Study
PLCO
(Learn more about this study)
Project ID
PLCO-50
Initial CDAS Request Approval
Oct 25, 2013
Title
Head and Neck Cancer Risk Prediction Model
Summary
The International Head and Neck Cancer Epidemiology (INHANCE) Consortium reported that tobacco and alcohol use account for a large proportion of cases in Latin America (83%) and Europe (84%) but only 51% of head and neck cancer (HNC) cases in the US. The population attributable risks for tobacco and alcohol were lower for women (57%) than men (74%), and lower for younger subjects (33%) than older subjects (73%). In recent years, tonsillar and tongue cancer incidence, and oral cavity and pharyngeal cancer incidence in young women have been increasing at alarming rates in the US, which may partly be due to human papillomavirus (HPV) infection. Since it is unlikely that tobacco, alcohol and HPV account for all HNC cases, other risk factors must continue to be investigated. Furthermore, cancer risk prediction models may aid in personalized prevention strategies targeted to high risk populations such as young HPV positive individuals and individuals with family history of HNC. We propose to develop HNC risk prediction models within the INHANCE Consortium database of 18,000 cases and 28,000 controls from 27 studies. Our specific aims are: 1) to conduct a comprehensive analysis of risk factors by pooling additional data on occupational exposures and oral hygiene, and to estimate their corresponding attributable risks by geographic region in the INHANCE Consortium; 2) to develop the first risk prediction models for HNC and for oral cavity, oropharyngeal, hypopharyngeal, and laryngeal cancer for the US, using aim 1 results and baseline cancer hazard rates derived from SEER data; and 3) to validate the HNC risk prediction model in US cohorts, including the Prostate, Lung, Colon, and Ovarian (PLCO) Cancer Screening Trial, from the Cohort Consortium. Competing risk models will be used to estimate the probability of developing these cancers over a 5-, 10-, and 20-year time interval. The risk factors under investigation will include tobacco smoking, alcohol drinking, family history of HNC, HPV infection, select genetic variants, involuntary smoking, sexual behaviors, oral hygiene, dietary factors, physical activity, socioeconomic status, occupational exposures, and BMI. We will validate the prediction models by comparing the expected and observed number of cancer cases, and ROC curves using cohort data from the Cohort Consortium. Our HNC risk prediction models will have the potential for: 1) clinical use within the context of personalized medicine in the future, 2) for dentists in decision making on whether to proceed with oral cancer exams beyond conventional visualization examination, 3) for public use by individuals interested in understanding their HNC risk (ie, individuals who have family history of HNC), and 4) for research use to select high risk populations for studies of primary prevention. Additionally, the prediction model can serve as a tool to estimate risk reduction for individuals to evaluate the impact of behavior modification (i.e., quitting cigarette smoking or reducing alcohol drinking). The large collaboration of multidisciplinary investigators in the INHANCE Consortium provides an exceptional opportunity for this project on comprehensive risk factor evaluation and risk prediction models for HNC.
Aims
Our specific aims are:
Aim 1: To estimate odds ratios (OR) and attributable risks (AR) for established and suspected HNC risk factors, including lifestyle, environmental, genetic and infection-related factors within each region covered by the INHANCE Consortium (North America, South America, Europe and East Asia). We will pool in additional data on occupational exposures and oral hygiene to the core INHANCE pooled data. We will assess potential interactions among the risk factors identified.
Aim 2: To develop HNC risk prediction models for the US using the results from aim 1 and baseline cancer hazard rates derived from the Surveillance, Epidemiology, and End Results (SEER) data. We will establish the cancer prediction models for HNC overall and for each cancer subsite (oral cavity, oropharynx, hypopharynx, and larynx) for the US.
Aim 3: To validate the HNC risk prediction model within US cohort studies, including PLCO, from the Cohort Consortium. We will pool data on at least five US-based cohort studies for approximately 2,660 head and neck cancer cases. Within a prospective setting, we will evaluate how well the HNC risk prediction model predicts who will be diagnosed with HNC, based on various combinations of HNC risk factors.
Collaborators
Mia Hashibe, PhD, University of Utah and Huntsman Cancer Institute
Saundra Buys, MD, Huntsman Cancer Institute
Lisa Gren, PhD, University of Utah
