiting 72,202 participants, published between 1986 and 2019. These studies evaluated the diagnostic test accuracy of conventional oral examination (10 studies, none new to this update), mouth self-examination (four studies, two new to this update), and remote screening (three studies, all new to this update). One randomised controlled trial of test accuracy directly evaluated conventional oral examination plus vital rinsing versus conventional oral examination alone. There were no eligible studies evaluating light-based detection or blood or salivary sample analysis (which tests for the presence of biomarkers for OPMD and OSCC). Only one study of conventional oral examination was judged as at overall low risk of bias and overall low concern regarding applicability. Given the clinical heterogeneity of the included studies in terms of the participants recruited, setting, prevalence of the target condition, the application of the index test and reference standard, and the flow and timing of the process, the data could not be pooled within the broader categories of index test. For conventional oral examination (10 studies, 25,568 participants), prevalence in the test accuracy sample ranged from 1% to 51%. For the seven studies with prevalence of 10% or lower, a prevalence more comparable to the general population, the sensitivity estimates were variable, and ranged from 0.50 (95% confidence interval (CI) 0.07 to 0.93) to 0.99 (95% CI 0.97 to 1.00); the specificity estimates were more consistent and ranged from 0.94 (95% CI 0.88 to 0.97) to 0.99 (95% CI 0.98 to 1.00). We judged the overall certainty of the evidence to be low, and downgraded for inconsistency and indirectness. Evidence for mouth self-examination and remote screening was more limited. We judged the overall certainty of the evidence for these index tests to be very low, and downgraded for imprecision, inconsistency, and indirectness. We judged the evidence for vital rinsing (toluidine blue) as an adjunct to conventional oral examination compared to conventional oral examination to be moderate, and downgraded for indirectness as the trial was undertaken in a high-risk population.

AUTHORS' CONCLUSIONS

There is a lack of high-certainty evidence to support the use of screening programmes for oral cavity cancer and OPMD in the general population. Frontline screeners such as general dentists, dental hygienists, other allied professionals, and community healthcare workers should remain vigilant for signs of OPMD and OSCC.

Deep Learning Predicts the Malignant-Transformation-Free Survival of Oral Potentially Malignant Disorders

Simple Summary

Mouth cancer is the most common malignancy in the head-and-neck region. Usually, these tumors develop from white lesions in the mouth that appear long before cancer diagnosis. However, platforms that can estimate the time-factored risk of cancer occurring from these diseases and guide treatment and monitoring approaches are elusive. To this end, our study presents time-to-event models that are based on machine learning for prediction of the risk of malignancy from oral white lesions following pathological diagnosis as a function of time. These models displayed very satisfactory discrimination and calibration after multiple tests. To facilitate their preliminary use in clinical practice and further validation, we created a website supporting the use of these models to aid decision making.

Abstract

Machine-intelligence platforms for the prediction of the probability of malignant transformation of oral potentially malignant disorders are required as adjunctive decision-making platforms in contemporary clinical practice. This study utilized time-to-event learning models to predict malignant transformation in oral leukoplakia and oral lichenoid lesions. A total of 1098 patients with oral white lesions from two institutions were included in this study. In all, 26 features available from electronic health records were used to train four learning algorithms—Cox-Time, DeepHit, DeepSurv, random survival forest (RSF)—and one standard statistical method—Cox proportional hazards model. Discriminatory performance, calibration of survival estimates, and model stability were assessed using a concordance index (c-index), integrated Brier score (IBS), and standard deviation of the averaged c-index and IBS following training cross-validation. This study found that DeepSurv (c-index: 0.95, IBS: 0.04) and RSF (c-index: 0.91, IBS: 0.03) were the two outperforming models based on discrimination and calibration following internal validation. However, DeepSurv was more stable than RSF upon cross-validation. External validation confirmed the utility of DeepSurv for discrimination (c-index—0.82 vs. 0.73) and RSF for individual survival estimates (0.18 vs. 0.03). We deployed the DeepSurv model to encourage incipient application in clinical practice. Overall, time-to-event models are successful in predicting the malignant transformation of oral leukoplakia and oral lichenoid lesions.

Predicting the clinical outcome of oral potentially malignant disorders using transcriptomic-based molecular pathology

BACKGROUND

This study was undertaken to develop and validate a gene expression signature that characterises oral potentially malignant disorders (OPMD) with a high risk of undergoing malignant transformation.

METHODS

Patients with oral epithelial dysplasia at one hospital were selected as the 'training set' (n = 56) whilst those at another hospital were selected for the 'test set' (n = 66). RNA was extracted from formalin-fixed paraffin-embedded (FFPE) diagnostic biopsies and analysed using the NanoString nCounter platform. A targeted panel of 42 genes selected on their association with oral carcinogenesis was used to develop a prognostic gene signature. Following data normalisation, uni- and multivariable analysis, as well as prognostic modelling, were employed to develop and validate the gene signature.

RESULTS

A prognostic classifier composed of 11 genes was developed using the training set. The multivariable prognostic model was used to predict patient risk scores in the test set. The prognostic gene signature was an independent predictor of malignant transformation when assessed in the test set, with the high-risk group showing worse prognosis [Hazard ratio = 12.65, p = 0.0003].

CONCLUSIONS

This study demonstrates proof of principle that RNA extracted from FFPE diagnostic biopsies of OPMD, when analysed on the NanoString nCounter platform, can be used to generate a molecular classifier that stratifies the risk of malignant transformation with promising clinical utility.

Diagnostic tests for oral cancer and potentially malignant disorders in patients presenting with clinically evident lesions

BACKGROUND

Squamous cell carcinoma is the most common form of malignancy of the oral cavity, and is often proceeded by oral potentially malignant disorders (OPMD). Early detection of oral cavity squamous cell carcinoma (oral cancer) can improve survival rates. The current diagnostic standard of surgical biopsy with histology is painful for patients and involves a delay in order to process the tissue and render a histological diagnosis; other diagnostic tests are available that are less invasive and some are able to provide immediate results. This is an update of a Cochrane Review first published in 2015.

OBJECTIVES

Primary objective: to estimate the diagnostic accuracy of index tests for the detection of oral cancer and OPMD, in people presenting with clinically evident suspicious and innocuous lesions.

SECONDARY OBJECTIVE

to estimate the relative accuracy of the different index tests.

SEARCH METHODS

Cochrane Oral Health's Information Specialist searched the following databases: MEDLINE Ovid (1946 to 20 October 2020), and Embase Ovid (1980 to 20 October 2020). The US National Institutes of Health Ongoing Trials Register (ClinicalTrials.gov) and the World Health Organization International Clinical Trials Registry Platform were also searched for ongoing trials to 20 October 2020. No restrictions were placed on the language or date of publication when searching the electronic databases. We conducted citation searches, and screened reference lists of included studies for additional references.

SELECTION CRITERIA

We selected studies that reported the diagnostic test accuracy of the following index tests when used as an adjunct to conventional oral examination in detecting OPMD or oral cavity squamous cell carcinoma: vital staining (a dye to stain oral mucosa tissues), oral cytology, light-based detection and oral spectroscopy, blood or saliva analysis (which test for the presence of biomarkers in blood or saliva).

DATA COLLECTION AND ANALYSIS

Two review authors independently screened titles and abstracts for relevance. Eligibility, data extraction and quality assessment were carried out by at least two authors, independently and in duplicate. Studies were assessed for methodological quality using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2). Meta-analysis was used to combine the results of studies for each index test using the bivariate approach to estimate the expected values of sensitivity and specificity.

MAIN RESULTS

This update included 63 studies (79 datasets) published between 1980 and 2020 evaluating 7942 lesions for the quantitative meta-analysis. These studies evaluated the diagnostic accuracy of conventional oral examination with: vital staining (22 datasets), oral cytology (24 datasets), light-based detection or oral spectroscopy (24 datasets). Nine datasets assessed two combined index tests. There were no eligible diagnostic accuracy studies evaluating blood or salivary sample analysis. Two studies were classed as being at low risk of bias across all domains, and 33 studies were at low concern for applicability across the three domains, where patient selection, the index test, and the reference standard used were generalisable across the population attending secondary care. The summary estimates obtained from the meta-analysis were: - vital staining: sensitivity 0.86 (95% confidence interval (CI) 0.79 to 0.90) specificity 0.68 (95% CI 0.58 to 0.77), 20 studies, sensitivity low-certainty evidence, specificity very low-certainty evidence; - oral cytology: sensitivity 0.90 (95% CI 0.82 to 0.94) specificity 0.94 (95% CI 0.88 to 0.97), 20 studies, sensitivity moderate-certainty evidence, specificity moderate-certainty evidence; - light-based: sensitivity 0.87 (95% CI 0.78 to 0.93) specificity 0.50 (95% CI 0.32 to 0.68), 23 studies, sensitivity low-certainty evidence, specificity very low-certainty evidence; and - combined tests: sensitivity 0.78 (95% CI 0.45 to 0.94) specificity 0.71 (95% CI 0.53 to 0.84), 9 studies, sensitivity very low-certainty evidence, specificity very low-certainty evidence.

AUTHORS' CONCLUSIONS

At present none of the adjunctive tests can be recommended as a replacement for the currently used standard of a surgical biopsy and histological assessment. Given the relatively high values of the summary estimates of sensitivity and specificity for oral cytology, this would appear to offer the most potential. Combined adjunctive tests involving cytology warrant further investigation. Potentially eligible studies of blood and salivary biomarkers were excluded from the review as they were of a case-control design and therefore ineligible. In the absence of substantial improvement in the tests evaluated in this updated review, further research into biomarkers may be warranted.

Dysplasia and DNA ploidy to prognosticate clinical outcome in oral potentially malignant disorders

BACKGROUND

Oral potentially malignant disorders are a clinical conundrum as there are no reliable methods to predict their behaviour. We combine conventional oral epithelial dysplasia grading with DNA ploidy analysis to examine the validity of this approach to risk assessment in a cohort of patients with known clinical outcomes.

METHODS

Sections from diagnostic biopsies were assessed for oral epithelial dysplasia using the WHO grading system, and DNA ploidy analysis was performed using established methods. Patients reviewed for a minimum of 5 years who did not develop oral squamous cell carcinoma were classified as "non-transforming" cases. Patients that developed oral squamous cell carcinoma ≥ 6 months after the initial diagnostic biopsy were classified as having "malignant transformation."

RESULTS

Ninety cases were included in the study. Seventy cases yielded informative DNA ploidy results. Of these 70 cases, 31 progressed to cancer. Oral epithelial dysplasia grading and DNA ploidy status were both significantly associated with clinical outcome (P < 0.05). Severe dysplasia had a hazard ratio of 3.50 (CI: 1.46, 8.45; P = 0.005) compared to cases with mild dysplasia. Aneuploidy had a hazard ratio of 2.09 (CI: 1.01, 4.32; P = 0.046) compared to cases with a diploid/tetraploid status. Receiver operating characteristic analysis gave an area under the curve of 0.617 for DNA ploidy status and 0.688 when DNA ploidy status was combined with dysplasia grading.

CONCLUSION

Our findings suggest that combining dysplasia grading with DNA ploidy status has clinical utility which could be used to develop novel management algorithms.

Oral potentially malignant disorders: A consensus report from an international seminar on nomenclature and classification, convened by the WHO Collaborating Centre for Oral Cancer

Oral potentially malignant disorders (OPMDs) are associated with an increased risk of occurrence of cancers of the lip or oral cavity. This paper presents an updated report on the nomenclature and the classification of OPMDs, based predominantly on their clinical features, following discussions by an expert group at a workshop held by the World Health Organization (WHO) Collaborating Centre for Oral Cancer in the UK. The first workshop held in London in 2005 considered a wide spectrum of disorders under the term "potentially malignant disorders of the oral mucosa" (PMD) (now referred to as oral potentially malignant disorders: OPMD) including leukoplakia, erythroplakia, proliferative verrucous leukoplakia, oral lichen planus, oral submucous fibrosis, palatal lesions in reverse smokers, lupus erythematosus, epidermolysis bullosa, and dyskeratosis congenita. Any new evidence published in the intervening period was considered to make essential changes to the 2007 classification. In the current update, most entities were retained with minor changes to their definition. There is sufficient evidence for an increased risk of oral cancer among patients diagnosed with "oral lichenoid lesions" and among those diagnosed with oral manifestations of 'chronic graft-versus-host disease'. These have now been added to the list of OPMDs. There is, to date, insufficient evidence concerning the malignant potential of chronic hyperplastic candidosis and of oral exophytic verrucous hyperplasia to consider these conditions as OPMDs. Furthermore, due to lack of clear evidence of an OPMD in epidermolysis bullosa this was moved to the category with limited evidence. We recommend the establishment of a global research consortium to further study the natural history of OPMDs based on the classification and nomenclature proposed here. This will require multi-center longitudinal studies with uniform diagnostic criteria to improve the identification and cancer risk stratification of patients with OPMDs, link them to evidence-based interventions, with a goal to facilitate the prevention and management of lip and oral cavity cancer.

Architectural Alterations in Oral Epithelial Dysplasia are Similar in Unifocal and Proliferative Leukoplakia

The current WHO histopathologic criteria for oral epithelial dysplasia (ED) are based on architectural and cytologic alterations, and do not address other histopathologic features of ED. Here we propose new diagnostic criteria including architectural, organizational, and cytologic features for oral ED. Cases of unifocal leukoplakia (UL) and proliferative leukoplakia (PL) with clinical photographs and follow-up information were identified. Only cases that showed minimal cytologic atypia or mild ED were used to demonstrate critical architectural changes as defined in this study. Eight biopsies from eight UL patients and 34 biopsies from four PL patients were included. The biopsies showed (a) corrugated, verrucous or papillary architecture, (b) hyperkeratosis with epithelial atrophy, (c) bulky squamous epithelial proliferation, and (d) demarcated hyperkeratosis and "skip" segments. The architectural alterations defined here are as important as the currently used criteria for the diagnosis of ED. Clinicopathologic correlation when diagnosing oral ED is also of the utmost importance in accurate diagnosis.

The "Peter Principle" revisited-Reflections on science, surgery and research

Revisiting the eponymous "Peter Principle," in this article the author reflects upon a professional lifetime as a clinician, surgeon and researcher, and summarises his efforts to improve contemporary understanding of oral cancer development, refine diagnostic procedure and facilitate efficacious, early treatment intervention to halt the process of carcinogenesis. The author reviews the principles underpinning effective translational research. Utilising his own work in oral epithelial cell science, interventional surgery and potentially malignant disorder patient management and surveillance, a number of significant advances in our knowledge base and their specific clinical application and potential impact are discussed. Research remains an active and ongoing process, however, and the author believes it essential for future relevance that hypotheses should always be initiated, led and mentored by experienced clinical practitioners.

A personalized computational model predicts cancer risk level of oral potentially malignant disorders and its web application for promotion of non-invasive screening

BACKGROUND

Despite their high accuracy to recognize oral potentially malignant disorders (OPMDs) with cancer risk, non-invasive oral assays are poor in discerning whether the risk is high or low. However, it is critical to identify the risk levels, since high-risk patients need active intervention, while low-risk ones simply need to be follow-up. This study aimed at developing a personalized computational model to predict cancer risk level of OPMDs and explore its potential web application in OPMDs screening.

METHODS

Each enrolled patient was subjected to the following procedure: personal information collection, non-invasive oral examination, oral tissue biopsy and histopathological analysis, treatment, and follow-up. Patients were randomly divided into a training set (N = 159) and a test set (N = 107). Random forest was used to establish classification models. A baseline model (model-B) and a personalized model (model-P) were created. The former used the non-invasive scores only, while the latter was incremented with appropriate personal features.

RESULTS

We compared the respective performance of cancer risk level prediction by model-B, model-P, and clinical experts. Our data suggested that all three have a similar level of specificity around 90%. In contrast, the sensitivity of model-P is beyond 80% and superior to the other two. The improvement of sensitivity by model-P reduced the misclassification of high-risk patients as low-risk ones. We deployed model-P in web.opmd-risk.com, which can be freely and conveniently accessed.

CONCLUSION

We have proposed a novel machine-learning model for precise and cost-effective OPMDs screening, which integrates clinical examinations, machine learning, and information technology.

Differential expression of ghrelin and GHSR via the mTOR pathway during the dynamic carcinogenic process involving oral, potentially malignant disorders

The purpose was to explore the sequence changes in ghrelin and GHSR in the mTOR signaling pathway during carcinogenesis involving oral, potentially malignant disorders (OPMD). The samples were confirmed through in vivo pathologic tissue screening and diagnosis. The immunohistochemical method was used to detect the expression of the ghrelin/growth hormone secretagogue receptor (GHSR) protein. The expression of ghrelin, GHSR 1α, GHSR 1β, and mammalian target of rapamycin (mTOR) RNA were detected by real-time PCR. The expression of ghrelin, GHSR, mTOR, and phosphorylated mTOR (phosphor-mTOR) protein were detected by Western blot. The expression of ghrelin/GHSR increased gradually in the dynamic process of OPMD carcinogenesis. There was a correlation between the increase in ghrelin, GHSR, mTOR, and phospho-mTOR. The in vivo expression of ghrelin/GHSR protein was the most apparent pathologic change from normal-to-mild, moderate, and severe dysplasia, and finally to the dynamic process from normal-to-mild-to-moderate dysplasia. The in vitro cell experiments based on QPCR results also proved that GHSR 1a functional receptor of ghrelin had a peak expression in LEUK-1 cells. In conclusioin, the close relationship between ghrelin and OPMD carcinogenesis can be used as a new biological target to assess the carcinogenesis of OPMD.

The"Newcastle Nomogram"-Statistical modelling predicts malignant transformation in potentially malignant disorders

BACKGROUND

Nomograms are graphical calculating devices used to predict risk of malignant transformation (MT) or response to treatment during cancer management. To date, a nomogram has not been used to predict clinical outcome during oral potentially malignant disorder (PMD) treatment. The aim of this study was to create a nomogram for use by clinicians to predict the probability of MT, thereby facilitating accurate assessment of risk and objective decision-making during individual patient management.

METHODS

Clinico-pathological data from a previously treated cohort of 590 newly presenting PMD patients were reviewed and clinical outcomes categorized as disease free, persistent PMD or MT. Multiple logistic regression was used to predict the probability of MT in the cohort using age, gender, lesion type, site and incision biopsy histopathological diagnoses. Internal validation and calibration of the model was performed using the bootstrap method (n = 1000), and bias-corrected indices of model performance were computed.

RESULTS

Potentially malignant disorders were predominantly leukoplakias (79%), presenting most frequently at floor of mouth and lateral tongue sites (51%); 99 patients (17%) developed oral squamous cell carcinoma during the study period. The nomogram performed well when MT predictions were compared with patient outcome data, demonstrating good bias-corrected discrimination and calibration (D_xy  = 0.58; C = 0.790), with a sensitivity of 87% and specificity 63%, and a positive predictive value of 32% and negative predictive value 96%.

CONCLUSION

The "Newcastle Nomogram" has been developed to predict the probability of MT in PMD, based on an internally validated statistical model. Based upon readily available and patient-specific clinico-pathological data, it provides clinicians with a pragmatic diagrammatic aid for clinical decision-making during diagnosis and management of PMD.

Perspectives on oral squamous cell carcinoma prevention-proliferation, position, progression and prediction

Squamous cell carcinoma arising from oral mucosal epithelium remains a lethal and deforming disease due to tumour invasion, oro-facial destruction, cervical lymph node metastasis and ultimate blood-borne dissemination. Worldwide, 300 000 new cases are seen each year, with a recent and significant rise in incidence affecting particularly the young. To rationalize perspectives on preventive strategies in oral cancer management, this study addresses a number of fundamental questions regarding carcinogenesis: proliferation-what epithelial cell changes precede tumour development? Position-why are certain oral sites so predisposed to cancer? Progression-why do some precursor lesions progress to invasive carcinoma and others do not? Prediction-how can we predict individual patient and/or lesion behaviour to prevent disease progression? By improving our understanding of oral carcinogenesis, can we thereby facilitate more effective primary, secondary and tertiary preventive strategies and ultimately reduce the global burden of oral squamous cell carcinoma (OSCC)?