Feasibility, Reliability, and Effectiveness of Oral cancer screening in South Asia and Southeast Asian countries: A systematic review and Meta-analysis.

OBJECTIVES: The incidence of oral cancer is significantly high in South Asia and Southeast Asia. Organized screening is an effective approach to early detection. The aim of this systematic review and meta-analysis was to evaluate the reliability, diagnostic accuracy, and effectiveness of visual oral screening by community health workers (CHWs) in identifying oral cancer/oral potentially malignant disorders (OPMDs) in this region. MATERIALS AND METHODS: We conducted a bibliographic search in PubMed, Scopus, the gray literature of Google Scholar, ProQuest dissertations, and additional manual searches. Twelve articles were included for qualitative synthesis and six for meta-analysis. Pooled sensitivity, specificity, diagnostic odds ratio (DOR), and forest plot analysis were performed. RESULTS: Meta-analysis showed CHWs identified 8% (n = 6365) as suspicious and 92% (n = 74,140) as normal. The diagnostic accuracy of visual oral screening by CHWs showed a sensitivity of 75% (CI: 74-76) and specificity of 97% (CI: 97-97) in the detection of OPMDs/oral cancer. Forest plots were obtained using a random effects model (DOR: 24.52 (CI: 22.65-26.55)) and SAUC: 0.96 (SE = 0.05). CONCLUSIONS: Oral visual examination by trained CHWs can be utilized for community screenings to detect oral cancer early. This approach can be used in primary healthcare to triage patients for further referral and management.

Inter-observer agreement among specialists in the diagnosis of oral potentially malignant disorders and oral cancer using store-and-forward technology.

OBJECTIVES: Oral cancer is a leading cause of morbidity and mortality. Screening and mobile Health (mHealth)-based approach facilitates early detection remotely in a resource-limited settings. Recent advances in eHealth technology have enabled remote monitoring and triage to detect oral cancer in its early stages. Although studies have been conducted to evaluate the diagnostic efficacy of remote specialists, to our knowledge, no studies have been conducted to evaluate the consistency of remote specialists. The aim of this study was to evaluate interobserver agreement between specialists through telemedicine systems in real-world settings using store-and-forward technology. MATERIALS AND METHODS: The two remote specialists independently diagnosed clinical images (n=822) from image archives. The onsite specialist diagnosed the same participants using conventional visual examination, which was tabulated. The diagnostic accuracy of two remote specialists was compared with that of the onsite specialist. Images that were confirmed histopathologically were compared with the onsite diagnoses and the two remote specialists. RESULTS: There was moderate agreement (k= 0.682) between two remote specialists and (k= 0.629) between the onsite specialist and two remote specialists in the diagnosis of oral lesions. The sensitivity and specificity of remote specialist 1 were 92.7% and 83.3%, respectively, and those of remote specialist 2 were 95.8% and 60%, respectively, each compared with histopathology. CONCLUSION: The diagnostic accuracy of the two remote specialists was optimal, suggesting that "store and forward" technology and telehealth can be an effective tool for triage and monitoring of patients. CLINICAL RELEVANCE: Telemedicine is a good tool for triage and enables faster patient care in real-world settings.

Cancer stem cell markers in patterning differentiation and in prognosis of oral squamous cell carcinoma.

Differentiation is a major histological parameter determining tumor aggressiveness and prognosis of the patient; cancer stem cells with their slow dividing and undifferentiated nature might be one of the factors determining the same. This study aims to correlate cancer stem cell markers (CD44 and CD147) with tumor differentiation and evaluate their subsequent effect on prognosis. Immunohistochemical analysis in treatment naïve oral cancer patients (n = 53) indicated that the expression of CD147 was associated with poorly differentiated squamous cell carcinoma and moderately differentiated squamous cell carcinoma (p < 0.01). Furthermore, co-expression analysis showed that 45% each of moderately differentiated squamous cell carcinoma and poorly differentiated squamous cell carcinoma patients were CD44high/CD147high as compared to only 10% of patients with well-differentiated squamous cell carcinoma. A three-way analysis indicated that differentiation correlated with recurrence and survival (p < 0.05) in only the patients with CD44high/CD147high cohort. Subsequently, relevance of these cancer stem cell markers in patterning the differentiation characteristics was evaluated in oral squamous cell carcinoma cell lines originating from different grades of oral cancer. Flowcytometry-based analysis indicated an increase in CD44+/CD147+ cells in cell lines of poorly differentiated squamous cell carcinoma (94.35 ± 1.14%, p < 0.001) and moderately differentiated squamous cell carcinoma origin (93.49 ± 0.47%, p < 0.001) as compared to cell line of well-differentiated squamous cell carcinoma origin (23.12% ± 0.49%). Expression profiling indicated higher expression of cancer stem cell and epithelial-mesenchymal transition markers in SCC029B (poorly differentiated squamous cell carcinoma originated; p ≤ 0.001), which was further translated into increased spheroid formation, migration, and invasion (p < 0.001) as compared to cell line of well-differentiated squamous cell carcinoma origin. This study suggests that CD44 and CD147 together improve the prognostic efficacy of tumor differentiation; in vitro results further point out that these markers might be determinant of differentiation characteristics, imparting properties of increased self-renewal, migration, and invasion.

Reference gene evaluation for normalization of gene expression studies with lymph tissue and node­derived stromal cells of patients with oral squamous cell carcinoma.

Profiling studies using reverse transcription quantitative PCR (RT-qPCR) require reliable normalization to reference genes to accurately interpret the results. A stable reference gene panel was established to profile metastatic and non-metastatic lymph nodes in patients with oral squamous cell carcinoma. The stability of 18S ribosomal RNA (18SrRNA), ribosomal Protein Lateral Stalk Subunit P0 (RPLP0), ribosomal Protein L27 (RPL27), TATA-box binding protein (TBP), hypoxanthine phosphoribosyl-transferase 1 (HPRT1), beta-actin (ACTB), glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH) and vimentin (VIM) was evaluated, as reference genes for profiling patient-derived lymph node stromal cells (LNSCs; N=8; N0:6, N+:2) and lymph node tissues (Patients:14, Nodes=20; N0:7; N+:13). The genes were initially assessed based on their expression levels, specificity, and stability rankings to identify the best combination of reference genes. VIM was excluded from the final analysis because of its low expression (high quantification cycle >32) and multiple peaks in the melting curve. The stability analysis was performed using Reffinder, which utilizes four tools; geNorm, NormFinder, BestKeeper and Comparative ∆Ct methods, thereby enabling the computing of a comprehensive ranking. Evaluation of the gene profiles indicated that while RPLP0 and 18SrRNA were stable in both lymph node tissues and LNSCs, HPRT1, RPL27 were uniquely stable in these tissues whereas ACTB and TBP were most stable in LNSCs. The present study identified the most stable reference gene panel for the RT-qPCR profiling of lymph node tissues and patient-derived LNSCs. The observation that the gene panel differed between the two model systems further emphasized the need to evaluate the reference gene subset based on the disease and cellular context.

Contrasting clinical outcomes and socio-economic impact of young versus elderly-onset oral squamous cell carcinoma, a novel health economic analysis.

OBJECTIVES: The incidence of young-onset oral squamous cell carcinoma (OSCC) is growing, even among non-smokers/drinkers. The effects of adverse histopathological features on long-term oncologic outcomes between the young and old are controversial and confounded by significant heterogeneity. Few studies have evaluated the socio-economic impact of premature mortality from OSCC. Our study seeks to quantify these differences and their economic impact on society. MATERIALS AND METHODS: Four hundred and seventy-eight young (<45 years) and 1660 old patients (≥45 years) with OSCC were studied. Logistic regression determined predictors of recurrence and death. Survival analysis was calculated via the Kaplan-Meier method. A separate health economic analysis was conducted for India and Singapore. Years of Potential Productive Life Lost (YPPLL) were estimated with the Human Capital Approach, and premature mortality cost was derived using population-level data. RESULTS: Adverse histopathological features were seen more frequently in young OSCC: PNI (42.9% vs. 35%, p = 0.002), LVI (22.4% vs. 17.3%, p = 0.013) and ENE (36% vs. 24.5%, p < 0.001). Although 5-year OS/DSS were similar, the young cohort had received more intensive adjuvant therapy (CCRT 26.9% vs. 16.6%, p < 0.001). Among Singaporean males, the premature mortality cost per death was US $396,528, and per YPPLL was US $45,486. This was US $397,402 and US $38,458 for females. Among Indian males, the premature mortality cost per death was US $30,641, and per YPPLL was US $595. This was US $ 21,038 and US $305 for females. CONCLUSION: Young-onset OSCC is an aggressive disease, mitigated by the ability to receive intensive adjuvant treatment. From our loss of productivity analysis, the socio-economic costs from premature mortality are substantial. Early cancer screening and educational outreach campaigns should be tailored to this cohort. Alongside, more funding should be diverted to genetic research, developing novel biomarkers and improving the efficacy of adjuvant treatment in OSCC.

Holistic Approach for the Early Detection of Oral Cancer: A Comprehensive Training Module.

Objectives: Oral cancer is significantly high in India, and screening is an effective approach to downstage the disease. Educating Community Health Workers (CHWs) on early oral cancer detection is an effective step toward reducing the burden and serves as a first step toward facilitating the transfer of knowledge. Therefore, the purpose of this hands-on education was to equip CHWs with insight on the advanced diagnostics, preventive techniques, and innovations for the early detection of oral cancer. Materials and Methods: A total of 178 participants were trained in two groups: Group 1 received training for screening and primary prevention, while group 2 received training on updates in recent diagnostic adjuncts and innovations, AI-enabled point-of-care diagnostics, and essential patient care in management of Oral Potentially Malignant Disorders (OPMDs). Pre- and post-assessment questionnaires were used to evaluate the participants. Results: The knowledge assessment scores between the pre- and post-tests showed a statistically significant difference (p < 0.001), with rise in mean score of 3.99 from baseline. Six months following training, knowledge retention revealed a statistically significant difference (p < 0.001) in the participants' ability to recall the information. Conclusion: A well-structured training module can create awareness, impart knowledge and upskill the CHWs for early detection of oral cancer. Retraining of CHWs is required for knowledge retention post-training.

Rapid and Label-Free Histopathology of Oral Lesions Using Deep Learning Applied to Optical and Infrared Spectroscopic Imaging Data.

Oral potentially malignant disorders (OPMDs) are precursors to over 80% of oral cancers. Hematoxylin and eosin (H&E) staining, followed by pathologist interpretation of tissue and cellular morphology, is the current gold standard for diagnosis. However, this method is qualitative, can result in errors during the multi-step diagnostic process, and results may have significant inter-observer variability. Chemical imaging (CI) offers a promising alternative, wherein label-free imaging is used to record both the morphology and the composition of tissue and artificial intelligence (AI) is used to objectively assign histologic information. Here, we employ quantum cascade laser (QCL)-based discrete frequency infrared (DFIR) chemical imaging to record data from oral tissues. In this proof-of-concept study, we focused on achieving tissue segmentation into three classes (connective tissue, dysplastic epithelium, and normal epithelium) using a convolutional neural network (CNN) applied to three bands of label-free DFIR data with paired darkfield visible imaging. Using pathologist-annotated H&E images as the ground truth, we demonstrate results that are 94.5% accurate with the ground truth using combined information from IR and darkfield microscopy in a deep learning framework. This chemical-imaging-based workflow for OPMD classification has the potential to enhance the efficiency and accuracy of clinical oral precancer diagnosis.

Two by Two Factorial Design using Metformin and Curcumin for Second Primary Head and Neck Cancer Prevention Trial.

OBJECTIVE: The 2x2 factorial design is an effective method that allows for multiple comparisons, especially in the context of interactions between different interventions, without substantially increasing the required sample size. In view of the considerable preclinical evidence for Curcumin and Metformin in preventing the development and progression of head and neck squamous cell carcinoma (HNSCC), this study describes the protocol of the clinical trial towards applying the drug combination in prevention of second primary tumors. METHODS: We have applied the trial design to a large phase IIB/III double-blind, multi-centric, placebo-controlled, randomized clinical trial to determine the safety and efficacy of Metformin and Curcumin in the prevention of second primary tumours (SPT) of the aerodigestive tract following treatment of HNSCC (n=1,500) [Clinical Registry of India, CTRI/2018/03/012274]. Patients recruited in this trial will receive Metformin (with placebo), Curcumin (with placebo), Metformin, and Curcumin or placebo alone for a period of 36 months. The primary endpoint of this trial is the development of SPT, while the secondary endpoints are toxicities associated with the agents, incidence of recurrence, and identifying potential biomarkers. In this article, we discuss the 2x2 factorial design and how it applies to the head and neck cancer chemoprevention trial. CONCLUSION: 2x2 factorial design is an effective trial design for chemoprevention clinical trials where the effectiveness of multiple interventions needs to be tested parallelly.

Selective neck dissection is effective in N1/N2 nodal stage oral cavity squamous cell carcinoma.

PURPOSE: Although the efficacy of selective neck dissection (SND) in the management of a node-negative neck is established, its utility in the management of node-positive disease remains controversial. The objective of this study was to evaluate the oncologic safety of SND in the management of N1/N2 oral cavity squamous cell carcinoma. MATERIALS AND METHODS: From a prospectively collected electronic database of patients with oral cavity cancer, a retrospective analysis was conducted of patients with nonrecurrent, clinical, and/or pathologic N1/N2 oral cavity squamous cell carcinoma who underwent SND of levels I to III/IV. The patients were stratified into 2 groups: clinical N0 but pathologic N1/N2 (cN0-pN1/N2) and clinical N1/N2 (cN1/N2). The primary outcome variable of the study was the ipsilateral regional recurrence rate. Categorical data were analyzed by the 2-sided Fisher exact test, and 3-year Kaplan-Meier ipsilateral regional control rate, regional recurrence-free survival, disease-free survival, and overall survival were estimated. RESULTS: Forty-nine patients constituted the study sample, with 37 patients in the cN1/N2 group and 12 patients in the cN0-pN1/N2 group. During the follow-up period of 3 years, 2 patients (∼4%) developed ipsilateral neck recurrence, and these patients were in the cN1/N2 group. The 3-year Kaplan-Meier ipsilateral regional control rate was 95%. CONCLUSIONS: SND may produce a satisfactory ipsilateral regional control rate in patients with early-stage node-positive oral squamous cell carcinoma. A prospective, randomized clinical trial comparing SND with modified radical neck dissection may be required for a categorical conclusion of these findings.

Reconstruction of large composite buccal defects using single soft tissue flap--analysis of functional outcome.

Resection of advanced gingivo-buccal tumors results in a posterolateral mandibular and large soft tissue defect. Because of large soft tissue requirement, these defects are difficult to reconstruct using a single osteocutaneous flap. A double free flap reconstruction of such defects is recommended. However, double flap may not be feasible in certain situations. In this study, we objectively evaluated functional and cosmetic outcomes following single soft-tissue flap reconstruction in a group of patients where double flap reconstruction was not feasible. Patient and defect characteristics were obtained from charts. The speech and swallowing functions of patients were prospectively assessed by a dedicated therapist. The cosmetic outcome of reconstruction was evaluated by an independent observer. Fifty-six patients with large soft tissue and segmental posterolateral mandible defect, reconstructed with anterolateral thigh or pectoralis major flap from May 2009 till December 2010 were included. In this series, none of the flaps were lost; two patients with pectoralis major flap developed partial skin paddle loss. Most of the patients developed mandibular drift; however, majority of these patients had no postoperative trismus. All patients resumed regular or soft solid oral diet. The mean speech intelligibility was more than 70%. Majority of patients had satisfactory cosmetic outcome. The defects were classified into regions resected to develop a reconstruction algorithm for optimal reconstruction using a free or pedicle flap. In conclusion, patients with large oro-mandibular defect undergoing single soft tissue flap reconstruction have satisfactory functional and cosmetic outcome.

Extended Level IV Neck Dissection for Limited Prophylactic Clearance of Level V in Node-Positive Papillary Thyroid Carcinoma: a Homogenous Study Population.

There is near consensus that prophylactic lateral neck dissection has no role in the management of differentiated thyroid cancer, but the extent of lateral neck dissection in differentiated thyroid cancer remains controversial, especially whether level V should be addressed or not. There is lot of heterogeneity in reporting of the management of level V in papillary thyroid cancer. We at our Institute address the lateral neck positive papillary thyroid cancer with selective neck dissection involving levels II-IV, performing extended level IV dissection with inclusion of the triangular area delineated by the sternocleidomastoid muscle, the clavicle, and the perpendicular line drawn to the clavicle from the point where the horizontal line at the level of cricoid cuts the posterior border of sternocleidomastoid muscle. Retrospective analysis of the departmental data set related to thyroidectomy with lateral neck dissection from 2013 to mid-2019 for papillary thyroid cancer, was carried out. Patients with recurrent papillary thyroid cancer were excluded as were patients with involvement of level V. Data related to the demography of patients, histological diagnosis, and postoperative complications were compiled and summarized. Note was made of the incidence of ipsilateral neck recurrence and the neck level involved with recurrence noted. Data was analyzed for fifty-two patients of non-recurrent papillary thyroid cancer who had undergone total thyroidectomy and lateral neck dissection involving levels II-IV, with extended dissection at level IV. It should be noted that none of the patients had clinical involvement of level V. Only two patients had lateral neck recurrence, both the recurrences were in level III, one on the ipsilateral side and the other on the contralateral side. Recurrence in the central compartment was noted in two patients, with one of these patients also having ipsilateral level III recurrence. One of the patients had distal metastasis to the lungs. Transient paresis of the unilateral vocal cords was noted in seven patients which got resolved within 2 months in all of them. Transient hypocalcemia was noted in four patients. Although our series has a small sample size with limited follow-up, it is one of the few studies in which prophylactic level V dissection has been studied in a homogenous study population of non-recurrent papillary thyroid cancer. Our study has shown that prophylactic dissection of level V may have a limited role, but further large multi-institutional studies need to be carried out to come up with a definite answer.

Interpretable and Reliable Oral Cancer Classifier with Attention Mechanism and Expert Knowledge Embedding via Attention Map.

Convolutional neural networks have demonstrated excellent performance in oral cancer detection and classification. However, the end-to-end learning strategy makes CNNs hard to interpret, and it can be challenging to fully understand the decision-making procedure. Additionally, reliability is also a significant challenge for CNN based approaches. In this study, we proposed a neural network called the attention branch network (ABN), which combines the visual explanation and attention mechanisms to improve the recognition performance and interpret the decision-making simultaneously. We also embedded expert knowledge into the network by having human experts manually edit the attention maps for the attention mechanism. Our experiments have shown that ABN performs better than the original baseline network. By introducing the Squeeze-and-Excitation (SE) blocks to the network, the cross-validation accuracy increased further. Furthermore, we observed that some previously misclassified cases were correctly recognized after updating by manually editing the attention maps. The cross-validation accuracy increased from 0.846 to 0.875 with the ABN (Resnet18 as baseline), 0.877 with SE-ABN, and 0.903 after embedding expert knowledge. The proposed method provides an accurate, interpretable, and reliable oral cancer computer-aided diagnosis system through visual explanation, attention mechanisms, and expert knowledge embedding.

BCA101 Is a Tumor-Targeted Bifunctional Fusion Antibody That Simultaneously Inhibits EGFR and TGFß Signaling to Durably Suppress Tumor Growth.

The EGFR and TGFß signaling pathways are important mediators of tumorigenesis, and cross-talk between them contributes to cancer progression and drug resistance. Therapies capable of simultaneously targeting EGFR and TGFß could help improve patient outcomes across various cancer types. Here, we developed BCA101, an anti-EGFR IgG1 mAb linked to an extracellular domain of human TGFßRII. The TGFß "trap" fused to the light chain in BCA101 did not sterically interfere with its ability to bind EGFR, inhibit cell proliferation, or mediate antibody-dependent cellular cytotoxicity. Functional neutralization of TGFß by BCA101 was demonstrated by several in vitro assays. BCA101 increased production of proinflammatory cytokines and key markers associated with T-cell and natural killer-cell activation, while suppressing VEGF secretion. In addition, BCA101 inhibited differentiation of naïve CD4+ T cells to inducible regulatory T cells (iTreg) more strongly than the anti-EGFR antibody cetuximab. BCA101 localized to tumor tissues in xenograft mouse models with comparable kinetics to cetuximab, both having better tumor tissue retention over TGFß "trap." TGFß in tumors was neutralized by approximately 90% in animals dosed with 10 mg/kg of BCA101 compared with 54% in animals dosed with equimolar TGFßRII-Fc. In patient-derived xenograft mouse models of head and neck squamous cell carcinoma, BCA101 showed durable response after dose cessation. The combination of BCA101 and anti-PD1 antibody improved tumor inhibition in both B16-hEGFR-expressing syngeneic mouse models and in humanized HuNOG-EXL mice bearing human PC-3 xenografts. Together, these results support the clinical development of BCA101 as a monotherapy and in combination with immune checkpoint therapy. SIGNIFICANCE: The bifunctional mAb fusion design of BCA101 targets it to the tumor microenvironment where it inhibits EGFR and neutralizes TGFß to induce immune activation and to suppress tumor growth.

Reconstructive indications of simultaneous double free flaps in the head and neck: a case series and literature review.

Extensive and complex defects of the head and neck involving multiple anatomical and functional subunits are a reconstructive challenge. The purpose of this study is to elucidate the reconstructive indications of the use of simultaneous double free flaps in head and neck oncological surgery. This is a retrospective review of 21 consecutive cases of head and neck malignancies treated surgically with resection and reconstruction with simultaneous use of double free flaps. Nineteen of 21 patients had T4 primary tumor stage. Eleven patients had prior history of radiotherapy or chemo-radiotherapy. Forty-two free flaps were used in these patients. The predominant combination was that of free fibula osteo-cutaneous flap with free anterolateral thigh (ALT) fascio-cutaneous flap. The indications of the simultaneous use of double free flaps can be broadly classified as: (a) large oro-mandibular bone and soft tissue defects (n = 13), (b) large oro-mandibular soft tissue defects (n = 4), (c) complex skull-base defects (n = 2), and (d) dynamic total tongue reconstruction (n = 2). Flap survival rate was 95%. Median follow-up period was 11 months. Twelve patients were alive and free of disease at the end of the follow-up. Eighteen of 19 patients with oro-mandibular and glossectomy defects were able to resume an oral diet within two months while one patient remained gastrostomy dependant till his death due to disease not related to cancer. This patient had a combination of free fibula flap with free ALT flap, for an extensive oro-mandibular defect. The associated large defect involving the tongue accounted for the swallowing difficulty. Simultaneous use of double free flap aided the reconstruction in certain large complex defects after head and neck oncologic resections. Such combination permits better complex multiaxial subunit reconstruction. An algorithm for choice of flap combination for the appropriate indications is proposed.

Incorporation of adverse features in advanced oral cancer improves precision in staging and patient prognostication.

BACKGROUND: Despite revised staging criteria, stratification of patients with advanced oral squamous cell carcinoma (OSCC) remains difficult. Well-established features like perineural invasion (PNI), differentiation, and lymphovascular-invasion (LVI) are controversial, and hence omitted from staging. We endeavor to better stratify this cohort by identifying predictors of survival in advanced OSCC (T3-4). METHODS: Seven hundred and forty-two patients with T3-4 OSCC underwent surgery from 2006 to 2013. Cox regression was performed to determine predictors of overall survival (OS). RESULTS: OS was adversely impacted by PNI (p = 0.046), LVI (p = 0.038), moderate/poor differentiation (p = 0.001), close/involved surgical margins (p = 0.002), pT (p = 0.034), and pN (p < 0.001). The cumulative number of adverse histopathological features predicted poorer OS; HR 2.64 (CI 1.42-4.90) for one adverse feature and HR 4.23 (CI 2.34-7.67) for ≥2. CONCLUSION: In advanced OSCC, stratification with histopathologic risk factors can predict survival even in maximally treated patients; adjuvant therapies are unable to entirely mitigate this risk. Incorporation of adverse features into future editions of TNM can improve precision in staging and identify candidates for treatment escalation.

A Distributed Cancer Care Model with a Technology-Driven Hub-and-Spoke and further Spoke Hierarchy: Findings from a Pilot Implementation Programme in Kerala, India.

BACKGROUND: The technology enabled distributed model in Kerala is based on an innovative partnership model between Karkinos Healthcare and private health centers. The model is designed to address the barriers to cancer screening by generating demand and by bringing together the private health centers and service providers at various levels to create a network for continued care. This paper describes the implementation process and presents some preliminary findings.  Methods: The model follows the hub-and-spoke and further spoke framework. In the pilot phases, from July 2021 to December 2021, five private health centers (partners) collaborated with Karkinos Healthcare across two districts in Kerala. Screening camps were organized across the districts at the community level where the target groups were administered a risk assessment questionnaire followed by screening tests at the spoke hospitals based on a defined clinical protocol. The screened positive patients were examined further for confirmatory diagnosis at the spoke centers. Patients requiring chemotherapy or minor surgeries were treated at the spokes. For radiation therapy and complex surgeries the patients were referred to the hubs. RESULTS: A total of 2,459 individuals were screened for cancer at the spokes and 299 were screened positive. Capacity was built at the spokes for cancer surgery and chemotherapy. A total of 189 chemotherapy sessions and 17 surgeries were performed at the spokes for cancer patients. 70 patients were referred to the hub. CONCLUSION: Initial results demonstrate the ability of the technology Distributed Cancer Care Network (DCCN) system to successfully screen and detect cancer and to converge the actions of various private health facilities towards providing a continuum of cancer care. The lessons learnt from this study will be useful for replicating the process in other States.

Exploring uncertainty measures in convolutional neural network for semantic segmentation of oral cancer images.

Significance: Oral cancer is one of the most prevalent cancers, especially in middle- and low-income countries such as India. Automatic segmentation of oral cancer images can improve the diagnostic workflow, which is a significant task in oral cancer image analysis. Despite the remarkable success of deep-learning networks in medical segmentation, they rarely provide uncertainty quantification for their output. Aim: We aim to estimate uncertainty in a deep-learning approach to semantic segmentation of oral cancer images and to improve the accuracy and reliability of predictions. Approach: This work introduced a UNet-based Bayesian deep-learning (BDL) model to segment potentially malignant and malignant lesion areas in the oral cavity. The model can quantify uncertainty in predictions. We also developed an efficient model that increased the inference speed, which is almost six times smaller and two times faster (inference speed) than the original UNet. The dataset in this study was collected using our customized screening platform and was annotated by oral oncology specialists. Results: The proposed approach achieved good segmentation performance as well as good uncertainty estimation performance. In the experiments, we observed an improvement in pixel accuracy and mean intersection over union by removing uncertain pixels. This result reflects that the model provided less accurate predictions in uncertain areas that may need more attention and further inspection. The experiments also showed that with some performance compromises, the efficient model reduced computation time and model size, which expands the potential for implementation on portable devices used in resource-limited settings. Conclusions: Our study demonstrates the UNet-based BDL model not only can perform potentially malignant and malignant oral lesion segmentation, but also can provide informative pixel-level uncertainty estimation. With this extra uncertainty information, the accuracy and reliability of the model's prediction can be improved.

Risk Stratification in Oral Cancer: A Novel Approach.

Background: Oral squamous cell carcinoma (OSCC) is a common head and neck cancer with high morbidity and mortality. Currently, treatment decisions are guided by TNM staging, which omits important negative prognosticators such as lymphovascular invasion, perineural invasion (PNI), and histologic differentiation. We proposed nomogram models based on adverse pathological features to identify candidates suitable for treatment escalation within each risk group according to the National Comprehensive Cancer Network (NCCN) guidelines. Methods: Anonymized clinicopathologic data of OSCC patients from 5 tertiary healthcare institutions in Asia were divided into 3 risk groups according to the NCCN guidelines. Within each risk group, nomograms were built to predict overall survival based on histologic differentiation, histologic margin involvement, depth of invasion (DOI), extranodal extension, PNI, lymphovascular, and bone invasion. Nomograms were internally validated with precision-recall analysis and the Kaplan-Meier survival analysis. Results: Low-risk patients with positive pathological nodal involvement and/or positive PNI should be considered for adjuvant radiotherapy. Intermediate-risk patients with gross bone invasion may benefit from concurrent chemotherapy. High-risk patients with positive margins, high DOI, and a high composite score of histologic differentiation, PNI, and the American Joint Committee on Cancer (AJCC) 8th edition T staging should be considered for treatment escalation to experimental therapies in clinical trials. Conclusion: Nomograms built based on prognostic adverse pathological features can be used within each NCCN risk group to fine-tune treatment decisions for OSCC patients.

Interpretable deep learning approach for oral cancer classification using guided attention inference network.

SIGNIFICANCE: Convolutional neural networks (CNNs) show the potential for automated classification of different cancer lesions. However, their lack of interpretability and explainability makes CNNs less than understandable. Furthermore, CNNs may incorrectly concentrate on other areas surrounding the salient object, rather than the network's attention focusing directly on the object to be recognized, as the network has no incentive to focus solely on the correct subjects to be detected. This inhibits the reliability of CNNs, especially for biomedical applications. AIM: Develop a deep learning training approach that could provide understandability to its predictions and directly guide the network to concentrate its attention and accurately delineate cancerous regions of the image. APPROACH: We utilized Selvaraju et al.'s gradient-weighted class activation mapping to inject interpretability and explainability into CNNs. We adopted a two-stage training process with data augmentation techniques and Li et al.'s guided attention inference network (GAIN) to train images captured using our customized mobile oral screening devices. The GAIN architecture consists of three streams of network training: classification stream, attention mining stream, and bounding box stream. By adopting the GAIN training architecture, we jointly optimized the classification and segmentation accuracy of our CNN by treating these attention maps as reliable priors to develop attention maps with more complete and accurate segmentation. RESULTS: The network's attention map will help us to actively understand what the network is focusing on and looking at during its decision-making process. The results also show that the proposed method could guide the trained neural network to highlight and focus its attention on the correct lesion areas in the images when making a decision, rather than focusing its attention on relevant yet incorrect regions. CONCLUSIONS: We demonstrate the effectiveness of our approach for more interpretable and reliable oral potentially malignant lesion and malignant lesion classification.

Development of a Low-cost NGS Test for the Evaluation of Thyroid Nodules.

Ultrasound-guided fine needle aspiration cytology (FNAC) is the preferred method of identifying malignancy in palpable thyroid nodules using the Bethesda reporting system. However, in around 30-40% of FNACs (Bethesda categories III, IV, and V), the results are indeterminate and surgery is required to confirm malignancy. Out of those who undergo surgery, only 10-40% of patients in these categories are found to have malignancies, thus proving surgery to be unnecessary for some patients or to be incomplete in others. While molecular testing on thyroid FNAC material is part of the American Thyroid Association (ATA) guidelines in evaluating thyroid nodules, it is currently unavailable in India due to cost constraints. In this study, we prospectively collected FNAC samples from sixty-nine patients who presented with palpable thyroid nodules. We designed a cost-effective next-generation sequencing (NGS) test to query multiple variants in the DNA and RNA isolated from the fine needle aspirate. The identification of oncogenic variants was considered to be indicative of malignancy, and confirmed by surgical histopathology. The panel showed an overall sensitivity of 81.25% and a specificity of 100%, while in the case of Bethesda categories III, IV, and V, the sensitivity was higher (87.5%) and the specificity was established at 100%. The panel could thereby serve as a rule-in test for the diagnosis of thyroid cancer and therefore help identify patients who require surgery, especially in the indeterminate Bethesda categories III, IV, and V.