| To Register SMDM Homepage | |
|
|
|
||||
Purpose. The sharp rise in early stage breast cancer incidence in the past 20 years is generally attributed to screening mammography. However sustained high incidence levels with plateaus in late-stage incidence led us to ask where all these cancers were before the advent of widespread screening. Accounting for this, we can now flesh out an explanation sketched over a decade ago: the Limited Malignant Potential (LMP) breast cancer hypothesis.
Method. A discrete event simulation of female breast cancer epidemiology from 1975-2000 was constructed. The simulation incorporates data about mammography dissemination, dissemination of adjuvant therapy, and improvements in mammography operating characteristics as well as secular improvements in female longevity over this time period. The simulation parameters for breast cancer natural history were heuristically manipulated in the context of these other simulation components in order to fit observed age- and historical stage-specific breast cancer surveillance data across the 25 years from 1975-2000. Computationally intensive parameter sampling experiments were conducted to assess likelihood of alternatives to the heuristically “best” solution.
Results. A large pool of occult breast cancers must exist to become incident breast cancers. The simulation balances the size of this pool and the assumed natural history of breast cancer against long run patterns of care data for diffusion of screening mammography and surveillance data regarding disease incidence and mortality. The “best fit” lead us to conclude 42% of all biologically initiated breast cancer is LMP; sampling experiments rule out substantial likelihood this fraction is under 30% or more than 55%. LMP tumors start with small focus (<.2 cm diameter) and grow to a maximum of approximately 1 cm diameter. If not detected in a span of about 2 years, they will recede and disappear. They never present a lethal threat to the host. In the year 2000 LMP breast cancers accounted for 30% of incident, clinically localized breast cancers and 44% of incident in situ disease. LMP breast cancers presently cannot be discerned histologically from in situ or early localized invasive cancer.
Discussion. Our calculations support the conclusion that a substantial fraction of diagnosed early stage breast cancer is LMP representing over-diagnosis. If true, the LMP hypothesis makes the value of developing a test to distinguish LMP from non-LMP breast cancers large.
See more of Oral Concurrent Session A - Simulation
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)