Modelling the early detection of breast cancer

Abstract

A mathematical model was developed to predict the outcome of early detection clinical trials or programs targeted at evaluating mortality benefit from earlier diagnosis of breast cancer. The model was applied to eight randomized breast cancer trials, which were carried out to evaluate the benefits of mammography, physical examination or their combination. The model assumes that breast cancer is a progressive disease and any mortality benefit from earlier diagnosis is generated from a favorable shift in the stage at diagnosis relative to usual care. The model predicted the reduction in mortality for seven of the eight trials within the reported confidence intervals. Input data required by the models are: stage shift distribution, examination schedules, population age distribution, follow up time, and survival conditional on stage at diagnosis. Survival distributions were obtained from the 1973-82 SEER database whereas the remaining data was obtained for each of the trials. Information on sensitivity and stage was ordinarily available during the early phase of the trials. The theoretical model has the promise of being able to predict the long-term outcome of early detection trials or programs during the initial examination phase. The theoretical model is general and may be applied to other chronic diseases, which satisfy the basic assumptions.

Mortality Modeling of Early Detection Programs

“…Other applications of the model to breast cancer deal with finding optimal examination schedules (Lee et al, 2004), predicting long‐term mortality reduction only using data available in the early phase of a study (Lee and Zelen, 2003), and using the model to estimate the individual contributions of early detection and advances in therapy considering the recent observation that breast cancer mortality has been decreasing in the United States (Berry et al, 2005). Another important use of the model is to determine screening programs for subpopulations that may be at elevated risks.…”

The Dana-Farber CISNET Model for Breast Cancer Screening Strategies: An Update

“…The invasive breast cancer component of the natural history (Figure 1) was previously developed and validated [1–3, 16]. Briefly, the model assumes that disease progresses to worse states, S du → S p → S c. Sojourn time in S p follows an exponential distribution (Zelen 1969) with an age-dependent mean sojourn time of 2–4 years [18].…”

“…The invasive breast cancer component of the natural history (Figure 1) was previously developed and validated. 1–3,16 Briefly, the model assumes that disease progresses to worse states: S du → S p → S c. Sojourn time in S p follows an exponential distribution, 17 with an age-dependent mean sojourn time of 2 to 4 years. 18 For early-stage DCIS in the preclinical, screen-detectable state: 1) some cases will stay in the early stage and eventually regress to the preclinical, undetectable DCIS state, 2) some will progress to invasive breast cancer, and 3) some will progress to the clinical DCIS state when symptoms appear.…”

Evaluating markers for the early detection of cancer: overview of study designs and methods

“…Somne mathematical models for cancer screening assuine that cancer fatality rates after cancer detection, conditional on observed features such as stage, tumor size, and node status, are the same for cancers detected on screening and cancers detected clinically [64,[67][68][69]. However, bias may arise due to unobserved differences in the characteristics of cancers.…”