Table 1. Characteristics of mathematical modeling studies.
| Study ID | Type of modeling | Reference population | Risk factors 1 | Comparison groups | Strategies | Outcome measures |
|---|---|---|---|---|---|---|
| Trentham-Dietz 2016 | CISNET (microsimulation). Three different models | Women 50–74 years | Combination of mammographic density, and 4 levels of relative risk (RR: 1.0, 1.3, 2.0, 4.0) based on previously published evidence. | Reference relative risk (RR = 1) vs. RR> 1.3, RR>2.0, RR>4.0 | • Annual, biennial, triennial screening age 50–74 years vs no screening • Biennial screening age 50–64 years vs biennial screening 65–74 years |
Lifetime cost of breast cancer deaths, life expectancy and number of QALY, false-positive results, biopsies with a benign result, overdiagnosis, cost-effectiveness, and ratio of false-positive results among breast cancer deaths avoided by screening. |
| O'Mahony 2014 | Cost-effectiveness microsimulation, and MISCAN (Monte Carlo microsimulation) | Women 50–70 years | Increase or decrease in breast cancer incidence in the population (continuous value) | Different screening periodicities based on breast cancer annul incidence. Taking as reference a cost-effectiveness threshold of €20,000 per QALY for an average incidence of 0.00225 per women-year (1.9 years screening interval) | Screening periodicity (continuous time measure) according to breast cancer risk (continuous risk measure) | ICER, cost per QALY |
| Vilaprinyo 2014 | Lee-Zelen probabilistic model (multiestate model) | Women 40–79 years | Mammographic density, family history, previous biopsy | Low: BI-RADS A + one risk factor (RF) among: family history, or previous biopsies, or BI-RADS B without RF Medium-Low: BIRADS A + 2 RF; or BIRADS B + 1 RF; or BIRADS C or D without RF Medium-high: BIRADS B + 2 RF; or BIRADS C or D + 1 RF High: BIRADS C or D + 2 RF |
2624 strategies: • Screening start age (40, 45, 50 years) • Periodicity (annual, biennial, triennial, and quinquenial) • Screening stop age (69, 74 years) |
Benefits: Number of lives extended, and number of QALY gained. Adverse effects: False-positive results, interval cancers and false-negatives, overdiagnosis, DCIS due to screening Costs: ICER, incremental benefit-harm ratio. |
| Wu 2013 | Markov (microsimulation) | Women ≥ 50 years | BRCA, mammographic density, SNPs, BMI, and age at 1st pregnancy | Deciles of risk according to risk score distribution. Percentile 50–60 as reference. | a) Start age based on age at which the 10-year risk equals 1% of the 10-year risk of the 50th percentile of the risk score at age 50 (29 to 69 years). b) Screening interval (0.4 to 8 years) based on interval cancer rate that equals the threshold of triennial mammography for the 50th percentil of risk score. c) Mammography and MRI, or mammography and US based on the improvement in sensitivity obtained from decreasing the interval cancer rate until the percentile equals the median value of the population with triennial mammography alone. |
Number of mammograms, incidence of screen-detected cancer, incidence of interval cancer, proportion of interval cancers among breast cancer cases |
| Schousboe 2011 | Markov (cost-utility model) | Women 40–79 years | Mammographic density, family history, previous biopsy | Risk groups based on cost-effectiveness thresholds ($100,000 and $ 50,000 per QALY), and 10-year age groups (40–49, 50–59, 60–69, 70–79), breast density (BI-RADS), and number of risk factors (family history, previous biopsy). | Periodicity (no screening, annual, biennial, every 3–4 years) Strategy re-evaluation every ten years (40–49, 50–59, 60–69, 70–79) |
Cost per QALY gained. Number of women screened in a 10-year period to prevent one breast cancer death. |
| Ahern 2014 | Markov (Monte Carlo microsimulation) | Women 30–90 years with > 25% lifetime breast cancer risk | N.S. | Women with a lifetime breast cancer risk ≥ 25%, vs. women with a lifetime risk ≥ 50% and ≥ 75%. | 12 strategies: MRI (annual, biennial), mammography + clinic examination (none, 6 months, 1 year, 2 years), screening stop age (50, 74). |
Cost, survival (life years), and QALY ICERs |
| Pashayan 2011 | Probabilistic model | Women 35–79 years | Polygenic risk score (18 loci) | Women aged 47–79 years with 10 years absolute risk ≥ 2.5% vs women 35–79 years with 10 years absolute risk = 2.5%. | Mamography in women 47–79 years (absolute 10-year risk ≥ 2.5%) vs. Mammography age 35–79 years with a 10-year absolute risk = 2.5% based on age + SNPs | Number of women in the target population, number of breast cancers potentitally detectable at screening |
| Gray 2017 | Discrete Event Simulation | Women 50–70 years | Cuzick-Tyrer IBIS risk calculator (phenotype, age at menarche, number of pregnancies, age at first delivery, age at menopause, atypical hyperplasia, lubular carcinoma in sit, BMI) improved with mammographic density. |
Four interventions: 1) 3 strata based on 10-year risk: < 3.5%, 3.5%-8%, and >8%; 2) 3 strata based on 10-year risk terciles: lowest risk, intermediate risk, highest risk; 3) Masking in women with high mammographic density (Volpara density 3 or 4); 4) Masking in women with high mammographic density + 3 strata based on intervention 1. Comparison group: Current screening mammmogram every 3 years in women aged 50–70 years, and a no-screening strategy. |
Interventions 1 and 2: Mammography every 3, 2, or 1 year for the low, intermediate, and high risk groups, respectively; Intervention 3: Additional US for women with high breast density. If high risk woman (10-year risk >8%) additional MRI instead of US; Intervention 4: triennial, biennial, yearly mammography based on risk group for intervention 1. Additional US If high breast density. | QALY of each strategy, Cost, and ICERs |
| Van Dyck 2012 | Markov (cost-efectiveness) | Women ≥ 50 years | SNPs, breast cancer risk calculator, and risk factors available through the electronic health records system. | High and low risk 2 | High frequency vs low frequency screening strategy 3 | Total cost, and QALYs |
NS: Not specified; RR: Relative Risk; QALY: Quality-adjusted life years; SNPs: Single Nucleotide Polymorphism; ASSURE (Adapting Breast Cancer Screening Strategy Using Personalised Risk Estimation)
BI-RADS, Breast Imaging Reporting and Data System: A, almost entirely fat; B, scattered fibroglandular density; C, heterogeneously dense; D, extremely dense
1 Age is a risk factor in all models, except in the model by Omahony et al, which assumes a constant incidence rate of breast cancer in the group aged 50–70 years
2 The study does not specify how the risk groups were stratified
3 The study does not specify the high and low frequency strategies.