UNCOVER: underlying novel causes of onset of very early cancer research

Recent studies, including our own, have shown dramatic increases in selected cancers among young adults. Using U.S. SEER data, we showed that incidence rates of three key cancers have been increasing in younger adults in the past decade: 318,781 adults aged 25–49 years were diagnosed with colorectal (n=114,220), thyroid (n=146,977), and kidney (n=57,584) cancer with increases per year at 2.44%, 4.81%, and 3.83%, respectively, during 2006–2015. Annual percent change over 1% per year is usually classified as an epidemic of cancer. Thus, there is urgent public health need to understand the drivers of these early-onset cancers. We have also modelled age-period-cohort effects and found that risks of developing these three cancers at younger ages (25–49 years) have increased significantly by birth year for the last several decades. In addition, these increases were consistently seen in most racial subgroups and in both localized and advanced cancers. These findings suggest increased risks for these cancers due to increasing extrinsic (i.e., environmental and lifestyle) risk factors, which, if identified, can be prevented. However, despite the dramatic increases in these early-onset cancers, their relative rarity coupled with a long induction time make traditional epidemiologic approaches inefficient to identify the underlying causes. For example, a population sample of more than 0.76 million would be needed to observe 500 cases of colorectal cancer in adults under 50 over a 5-year period. We thus propose to utilize a novel, and efficient, mathematical framework to investigate drivers of the three earlyonset cancers (i.e. colorectal, thyroid, and kidney). This framework will use mechanistic mathematical models to capture the underlying cancer biology over the life course, and further couple these models with advanced data assimilation methods to test various hypothesized risk mechanisms based on cancer incidence and exposures recorded in nationally representative datasets. The project team has extensive experience using data assimilation methods and mathematical models to understand disease systems including cancers. The framework will also innovatively account for changing cancer detection rates over time, observation errors, interactions among multiple risk factors, and changes in risk impact over the life course. Using this new framework, for each of the three key cancers, we aim to 1) systematically examine and identify key risk factors, in particular, in younger adults <50 years, 2) infer the risk mechanisms (e.g., cancer initiation vs. promotion), 3) examine the dynamic interactions among co-exposures (e.g., risk amplification due to synergistic effect of coexposures), and 4) quantify the impact of individual risk factors and co-exposures in different stages of life and particularly to examine critical windows of susceptibility. Further, we will quantify whether the identified etiological factors when combined are of high enough magnitude to be used to develop a risk calculator for starting earlier cancer screening. Completing these aims will thus provide important etiologic evidence for primary prevention at the population-level as well as aid clinicians and their patients in devising cancer screening strategies.