Abstract 1135P
Background
Cancer is a leading cause of death worldwide, but at present, screening is recommended for few cancer types. Multi-cancer early detection (MCED) tests can screen for many cancers simultaneously. We use a model1 of a cell-free DNA (cfDNA) MCED test to examine impact of screening intervals on stage at diagnosis and mortality outcomes.
Methods
Using US Surveillance, Epidemiology and End Results (SEER) data describing stage-specific incidence and cancer-specific survival of persons aged 50-79, published MCED performance measures, and previously published state-transition model,1 we computed potential diagnostic yield, stage shift, and effect on mortality. Extending this work to understand the influence of screening interval (annual vs biennial), we performed a sensitivity analysis for interval interacting with different tumor growth rate scenarios. We present summary statistics for performance as the expected rate in a sample of population-per-year of cancer incidence.
Results
Our model estimated intercepting 370 cancers/year/100,000 persons and a 49% reduction in late-stage cancers for annual screening, vs 292 cancers/year/100,000 and a 39% reduction for biennial screening (assuming fast tumor growth, Table). Compared with usual care,1 the model estimated 85 cancer deaths averted with annual screening vs 69 with biennial (Table). This differential results from 20% lower per-year diagnostic yield of biennial screening. This ratio became worse with faster growth rate scenarios. Table: 1135P
Reductions in estimated late-stage cancer diagnoses and deaths by adding MCED to usual carea
| MCED Screening Interval | Intensity (%)b | Cancer cfDNA Detected | Diagnoses at Late-Stage (III/IV) | Deaths within 5 yearsd | ||||
| N | % vs Annual | N | Benefit vs No MCED (% reduction)c | N | Averted vs No MCED | |||
| N | % | |||||||
| Annual | 100 | 370 | 100 | 210 | 49 | 308 | 85 | 21 |
| Biennial | 50 | 292 | 79 | 248 | 39 | 324 | 69 | 17 |
| None | 0 | 0 | 0 | 409 | 0 | 393 | 0 | 0 |
aFast tumor growth rate scenario. b% of pts screened/year in each scenario. c% of pts diagnosed at an earlier stage with each screening interval scenario; ie, no change in current practice would leave many cancers without any screening options and no net benefit. dDeaths within 5-years of original diagnosis to account for lead time.
Conclusions
Both annual and biennial screening potentially intercept a large fraction of all cancers before late stage, but annual screening is associated with more favorable per-year diagnostic yield and mortality. Though tumor growth rates for cfDNA-shedding cancers are poorly understood, this analysis suggests that annual and biennial intervals have noticeable differences in expected mortality, which should be considered in design of MCED screening programs. 1Hubbell, et al; 2020; CEBP;1134.2020.
Clinical trial identification
Editorial acknowledgement
Writing and editorial assistance provided by Neva West, Ph.D., Prescott Medical Communications Group (Chicago, IL).
Legal entity responsible for the study
GRAIL, Inc.
Funding
GRAIL, Inc.
Disclosure
P. Sasieni: Financial Interests, Advisory Board, Peter Sasieni is a paid member of the Scientific Advisory Board: GRAIL, Inc. C.A. Clarke, E. Hubbell: Financial Interests, Full or part-time Employment: GRAIL, Inc; Financial Interests, Stocks/Shares: GRAIL, Inc.