Chapter 7: The Wisconsin Breast Cancer Epidemiology Simulation Model

Examining breast cancer screening recommendations in Canada: The projected resource impact of screening among women aged 40-49

OBJECTIVE

To quantify the resource use of revising breast cancer screening guidelines to include average-risk women aged 40-49 years across Canada from 2024 to 2043 using a validated microsimulation model.

SETTING

OncoSim-Breast microsimulation platform was used to simulate the entire Canadian population in 2015-2051.

METHODS

We compared resource use between current screening guidelines (biennial screening ages 50-74) and alternate screening scenarios, which included annual and biennial screening for ages 40-49 and ages 45-49, followed by biennial screening ages 50-74. We estimated absolute and relative differences in number of screens, abnormal screening recalls without cancer, total and negative biopsies, screen-detected cancers, stage of diagnosis, and breast cancer deaths averted.

RESULTS

Compared with current guidelines in Canada, the most intensive screening scenario (annual screening ages 40-49) would result in 13.3% increases in the number of screens and abnormal screening recalls without cancer whereas the least intensive scenario (biennial screening ages 45-49) would result in a 3.4% increase in number of screens and 3.8% increase in number of abnormal screening recalls without cancer. More intensive screening would be associated with fewer stage II, III, and IV diagnoses, and more breast cancer deaths averted.

CONCLUSIONS

Revising breast cancer screening in Canada to include average-risk women aged 40-49 would detect cancers earlier leading to fewer breast cancer deaths. To realize this potential clinical benefit, a considerable increase in screening resources would be required in terms of number of screens and screen follow-ups. Further economic analyses are required to fully understand cost and budget implications.

Effect of Breast Screening Regimen on Breast Cancer Outcomes: A Modeling Study

Guidelines vary for the age at which to begin breast cancer screening and the interval between examinations. A validated computer model was used to compare estimated outcomes between various screening regimens. The OncoSim-Breast microsimulation model (Canadian Partnership Against Cancer) was used to simulate a cohort of 1.53 million Canadian women born in 1975. The effect of screening regimen on absolute breast cancer mortality rates, stage at diagnosis, number needed to be screened to avert a breast cancer death or save a life year, abnormal recall rates and negative biopsy rates was examined for unscreened women or those entering screening at age 40 or 50 and screened annually or biennially to age 74. Compared to no screening, absolute mortality reduction was 4.6 (biennial 50-74), 5.9 (biennial 40-74) and 7.9 (annual 40-74) fewer deaths per 1000 women. The absolute rate of diagnosis of advanced cancers (Stage 2, 3 and 4) falls in favor of earlier stages as the number of lifetime screens increases. Annual screening beginning at age 40 until age 74 would provide an additional reduction of 2 and 3.3 breast cancer deaths per 1000 women compared to biennial screening beginning at ages 40 and 50, respectively. There is a corresponding drop in the absolute number of Stage 2, 3 and 4 cancers diagnosed.

Non-progressive breast carcinomas detected at mammography screening: a population study

BACKGROUND

Some breast carcinomas detected at screening, especially ductal carcinoma in situ, may have limited potential for progression to symptomatic disease. To determine non-progression is a challenge, but if all screening-detected breast tumors eventually reach a clinical stage, the cumulative incidence at a reasonably high age would be similar for women with or without screening, conditional on the women being alive.

METHODS

Using high-quality population data with 24 years of follow-up from the gradually introduced BreastScreen Norway program, we studied whether all breast carcinomas detected at mammography screening 50-69 years of age would progress to clinical symptoms within 85 years of age. First, we estimated the incidence rates of breast carcinomas by age in scenarios with or without screening, based on an extended age-period-cohort incidence model. Next, we estimated the frequency of non-progressive tumors among screening-detected cases, by calculating the difference in the cumulative rate of breast carcinomas between the screening and non-screening scenarios at 85 years of age.

RESULTS

Among women who attended BreastScreen Norway from the age of 50 to 69 years, we estimated that 1.1% of the participants were diagnosed with a breast carcinoma without the potential to progress to symptomatic disease by 85 years of age. This proportion of potentially non-progressive tumors corresponded to 15.7% [95% CI 3.3, 27.1] of breast carcinomas detected at screening.

CONCLUSIONS

Our findings suggest that nearly one in six breast carcinomas detected at screening may be non-progressive.

A mechanistic analysis of spontaneous cancer remission phenomenon: identification of genomic basis and effector biomolecules for therapeutic applicability

Based on the well-documented studies, numerous tumors episodically regress permanently without treatment. Knowing the host tissue-initiated causative factors would offer considerable translational applicability, as a permanent regression process may be therapeutically replicated on patients. For this, we developed a systems biological formulation of the regression process with experimental verification and identified the relevant candidate biomolecules for therapeutic utility. We devised a cellular kinetics-based quantitative model of tumor extinction in terms of the temporal behavior of three main tumor-lysis entities: DNA blockade factor, cytotoxic T-lymphocyte and interleukin-2. As a case study, we analyzed the time-wise biopsy and microarrays of spontaneously regressing melanoma and fibrosarcoma tumors in mammalian/human hosts. We analyzed the differentially expressed genes (DEGs), signaling pathways, and bioinformatics framework of regression. Additionally, prospective biomolecules that could cause complete tumor regression were investigated. The tumor regression process follows a first-order cellular dynamics with a small negative bias, as verified by experimental fibrosarcoma regression; the bias is necessary to eliminate the residual tumor. We identified 176 upregulated and 116 downregulated DEGs, and enrichment analysis showed that the most significant were downregulated cell-division genes: TOP2A-KIF20A-KIF23-CDK1-CCNB1. Moreover, Topoisomerase-IIA inhibition might actuate spontaneous regression, with collateral confirmation provided from survival and genomic analysis of melanoma patients. Candidate molecules such as Dexrazoxane/Mitoxantrone, with interleukin-2 and antitumor lymphocytes, may potentially replicate permanent tumor regression process of melanoma. To conclude, episodic permanent tumor regression is a unique biological reversal process of malignant progression, and signaling pathway understanding, with candidate biomolecules, may plausibly therapeutically replicate the regression process on tumors clinically.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13205-023-03515-0.

Computational systems biology approach for permanent tumor elimination and normal tissue protection using negative biasing: Experimental validation in malignant melanoma as case study

Complete spontaneous tumor regression (without treatment) is well documented to occur in animals and humans as epidemiological analysis show, whereby the malignancy is permanently eliminated. We have developed a novel computational systems biology model for this unique phenomenon to furnish insight into the possibility of therapeutically replicating such regression processes on tumors clinically, without toxic side effects. We have formulated oncological informatics approach using cell-kinetics coupled differential equations while protecting normal tissue. We investigated three main tumor-lysis components: (ⅰ) DNA blockade factors, (ⅱ) Interleukin-2 (IL-2), and (ⅲ) Cytotoxic T-cells (CD8⁺ T). We studied the temporal variations of these factors, utilizing preclinical experimental investigations on malignant tumors, using mammalian melanoma microarray and histiocytoma immunochemical assessment. We found that permanent tumor regression can occur by: 1) Negative-Bias shift in population trajectory of tumor cells, eradicating them under first-order asymptotic kinetics, and 2) Temporal alteration in the three antitumor components (DNA replication-blockade, Antitumor T-lymphocyte, IL-2), which are respectively characterized by the following patterns: (a) Unimodal Inverted-U function, (b) Bimodal M-function, (c) Stationary-step function. These provide a time-wise orchestrated tri-phasic cytotoxic profile. We have also elucidated gene-expression levels corresponding to the above three components: (ⅰ) DNA-damage G2/M checkpoint regulation [genes: CDC2-CHEK], (ⅱ) Chemokine signaling: IL-2/15 [genes: IL2RG-IKT3], (ⅲ) T-lymphocyte signaling (genes: TRGV5-CD28). All three components quantitatively followed the same activation profiles predicted by our computational model (Smirnov-Kolmogorov statistical test satisfied, α = 5%). We have shown that the genes CASP7-GZMB are signatures of Negative-bias dynamics, enabling eradication of the residual tumor. Using the negative-biasing principle, we have furnished the dose-time profile of equivalent therapeutic agents (DNA-alkylator, IL-2, T-cell input) so that melanoma tumor may therapeutically undergo permanent extinction by replicating the spontaneous tumor regression dynamics.

Patient Navigation Can Improve Breast Cancer Outcomes among African American Women in Chicago: Insights from a Modeling Study

African American (AA) women experience much greater mortality due to breast cancer (BC) than non-Latino Whites (NLW). Clinical patient navigation is an evidence-based strategy used by healthcare institutions to improve AA women's breast cancer outcomes. While empirical research has demonstrated the potential effect of navigation interventions for individuals, the population-level impact of navigation on screening, diagnostic completion, and stage at diagnosis has not been assessed. An agent-based model (ABM), representing 50-74-year-old AA women and parameterized with locally sourced data from Chicago, is developed to simulate screening mammography, diagnostic resolution, and stage at diagnosis of cancer. The ABM simulated three counterfactual scenarios: (1) a control setting without any navigation that represents the "standard of care"; (2) a clinical navigation scenario, where agents receive navigation from hospital-affiliated staff; and (3) a setting with network navigation, where agents receive clinical navigation and/or social network navigation (i.e., receiving support from clinically navigated agents for breast cancer care). In the control setting, the mean population-level screening mammography rate was 46.3% (95% CI: 46.2%, 46.4%), the diagnostic completion rate was 80.2% (95% CI: 79.9%, 80.5%), and the mean early cancer diagnosis rate was 65.9% (95% CI: 65.1%, 66.7%). Simulation results suggest that network navigation may lead up to a 13% increase in screening completion rate, 7.8% increase in diagnostic resolution rate, and a 4.9% increase in early-stage diagnoses at the population-level. Results suggest that systems science methods can be useful in the adoption of clinical and network navigation policies to reduce breast cancer disparities.

Active Learning for Multi-way Sensitivity Analysis with Application to Disease Screening Modeling

Sensitivity analysis is an important aspect of model development as it can be used to assess the level of confidence that is associated with the outcomes of a study. In many practical problems, sensitivity analysis involves evaluating a large number of parameter combinations which may require an extensive amount of time and resources. However, such a computational burden can be avoided by identifying smaller subsets of parameter combinations that can be later used to generate the desired outcomes for other parameter combinations. In this study, we investigate machine learning-based approaches for speeding up the sensitivity analysis. Furthermore, we apply feature selection methods to identify the relative importance of quantitative model parameters in terms of their predictive ability on the outcomes. Finally, we highlight the effectiveness of active learning strategies in improving the sensitivity analysis processes by reducing the total number of quantitative model runs required to construct a high-performance prediction model. Our experiments on two datasets obtained from the sensitivity analysis performed for two disease screening modeling studies indicate that ensemble methods such as Random Forests and XGBoost consistently outperform other machine learning algorithms in the prediction task of the associated sensitivity analysis. In addition, we note that active learning can lead to significant speed-ups in sensitivity analysis by enabling the selection of more useful parameter combinations (i.e., instances) to be used for prediction models.

Breast Cancer Screening Strategies for Women With ATM, CHEK2, and PALB2 Pathogenic Variants: A Comparative Modeling Analysis

IMPORTANCE

Screening mammography and magnetic resonance imaging (MRI) are recommended for women with ATM, CHEK2, and PALB2 pathogenic variants. However, there are few data to guide screening regimens for these women.

OBJECTIVE

To estimate the benefits and harms of breast cancer screening strategies using mammography and MRI at various start ages for women with ATM, CHEK2, and PALB2 pathogenic variants.

DESIGN, SETTING, AND PARTICIPANTS

This comparative modeling analysis used 2 established breast cancer microsimulation models from the Cancer Intervention and Surveillance Modeling Network (CISNET) to evaluate different screening strategies. Age-specific breast cancer risks were estimated using aggregated data from the Cancer Risk Estimates Related to Susceptibility (CARRIERS) Consortium for 32 247 cases and 32 544 controls in 12 population-based studies. Data on screening performance for mammography and MRI were estimated from published literature. The models simulated US women with ATM, CHEK2, or PALB2 pathogenic variants born in 1985.

INTERVENTIONS

Screening strategies with combinations of annual mammography alone and with MRI starting at age 25, 30, 35, or 40 years until age 74 years.

MAIN OUTCOMES AND MEASURES

Estimated lifetime breast cancer mortality reduction, life-years gained, breast cancer deaths averted, total screening examinations, false-positive screenings, and benign biopsies per 1000 women screened. Results are reported as model mean values and ranges.

RESULTS

The mean model-estimated lifetime breast cancer risk was 20.9% (18.1%-23.7%) for women with ATM pathogenic variants, 27.6% (23.4%-31.7%) for women with CHEK2 pathogenic variants, and 39.5% (35.6%-43.3%) for women with PALB2 pathogenic variants. Across pathogenic variants, annual mammography alone from 40 to 74 years was estimated to reduce breast cancer mortality by 36.4% (34.6%-38.2%) to 38.5% (37.8%-39.2%) compared with no screening. Screening with annual MRI starting at 35 years followed by annual mammography and MRI at 40 years was estimated to reduce breast cancer mortality by 54.4% (54.2%-54.7%) to 57.6% (57.2%-58.0%), with 4661 (4635-4688) to 5001 (4979-5023) false-positive screenings and 1280 (1272-1287) to 1368 (1362-1374) benign biopsies per 1000 women. Annual MRI starting at 30 years followed by mammography and MRI at 40 years was estimated to reduce mortality by 55.4% (55.3%-55.4%) to 59.5% (58.5%-60.4%), with 5075 (5057-5093) to 5415 (5393-5437) false-positive screenings and 1439 (1429-1449) to 1528 (1517-1538) benign biopsies per 1000 women. When starting MRI at 30 years, initiating annual mammography starting at 30 vs 40 years did not meaningfully reduce mean mortality rates (0.1% [0.1%-0.2%] to 0.3% [0.2%-0.3%]) but was estimated to add 649 (602-695) to 650 (603-696) false-positive screenings and 58 (41-76) to 59 (41-76) benign biopsies per 1000 women.

CONCLUSIONS AND RELEVANCE

This analysis suggests that annual MRI screening starting at 30 to 35 years followed by annual MRI and mammography at 40 years may reduce breast cancer mortality by more than 50% for women with ATM, CHEK2, and PALB2 pathogenic variants. In the setting of MRI screening, mammography prior to 40 years may offer little additional benefit.

Cancer Modeling as Learning Experience

Cancer modeling has become an accepted method for generating evidence about comparative effectiveness and cost-effectiveness of candidate cancer control policies across the continuum of care. Models of early detection policies require inputs concerning disease natural history and screening test performance, which are often subject to considerable uncertainty. Model validation against an external data source can increase confidence in the reliability of assumed or calibrated inputs. When a model fails to validate, this presents an opportunity to revise these inputs, thereby learning new information about disease natural history or diagnostic performance that could both enhance the model results and inform real-world practices. We discuss the conditions necessary for validly drawing conclusions about specific inputs such as diagnostic performance from model validation studies. Doing so requires being able to faithfully replicate the validation study in terms of its design and implementation and being alert to the problem of non-identifiability, which could lead to explanations for failure to validate other than those identified. See related article by Rutter et al., p. 775.

Estimation of Breast Cancer Overdiagnosis in a U.S. Breast Screening Cohort

BACKGROUND

Mammography screening can lead to overdiagnosis-that is, screen-detected breast cancer that would not have caused symptoms or signs in the remaining lifetime. There is no consensus about the frequency of breast cancer overdiagnosis.

OBJECTIVE

To estimate the rate of breast cancer overdiagnosis in contemporary mammography practice accounting for the detection of nonprogressive cancer.

DESIGN

Bayesian inference of the natural history of breast cancer using individual screening and diagnosis records, allowing for nonprogressive preclinical cancer. Combination of fitted natural history model with life-table data to predict the rate of overdiagnosis among screen-detected cancer under biennial screening.

SETTING

Breast Cancer Surveillance Consortium (BCSC) facilities.

PARTICIPANTS

Women aged 50 to 74 years at first mammography screen between 2000 and 2018.

MEASUREMENTS

Screening mammograms and screen-detected or interval breast cancer.

RESULTS

The cohort included 35 986 women, 82 677 mammograms, and 718 breast cancer diagnoses. Among all preclinical cancer cases, 4.5% (95% uncertainty interval [UI], 0.1% to 14.8%) were estimated to be nonprogressive. In a program of biennial screening from age 50 to 74 years, 15.4% (UI, 9.4% to 26.5%) of screen-detected cancer cases were estimated to be overdiagnosed, with 6.1% (UI, 0.2% to 20.1%) due to detecting indolent preclinical cancer and 9.3% (UI, 5.5% to 13.5%) due to detecting progressive preclinical cancer in women who would have died of an unrelated cause before clinical diagnosis.

LIMITATIONS

Exclusion of women with first mammography screen outside BCSC.

CONCLUSION

On the basis of an authoritative U.S. population data set, the analysis projected that among biennially screened women aged 50 to 74 years, about 1 in 7 cases of screen-detected cancer is overdiagnosed. This information clarifies the risk for breast cancer overdiagnosis in contemporary screening practice and should facilitate shared and informed decision making about mammography screening.

PRIMARY FUNDING SOURCE

National Cancer Institute.

The impact of episodic screening interruption: COVID-19 and population-based cancer screening in Canada

BACKGROUND

Population-based cancer screening can reduce cancer burden but was interrupted temporarily due to the COVID-19 pandemic. We estimated the long-term clinical impact of breast and colorectal cancer screening interruptions in Canada using a validated mathematical model.

METHODS

We used the OncoSim breast and colorectal cancers microsimulation models to explore scenarios of primary screening stops for 3, 6, and 12 months followed by 6-24-month transition periods of reduced screening volumes. For breast cancer, we estimated changes in cancer incidence over time, additional advanced-stage cases diagnosed, and excess cancer deaths in 2020-2029. For colorectal cancer, we estimated changes in cancer incidence over time, undiagnosed advanced adenomas and colorectal cancers in 2020, and lifetime excess cancer incidence and deaths.

RESULTS

Our simulations projected a surge of cancer cases when screening resumes. For breast cancer screening, a three-month interruption could increase cases diagnosed at advanced stages (310 more) and cancer deaths (110 more) in 2020-2029. A six-month interruption could lead to 670 extra advanced cancers and 250 additional cancer deaths. For colorectal cancers, a six-month suspension of primary screening could increase cancer incidence by 2200 cases with 960 more cancer deaths over the lifetime. Longer interruptions, and reduced volumes when screening resumes, would further increase excess cancer deaths.

CONCLUSIONS

Interruptions in cancer screening will lead to additional cancer deaths, additional advanced cancers diagnosed, and a surge in demand for downstream resources when screening resumes. An effective strategy is needed to minimize potential harm to people who missed their screening.

Bayesian versus Empirical Calibration of Microsimulation Models: A Comparative Analysis

Calibration of a microsimulation model (MSM) is a challenging but crucial step for the development of a valid model. Numerous calibration methods for MSMs have been suggested in the literature, most of which are usually adjusted to the specific needs of the model and based on subjective criteria for the selection of optimal parameter values. This article compares 2 general approaches for calibrating MSMs used in medical decision making, a Bayesian and an empirical approach. We use as a tool the MIcrosimulation Lung Cancer (MILC) model, a streamlined, continuous-time, dynamic MSM that describes the natural history of lung cancer and predicts individual trajectories accounting for age, sex, and smoking habits. We apply both methods to calibrate MILC to observed lung cancer incidence rates from the Surveillance, Epidemiology and End Results (SEER) database. We compare the results from the 2 methods in terms of the resulting parameter distributions, model predictions, and efficiency. Although the empirical method proves more practical, producing similar results with smaller computational effort, the Bayesian method resulted in a calibrated model that produced more accurate outputs for rare events and is based on a well-defined theoretical framework for the evaluation and interpretation of the calibration outcomes. A combination of the 2 approaches is an alternative worth considering for calibrating complex predictive models, such as microsimulation models.

Cost-effectiveness of risk-based breast cancer screening: A systematic review

To analyse published evidence on the economic evaluation of risk-based screening (RBS), a full systematic literature review was conducted. After a quality appraisal, we compared the cost-effectiveness of risk-based strategies (low-risk, medium-risk and high-risk) with no screening and age-based screening. Studies were also analysed for modelling, risk stratification methods, input parameters, data sources and harms and benefits. The 10 modelling papers analysed were based on screening performance of film-based mammography (FBM) (three); digital mammography (DM) and FBM (two); DM alone (three); DM, ultrasound (US) and magnetic resonance imaging (one) and DM and US (one). Seven studies did not include the cost of risk-stratification, and one did not consider the cost of diagnosis. Disutility was incorporated in only six studies (one for screening and five for diagnosis). None of the studies reported disutility of risk-stratification (being considered as high-risk). Risk-stratification methods varied from only breast density (BD) to the combination of familial risk, genetic susceptibility, lifestyle, previous biopsies, Jewish ancestry and reproductive history. Less or no screening in low-risk women and more frequent mammography screening in high-risk women was more cost-effective compared to no screening and age-based screening. High-risk women screened annually yielded a higher mortality rate reduction and more quality-adjusted life years at the expense of higher cost and false positives. RBS can be cost effective compared to the alternatives. However, heterogeneity among risk-stratification methods, input parameters, and weaknesses in the methodologies hinder the derivation of robust conclusions. Therefore, further studies are warranted to assess newer technologies and innovative risk-stratification methods.

Simulation model of disease incidence driven by diagnostic activity

It is imperative to understand the effects of early detection and treatment of chronic diseases, such as prostate cancer, regarding incidence, overtreatment and mortality. Previous simulation models have emulated clinical trials, and relied on extensive assumptions on the natural history of the disease. In addition, model parameters were typically calibrated to a variety of data sources. We propose a model designed to emulate real‐life scenarios of chronic disease using a proxy for the diagnostic activity without explicitly modeling the natural history of the disease and properties of clinical tests. Our model was applied to Swedish nation‐wide population‐based prostate cancer data, and demonstrated good performance in terms of reconstructing observed incidence and mortality. The model was used to predict the number of prostate cancer diagnoses with a high or limited diagnostic activity between 2017 and 2060. In the long term, high diagnostic activity resulted in a substantial increase in the number of men diagnosed with lower risk disease, fewer men with metastatic disease, and decreased prostate cancer mortality. The model can be used for prediction of outcome, to guide decision‐making, and to evaluate diagnostic activity in real‐life settings with respect to overdiagnosis and prostate cancer mortality.

The Value of All-Cause Mortality as a Metric for Assessing Breast Cancer Screening

Although screening mammography has been demonstrated to contribute to reducing mortality due to breast cancer, some have suggested that reduced all-cause mortality should constitute the burden of proof for effectiveness. Using a microsimulation model of the development, detection, and treatment of breast cancer, it is straightforward to demonstrate that this is an unrealistic expectation for trials of practical size and period of observation, even where the reduction of breast cancer mortality is substantial. Estimates of all-cause mortality will depend not only on the efficacy of the screening intervention but also on the alignment between the age distribution of the effect of screening on reduction of deaths and that of the other major causes of death. The size of a randomized trial required to demonstrate a reduction in all-cause mortality will, therefore, depend on the length and timing of the observation period and will typically be at least 10 times larger than the size of a trial powered to test for a reduction in deaths due to breast cancer. For breast cancer, which represents a small fraction of overall deaths, all-cause mortality is neither a practical nor informative metric for assessing the effectiveness of screening.

Estimating the natural progression of non-invasive ductal carcinoma in situ breast cancer lesions using screening data

OBJECTIVES

In addition to invasive breast cancer, mammography screening often detects preinvasive ductal carcinoma in situ (DCIS) lesions. The natural progression of DCIS is largely unknown, leading to uncertainty regarding treatment. The natural history of invasive breast cancer has been studied using screening data. DCIS modeling is more complicated because lesions might progress to clinical DCIS, preclinical invasive cancer, or may also regress to a state undetectable by screening. We have here developed a Markov model for DCIS progression, building on the established invasive breast cancer model.

METHODS

We present formulas for the probability of DCIS detection by time since last screening under a Markov model of DCIS progression. Progression rates were estimated by maximum likelihood estimation using BreastScreen Norway data from 1995-2002 for 336,533 women (including 399 DCIS cases) aged 50-69. As DCIS incidence varies by age, county, and mammography modality (digital vs. analog film), a Poisson regression approach was used to align the input data.

RESULTS

Estimated mean sojourn time in preclinical, screening-detectable DCIS phase was 3.1 years (95% confidence interval: 1.3, 7.6) with a screening sensitivity of 60% (95% confidence interval: 32%, 93%). No DCIS was estimated to be non-progressive.

CONCLUSION

Most preclinical DCIS lesions progress or regress with a moderate sojourn time in the screening-detectable phase. While DCIS mean sojourn time could be deduced from DCIS data, any estimate of preclinical DCIS progressing to invasive breast cancer must include data on invasive cancers to avoid strong, probably unrealistic, assumptions.

Analysis of Mammography Screening Schedules under Varying Resource Constraints for Planning Breast Cancer Control Programs in Low- and Middle-Income Countries: A Mathematical Study

Background. Low-and-middle-income countries (LMICs) have higher mortality-to-incidence ratio for breast cancer compared to high-income countries (HICs) because of late-stage diagnosis. Mammography screening is recommended for early diagnosis, however, the infrastructure capacity in LMICs are far below that needed for adopting current screening guidelines. Current guidelines are extrapolations from HICs, as limited data had restricted model development specific to LMICs, and thus, economic analysis of screening schedules specific to infrastructure capacities are unavailable. Methods. We applied a new Markov process method for developing cancer progression models and a Markov decision process model to identify optimal screening schedules under a varying number of lifetime screenings per person, a proxy for infrastructure capacity. We modeled Peru, a middle-income country, as a case study and the United States, an HIC, for validation. Results. Implementing 2, 5, 10, and 15 lifetime screens would require about 55, 135, 280, and 405 mammography machines, respectively, and would save 31, 62, 95, and 112 life-years per 1000 women, respectively. Current guidelines recommend 15 lifetime screens, but Peru has only 55 mammography machines nationally. With this capacity, the best strategy is 2 lifetime screenings at age 50 and 56 years. As infrastructure is scaled up to accommodate 5 and 10 lifetime screens, screening between the ages of 44-61 and 41-64 years, respectively, would have the best impact. Our results for the United States are consistent with other models and current guidelines. Limitations. The scope of our model is limited to analysis of national-level guidelines. We did not model heterogeneity across the country. Conclusions. Country-specific optimal screening schedules under varying infrastructure capacities can systematically guide development of cancer control programs and planning of health investments.

Microsimulation Modeling in Oncology

PURPOSE

Microsimulation is a modeling technique that uses a sample size of individual units (microunits), each with a unique set of attributes, and allows for the simulation of downstream events on the basis of predefined states and transition probabilities between those states over time. In this article, we describe the history of the role of microsimulation in medicine and its potential applications in oncology as useful tools for population risk stratification and treatment strategy design for precision medicine.

METHODS

We conducted a comprehensive and methodical search of the literature using electronic databases-Medline, Embase, and Cochrane-for works published between 1985 and 2016. A medical subject heading search strategy was constructed for Medline searches by using a combination of relevant search terms, such as "microsimulation model medicine," "multistate modeling cancer," and "oncology."

RESULTS

Microsimulation modeling is particularly useful for the study of optimal intervention strategies when randomized control trials may not be feasible, ethical, or practical. Microsimulation models can retain memory of prior behaviors and states. As such, it allows an explicit representation and understanding of how various processes propagate over time and affect the final outcomes for an individual or in a population.

CONCLUSION

A well-calibrated microsimulation model can be used to predict the outcome of the event of interest for a new individual or subpopulations, assess the effectiveness and cost effectiveness of alternative interventions, and project the future disease burden of oncologic diseases. In the growing field of oncology research, a microsimulation model can serve as a valuable tool among the various facets of methodology available.

Breast Cancer Screening: Beyond Mortality

Traditionally, the effectiveness of breast cancer screening has been measured in terms of reducing the number of deaths attributable to breast cancer. Other metrics such as the number of life-years or quality-adjusted life-years gained through screening may be more relevant and certainly may better reflect the important burden of the disease on younger women, their families, and society. The effects of earlier detection of breast cancer in reducing morbidities associated with treatment have often also been neglected. In addition, the harms and limitations associated with cancer screening have been poorly quantified and are seldom put into perspective vis-à-vis the benefits. Here, these alternative measures will be discussed and quantified.

Early detection and treatment strategies for breast cancer in low-income and upper middle-income countries: a modelling study

BACKGROUND

Poor breast cancer survival in low-income and middle-income countries (LMICs) can be attributed to advanced-stage presentation and poor access to systemic therapy. We aimed to estimate the outcomes of different early detection strategies in combination with systemic chemotherapy and endocrine therapy in LMICs.

METHODS

We adapted a microsimulation model to project outcomes of three early detection strategies alone or in combination with three systemic treatment programmes beyond standard of care (programme A): programme B was endocrine therapy for all oestrogen-receptor (ER)-positive cases; programme C was programme B plus chemotherapy for ER-negative cases; programme D was programme C plus chemotherapy for advanced ER-positive cases. The main outcomes were reductions in breast cancer-related mortality and lives saved per 100 000 women relative to the standard of care for women aged 30-49 years in a low-income setting (East Africa; using incidence data and life tables from Uganda and data on tumour characteristics from various East African countries) and for women aged 50-69 years in a middle-income setting (Colombia).

FINDINGS

In the East African setting, relative mortality reductions were 8-41%, corresponding to 23 (95% uncertainty interval -12 to 49) to 114 (80 to 138) lives saved per 100 000 women over 10 years. In Colombia, mortality reductions were 7-25%, corresponding to 32 (-29 to 70) to 105 (61 to 141) lives saved per 100 000 women over 10 years.

INTERPRETATION

The best projected outcomes were in settings where access to both early detection and adjuvant therapy is improved. Even in the absence of mammographic screening, improvements in detection can provide substantial benefit in settings where advanced-stage presentation is common.

FUNDING

Fred Hutchinson Cancer Research Center/University of Washington Cancer Consortium Cancer Center Support Grant of the US National Institutes of Health.

[Discrepancies and overdiagnosis in breast cancer organized screening. A "methodology" systematic review]

BACKGROUND

The risk-benefit ratio of breast cancer organized screening is the focus of much scientific controversy, especially about overdiagnosis. The aim of this study was to relate methodological discrepancies to variations in rates of overdiagnosis to help build future decision aids and to better communicate with patients.

METHODS

A systematic review of methodology was conducted by two investigators who searched Medline and Cochrane databases from 01/01/2004 to 12/31/2016. Results were restricted to randomized controlled trials (RCTs) and observational studies in French or English that examined the question of the overdiagnosis computation.

RESULTS

Twenty-three observational studies and four RCTs were analyzed. The methods used comparisons of annual or cumulative incidence rates (age-cohort model) in populations invited to screen versus non-invited populations. Lead time and ductal carcinoma in situ (DCIS) were often taken into account. Some studies used statistical modeling based on the natural history of breast cancer and gradual screening implementation. Adjustments for lead time lowered the rate of overdiagnosis. Rate discrepancies, ranging from 1 to 15 % for some authors and around 30 % for others, could be explained by the hypotheses accepted concerning very slow growing tumors or tumors that regress spontaneously.

CONCLUSION

Apparently, research has to be centered on the natural history of breast cancer in order to provide responses concerning the questions raised by the overdiagnosis controversy.

The effect of mammography screening regimen on incidence-based breast cancer mortality

OBJECTIVES

Incidence-based mortality quantifies the distribution of cancer deaths and life-years lost, according to age at detection. We investigated the temporal distribution of the disease burden, and the effect of starting and stopping ages and interval between screening mammography examinations, on incidence-based mortality.

METHODS

Incidence-based mortality was estimated using an established breast cancer simulation model, adapted and validated to simulate breast cancer incidence, screening performance, and delivery of therapies in Canada. Ten strategies were examined, with varying starting age (40 or 50), stopping age (69 or 74), and interval (1, 2, 3 years), and "No Screening." Life-years lost were computed as the difference between model predicted time of breast cancer death and that estimated from life tables.

RESULTS

Without screening, 70% of the burden in terms of breast cancer deaths extends between ages 45 and 75. The mean of the distribution of ages of detection of breast cancers that will be fatal in an unscreened population is 61.8 years, while the mean age of detection weighted by the number of life-years lost is 55, a downward shift of 6.8 years. Similarly, the mean age of detection for the distribution of life-years gained through screening is lower than that for breast cancer deaths averted.

CONCLUSION

Incidence-based mortality predictions from modeling elucidate the age dependence of the breast cancer burden and can provide guidance for optimizing the timing of screening regimens to achieve maximal impact. Of the regimens studied, the greatest lifesaving effect was achieved with annual screening beginning at age 40.

The University of Wisconsin Breast Cancer Epidemiology Simulation Model: An Update

The University of Wisconsin Breast Cancer Epidemiology Simulation Model (UWBCS), also referred to as Model W, is a discrete-event microsimulation model that uses a systems engineering approach to replicate breast cancer epidemiology in the US over time. This population-based model simulates the lifetimes of individual women through 4 main model components: breast cancer natural history, detection, treatment, and mortality. A key feature of the UWBCS is that, in addition to specifying a population distribution in tumor growth rates, the model allows for heterogeneity in tumor behavior, with some tumors having limited malignant potential (i.e., would never become fatal in a woman's lifetime if left untreated) and some tumors being very aggressive based on metastatic spread early in their onset. The model is calibrated to Surveillance, Epidemiology, and End Results (SEER) breast cancer incidence and mortality data from 1975 to 2010, and cross-validated against data from the Wisconsin cancer reporting system. The UWBCS model generates detailed outputs including underlying disease states and observed clinical outcomes by age and calendar year, as well as costs, resource usage, and quality of life associated with screening and treatment. The UWBCS has been recently updated to account for differences in breast cancer detection, treatment, and survival by molecular subtypes (defined by ER/HER2 status), to reflect the recent advances in screening and treatment, and to consider a range of breast cancer risk factors, including breast density, race, body-mass-index, and the use of postmenopausal hormone therapy. Therefore, the model can evaluate novel screening strategies, such as risk-based screening, and can assess breast cancer outcomes by breast cancer molecular subtype. In this article, we describe the most up-to-date version of the UWBCS.

A Two-Step Markov Processes Approach for Parameterization of Cancer State-Transition Models for Low- and Middle-Income Countries

Implementation of organized cancer screening and prevention programs in high-income countries (HICs) has considerably decreased cancer-related incidence and mortality. In low- and middle-income countries (LMICs), screening and early diagnosis programs are generally unavailable, and most cancers are diagnosed in late stages when survival is very low. Analyzing the cost-effectiveness of alternative cancer control programs and estimating resource needs will help prioritize interventions in LMICs. However, mathematical models of natural cancer onset and progression needed to conduct the economic analyses are predominantly based on populations in HICs because the longitudinal data on screening and diagnoses required for parameterization are unavailable in LMICs. Models currently used for LMICs mostly concentrate on directly calculating the shift in distribution of cancer diagnosis as an evaluative measure of screening. We present a mathematical methodology for the parameterization of natural cancer onset and progression, specifically for LMICs that do not have longitudinal data. This full onset and progression model can help conduct comprehensive analyses of cancer control programs, including cancer screening, by considering both the positive impact of screening as well as any adverse consequences, such as over-diagnosis and false-positive results. The methodology has been applied to breast, cervical, and colorectal cancers for 2 regions, under the World Health Organization categorization: Eastern Sub-Saharan Africa (AFRE) and Southeast Asia (SEARB). The cancer models have been incorporated into the Spectrum software and interfaced with country-specific demographic data through the demographic projections (DemProj) module and costing data through the OneHealth tool. These software are open-access and can be used by stakeholders to analyze screening strategies specific to their country of interest.

Implementing parallel spreadsheet models for health policy decisions: The impact of unintentional errors on model projections

Background

Spreadsheet software is increasingly used to implement systems science models informing health policy decisions, both in academia and in practice where technical capacity may be limited. However, spreadsheet models are prone to unintentional errors that may not always be identified using standard error-checking techniques. Our objective was to illustrate, through a methodologic case study analysis, the impact of unintentional errors on model projections by implementing parallel model versions.

Methods

We leveraged a real-world need to revise an existing spreadsheet model designed to inform HIV policy. We developed three parallel versions of a previously validated spreadsheet-based model; versions differed by the spreadsheet cell-referencing approach (named single cells; column/row references; named matrices). For each version, we implemented three model revisions (re-entry into care; guideline-concordant treatment initiation; immediate treatment initiation). After standard error-checking, we identified unintentional errors by comparing model output across the three versions. Concordant model output across all versions was considered error-free. We calculated the impact of unintentional errors as the percentage difference in model projections between model versions with and without unintentional errors, using +/-5% difference to define a material error.

Results

We identified 58 original and 4,331 propagated unintentional errors across all model versions and revisions. Over 40% (24/58) of original unintentional errors occurred in the column/row reference model version; most (23/24) were due to incorrect cell references. Overall, >20% of model spreadsheet cells had material unintentional errors. When examining error impact along the HIV care continuum, the percentage difference between versions with and without unintentional errors ranged from +3% to +16% (named single cells), +26% to +76% (column/row reference), and 0% (named matrices).

Conclusions

Standard error-checking techniques may not identify all errors in spreadsheet-based models. Comparing parallel model versions can aid in identifying unintentional errors and promoting reliable model projections, particularly when resources are limited.

Evaluating different breast tumor progression models using screening data

Background

Mammography screening is used to detect breast cancer at an early treatable stage, reducing breast cancer mortality. Traditionally, breast cancer has been seen as a disease with only progressive lesions, and here we examine the validity of this assumption by testing if incidence levels after introducing mammography screening can be reproduced assuming only progressive tumors.

Methods

Breast cancer incidence data 1990–2009 obtained from the initially screened Norwegian counties (Akershus, Oslo, Rogaland and Hordaland) was included, covering the time-period before, during and after the introduction of mammography screening. From 1996 women aged 50–69 were invited for biennial public screening. Using estimates of tumor growth and screening sensitivity based on pre-screening and prevalence screening data (1990–1998), we simulated incidence levels during the following period (1999–2009).

Results

The simulated incidence levels during the period with repeated screenings were markedly below the observed levels. The results were robust to changes in model parameters. Adjusting for hormone replacement therapy use, we obtained levels closer to the observed levels. However, there was still a marked gap, and only by assuming some tumors that undergo regressive changes or enter a markedly less detectable state, was our model able to reproduce the observed incidence levels.

Conclusions

Models with strictly progressive tumors are only able to partly explain the changes in incidence levels observed after screening introduction in the initially screened Norwegian counties. More complex explanations than a time shift in detection of future clinical cancers seem to be needed to reproduce the incidence trends, questioning the basis for many over-diagnosis calculations. As data are not randomized, similar studies in other populations are wanted to exclude effect of unknown confounders.

Electronic supplementary material

The online version of this article (10.1186/s12885-018-4130-2) contains supplementary material, which is available to authorized users.

Cost-effectiveness of mammography from a publicly funded health care system perspective

BACKGROUND

The implementation of population-wide breast cancer screening programs has important budget implications. We evaluated the cost-effectiveness of various breast cancer screening scenarios in Canada from a publicly funded health care system perspective using an established breast cancer simulation model.

METHODS

Breast cancer incidence, outcomes and total health care system costs (screening, investigation, diagnosis and treatment) for the Canadian health care environment were modelled. The model predicted costs (in 2012 dollars), life-years gained and quality-adjusted life-years (QALYs) gained for 11 active screening scenarios that varied by age range for starting and stopping screening (40-74 yr) and frequency of screening (annual, biennial or triennial) relative to no screening. All outcomes were discounted. Marginal and incremental cost-effectiveness analyses were conducted. One-way sensitivity analyses of key parameters assessed robustness.

RESULTS

The lifetime overall costs (undiscounted) to the health care system for annual screening per 1000 women ranged from $7.4 million (for women aged 50-69 yr) to $10.7 million (40-74 yr). For biennial and triennial screening per 1000 women (aged 50-74 yr), costs were less, at about $6.1 million and $5.3 million, respectively. The incremental cost-utility ratio varied from $36 981/QALY for triennial screening in women aged 50-69 versus no screening to $38 142/QALY for biennial screening in those aged 50-69 and $83 845/QALY for annual screening in those aged 40-74.

INTERPRETATION

Our economic analysis showed that both benefits of mortality reduction and costs rose together linearly with the number of lifetime screens per women. The decision on how to screen is related mainly to willingness to pay and additional considerations such as the number of women recalled after a positive screening result.

Estimating the frequency of indolent breast cancer in screening trials

Cancer screening can detect cancer that would not have been detected in a patient's lifetime without screening. Standard methods for analyzing screening data do not explicitly account for the possibility that a fraction of tumors may remain latent indefinitely. We extend these methods by representing cancers as a mixture of those that progress to symptoms (progressive) and those that remain latent (indolent). Given sensitivity of the screening test, we derive likelihood expressions to simultaneously estimate (1) the rate of onset of preclinical cancer, (2) the average preclinical duration of progressive cancers, and (3) the fraction of preclinical cancers that are indolent. Simulations demonstrate satisfactory performance of the estimation approach to identify model parameters subject to precise specifications of input parameters and adequate numbers of interval cancers. In application to four breast cancer screening trials, the estimated indolent fraction among preclinical cancers varies between 2% and 35% when assuming 80% test sensitivity and varying specifications for the earliest time that participants could plausibly have developed cancer. We conclude that standard methods for analyzing screening data can be extended to allow some indolent cancers, but accurate estimation depends on correctly specifying key inputs that may be difficult to determine precisely in practice.

Effect of Time to Diagnostic Testing for Breast, Cervical, and Colorectal Cancer Screening Abnormalities on Screening Efficacy: A Modeling Study

Background: Patients who receive an abnormal cancer screening result require follow-up for diagnostic testing, but the time to follow-up varies across patients and practices.Methods: We used a simulation study to estimate the change in lifetime screening benefits when time to follow-up for breast, cervical, and colorectal cancers was increased. Estimates were based on four independently developed microsimulation models that each simulated the life course of adults eligible for breast (women ages 50-74 years), cervical (women ages 21-65 years), or colorectal (adults ages 50-75 years) cancer screening. We assumed screening based on biennial mammography for breast cancer, triennial Papanicolaou testing for cervical cancer, and annual fecal immunochemical testing for colorectal cancer. For each cancer type, we simulated diagnostic testing immediately and at 3, 6, and 12 months after an abnormal screening exam.Results: We found declines in screening benefit with longer times to diagnostic testing, particularly for breast cancer screening. Compared to immediate diagnostic testing, testing at 3 months resulted in reduced screening benefit, with fewer undiscounted life years gained per 1,000 screened (breast: 17.3%, cervical: 0.8%, colorectal: 2.0% and 2.7%, from two colorectal cancer models), fewer cancers prevented (cervical: 1.4% fewer, colorectal: 0.5% and 1.7% fewer, respectively), and, for breast and colorectal cancer, a less favorable stage distribution.Conclusions: Longer times to diagnostic testing after an abnormal screening test can decrease screening effectiveness, but the impact varies substantially by cancer type.Impact: Understanding the impact of time to diagnostic testing on screening effectiveness can help inform quality improvement efforts. Cancer Epidemiol Biomarkers Prev; 27(2); 158-64. ©2017 AACR.

Insights Into Breast Cancer Screening: A Computer Simulation of Two Contemporary Screening Strategies

OBJECTIVE

The debate over the value of screening mammography is rekindled with each new published study or guideline. Central to the discussion are the uncertainties about screening benefits and harms and the criteria used to assess them. Today, the magnitude of benefits for a population is less certain, and the evolving concept of harm has come to encompass false-positives (FPs), unnecessary biopsies, overdiagnosis, and overtreatment. This study uses a Monte Carlo computer simulation to study the balance of benefits and harms of mammographic breast cancer screening for average-risk women.

MATERIALS AND METHODS

This investigation compares the American Cancer Society's 2015 mixed annual-biennial guideline with the U.S. Preventive Services Task Force's 2016 fixed biennial guideline. Screening strategies are compared using cost-effectiveness acceptability curves, an economic analysis describing uncertainty in evaluating costs and health outcomes. Strategy preference is examined under changing assumptions of willingness to pay for a quality-adjusted life-year. Additionally, comparative effectiveness analysis is performed using FP screens and unnecessary biopsies per life-year gained. Alternative scenarios are compared assuming a reduced mortality benefit of screening.

RESULTS

In general, results using both cost-effectiveness and clinical measures indicate that American Cancer Society's 2015 mixed annual-biennial guideline is preferred. Assuming decreases in the mortality benefit of mammography, no screening may be reasonable.

CONCLUSION

The use of a mixed annual-biennial strategy for population screening takes advantage of the nonuniformity of occurrence of mammography benefits and harms over the duration of screening. This approach represents a step toward improving guidelines by exploiting age dependencies at which benefits and harms accrue.

Comparative effectiveness of incorporating a hypothetical DCIS prognostic marker into breast cancer screening

PURPOSE

Due to limitations in the ability to identify non-progressive disease, ductal carcinoma in situ (DCIS) is usually managed similarly to localized invasive breast cancer. We used simulation modeling to evaluate the potential impact of a hypothetical test that identifies non-progressive DCIS.

METHODS

A discrete-event model simulated a cohort of U.S. women undergoing digital screening mammography. All women diagnosed with DCIS underwent the hypothetical DCIS prognostic test. Women with test results indicating progressive DCIS received standard breast cancer treatment and a decrement to quality of life corresponding to the treatment. If the DCIS test indicated non-progressive DCIS, no treatment was received and women continued routine annual surveillance mammography. A range of test performance characteristics and prevalence of non-progressive disease were simulated. Analysis compared discounted quality-adjusted life years (QALYs) and costs for test scenarios to base-case scenarios without the test.

RESULTS

Compared to the base case, a perfect prognostic test resulted in a 40% decrease in treatment costs, from $13,321 to $8005 USD per DCIS case. A perfect test produced 0.04 additional QALYs (16 days) for women diagnosed with DCIS, added to the base case of 5.88 QALYs per DCIS case. The results were sensitive to the performance characteristics of the prognostic test, the proportion of DCIS cases that were non-progressive in the model, and the frequency of mammography screening in the population.

CONCLUSION

A prognostic test that identifies non-progressive DCIS would substantially reduce treatment costs but result in only modest improvements in quality of life when averaged over all DCIS cases.

Using Collaborative Simulation Modeling to Develop a Web-Based Tool to Support Policy-Level Decision Making About Breast Cancer Screening Initiation Age

BACKGROUND

There are no publicly available tools designed specifically to assist policy makers to make informed decisions about the optimal ages of breast cancer screening initiation for different populations of US women.

OBJECTIVE

To use three established simulation models to develop a web-based tool called Mammo OUTPuT.

METHODS

The simulation models use the 1970 US birth cohort and common parameters for incidence, digital screening performance, and treatment effects. Outcomes include breast cancers diagnosed, breast cancer deaths averted, breast cancer mortality reduction, false-positive mammograms, benign biopsies, and overdiagnosis. The Mammo OUTPuT tool displays these outcomes for combinations of age at screening initiation (every year from 40 to 49), annual versus biennial interval, lifetime versus 10-year horizon, and breast density, compared to waiting to start biennial screening at age 50 and continuing to 74. The tool was piloted by decision makers (n = 16) who completed surveys.

RESULTS

The tool demonstrates that benefits in the 40s increase linearly with earlier initiation age, without a specific threshold age. Likewise, the harms of screening increase monotonically with earlier ages of initiation in the 40s. The tool also shows users how the balance of benefits and harms varies with breast density. Surveys revealed that 100% of users (16/16) liked the appearance of the site; 94% (15/16) found the tool helpful; and 94% (15/16) would recommend the tool to a colleague.

CONCLUSIONS

This tool synthesizes a representative subset of the most current CISNET (Cancer Intervention and Surveillance Modeling Network) simulation model outcomes to provide policy makers with quantitative data on the benefits and harms of screening women in the 40s. Ultimate decisions will depend on program goals, the population served, and informed judgments about the weight of benefits and harms.

Cost-Effectiveness Models in Breast Cancer Screening in the General Population: A Systematic Review

BACKGROUND

Many Western countries have long-established population-based mammography screening programs. Prior to implementing these programs, decision-analytic modeling was widely used to inform decisions.

OBJECTIVE

The aim of this study was to perform a systematic review of cost-effectiveness models in breast cancer screening in the general population to analyze their structural and methodological approaches.

METHODS

A systematic literature search for health economic models was performed in the electronic databases MEDLINE (Ovid), EMBASE, CRD Databases, Cochrane Library, and EconLit in August 2011 with updates in June 2013, April 2015, and November 2016. To assess studies systematically, a standardized form was applied to extract relevant information that was then summarized in evidence tables.

RESULTS

Thirty-five studies were included; 27 state-transition models were analyzed using cohort (n = 12) and individual-level simulation (n = 15). Twenty-one studies modeled the natural history of breast cancer and predicted mortality as a function of the early detection modality. The models employed different assumptions regarding ductal carcinoma in situ. Thirteen studies performed cost-utility analyses with different sources for utility values, but assumptions were often made about utility weights. Twenty-two models did not report any validation.

CONCLUSION

State-transition modeling was the most frequently applied analytic approach. Different methods in modeling the progression of ductal carcinoma in situ to invasive cancer were identified because there is currently no agreement on the biological behavior of noninvasive breast cancer. Main weaknesses were the lack of precise utility estimates and insufficient reporting of validation. Sensitivity analyses of assumptions regarding ductal carcinoma in situ and in particular adequate validation are critical to minimize the risk of biased model outcomes.

Missteps in Current Estimates of Cancer Overdiagnosis

The balance between the benefits and harms of imaging-based cancer screening continues to be an area of controversy and widespread media attention. Of the potential harms, overdiagnosis from screening is likely the most elusive in estimating and quantifying. This article describes the major methodological issues with recently reported estimates of overdiagnosis that are based on excess cancer incidence, and suggests that modeling focused on tumor lead-time can serve as a complementary method for excess incidence-based overdiagnosis estimates. Radiologists should be conversant on the topic of overdiagnosis and understand the limitations of different methods used to estimate its magnitude.

Tailoring Breast Cancer Screening Intervals by Breast Density and Risk for Women Aged 50 Years or Older: Collaborative Modeling of Screening Outcomes

BACKGROUND

Biennial screening is generally recommended for average-risk women aged 50 to 74 years, but tailored screening may provide greater benefits.

OBJECTIVE

To estimate outcomes for various screening intervals after age 50 years based on breast density and risk for breast cancer.

DESIGN

Collaborative simulation modeling using national incidence, breast density, and screening performance data.

SETTING

United States.

PATIENTS

Women aged 50 years or older with various combinations of breast density and relative risk (RR) of 1.0, 1.3, 2.0, or 4.0.

INTERVENTION

Annual, biennial, or triennial digital mammography screening from ages 50 to 74 years (vs. no screening) and ages 65 to 74 years (vs. biennial digital mammography from ages 50 to 64 years).

MEASUREMENTS

Lifetime breast cancer deaths, life expectancy and quality-adjusted life-years (QALYs), false-positive mammograms, benign biopsy results, overdiagnosis, cost-effectiveness, and ratio of false-positive results to breast cancer deaths averted.

RESULTS

Screening benefits and overdiagnosis increase with breast density and RR. False-positive mammograms and benign results on biopsy decrease with increasing risk. Among women with fatty breasts or scattered fibroglandular density and an RR of 1.0 or 1.3, breast cancer deaths averted were similar for triennial versus biennial screening for both age groups (50 to 74 years, median of 3.4 to 5.1 vs. 4.1 to 6.5 deaths averted; 65 to 74 years, median of 1.5 to 2.1 vs. 1.8 to 2.6 deaths averted). Breast cancer deaths averted increased with annual versus biennial screening for women aged 50 to 74 years at all levels of breast density and an RR of 4.0, and those aged 65 to 74 years with heterogeneously or extremely dense breasts and an RR of 4.0. However, harms were almost 2-fold higher. Triennial screening for the average-risk subgroup and annual screening for the highest-risk subgroup cost less than $100 000 per QALY gained.

LIMITATION

Models did not consider women younger than 50 years, those with an RR less than 1, or other imaging methods.

CONCLUSION

Average-risk women with low breast density undergoing triennial screening and higher-risk women with high breast density receiving annual screening will maintain a similar or better balance of benefits and harms than average-risk women receiving biennial screening.

PRIMARY FUNDING SOURCE

National Cancer Institute.

Improvement in risk prediction, early detection and prevention of breast cancer in the NHS Breast Screening Programme and family history clinics: a dual cohort study

Background

In the UK, women are invited for 3-yearly mammography screening, through the NHS Breast Screening Programme (NHSBSP), from the ages of 47–50 years to the ages of 69–73 years. Women with family histories of breast cancer can, from the age of 40 years, obtain enhanced surveillance and, in exceptionally high-risk cases, magnetic resonance imaging. However, no NHSBSP risk assessment is undertaken. Risk prediction models are able to categorise women by risk using known risk factors, although accurate individual risk prediction remains elusive. The identification of mammographic breast density (MD) and common genetic risk variants [single nucleotide polymorphisms (SNPs)] has presaged the improved precision of risk models.

Objectives

To (1) identify the best performing model to assess breast cancer risk in family history clinic (FHC) and population settings; (2) use information from MD/SNPs to improve risk prediction; (3) assess the acceptability and feasibility of offering risk assessment in the NHSBSP; and (4) identify the incremental costs and benefits of risk stratified screening in a preliminary cost-effectiveness analysis.

Design

Two cohort studies assessing breast cancer incidence.

Setting

High-risk FHC and the NHSBSP Greater Manchester, UK.

Participants

A total of 10,000 women aged 20–79 years [Family History Risk Study (FH-Risk); UK Clinical Research Network identification number (UKCRN-ID) 8611] and 53,000 women from the NHSBSP [aged 46–73 years; Predicting the Risk of Cancer At Screening (PROCAS) study; UKCRN-ID 8080].

Interventions

Questionnaires collected standard risk information, and mammograms were assessed for breast density by a number of techniques. All FH-Risk and 10,000 PROCAS participants participated in deoxyribonucleic acid (DNA) studies. The risk prediction models Manual method, Tyrer–Cuzick (TC), BOADICEA (Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm) and Gail were used to assess risk, with modelling based on MD and SNPs. A preliminary model-based cost-effectiveness analysis of risk stratified screening was conducted.

Main outcome measures

Breast cancer incidence.

Data sources

The NHSBSP; cancer registration.

Results

A total of 446 women developed incident breast cancers in FH-Risk in 97,958 years of follow-up. All risk models accurately stratified women into risk categories. TC had better risk precision than Gail, and BOADICEA accurately predicted risk in the 6268 single probands. The Manual model was also accurate in the whole cohort. In PROCAS, TC had better risk precision than Gail [area under the curve (AUC) 0.58 vs. 0.54], identifying 547 prospective breast cancers. The addition of SNPs in the FH-Risk case–control study improved risk precision but was not useful inBRCA1(breast cancer 1 gene) families. Risk modelling of SNPs in PROCAS showed an incremental improvement from using SNP18 used in PROCAS to SNP67. MD measured by visual assessment score provided better risk stratification than automatic measures, despite wide intra- and inter-reader variability. Using a MD-adjusted TC model in PROCAS improved risk stratification (AUC = 0.6) and identified significantly higher rates (4.7 per 10,000 vs. 1.3 per 10,000;p < 0.001) of high-stage cancers in women with above-average breast cancer risks. It is not possible to provide estimates of the incremental costs and benefits of risk stratified screening because of lack of data inputs for key parameters in the model-based cost-effectiveness analysis.

Conclusions

Risk precision can be improved by using DNA and MD, and can potentially be used to stratify NHSBSP screening. It may also identify those at greater risk of high-stage cancers for enhanced screening. The cost-effectiveness of risk stratified screening is currently associated with extensive uncertainty. Additional research is needed to identify data needed for key inputs into model-based cost-effectiveness analyses to identify the impact on health-care resource use and patient benefits.

Future work

A pilot of real-time NHSBSP risk prediction to identify women for chemoprevention and enhanced screening is required.

Funding

The National Institute for Health Research Programme Grants for Applied Research programme. The DNA saliva collection for SNP analysis for PROCAS was funded by the Genesis Breast Cancer Prevention Appeal.

Overdiagnosis and overtreatment of breast cancer

Breast cancer is the most common cancer in women. Through greater awareness, mammographic screening, and aggressive biopsy of calcifications, the proportion of low-grade, early stage cancers and in situ lesions among all breast cancers has risen substantially. The introduction of molecular testing has increased the recognition of lower risk subtypes, and less aggressive treatments are more commonly recommended for these subtypes. Mammographically detected breast cancers are much more likely to have low-risk biology than symptomatic tumors found between screenings (interval cancers) or that present as clinical masses. Recognizing the lower risk associated with these lesions and the ability to confirm the risk with molecular tests should safely enable the use of less aggressive treatments. Importantly, ductal carcinoma in situ (DCIS) lesions, or what have been called stage I cancers, in and of themselves are not life-threatening. In situ lesions have been treated in a manner similar to that of invasive cancer, but there is little evidence to support that this practice has improved mortality. It is also being recognized that DCIS lesions are heterogeneous, and a substantial proportion of them may in fact be precursors of more indolent invasive cancers. Increasing evidence suggests that these lesions are being overtreated. The introduction of molecular tests should be able to help usher in a change in approach to these lesions. Reclassifying these lesions as part of the spectrum of high-risk lesions enables the use of a prevention approach. Learning from the experience with active surveillance in prostate cancer should empower the introduction of new approaches, with a focus on preventing invasive cancer, especially given that there are effective, United States Food and Drug Administration (FDA)-approved breast cancer preventive interventions.

Collaborative Modeling of the Benefits and Harms Associated With Different U.S. Breast Cancer Screening Strategies

BACKGROUND

Controversy persists about optimal mammography screening strategies.

OBJECTIVE

To evaluate screening outcomes, taking into account advances in mammography and treatment of breast cancer.

DESIGN

Collaboration of 6 simulation models using national data on incidence, digital mammography performance, treatment effects, and other-cause mortality.

SETTING

United States.

PATIENTS

Average-risk U.S. female population and subgroups with varying risk, breast density, or comorbidity.

INTERVENTION

Eight strategies differing by age at which screening starts (40, 45, or 50 years) and screening interval (annual, biennial, and hybrid [annual for women in their 40s and biennial thereafter]). All strategies assumed 100% adherence and stopped at age 74 years.

MEASUREMENTS

Benefits (breast cancer-specific mortality reduction, breast cancer deaths averted, life-years, and quality-adjusted life-years); number of mammograms used; harms (false-positive results, benign biopsies, and overdiagnosis); and ratios of harms (or use) and benefits (efficiency) per 1000 screens.

RESULTS

Biennial strategies were consistently the most efficient for average-risk women. Biennial screening from age 50 to 74 years avoided a median of 7 breast cancer deaths versus no screening; annual screening from age 40 to 74 years avoided an additional 3 deaths, but yielded 1988 more false-positive results and 11 more overdiagnoses per 1000 women screened. Annual screening from age 50 to 74 years was inefficient (similar benefits, but more harms than other strategies). For groups with a 2- to 4-fold increased risk, annual screening from age 40 years had similar harms and benefits as screening average-risk women biennially from 50 to 74 years. For groups with moderate or severe comorbidity, screening could stop at age 66 to 68 years.

LIMITATION

Other imaging technologies, polygenic risk, and nonadherence were not considered.

CONCLUSION

Biennial screening for breast cancer is efficient for average-risk populations. Decisions about starting ages and intervals will depend on population characteristics and the decision makers' weight given to the harms and benefits of screening.

PRIMARY FUNDING SOURCE

National Institutes of Health.

Cancer Models and Real-world Data: Better Together

Decision-analytic models are increasingly used to inform health policy decisions. These models synthesize available data on disease burden and intervention effectiveness to project estimates of the long-term consequences of care, which are often absent when clinical or policy decisions must be made. While models have been influential in informing US cancer screening guidelines under ideal conditions, incorporating detailed data on real-world screening practice has been limited given the complexity of screening processes and behaviors throughout diverse health delivery systems in the United States. We describe the synergies that exist between decision-analytic models and health care utilization data that are increasingly accessible through research networks that assemble data from the growing number of electronic medical record systems. In particular, we present opportunities to enrich cancer screening models by grounding analyses in real-world data with the goals of projecting the harms and benefits of current screening practices, evaluating the value of existing and new technologies, and identifying the weakest links in the cancer screening process where efforts for improvement may be most productively focused. We highlight the example of the National Cancer Institute-funded consortium Population-based Research Optimizing Screening through Personalized Regimens (PROSPR), a collaboration to harmonize and analyze screening process and outcomes data on breast, colorectal, and cervical cancers across seven research centers. The pairing of models with such data can create more robust models to not only better inform policy but also inform health care systems about best approaches to improve the provision of cancer screening in the United States.

Outcomes of Active Surveillance for Ductal Carcinoma in Situ: A Computational Risk Analysis

BACKGROUND

Ductal carcinoma in situ (DCIS) is a noninvasive breast lesion with uncertain risk for invasive progression. Usual care (UC) for DCIS consists of treatment upon diagnosis, thus potentially overtreating patients with low propensity for progression. One strategy to reduce overtreatment is active surveillance (AS), whereby DCIS is treated only upon detection of invasive disease. Our goal was to perform a quantitative evaluation of outcomes following an AS strategy for DCIS.

METHODS

Age-stratified, 10-year disease-specific cumulative mortality (DSCM) for AS was calculated using a computational risk projection model based upon published estimates for natural history parameters, and Surveillance, Epidemiology, and End Results data for outcomes. AS projections were compared with the DSCM for patients who received UC. To quantify the propagation of parameter uncertainty, a 95% projection range (PR) was computed, and sensitivity analyses were performed.

RESULTS

Under the assumption that AS cannot outperform UC, the projected median differences in 10-year DSCM between AS and UC when diagnosed at ages 40, 55, and 70 years were 2.6% (PR = 1.4%-5.1%), 1.5% (PR = 0.5%-3.5%), and 0.6% (PR = 0.0%-2.4), respectively. Corresponding median numbers of patients needed to treat to avert one breast cancer death were 38.3 (PR = 19.7-69.9), 67.3 (PR = 28.7-211.4), and 157.2 (PR = 41.1-3872.8), respectively. Sensitivity analyses showed that the parameter with greatest impact on DSCM was the probability of understaging invasive cancer at diagnosis.

CONCLUSION

AS could be a viable management strategy for carefully selected DCIS patients, particularly among older age groups and those with substantial competing mortality risks. The effectiveness of AS could be markedly improved by reducing the rate of understaging.

To screen or not to screen for breast cancer? How do modelling studies answer the question?

Breast cancer screening is a topic of hot debate, and currently no general consensus has been reached on starting and ending ages and screening intervals, in part because of a lack of precise estimations of the benefit-harm ratio. Simulation models are often applied to account for the expected benefits and harms of regular screening; however, the degree to which the model outcomes are reliable is not clear. In a recent systematic review, we therefore aimed to assess the quality of published simulation models for breast cancer screening of the general population. The models were scored according to a framework for qualitative assessment. We distinguished seven original models that utilized a common model type, modelling approach, and input parameters. The models predicted the benefit of regular screening in terms of mortality reduction; and overall, their estimates compared well to estimates of mortality reduction from randomized controlled trials. However, the models did not report on the expected harms associated with regular screening. We found that current simulation models for population breast cancer screening are prone to many pitfalls; their outcomes bear a high overall risk of bias, mainly because of a lack of systematic evaluation of evidence to calibrate the input parameters and a lack of external validation. Our recommendations concerning future modelling are therefore to use systematically evaluated data for the calibration of input parameters, to perform external validation of model outcomes, and to account for both the expected benefits and the expected harms so as to provide a clear balance and cost-effectiveness estimation and to adequately inform decision-makers.

Using Active Learning for Speeding up Calibration in Simulation Models

BACKGROUND

Most cancer simulation models include unobservable parameters that determine disease onset and tumor growth. These parameters play an important role in matching key outcomes such as cancer incidence and mortality, and their values are typically estimated via a lengthy calibration procedure, which involves evaluating a large number of combinations of parameter values via simulation. The objective of this study is to demonstrate how machine learning approaches can be used to accelerate the calibration process by reducing the number of parameter combinations that are actually evaluated.

METHODS

Active learning is a popular machine learning method that enables a learning algorithm such as artificial neural networks to interactively choose which parameter combinations to evaluate. We developed an active learning algorithm to expedite the calibration process. Our algorithm determines the parameter combinations that are more likely to produce desired outputs and therefore reduces the number of simulation runs performed during calibration. We demonstrate our method using the previously developed University of Wisconsin breast cancer simulation model (UWBCS).

RESULTS

In a recent study, calibration of the UWBCS required the evaluation of 378 000 input parameter combinations to build a race-specific model, and only 69 of these combinations produced results that closely matched observed data. By using the active learning algorithm in conjunction with standard calibration methods, we identify all 69 parameter combinations by evaluating only 5620 of the 378 000 combinations.

CONCLUSION

Machine learning methods hold potential in guiding model developers in the selection of more promising parameter combinations and hence speeding up the calibration process. Applying our machine learning algorithm to one model shows that evaluating only 1.49% of all parameter combinations would be sufficient for the calibration.

Developing a utility decision framework to evaluate predictive models in breast cancer risk estimation

Combining imaging and genetic information to predict disease presence and progression is being codified into an emerging discipline called "radiogenomics." Optimal evaluation methodologies for radiogenomics have not been well established. We aim to develop a decision framework based on utility analysis to assess predictive models for breast cancer diagnosis. We garnered Gail risk factors, single nucleotide polymorphisms (SNPs), and mammographic features from a retrospective case-control study. We constructed three logistic regression models built on different sets of predictive features: (1) Gail, (2) Gail + Mammo, and (3) Gail + Mammo + SNP. Then we generated receiver operating characteristic (ROC) curves for three models. After we assigned utility values for each category of outcomes (true negatives, false positives, false negatives, and true positives), we pursued optimal operating points on ROC curves to achieve maximum expected utility of breast cancer diagnosis. We performed McNemar's test based on threshold levels at optimal operating points, and found that SNPs and mammographic features played a significant role in breast cancer risk estimation. Our study comprising utility analysis and McNemar's test provides a decision framework to evaluate predictive models in breast cancer risk estimation.

Viewpoint: It is time to reconsider policy for population-based mammography screening

Breast cancer in women is an important cause of morbidity and mortality. Many countries in the Western world have widely promoted early detection through mammography screening and established population-based screening programs. Over the past 15 years, there has been growing debate about the benefits and harms of universal mammography screening. This article presents findings from the latest systematic review conducted by the Cochrane Collaboration and from the Canadian National Breast Screening Study 25-year follow up. The authors of both reports conclude there is no reliable evidence that population-based mammography screening reduces mortality, but there is good evidence of harm in the form of false positive findings, over-diagnosis and unnecessary treatment, and associated psychological distress. It is time for policymakers to discontinue universal population-based mammography screening and shift to a more selective approach to early detection.

Meta-Analytic Approaches for Multistressor Dose-Response Function Development: Strengths, Limitations, and Case Studies

For many policy analyses, including but not limited to cumulative risk assessments, it is important to characterize the individual and joint health effects of multiple stressors. With an increasing focus on psychosocial and other nonchemical stressors, this often includes epidemiological meta-analysis. Meta-analysis has limitations if epidemiological studies do not include all of the stressors of interest or do not provide multivariable outputs in a format necessary for risk assessment. Given these limitations, novel analytical methods are often needed to synthesize the published literature or to build upon available evidence. In this article, we discuss three recent case studies that highlight the strengths and limitations of meta-analytic approaches and other research synthesis techniques for human health risk assessment applications. First, a literature-based meta-analysis within a risk assessment context informed the design of a new epidemiological investigation of the differential toxicity of fine particulate matter constituents. Second, a literature synthesis for an effects-based cumulative risk assessment of hypertension risk factors led to a decision to develop new epidemiological associations using structural equation modeling. Third, discrete event simulation modeling was used to simulate the impact of changes in the built environment on environmental exposures and associated asthma outcomes, linking literature meta-analyses for key associations with a simulation model to synthesize all of the model components. These case studies emphasize the importance of conducting epidemiology with a risk assessment application in mind, the need for interdisciplinary collaboration, and the value of advanced analytical methods to synthesize epidemiological and other evidence for risk assessment applications.

Benefits and harms of mammography screening after age 74 years: model estimates of overdiagnosis

BACKGROUND

The aim of this study was to quantify the benefits and harms of mammography screening after age 74 years, focusing on the amount of overdiagnosis of invasive breast cancer and ductal carcinoma in situ (DCIS).

METHODS

Three well-established microsimulation models were used to simulate a cohort of American women born in 1960. All women received biennial screening starting at age 50 years with cessation ages varying from 74 up to 96 years. We estimated the number of life-years gained (LYG), quality-adjusted life-years, breast cancer deaths averted, false-positives, and overdiagnosed women per 1000 screens.

RESULTS

The models predicted that there were 7.8 to 11.4 LYG per 1000 screens at age 74 years (range across models), decreasing to 4.8 to 7.8 LYG per 1000 screens at age 80 years, and 1.4 to 2.4 LYG per 1000 screens at age 90 years. When adjusted for quality-of-life decrements, the LYG decreased by 5% to 13% at age 74 years and 11% to 22% at age 80 years. At age 90 to 92 years, all LYG were counterbalanced by a loss in quality-of-life, mainly because of the increasing number of overdiagnosed breast cancers per 1000 screens: 1.2 to 5.0 at age 74 years, 1.8 to 6.0 at age 80 years, and 3.7 to 7.5 at age 90 years. The age at which harms began to outweigh benefits shifted to a younger age when larger or longer utility losses because of a breast cancer diagnosis were assumed.

CONCLUSION

The balance between screening benefits and harms becomes less favorable after age 74 years. At age 90 years, harms outweigh benefits, largely as a consequence of overdiagnosis. This age was the same across the three models, despite important model differences in assumptions on DCIS.

Simulation models in population breast cancer screening: A systematic review

The aim of this review was to critically evaluate published simulation models for breast cancer screening of the general population and provide a direction for future modeling. A systematic literature search was performed to identify simulation models with more than one application. A framework for qualitative assessment which incorporated model type; input parameters; modeling approach, transparency of input data sources/assumptions, sensitivity analyses and risk of bias; validation, and outcomes was developed. Predicted mortality reduction (MR) and cost-effectiveness (CE) were compared to estimates from meta-analyses of randomized control trials (RCTs) and acceptability thresholds. Seven original simulation models were distinguished, all sharing common input parameters. The modeling approach was based on tumor progression (except one model) with internal and cross validation of the resulting models, but without any external validation. Differences in lead times for invasive or non-invasive tumors, and the option for cancers not to progress were not explicitly modeled. The models tended to overestimate the MR (11-24%) due to screening as compared to optimal RCTs 10% (95% CI - 2-21%) MR. Only recently, potential harms due to regular breast cancer screening were reported. Most scenarios resulted in acceptable cost-effectiveness estimates given current thresholds. The selected models have been repeatedly applied in various settings to inform decision making and the critical analysis revealed high risk of bias in their outcomes. Given the importance of the models, there is a need for externally validated models which use systematical evidence for input data to allow for more critical evaluation of breast cancer screening.

Cost-effectiveness of breast cancer screening policies using simulation

In this paper, we study breast cancer screening policies using computer simulation. We developed a multi-state Markov model for breast cancer progression, considering both the screening and treatment stages of breast cancer. The parameters of our model were estimated through data from the Canadian National Breast Cancer Screening Study as well as data in the relevant literature. Using computer simulation, we evaluated various screening policies to study the impact of mammography screening for age-based subpopulations in Canada. We also performed sensitivity analysis to examine the impact of certain parameters on number of deaths and total costs. The analysis comparing screening policies reveals that a policy in which women belonging to the 40-49 age group are not screened, whereas those belonging to the 50-59 and 60-69 age groups are screened once every 5 years, outperforms others with respect to cost per life saved. Our analysis also indicates that increasing the screening frequencies for the 50-59 and 60-69 age groups decrease mortality, and that the average number of deaths generally decreases with an increase in screening frequency. We found that screening annually for all age groups is associated with the highest costs per life saved. Our analysis thus reveals that cost per life saved increases with an increase in screening frequency.

Comparative effectiveness of combined digital mammography and tomosynthesis screening for women with dense breasts

PURPOSE

To evaluate the effectiveness of combined biennial digital mammography and tomosynthesis screening, compared with biennial digital mammography screening alone, among women with dense breasts.

MATERIALS AND METHODS

An established, discrete-event breast cancer simulation model was used to estimate the comparative clinical effectiveness and cost-effectiveness of biennial screening with both digital mammography and tomosynthesis versus digital mammography alone among U.S. women aged 50-74 years with dense breasts from a federal payer perspective and a lifetime horizon. Input values were estimated for test performance, costs, and health state utilities from the National Cancer Institute Breast Cancer Surveillance Consortium, Medicare reimbursement rates, and medical literature. Sensitivity analyses were performed to determine the implications of varying key model parameters, including combined screening sensitivity and specificity, transient utility decrement of diagnostic work-up, and additional cost of tomosynthesis.

RESULTS

For the base-case analysis, the incremental cost per quality-adjusted life year gained by adding tomosynthesis to digital mammography screening was $53 893. An additional 0.5 deaths were averted and 405 false-positive findings avoided per 1000 women after 12 rounds of screening. Combined screening remained cost-effective (less than $100 000 per quality-adjusted life year gained) over a wide range of incremental improvements in test performance. Overall, cost-effectiveness was most sensitive to the additional cost of tomosynthesis.

CONCLUSION

Biennial combined digital mammography and tomosynthesis screening for U.S. women aged 50-74 years with dense breasts is likely to be cost-effective if priced appropriately (up to $226 for combined examinations vs $139 for digital mammography alone) and if reported interpretive performance metrics of improved specificity with tomosynthesis are met in routine practice.