Modeling the natural history of ductal carcinoma in situ based on population data

Abstract

Background: The incidence of ductal carcinoma in situ (DCIS) has increased substantially since the introduction of mammography screening. Nevertheless, little is known about the natural history of preclinical DCIS in the absence of biopsy or complete excision.

Methods: Two well-established population models evaluated six possible DCIS natural history submodels. The submodels assumed 30%, 50%, or 80% of breast lesions progress from undetectable DCIS to preclinical screen-detectable DCIS; each model additionally allowed or prohibited DCIS regression. Preclinical screen-detectable DCIS could also progress to clinical DCIS or invasive breast cancer (IBC). Applying US population screening dissemination patterns, the models projected age-specific DCIS and IBC incidence that were compared to Surveillance, Epidemiology, and End Results data. Models estimated mean sojourn time (MST) in the preclinical screen-detectable DCIS state, overdiagnosis, and the risk of progression from preclinical screen-detectable DCIS.

Results: Without biopsy and surgical excision, the majority of DCIS (64-100%) in the preclinical screen-detectable state progressed to IBC in submodels assuming no DCIS regression (36-100% in submodels allowing for DCIS regression). DCIS overdiagnosis differed substantially between models and submodels, 3.1-65.8%. IBC overdiagnosis ranged 1.3-2.4%. Submodels assuming DCIS regression resulted in a higher DCIS overdiagnosis than submodels without DCIS regression. MST for progressive DCIS varied between 0.2 and 2.5 years.

Conclusions: Our findings suggest that the majority of screen-detectable but unbiopsied preclinical DCIS lesions progress to IBC and that the MST is relatively short. Nevertheless, due to the heterogeneity of DCIS, more research is needed to understand the progression of DCIS by grades and molecular subtypes.

Keywords: Breast carcinoma in situ; Breast neoplasms; Disease progression; Early detection of cancer; United States.

Fig. 1

Graphical representation of the DCIS model and its submodels. The figure depicts the states that are included in the model. The dotted arrow from the state preclinical screen-detectable DCIS to the state no breast cancer is included into the submodels where DCIS regression is allowed. The proportion of preclinical undetectable DCIS that progresses to preclinical screen-detectable DCIS can be 30%, 50%, or 80%. Different model assumptions on the natural history of DCIS are included for each submodel

Fig. 2

DCIS incidence and overdiagnosis. Each graph includes SEER data and 2 submodels (1 without DCIS regression and 1 with DCIS regression). The projections include women in the age group 30–79 years. noReg, model without DCIS regression; wReg, model with DCIS regression; OD, overdiagnosis

Fig. 3

IBC incidence and overdiagnosis. Each graph includes SEER data and 2 submodels (1 without DCIS regression and 1 with DCIS regression). The projections include women in the age group 30–79 years. noReg, model without DCIS regression; wReg, model with DCIS regression; OD, overdiagnosis

Fig. 4

Proportion of DCIS cases progressing to other states. Stacked bar plots showing the proportion of preclinical screen-detectable DCIS progressing to preclinical invasive breast cancer (P1), clinical DCIS (P2), or no breast cancer (P3; regression). P1, P2, and P3 are represented by blue, red, and green bars, respectively. Simulated birth cohort 1930

Fig. 5

Overdiagnosis by age. Calendar year 2010. Each graph includes overdiagnosis for DCIS and IBC. DCIS, ductal carcinoma in situ; IBC, invasive breast cancer; noReg, submodel without DCIS regression; wReg, with DCIS regression