Prostate Cancer

Background

The CISNET prostate working group originated with investigators from Fred Hutchinson Cancer Research Center (PSAPC model) and the University of California Davis (SCANS model) in the year 2000. Beginning in 2005, investigators from Erasmus University Medical Center (MISCAN-PRO model) joined, and shortly afterwards investigators from University of California Davis moved to the University of Michigan. Beginning in 2020, investigators from University College London (MSM-CanProg model) joined and shortly afterwards moved to the University of Cambridge.

Recent Studies

The prostate models were originally developed to study the plausible roles of screening and changes in treatment on prostate cancer mortality (Etzioni et al., 2008, 2012). The models were also extended to reconcile widely disparate estimates of lead times and overdiagnosis associated with prostate-specific antigen (PSA) screening (Draisma et al., 2009), to compare natural histories and risks of cancer progression (Gulati et al., 2011), to examine contamination in the prostate section of the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial (Gulati et al., 2012), and to project expected disease trends under discontinued PSA screening (Gulati et al., 2014). Subsequent work studied differential natural history in Black men vs men overall (Tsodikov et al., 2017) and implications for targeted screening (Nyame et al., 2021), reconciled effects of screening on mortality in the PLCO cancer screening trial and the European Randomized Study of Screening for Prostate Cancer (ERSPC) (Tsodikov et al., 2017; de Koning et al., 2018), investigated risk-stratified screening intervals and stopping ages (Heijnsdijk et al., 2020), and evaluated screening approaches in a low-risk high-risk population (Heijnsdijk et al., 2022).

Common structure and inputs

Each model contains the same basic components: (1) a natural history model, projecting latent onset and progression and clinical detection in the absence of screening and other-cause death; (2) a model for disease-specific and other-cause survival in the absence of screening; (3) a model for screening and, if PSA is explicitly modeled, for biopsy after screening; (4) a model for how screening impacts disease-specific survival; and (5) a model for how treatment impacts disease-specific survival.

Common inputs consist of disease incidence, prostate cancer screening patterns, and primary treatment trends in the US. The models also share estimates of disease-specific survival in the absence of screening and other-cause death rates from US life tables. Disease incidence is used primarily for calibration purposes. The calibration process is the process by which each model identifies natural history parameters that are most consistent with observed incidence. Calibrated CISNET prostate models have been applied to explain population trends in prostate cancer mortality (population models) and to project the outcomes of different screening and/or treatment interventions for policy development (cohort models).

Model frameworks

The PSAPC, MISCAN-PRO, SCOPE, and MSM-CanProg models are microsimulation models that generate individual life histories composed of cancer natural history, screening, and diagnosis using computer algorithms and random draws from parametric statistical distributions. The PSAPC, MISCAN-PRO, and SCOPE models also simulate treatment and cause-specific survival events. The MSM-CanProg model simulates cause-specific survival that represents an average across treatments. In contrast, the SCANS model is an analytic mathematical model that represents pre- and post-diagnosis events in an integrated set of analytic probability models. SCOPE is a microsimulation translation and extension of the SCANS model that preserves key SCANS components while explicitly simulating longitudinal PSA trajectories, latent tumor growth, and disease progression.

Stage progression

The PSAPC, MISCAN-PRO, and SCOPE models assume that all prostate cancers begin in localized stage and can progress to distant metastases. The MSM-CanProg model does not explicitly include stage. The SCANS model does not specify stage at onset and requires only that stage does not regress during the window of detectability by screening.

Grade progression

The PSAPC model assumes that prostate cancers can be low or high grade at onset but cannot progress over time. The MISCAN-PRO and SCOPE models assume that all prostate cancers begin in low grade and can progress over time. The MSM-CanProg model assumes that prostate cancers begin in low or high grade at onset and low grade can progress to high grade over time. The SCANS model does not specify grade at onset and requires only that grade does not regress during the window of detectability by screening.

Screening sensitivity

The PSAPC and SCOPE models generate PSA levels for each simulated individual at each screen. Simulated men with PSA level exceeding a specified threshold (e.g., 3 or 4 ng/mL) at a screen are referred for further evaluation. Further evaluation can include secondary testing (e.g., magnetic resonance imaging or biomarker testing) with subsequent biopsy. Performance characteristics of secondary testing and biopsy are externally sourced from systematic reviews and meta-analyses. The MISCAN-PRO and SCANS models estimate an effective test sensitivity which combines the probability of a positive PSA test, any secondary testing, receipt of biopsy, and sensitivity of the biopsy to detect latent cancer.

Screening benefit

The PSAPC, MISCAN-PRO, and SCOPE models are designed to consider multiple mechanisms of screening benefit, including stage shift (i.e., that individuals whose cancers are detected by screening at an earlier stage and/or grade receive a corresponding improvement in cancer-specific survival) and cure rates (a fraction of individuals detected earlier by screening have their cancer death prevented by early detection; the cure rate may be constant across the population or may depend on lead time). The MSM-CanProg model uses grade/stage- and age-dependent survival mechanisms for screening benefit.

References

1. Etzioni R, Tsodikov A, Mariotto A, Szabo A, Falcon S, Wegelin J, DiTommaso D, Karnofski K, Gulati R, Penson DF, Feuer E. Quantifying the role of PSA screening in the US prostate cancer mortality decline. Cancer Causes Control. Mar;19(2):175-81.
2. Draisma G, Etzioni R, Tsodikov A, Mariotto A, Wever E, Gulati R, Feuer E, de Koning H. Lead time and overdiagnosis in prostate-specific antigen screening: importance of methods and context. J Natl Cancer Inst. 2009 Mar 18;101(6):374-83.
3. Gulati R, Wever EM, Tsodikov A, Penson DF, Inoue LY, Katcher J, Lee SY, Heijnsdijk EA, Draisma G, de Koning HJ, Etzioni R. What if I don't treat my PSA-detected prostate cancer? Answers from three natural history models. Cancer Epidemiol Biomarkers Prev. 2011 May;20(5):740-50.
4. Gulati R, Tsodikov A, Wever EM, Mariotto AB, Heijnsdijk EA, Katcher J, de Koning HJ, Etzioni R. The impact of PLCO control arm contamination on perceived PSA screening efficacy. Cancer Causes Control. Jun;23(6):827-35.
5. Etzioni R, Gulati R, Tsodikov A, Wever EM, Penson DF, Heijnsdijk EA, Katcher J, Draisma G, Feuer EJ, de Koning HJ, Mariotto AB. The prostate cancer conundrum revisited: treatment changes and prostate cancer mortality declines. Cancer. 2012 Dec 1;118(23):5955-63.
6. Gulati R, Tsodikov A, Etzioni R, Hunter-Merrill RA, Gore JL, Mariotto AB, Cooperberg MR. Expected population impacts of discontinued prostate-specific antigen screening. Cancer. 2014 Nov 15;120(22):3519-26.
7. Tsodikov A, Gulati R, de Carvalho TM, Heijnsdijk EA, Hunter-Merrill RA, Mariotto AB, de Koning HJ, Etzioni R. Is prostate cancer different in black men? Answers from 3 natural history models. Cancer. 2017 Jun 15;123(12):2312-2319.
8. Nyame YA, Gulati R, Heijnsdijk EA, Tsodikov A, Mariotto AB, Gore JL, Etzioni R. The Impact of Intensifying Prostate Cancer Screening in Black Men: A Model-Based Analysis. J Natl Cancer Inst. 2021 Oct 1;113(10):1336-1342.
9. Tsodikov A, Gulati R, Heijnsdijk EA, Pinsky PF, Moss SM, Qiu S, de Carvalho TM, Hugosson J, Berg CD, Auvinen A, Andriole GL, Roobol MJ, Crawford ED, Nelen V, Kwiatkowski M, Zappa M, Luján M, Villers A, Feuer EJ, de Koning HJ, Mariotto AB, Etzioni R. Reconciling the Effects of Screening on Prostate Cancer Mortality in the ERSPC and PLCO Trials. Ann Intern Med. 2017 Oct 3;167(7):449-455.
10. de Koning HJ, Gulati R, Moss SM, Hugosson J, Pinsky PF, Berg CD, Auvinen A, Andriole GL, Roobol MJ, Crawford ED, Nelen V, Kwiatkowski M, Zappa M, Luján M, Villers A, de Carvalho TM, Feuer EJ, Tsodikov A, Mariotto AB, Heijnsdijk EA, Etzioni R. The efficacy of prostate-specific antigen screening: Impact of key components in the ERSPC and PLCO trials. Cancer. 2018 Mar 15;124(6):1197-1206.
11. Heijnsdijk EA, Gulati R, Tsodikov A, Lange JM, Mariotto AB, Vickers AJ, Carlsson SV, Etzioni R. Lifetime Benefits and Harms of Prostate-Specific Antigen-Based Risk-Stratified Screening for Prostate Cancer. J Natl Cancer Inst. 2020 Oct 1;112(10):1013-1020.
12. Heijnsdijk EA, Gulati R, Lange JM, Tsodikov A, Roberts R, Etzioni R. Evaluation of Prostate Cancer Screening Strategies in a Low-Resource, High-risk Population in the Bahamas. JAMA Health Forum. 2022 May 20;3(5):e221116.