Risk-based breast cancer screening in Austria: a systematic analysis of predictive models to assess the individual breast cancer risk, their utility and applicability in breast cancer screening programs

Research area: Prevention and screening

Irmgard Frühwirth
Project lead: Irmgard Frühwirth
Duration: March 2022 – August 2022
Language: German

Publication: HTA Project Report No. 145: https://eprints.aihta.at/1402/

Background: Organised mammography screening programs have been established worldwide based on several randomised controlled trials (RCTs) from 1963 to 1991 [1-4]. Although there is no uniform consensus on age and screening intervals, most European countries recommend biennial or triennial mammograms for women aged 50 to 69 [5]. In comparison, in the United States, there are recommendations for annual or biennial screening for women aged 40 to 74 years [6-9].

Since January 2014, a population-based breast cancer screening program has been implemented in Austria for women aged 45 to 69 who are invited to participate at two-year intervals. Younger women aged 40 and over and women aged 70 and over can take advantage of the program at their request [10].

There have been ongoing debates about the benefits and harms of breast cancer screening for a long time, with conflicting research findings [11-15]. Currently, age is the sole risk factor in defining the target population for breast cancer screening for women at average risk.

Predictive models of breast cancer risk incorporate, in addition to the classic risk factor of age, other risk factors, such as breast density, genetic factors, family history, lifestyle, and hormone therapy. They are designed to quantify the risk of whether an individual woman will develop breast cancer in a defined period. Modelling has shown that modifying the screening interval, modality, and initiation based on individual risk may provide more significant benefits than conventional screening strategies [16-19].

Prediction models of breast cancer risk are key elements to develop risk-based screening strategies [20, 21]. Several prediction models that include different risk factors are used in the clinical context [22]. However, it is unknown whether prediction models are already used in organised screening programs. All risk prediction models have limitations. Therefore, the quality and applicability of each of these models need to be assessed and evaluated in a screening setting in high-quality studies [21].

There are three main categories of predictive models used in the clinical setting for the risk assessment of breast cancer [23]:

1. Models that predict the individual breast cancer risk based on personal and hormonal risk factors, and/or radiologic data, and/or family history within a specified period (e.g., lifetime, 10 years).
2. Models that estimate the probability of detecting a germline mutation of specific genes, such as BRCA1 or BRCA2, in a given family or individual.
3. Models that estimate clinical outcomes of therapy or its response in a patient diagnosed with breast cancer.

Project aim: According to an agreement between the Austrian Medical Association and the social insurance, a breast cancer risk assessment is to be implemented in Austria. Consequently, general practitioners or gynaecologists should carry out risk assessments as part of the breast cancer screening program. Based on international evidence, the project aims to demonstrate the benefits of a risk-based breast cancer screening program compared to the current standard and evaluate which prediction models have the best predictive performance and could be used in a screening program. In addition, the organisational requirements for the implementation of risk assessments will be surveyed.

Research questions: To potentially improve the benefit-harm ratio of mammography screening, assessing the individual breast cancer risk could help screening become more effective [16]. Several risk assessment tools exist in different forms of application, e.g. as online self-assessment or assessment by healthcare professionals. This raises the question of which type of assessment lead to better clinical outcomes and are applicable in a breast cancer screening program, as there are numerous challenges associated with their use in practice (outside of a clinical trial setting). These include:

• time requirements,
• need for training on the interpretation of results and risk communication,
• impact on workflow processes,
• the acceptance of affected women and/or
• coordination in collecting different risk factors (e.g., laboratory parameters, genetic testing).

These factors may hinder proper implementation and reduce potential benefits in widespread use. In addition, risk assessments may also be associated with harm (e.g., through incorrect risk predictions, psychological distress for women, financial disadvantages in private insurance) or exacerbate health inequalities (e.g., through language barriers due to higher complexity in communication or different levels of literacy and numeracy) [24].

The following research questions (RQ) will be answered within this project:

• RQ 1: Does risk assessment in breast cancer screening lead to a better benefit-harm ratio than the conventional screening approach?
• RQ 2: Which prediction models are suitable for assessing the individual breast cancer risk in the context of a screening program? How do they differ in diagnostic accuracy, benefit-harm balance and application characteristics?
• RQ 3: Is there any experience with using predictive models in a screening setting?
• RQ 4: Which open questions arise from the analysis, especially regarding implementation and possible needs for further evidence analyses?

Methods: For RQ 1, a systematic literature search is performed in the Ovid MEDLINE, EMBASE, the Cochrane Library, and CRD databases. The search will be conducted according to the "Population-Intervention-Comparator-Outcome (PICO) scheme" (see Table 1). The search is restricted to RCTs, meta-analyses (MA), and systematic reviews (SR) in English and German, with no restriction on publication date. MA and SR are listed separately from RCTs, and only their references are screened for citations that may not have been found in the systematic search. Only high-quality RCTs answering the PICO question are included. Data extraction is performed on predefined outcomes, and results are presented in a narrative summary.

Searches for currently ongoing RCTs are conducted in the following databases: Clinical-Trials.gov, EU Clinical Trials Register, CENTRAL, International Clinical Trials Registry Platform (ICTRP), and ISRCTN Registry.

For RQ 2, a systematic literature search is performed in the databases Ovid MEDLINE, EMBASE, the Cochrane Library, and CRD. The search is performed according to the PICO scheme (see Table 2) and is limited to SRs in English and German without publication date restrictions for outcome 1. The risk prediction models cited in the SRs are summarised narratively in terms of predictive quality. For outcome 2, both SR and primary studies are included.

For RQ 3 and 4, a supplementary hand search is conducted, and for RQ 3, an additional survey of members of the international HTA network INAHTA is undertaken.

PICO-questions: The PICO scheme is used to specify the research questions. Tables 1 and 2 show PICOs defined for the assessment.

Table 1: PICO 1 (research question 1)

Population	Women between 25 and 75 years of age without suspected breast cancer
Intervention	Risk assessment as part of a screening program using a predictive model to assess an individual's risk of developing breast cancer within a specified period
Control	Screening programs with a one-size-fits-all strategy (all women are invited at defined intervals after a certain age)
Outcomes: Effectiveness and safety	Health benefit/harm: quality of life breast cancer morbidity breast cancer mortality all-cause mortality overdiagnoses adherence participation rates
Study design	RCTs

Table 2: PICO 2 (research questions 2 – 4)

Population	Women between 25 and 75 years of age without suspected breast cancer
Intervention	Predictive models to capture the individual breast cancer risk within a given period in a screening program
Control	Predictive models compared with each other
Outcome 1: Effectiveness and safety	Prognostic quality of the identified tools: Discriminatory accuracy1 Calibration accuracy2 Cut-off of the risk score (available? If yes, what is it?) Quality of life (e.g. mental stress due to the assessment result) Adverse events due to incorrect risk classification
Outcome 2: Relevant aspects for implementation	Target population: age ethnic groups general population high risk Included risk factors: parameters and methods collected sensitivity analysis of individual risk parameters Acceptance Qualification requirements for application (professional groups, training) Language(s) in which the instruments are available Setting in which the tools are used/tested Accompanying measures during application (consulting, risk communication) Examples of use/international experience
Study design	Outcome 1 (prognostic quality): high-quality SR All other outcomes: no restriction in study design

¹ the probability that a randomly chosen woman with the disease would be correctly categorised as higher risk compared to a randomly chosen woman without disease

² whether the model is more or less accurate in predicting the risk of specific individuals

Inclusion and exclusion criteria:

The literature search and selection will be based on predefined inclusion and exclusion criteria (see Table 3).

Table 3: Inclusion and exclusion criteria

Criteria	Inclusion	Exclusion
Language	English, German	Any other language
Quality of the study		Poor methodological quality (concerning the methods used and the insufficient description of it)
Study design	For PICO 1: RCTs For PICO 2, Outcome 1: SRs For PICO 2, Outcome 2: Any study type with relevant outcomes
Study population	Women in the general population aged 25 to 75 years with an average risk of breast cancer without breast cancer	Women diagnosed with breast cancer and men
Populations studied in models	Europe, USA, Canada, Australia, New Zealand	Asia, South America, Africa
Study intervention	Risk assessment in a screening setting
Predictive models	Prediction of individual breast cancer risk within a given period based on personal and hormonal risk factors, and/or radiologic data, and/or family history, and/or genetic information	Prediction models based solely on genetic information; detection of a high-risk germline mutation such as BRCA1 or BRCA2; prediction of clinical outcomes of therapy or its response
Outcomes	See outcomes in PICO-tables

PICO-question 1

Deficiencies in the design, conduct, analysis and evaluation of RCTs affect the intervention, which may subsequently be overestimated or underestimated [25]. The quality of the selected studies is assessed according to the Cochrane Collaboration's tool for assessing risk of bias [26] (see Appendix A-1).

Study characteristics of the included publications are presented in a data extraction table (see Table 4 for an example). One reviewer extracts the study characteristics, which are reviewed by another reviewer. Critical appraisal and data extraction are done in English.

Table 4: Preliminary study characteristics

First author, year, country

Study design

Study population

Intervention

Comparator

Outcome measure

Funding source

Conclusion

Critical appraisal

PICO-question 2

The quality of the selected systematic reviews is assessed according to the AMSTAR 2 tool [27] (see Appendix A-2).

Study characteristics of the included publications are presented in a data extraction table (see Table 5 for an example). One reviewer extracts the study characteristics, which are reviewed by another reviewer. Critical appraisal and data extraction are performed in English.

Table 5: Preliminary study characteristics

First author, year, country

Included studies

Targeted population

Risk factors

Discriminatory accuracy

Calibration accuracy

Funding source

Conclusion

Critical appraisal

Involvement of professional societies:

The Austrian Society of Gynaecology and Obstetrics (ÖGGG), the Federal Section of Radiology of the Austrian Medical Association (BURA) and the Austrian Society of General Medicine (ÖGAM) will be continuously involved in the overall project. The following steps are planned:

1. Commenting on the project protocol (mid-March).
2. Sending the list of selected studies for possible additions if relevant studies were not included (mid-April).
3. Comments on the summary of the preliminary results (mid-May)
4. Commenting on the preliminary report (end of July).

Schedule and milestones:

Time period	Task
Mid-March – end of March 2022	Project protocol and systematic literature search
April 2022	Data extraction
Mai 2022	Prepare summary/overview for negotiations
à 31.5.2022 (expected)	Presentation for the Medical Board: Overview of preliminary results
June - July 2022	Finalising report draft
August 2022	External review
à 31.8. 2022	Layout and publication

References:

[1]      Marmot MG, Altman DG, Cameron DA, Dewar JA, Thompson SG, Wilcox M. The benefits and harms of breast cancer screening: an independent review. Br J Cancer. 2013;108(11):2205-40.

[2]      The European Commission Initiative on Breast Cancer (ECIBC). Recommendations from European Breast Guidelines. 2016 [Available from: https://ecibc.jrc.ec.europa.eu/recommendations/].

[3]      Oeffinger KC, Fontham ET, Etzioni R, Herzig A, Michaelson JS, Shih YC, et al. Breast Cancer Screening for Women at Average Risk: 2015 Guideline Update From the American Cancer Society. Jama. 2015;314(15):1599-614.

[4]      Myers ER, Moorman P, Gierisch JM, Havrilesky LJ, Grimm LJ, Ghate S, et al. Benefits and Harms of Breast Cancer Screening: A Systematic Review. Jama. 2015;314(15):1615-34.

[5]      Perry N, Broeders M, de Wolf C, Törnberg S, Holland R, von Karsa L. European guidelines for quality assurance in breast cancer screening and diagnosis. 2006 [Available from: http://screening.iarc.fr/doc/ND7306954ENC_002.pdf].

[6]      Bevers TB, Helvie M, Bonaccio E, Calhoun KE, Daly MB, Farrar WB, et al. Breast Cancer Screening and Diagnosis, Version 3.2018, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. 2018;16(11):1362-89.

[7]      Siu AL. Screening for Breast Cancer: U.S. Preventive Services Task Force Recommendation Statement. Ann Intern Med. 2016;164(4):279-96.

[8]      Román M, Sala M, Domingo L, Posso M, Louro J, Castells X. Personalised breast cancer screening strategies: A systematic review and quality assessment. PLoS One. 2019;14(12):e0226352.

[9]      Committee on Practice Bulletins-Gynecology. Breast Cancer Risk Assessment and Screening in Average-Risk Women. Practice Bulletin Nr 179. 2017 [Available from: https://www.acog.org/clinical/clinical-guidance/practice-bulletin/articles/2017/07/breast-cancer-risk-assessment-and-screening-in-average-risk-women].

[10]    Versorgung CCI. Österreichisches Brustkrebsfrüherkennungsprogramm. 2022 [Available from: https://www.cciv.at/cdscontent/?contentid=10007.864466&portal=ccivportal].

[11]    Bleyer A, Welch HG. Effect of three decades of screening mammography on breast-cancer incidence. N Engl J Med. 2012;367(21):1998-2005.

[12]    Paci E, Broeders M, Hofvind S, Puliti D, Duffy SW. European breast cancer service screening outcomes: a first balance sheet of the benefits and harms. Cancer Epidemiol Biomarkers Prev. 2014;23(7):1159-63.

[13]    Welch HG, Passow HJ. Quantifying the benefits and harms of screening mammography. JAMA Intern Med. 2014;174(3):448-54.

[14]    Canelo-Aybar C, Posso M, Montero N, Solà I, Saz-Parkinson Z, Duffy SW, et al. Benefits and harms of annual, biennial, or triennial breast cancer mammography screening for women at average risk of breast cancer: a systematic review for the European Commission Initiative on Breast Cancer (ECIBC). Br J Cancer. 2022;126(4):673-88.

[15]    Gøtzsche PC, Jørgensen KJ. Screening for breast cancer with mammography. Cochrane Database Syst Rev. 2013;2013(6):Cd001877.

[16]    Vilaprinyo E, Forné C, Carles M, Sala M, Pla R, Castells X, et al. Cost-effectiveness and harm-benefit analyses of risk-based screening strategies for breast cancer. PLoS One. 2014;9(2):e86858.

[17]    Trentham-Dietz A, Kerlikowske K, Stout NK, Miglioretti DL, Schechter CB, Ergun MA, et al. Tailoring Breast Cancer Screening Intervals by Breast Density and Risk for Women Aged 50 Years or Older: Collaborative Modeling of Screening Outcomes. Ann Intern Med. 2016;165(10):700-12.

[18]    Schousboe JT, Kerlikowske K, Loh A, Cummings SR. Personalizing mammography by breast density and other risk factors for breast cancer: analysis of health benefits and cost-effectiveness. Ann Intern Med. 2011;155(1):10-20.

[19]    Louro J, Román M, Posso M, Vázquez I, Saladié F, Rodriguez-Arana A, et al. Developing and validating an individualised breast cancer risk prediction model for women attending breast cancer screening. PLoS One. 2021;16(3):e0248930.

[20]    Steyerberg EW. Clinical Prediction Models. A Practical Approach to Development, Validation, and Updating. New York: Springer Science; 2009.

[21]    Louro J, Posso M, Hilton Boon M, Román M, Domingo L, Castells X, et al. A systematic review and quality assessment of individualised breast cancer risk prediction models. Br J Cancer. 2019;121(1):76-85.

[22]    Anothaisintawee T, Teerawattananon Y, Wiratkapun C, Kasamesup V, Thakkinstian A. Risk prediction models of breast cancer: a systematic review of model performances. Breast Cancer Res Treat. 2012;133(1):1-10.

[23]    McGarrigle SA, Hanhauser YP, Mockler D, Gallagher DJ, Kennedy MJ, Bennett K, et al. Risk prediction models for familial breast cancer. Cochrane Database of Systematic Reviews 2018;2018(12).

[24]    Reyna VF, Nelson WL, Han PK, Dieckmann NF. How numeracy influences risk comprehension and medical decision making. Psychol Bull. 2009;135(6):943-73.

[25]    2Higgins JPT, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ (Clinical research ed). 2011;343:d5928.

[26]    Higgins J TJ. Cochrane Handbook for Systematic Reviews of Interventions, Version 6. 2019 [Available from: https://training.cochrane.org/handbook/current].

[27]    Shea BJ, Reeves BC, Wells G, Thuku M, Hamel C, Moran J, et al. AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both. Bmj. 2017;358:j4008.

Appendix

Table A-1: Cochrane Collaboration’s tool for assessing risk of bias [26]

Bias	Authors’ judgment
Random sequence generation (selection bias)	Low / high / unclear risk
Allocation concealment (selection bias)	Low / high / unclear risk
Blinding of participants and researchers (performance bias)	Low / high / unclear risk
Blinding of outcome assessment (detection bias)	Low / high / unclear risk
Incomplete outcome data (attrition bias)	Low / high / unclear risk
Selective reporting (reporting bias)	Low / high / unclear risk
Other bias	Low / high / unclear risk

Table A-2: AMSTAR 2: a critical appraisal tool for systematic reviews that include randomized or non-randomized studies of healthcare interventions, or both [27]

1	Did the research questions and inclusion criteria for the review include the components of PICO?
2	Did the report of the review contain an explicit statement that the review methods were established prior to the conduct of the review and did the report justify any significant deviations from the protocol?
3	Did the review authors explain their selection of the study designs for inclusion in the review?
4	Did the review authors use a comprehensive literature search strategy?
5	Did the review authors perform study selection in duplicate?
6	Did the review authors perform data extraction in duplicate?
7	Did the review authors provide a list of excluded studies and justify the exclusions?
8	Did the review authors describe the included studies in adequate detail?
9	Did the review authors use a satisfactory technique for assessing the risk of bias (RoB) in individual studies that were included in the review?
10	Did the review authors report on the sources of funding for the studies included in the review?
11	Did the review authors account for RoB in individual studies when interpreting/ discussing the results of the review?
12	Did the review authors report any potential sources of conflict of interest, including any funding they received for conducting the review?