Anil Chaturvedi

Ph.D.

National Cancer Institute

Senior Investigator

PLCO (Learn more about this study)

PLCO-402

Sep 28, 2018

Design and validation of an oral cancer risk prediction model

In 2016, an estimated 31,910 new cases of oral cavity cancer (6,490 deaths) will occur in the United States [1]. Although oral cavity cancers (OCCs) and oropharyngeal cancers (OPCs) have traditionally been grouped together, a much smaller proportion of OCCs are associated with human papillomavirus (HPV) infection [2, 3]. As a result, HPV vaccination would not be expected to have a major impact on the burden of OCCs. The majority of OCC cases (~75%) are attributable to tobacco and heavy alcohol use [3, 4]. OCCs are ideal candidates for screening, secondary prevention, and early detection given the amenability of the oral cavity for visual inspection as well as the availability of recognized precursor lesions, such as oral leukoplakia. But evidence is currently lacking to assess the benefits and harms of screening, which is reflected in the US Preventive Services Task Force recommendation statement concerning oral cancer screening [5]. This lack of evidence arises from several gaps including the identification of high-risk populations/risk stratification tools, methods for screening and intervention, and natural history of OCCs [5]. Ongoing studies by our group and others are addressing several of these questions. However, there are currently no risk stratification tools for OCCs.

Eventually, answers to whether OCC screening is effective will come from high quality randomized controlled trials (RCTs). When investigators decide to plan RCTs, results from risk prediction models could be useful for efficient study design, to selectively include high-risk individuals.

The overall goal of this study is to develop and validate a risk prediction model for oral cavity cancer (incidence) using data from US Cohorts within the Consortium. The models will be validated internally within the Consortium by randomly selecting development and validation cohorts. We will also apply the model to US national surveys to investigate the distribution of risk in the US population.

1. To develop risk prediction models for oral cavity cancer incidence using data on socio-demographic factors (age, gender, race, education) anthropometrics (height, weight, and BMI), behaviors (smoking and alcohol use), and medical and family history.

2. To validate the risk prediction model internally and in US representative national health surveys.

Methods:

Cox regression models, adjusted for competing mortality, will be used to predict 5-year risk of oral cancer incidence. Models will be built using a range of predictors (noted below), and the best-fitting models will be selected using the AIC and -2LL. Final models will be validated internally (by randomly selecting development vs. validation cohorts within the Consortium), and externally using US national survey data. Model validation will be based on discrimination (AUC) and calibration (E/O ratio), both overall and across key demographic/risk subgroups.

Exposure data: Risk factors for oral cavity cancers. Socio-demographic: age, gender, race, education, marital status, income/poverty level; Anthropometric: height, weight, and BMI; Behavioral: tobacco and alcohol use, including duration, intensity, and time since cessation; Medical history: oral health, diabetes, heart disease, and medications; Family history of oral cancer.

Outcome data: Incidence of oral cavity cancers (ICD-O-3 codes: C000-C069, excluding oropharyngeal sites--C019, C024, C051, and C052).