European guidelines on breast cancer screening and diagnosis

The GDG agreed that the balance between the desirable and undesirable effects favours the use of DBT. The overall certainty of the evidence is low because of the uncertainty in estimates for interval breast cancers and false positives, and the absence of data for the downstream impact on breast cancer mortality. 

The evidence shows that when using DBT, there are likely greater number of breast cancers detected. In addition, there may be less overdiagnosis when using DBT since a higher proportion of invasive breast cancers are detected compared with non-invasive ductal carcinoma (DCIS) which are clinically less relevant cancers. Interval breast cancers (detected between regular screening visits) may also be reduced. These moderate benefits probably outweigh the higher number of false positives that may occur with DBT.  

There are moderate costs associated with DBT over DM, which could increase health inequities if implemented, but there are fewer women with high breast density in the target population for screening. Implementation of DBT would be facilitated by increased availability of DBT machines, adequate human resources (radiologists and technical personnel), and financial resources.

Potential challenges may exist: 

• The need for an infrastructure to archive data on breast density from previous screening exams and to share mammograms across centres as well as alerting that a woman with high breast density during the screening is required.
• The use of DBT might increase the overall time for reading mammograms (either due to an increased time for reading a single mammogram or due to the necessity for the radiologist to rest between reading different mammograms).
• In some countries, difficulties in recruiting radiologists for screening programmes exist.
• Referring women to another centre equipped with DBT might be not acceptable in certain countries.

Use of artificial intelligence (AI) algorithms might help overcome some challenges (please refer to the specific ECIBC recommendations on the use of AI in screening). 

A validated approach to assess breast density objectively will facilitate implementation. 

Quality assurance and improvement will be challenging given the need to divide women between the first or previous detection of high breast density.

Feasibility and acceptability could be assessed in the monitoring of programmes. Quality control procedures and quality standards should be further developed. Standards should be developed in particular for the image quality of synthesised 2D images from the tomosynthesis technology.

• Further research is needed to identify which high density group experiences the greatest net desirable consequences.
• There is a need for research examining the classification of mammographic breast density and standardisation of the classification systems used for breast density, including technology for the automation of the determination of breast density. Research should also aim at establishing the appropriate density threshold for changing imaging techniques.
• Further research is needed to build the evidence on benefits and harms of DBT vs DM through comparison of direct outcomes, including impacts of interval cancer incidence, stage of breast cancer at detection, and mortality reduction.
• There is also a need for research evidence on repeated DBT examinations since the current evidence is mainly restricted to a single surveillance episode.
• Further research should also assess the cost-effectiveness implications of tailored DBT screening for high mammographic breast density.
• Additional research should also assess the comparison between DBT and DM plus ultrasound for dense breast screening.
• Research is needed to define the quality parameters that need to be fulfilled for DBT-based breast cancer screening programmes to be implemented. 

The GDG agreed that the balance of desirable and undesirable effects favours not implementing additional MRI for screening women with high mammographic breast density with negative mammography (either DBT or DM), but the evidence is uncertain. The overall certainty in the evidence is very low because of the uncertainty in the estimates for breast cancer mortality, breast cancer incidence, interval breast cancers, and false positive recall rate. The evidence informing these conclusions is from studies of women with high mammographic breast density detected by DM, but the evidence also applies to women screened with DBT (and the effects may be smaller when DBT is used).

The evidence shows that there may be large benefits with the addition of MRI. There may be more breast cancers detected in the first round of screening (1.6 more per 100), of which 1.3 are invasive. The evidence is very uncertain for a reduction in breast cancer deaths (1 fewer per 100 women with high breast mammographic density), the number of breast cancers detected in the second round (0.6 more per 100), and whether cancers detected between screening (interval cancers) is reduced (0.4 fewer per 100). In addition, overdiagnosis may be reduced when adding MRI (0.4 per 100). However, there may be large harms: adding MRI may increase the number of women incorrectly called back for another exam due to false positive findings (9 more per 100), but this evidence is very uncertain. Overall, the large desirable effects of adding MRI may be equivalent to the undesirable effects.

There would be large costs associated with MRI, but the overall cost-effectiveness of adding MRI could vary across settings. The GDG agreed that most women would probably find additional MRI acceptable, but it is probably not acceptable to health care systems given limited human resource capacity, and the necessary specialised training. It is also probably not feasible to implement due to the increased number of MRIs that would be necessary, the reduced accessibility to MRI in many countries, the lack of quality assurance, and the scarcity of gadolinium. While the addition of MRI to current screening protocols (including frequency) is likely not feasible, it could be feasible in the future with other screening protocols, or with screening intervals that could potentially be longer than current intervals. Given the lack of accessibility and feasibility, the GDG agreed that adding MRI would increase health inequities.

The studies included mostly women with BI-RADS D. Given that there may be fewer benefits in women with BI-RADS C, the recommendation to not implement additional MRI might be considered even stronger in this group.

The acceptability of not providing an MRI may vary amongst women. Therefore, when asked, it will be important to share information about classification of breast density, availability of MRI, and the benefits and harms of MRI for women with high mammographic breast density and a negative mammography. 

Research priorities include:

• more research into measuring and classifying breast density,
• information to provide to women and communication,
• studies evaluating different screening frequencies and age to stop screening in women with high mammographic breast density,  and the use of abbreviated MRI,
• the role of Artificial Intelligence should be further investigated,
• the creation of an European consortium for a screening database would be beneficial in evaluating the efficacy of additional tests (e.g. MRI, ABUS, abMRI).

The GDG agreed that the balance of desirable and undesirable effects is unknown and therefore favours not implementing abbreviated MRI (abMRI) for screening women with high mammographic breast density with a negative mammography (either DBT or DM). The overall certainty in the evidence is very low because of the uncertainty in the estimates for breast cancer incidence and interval breast cancers. Other critical outcomes (including harms) were not measured in the studies and those effects are unknown. The evidence informing these conclusions is from a study including women with high mammographic breast density detected when receiving DBT, but the evidence applies also to women screened with DM.

The evidence shows that there may be more breast cancers detected in the first round of screening (2.7 more per 100), of which 1.7 are invasive. This benefit is large, but it is uncertain. There would be large costs associated with abbreviated MRI, but the overall cost-effectiveness of adding abbreviated MRI could vary across settings. While abbreviated MRI may be acceptable to some women, it is probably not acceptable to health care systems given limited human resource capacity and the necessary specialised training. It is probably not feasible to implement due to the increased number of MRIs that would be necessary, the reduced accessibility to MRI in many countries, the lack of quality assurance, and the scarcity of gadolinium. Currently, the addition of abbreviated MRI to current screening protocols (including frequency) is likely not feasible, but it could be feasible in future with other screening protocols or with screening intervals that could potentially be longer than current intervals. Given the lack of accessibility and feasibility, the GDG agreed that at a health care level, adding an abbreviated MRI would increase health inequities. 
 

The studies included women with BI-RADS C and D. There are likely fewer benefits in women with BI-RADS C and therefore the recommendation may be considered even stronger in this group. 

The acceptability of not providing an abbreviated MRI may vary amongst women. Therefore, when asked it will be important to share information about classification of breast density, availability of abbreviated MRI, and the benefits and harms of abbreviated MRI for women with high mammographic breast density and a negative mammography .

Research priorities include:

• more research into measuring and classifying breast density,
• information to provide to women and communication,
• studies evaluating different screening frequencies and age to stop screening in women with high mammographic breast density.
• Studies that directly compare MRI and abbreviated MRI are needed, including proof of principle studies to establish how sensitive it is compared to MRI.
• The role of Artificial Intelligence should be further investigated.
• The creation of an European consortium for a screening database would be beneficial in evaluating the efficacy of additional tests (e.g. MRI, ABUS, abMRI). 

The GDG agreed that the balance of desirable and undesirable effects favours not implementing automated breast ultrasound (ABUS) for screening women with high mammographic breast density with a negative mammography (either DBT or DM) but the evidence is uncertain. The overall certainty of the evidence is very low because of the uncertainty in the estimates for breast cancer incidence, interval breast cancers, and the number of women called back for another exam (recall rate). Other critical outcomes were not measured in the studies.

The evidence shows that there may be more breast cancers detected in the first round of screening (0.7 more per 100) of which 0.6 are invasive. This benefit is large, but it is uncertain. However, there may be a large number of women who are incorrectly called back for another exam or biopsy (15% more women when adding ABUS). There would also be large costs associated with ABUS, but the overall cost-effectiveness could vary across settings. While ABUS may be acceptable to some women, it is probably not acceptable to payers because of the high costs. It is probably not feasible to implement because of the greater needs for training and time to use ABUS and the lack of reimbursement. In the short and mid-term, the addition of ABUS to current screening protocols is likely not feasible, but it could be feasible in the long term if there are improvements in the technology. Given the lack of accessibility and feasibility, the GDG agreed that at a health care level, adding an ABUS would increase health inequities.
 

The studies included women with BI-RADS C and D. There are likely fewer benefits in women with BI-RADS C and therefore the recommendation may be considered even stronger in this group. 

The acceptability of not providing ABUS may vary amongst women. Therefore, when asked it will be important to share information about classification of breast density, availability of ABUS, and the benefits and harms of ABUS for women with high mammographic breast density and a negative mammography. 

Research priorities include:

• more research into measuring and classifying breast density,
• information to provide to women and communication,
• studies evaluating different screening frequencies and age to stop screening in women with high mammographic breast density.
• The role of Artificial Intelligence should be further investigated.
• The creation of an European consortium for a screening database would be beneficial in evaluating the efficacy of additional tests (e.g. MRI, ABUS, abMRI).