Rethinking risk and risk assessment for liquid biopsy-based single- & multi-cancer screening
In the ever-evolving landscape of cancer detection technologies, the emergence of novel multi-cancer screening tests presents a new wave of challenges. At the 2024 annual meeting of the American Association for Cancer Research (AACR), BLOODPAC’s Early Detection and Screening Working Group (ED&S WG) hosted a session focusing on cancer risk and risk assessment for liquid biopsy-based single- and multi-cancer screening.
Why Focus on Risk?
Bree Mitchell, PhD, Senior Director, Translational Medicine and Research at Natera and Co-Chair of the ED&S, explained that the Working Group is actively developing and building consensus around common definitions and standardized frameworks for evidence development in early cancer detection. During the group discussions, they found that some terms, such as risk, were more challenging to define than others.
This led to the session’s central question: What risk terminology should be used in the context of new blood-based screening tests, some of which seek to detect a single cancer type, others that seek to detect multiple types? To answer this question, the intended use population needs to be defined to operationalize risk assessment and target screening effectively. However, Mitchell emphasized that current evidence gaps present a significant barrier.
The WG is seeking to consolidate feedback on risk terminology. Even in the context of single-cancer screening, risk definitions vary, and the complexity increases with multi-cancer screening.
What is an Individual's Risk of Developing Cancer?
Christina Clarke Dur, PhD MPH, VP Epidemiology, Distinguished Scientist at GRAIL and ED&S Co-Chair, discussed the challenges of risk assessment in multi-cancer detection tests. While traditional single-cancer tests focus on finding cancer in specific organs, multi-cancer detection tests look for multiple cancers within a single person.
Absolute risk estimates (cancers/persons/time) are used to model benefit/harm and establish target populations. Once a screening paradigm is established, the focus often shifts to relative risk, which compares the risk to a reference or average population, identifying high-risk populations needing special screening guidelines.
Clarke explained that age-specific annual risk estimates are needed to best plan screening programs with annual or biannual intervals. The Surveillance, Epidemiology, and End Results Program (SEER) provides specific annual estimates for all cancers combined by age, sex, race, and geography. However, there are significant evidence gaps in age-specific annual risks for other important risk factors like smoking, diet and polygenic risk scores.
In single-cancer screening paradigms, absolute and relative risk concepts are more straightforward. Multi-cancer detection tests, on the other hand, introduce complexity. Literature from cancer epidemiologic cohort studies may detail age-specific annual risks for certain germline mutations associated with multiple cancer types, but tend to focus on selected single cancer types without providing the full picture of risk for the full cancer spectrum.
Designing a screening cadence for a population is challenging without understanding age-specific annual risks for environmental or behavioral factors. Therefore, defining and describing an intended use population and optimal screening intervals for multi-cancer screening tests requires a deeper understanding of age-specific, annual incidence rates for the full range of cancer types by key risk factors.
How Can Risk Define Intended-Use Populations?
During her talk, Wendy Rubinstein, PhD, Senior Scientific Officer of the Division of Cancer Prevention at the National Cancer Institute and member of the ED&S WG, highlighted two major bottlenecks in cancer screening strategies:
Evidence Bottleneck: Demonstrating the technology's promise is essential.
Disease Incidence Bottleneck: Low cancer incidence in screening trials affects the ability to achieve statistical power.
These bottlenecks impact three main cancer screening strategies that benefit public health:
Use several single cancer screening tests in specific situations. This strategy is the current paradigm, utilizing single cancer screening to detect cancer signals in specific organs.
Everyone is at risk, so screen everyone for any cancer they might develop using one screening test. Multi-cancer tests fall under this strategy, which raises the question: What is a person's risk to develop any cancer? Multi-cancer tests aggregate cancer incidence of different types of cancer, but don't resolve the incidence bottleneck.
Achieve greater power by studying populations with higher cancer incidence rate. This strategy consists in using risk factors to enroll people at higher risk in clinical trials, but the concern is ending with perhaps an older population that is not representative of the general population. An approach to follow this strategy while still looking into average risk population is to oversample an elevated risk population.
Rubinstein mentioned the existence of several risk models for single cancers, for a genetic syndrome with multiple specific cancers, and polygenic risk scores (PRS). Multi-cancer risk models are not currently available but are feasible to generate.
The Genome Informed Risk Assessment (GIRA) provides a platform that integrates several clinical risk factors to do risk stratification. Rubinstein explained that with the underlying data that is available it would be possible to do a multi-cancer PRS model but the evidence gaps impact analytical and clinical validity, and some tests might need to be tuned to the underlying population.
Risk in the Context of a Clinical Program
Elizabeth O’Donnell, MD, DipABLM, Director of Early Detection and Prevention of Malignant Conditions at Dana-Farber Cancer Institute, emphasized the need for focus and rigor to intercept cancers when they are most curable. O’Donnell talked about Dana Farber’s early cancer detection program which has four steps: 1. risk assessment, 2. interception, 3. novel screening, and 4. comprehensive care.
The program’s risk assessment is moving towards polygenic risk assessment, though any inherited risk or familial history is vital, and focuses on patients with:
Hereditary Risk (e.g., carriers of germline variants, family history, elevated cancer risk from prediction models)
Exposed Risk (e.g., childhood and adult cancer survivors, exposure to IARC carcinogens)
Precursor Conditions (e.g., obesity, gender, predisposing infections, UV exposure)
The INAdvance study is part of the clinical program and aims to understand risk with comprehensive data collection through surveys capturing lifestyle, dietary intake, exercise, and integration with electronic medical records. This information survey, as well as the sample collection, can be done remotely, minimizing the need for center visits.
A critical question for the program is managing patients with positive tests. For the 1% who test positive, it’s crucial to follow up, recognizing that about half will ultimately have a positive cancer diagnosis.
O’Donnell also talked about the INFORM study, a prospective clinical trial to study the Galleri® multi-cancer early detection (MCED) test in hereditary high-risk populations. The primary endpoint of the study is identifying the number and types of cancers diagnosed by MCED screening in a high-risk population. Lifetime risk can be as high as 50-60% for those with pathogenic germline mutations.These patients also face significant barriers to screening, including inconsistent practices, confusion about care coordination, and various structural barriers that need addressing to improve adherence and outcomes.
Conclusion
The BLOODPAC ED&S Working Group's first installment of the lexicon for evidence development in early cancer detection is expected to be published later this year. This panel highlighted some of the challenges and inconsistencies in our current risk terminology and invited a robust discussion with the audience. As the Working Group continues to gather feedback on risk terminology, their next goal is to publish a perspective paper on the topic of this session: cancer risk and risk assessment for liquid biopsy-based single- and multi-cancer screening.