A pilot study to develop an efficacious oral cancer screening strategy for India

Oral cancer imposes a huge burden globally, especially in low and middle income countries (LMICs) (8). This high burden of oral cancer is attributable to high prevalence of risk factors such as tobacco, areca nut (tamul) and alcohol. In most countries, cigarette smoking is the predominant form of tobacco use (9). However, in India, approximately 75% of tobacco consumption is in smokeless form (9). India has the highest burden of oral cancer globally (10). It is recognized that smokeless tobacco causes cancers of the mouth, gullet and pancreas (11). The proposed study will be conducted in Assam, where the prevalence of tobacco is 48% (12). In Assam, tamul is traditionally offered as a mark of respect and auspicious beginnings. We recently conducted a survey in Assam on tamul use. In all, 34% of subjects reported using tamul. Influencing factors were family (58%) and friends (34%). Majority of the subjects (60%) chewed tamul due to pressure at work, after food and during leisure with a mean age of initiation at 15 years. While knowledge about the ill-effects of tamul was low and willingness to quit high (77%), paradoxically, 81% of subjects had not attempted to quit the habit in the preceding 6 months.

The 5-year survival rate for oral cancer is 80% in those diagnosed at early stage and only 20% in those diagnosed at advanced stage, underscoring the need for early detection (13). In India, most oral cancer patients present with advanced disease and have survival rates as low as 3-5%. There is, therefore, a pressing need to diagnose early and ideally at a pre-cancerous stage (2).

In India and other LMICs, the current approach for oral cancer screening relies on visual inspection of the inner lining of the oral cavity (conventional oral examination [COE]) in tobacco/ alcohol users aged over 30 years (4). Yet, owing to issues with the method of screening (low sensitivity of COE) and size of the target population (ages 30+), this approach is not efficacious. For example, the eligible population for oral cancer screening in India based on COE is ~300 million, which reduces its operational feasibility and makes it cost prohibitive. Any screening system developed in LMICs should fulfill two features for optimal effectiveness: 1) The screening test should have high accuracy 2) The screening system (infrastructure, human resource and referral centres) has to be financially viable. Unfortunately, the current oral cancer screening strategy of visual inspection in India and several other LMICs fails at both these levels. Risk stratification tools for systematic identification of high-risk population could potentially drastically reduce the target population for screening.

We propose a 3-step oral cancer screening strategy to enable effective implementation of the screening program. First, a risk prediction model will be used for identification of those at greatest risk of oral cancer, thus reducing the number of screened individuals and enhancing cost effectiveness. Second, high-risk individuals will be screened using autoflourescence, a highly sensitive test for diagnosing oral cancer to rule-out disease by causing only the abnormal areas to light up. Finally, individuals positive on autofluorescence will be ruled-in through a brush biopsy to look for suspicious cells, a test that is highly specific for diagnosing oral cancer. We will validate the risk prediction model of the proposed 3-step strategy as part of this pump priming application which will subsequently lead to a large randomized controlled trial in the population of North East India to test its effectiveness. Our study will provide proof-of-principle of the 3-step screening strategy for implementation of the oral cancer screening in LMICs.

We propose conducting a study in North East India (Assam), to investigate the proof-of-performance of the proposed 3-step oral cancer screening strategy. First, we propose to use a risk prediction model to identify individuals at high-risk of oral pre-cancer/ early oral cancer for inclusion in the screening program. Second, we propose to rule-out disease in these high-risk individuals using a wide-field screening adjunct with high sensitivity - autofluorescence. Third, in individuals found to be positive by autofluorescence, we propose to rule-in disease using a narrow-field screening adjunct with high specificity - brush cytology.

The study will be conducted in two phases. All healthy individuals aged between 30 to 65 years, with exposure to tobacco, areca nut and alcohol, will be included in the study. Logistic regression analyses will be utilized for the development of the risk-prediction model using data on 15,000 participants recruited in Phase I. The developed model will be applied to 15,000 participants recruited into Phase II for validation. Model-validation will be evaluated using calibration ratios (ratio of observed to expected number of cases) and discrimination (area under the receiver operating characteristic curve [AUC]).

Successful completion of this study is anticipated to lead to the design and implementation of a large randomized controlled trial to investigate the efficacy, cost and population-level effectiveness of this oral cancer screening strategy.

The current approach for oral cancer screening and early detection relies on visual inspection of the oral mucosa (conventional oral examination [COE]) in tobacco/alcohol users aged over 30 years (2). COE has low sensitivity and low specificity for the identification of oral pre-cancer (e.g. oral leukoplakia, the most common oral pre-cancer) or early oral cancer (carcinoma in-situ or Stage I cancers). Owing to the low sensitivity of COE, as well as the size of the target population (ages 30+; ~300M), this approach is not expected to be efficacious.

This proposal aims to develop an oral cancer screening strategy that identifies high-risk individuals, rules-out disease through a highly sensitive screening modality, and then rules-in disease through a highly specific screening modality. The proposed 3-step approach for oral cancer screening is highly innovative because it efficiently accomplishes several desirable features of population-wide cancer screening programs, including efficiency, high negative reassurance for screen negative individuals, high positive predictive value for screen positive individuals, as well as the use of non-invasive, low-cost screening modalities. Hence, this strategy will ensure that the already strained public health systems in LMICs are able to utilize their precious resources in a more efficient manner.

Our study potentially has a high public health impact because it addresses the most common cancer in India. Further, our results would be broadly generalizable to other countries in Asia with similar high-risk exposures and high oral cancer incidence such as Pakistan, Nepal, Bangladesh, Sri Lanka and other LMICs around the world. Lastly, our study leverages investment through the existing field infrastructure in place for screening of chronic diseases (oral, breast and cervical cancer, diabetes and hypertension) led by the Government of Assam and the Tata Trusts.