DFCI (Dana-Farber) Dana-Farber Cancer Institute

Purpose

The main purpose of the CISNET-DFCI model, which was developed at Dana-Farber Cancer Institute, is to predict the mortality associated with female breast cancer in the presence of screening and treatment in the U.S. population. The predictions may be by chronological year and/or age. Mortality may change by advances in treatment and/or changing dissemination of screening. The model incorporates the possibility that these latter two factors will change by chronological time and age. The model is general and enables the prediction of changes in mortality if technical advances are made by radiology or the discovery of other disease markers.

Overview

CISNET-DFCI is a stochastic model that depicts the early detection process of screening and predicts breast cancer mortality as a function of disease natural history, detection, and treatment. This analytic approach was applied to estimate the impact of mammography screening and treatment on breast cancer incidence and mortality. A series of equations was derived to project the age-specific incidence of breast cancer and the probability of breast cancer deaths (mortality) in the presence of screening. Other outcomes associated with screening, such as life-years gained, quality adjusted life years (QALYs), cost-effectiveness, overdiagnosis, and false positive findings, are also generated.

For invasive breast cancer, the model characterizes the natural history of breast cancer by health states that transition from normal breast tissue to a pre-clinical undetectable DCIS state, and progression to a screen detectable DCIS or screen detectable invasive cancer state: S₀: A woman is disease free or has disease but it is asymptomatic and cannot be diagnosed by any modality; S_du: A woman has early-stage undetectable DCIS; S_dp: A woman has early-stage detectable DCIS; S_dc: A woman has symptomatic DCIS; S_p: A woman has breast cancer, but it is asymptomatic and may be diagnosed by a special examination or examination program; S_c: A woman, having usual care, is diagnosed with invasive breast cancer; S_d: Death attributed to breast cancer. There are two main model assumptions: i) invasive breast cancer is progressive and described by the transitions S₀ to S_p to S_c and some eventually progress to S_d; ii) the mortality benefit from screening is attributable to a stage shift in diagnosis. Updates to the model include incorporation of recurrence, molecular subtypes of estrogen receptor (ER) and human epidermal growth factor receptor 2 (HER2) status, and ER/HER2-specific natural history parameters.

The main goal of screening is to diagnose individuals in S_p where subjects have an early-stage disease with no symptoms. The stage shift implies that the subjects are diagnosed earlier (in S_p) before symptoms surface (in S_c).

Any specific screening patterns or combination of screening patterns as used in the US population, for example, can be applied to specific birth cohorts. The mortality benefit of the mammography screening is obtained by finding cases in an earlier stage. This is addressed through a stage shift in the model. Treatment benefits captured as hazard reductions are applied to the baseline (in the absence of screening and treatment) underlying breast cancer survival data. Screening dissemination patterns, treatment dissemination patterns, stage shift and hazard reductions from treatment are provided as common input parameters. Updates to the model include ER/HER2-specific baseline underlying survival and treatment efficacy up to the year 2015.

Figure 1. CISNET-DFCI Model: Natural History of Breast Cancer

References

1. Lee S, Li X, Huang H, Zelen M. The Dana-Farber CISNET Model for Breast Cancer Screening Strategies: An Update. Med Decis Making. 2018 Apr;38(1_suppl):44S-53S.
2. Lee S, Zelen M. A stochastic model for predicting the mortality of breast cancer. J Natl Cancer Inst Monogr. 2006;(36):79-86.
3. Lee S, Zelen M. Mortality modeling of early detection programs. Biometrics. June 1, 2008
4. Lee S, Zelen M. Modelling the early detection of breast cancer. Ann Oncol. Aug. 1, 2003
5. Lee S, Huang H, Zelen M. Early detection of disease and scheduling of screening examinations. Stat Methods Med Res. Dec. 1, 2004