MIcrosimulation SCreening Analysis (MISCAN) Prostate Cancer Model
The microsimulation screening analysis prostate cancer model (MISCAN-PRO) simulates individual life histories. The cancer progression process in individuals is modeled as a sequence of tumor states. There are 18 preclinical detectable states in the natural history of prostate cancer (Figure 1), which are derived from combinations of histologic grade (Surveillance, Epidemiology, and End Results [SEER] categories well, moderately, and poorly differentiated), clinical T-stages (American Joint Committee on Cancer stages T1, T2, and T3), and clinical M-stages (M0 and M1). The onset of disease is modeled as an age-dependent hazard. Progression through the clinical stages and grades is modeled as a semi-Markov process, and we assume stage- and grade-specific risks of transitions from earlier to later stages and grades. From each preclinical detectable state, the cancer can progress to the clinical disease state and be diagnosed. Screen detection depends on the probability of attendance, frequency of prostate-specific antigen (PSA) tests, the PSA threshold for a positive test, and, after a positive PSA test, biopsy compliance and stage-specific biopsy sensitivity.
Figure 1. In the MISCAN-PRO model, preclinical prostate cancer progresses through tumor states with risks of progression that depend on the current stage and grade. Each state can be metastatic (M1) or non-metastatic (M0), but this is not illustrated for simplicity. (From: New England Journal of Medicine, Heijnsdijk EAM, Wever EM, Auvinen A, et al. Quality-of-life effects of prostate-specific antigen screening, 367, S3. Copyright © 2012 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.)
The baseline parameters for the natural history of prostate cancer were estimated first using data from the Rotterdam section of the European Randomized Study of Screening for Prostate Cancer (ERSPC)(1, 2). Data from the Swedish section of the ERSPC have also been included(3). For calibration to the US situation, the model used US life tables and we re-estimated the sensitivity parameters and estimated an additional stage-specific risk of clinical diagnosis to capture different pre-PSA disease diagnosis patterns in the US as compared with Europe. US-specific estimates for the parameters were obtained by calibrating the model to the observed age-specific incidence and age-specific SEER stage distribution using maximum likelihood(4).
The model is capable of handling a variety of ways to model the benefit of early detection. This includes modeling benefit among screen-detected men using a stage-shift approach, a constant cure rate, or a cure rate dependent on the stage or the lead time(5).
Recent extensions include substantial changes to the original MISCAN-PRO model. We are now able to model PSA growth and progression of disease after diagnosis. We used a modified version of the PSA growth model in the PSAPC model and linked this to the disease progression in MISCAN-PRO. We calibrated the PSA growth parameters to the PSA distribution in the ERSPC trial and SEER incidence data. In this new model we are able to implement PSA-dependent screening policies: we can stop screening or change the screening frequency if PSA is below a certain value at a certain age or screening round in a trial. Additionally, we are able to study screening policies where PSA threshold for biopsy referral depends on age. We found that stopping screening at age 70 is a reasonable way to reduce overdiagnosis and retain the benefit of early detection(6), and decreasing the stopping age has a more pronounced impact on overdiagnosis reduction than reducing the screening frequency.
Tip: Hover your cursor over the dashed attribute links below for more information. View the details of this model in a grid with other prostate models.