What is the sensitivity, specificity, positive predictive value, negative predictive value and feasibility of MRI in addition to mammography, rather than mammography alone, for women with an increased risk of breast cancer due to family history?
Screening by MRI should be reserved for women at very high risk, specifically the BRCA1/2 mutation carriers.
Adding MRI to mammography for the screening of high-risk women results in a higher sensitivity for breast cancer.
Level 1: A2 Lord 2007, Bigenwald 2008, Kuhl 2010, Hoogerbrugge 2008, Weinstein 2009
The diagnostic accuracy of MRI as a screening method varies according to the cut-off value used.
Level 1: A2 Warner 2008, Bigenwald 2008
The diagnostic accuracy of MRI as a screening method decreases as breast tissue density increases.
Level 3: A2 Bigenwald 2008
No obvious differences have been observed among the various groups in the diagnostic accuracy of MRI as a screening method. The heterogeneity within each group is just as significant as the heterogeneity between the groups.
Level 1: A2 Bigenwald 2008, Kriege 2007, Hoogerbrugge 2008, Weinstein 2009
It is likely that the comparison between mammograms and MRI in a first round is distorted by the fact that prior to the study mammograms had already been performed. The sensitivity of MRI is lower in women who have not had a prior mammogram; the numer of positive MRI scans decreases in subsequent rounds.
Level 1: A2 Kriege 2007, Hoogerbrugge 2008
It is likely that with MRI screening there is a higher risk of being called back and of having more biopsies, fewer lymphadenectomies, less adjuvant chemotherapy and less radiation therapy.
Level 1: A2 Chereau 2010, Lord 2007
It is likely that screening by both MRI and mammography improves the five-year survival rate of women at high risk due to a family history of the disease.
Level 3: A2 Rijnsburger 2010
The characteristics of tumours in BRCA1 gene mutation carriers differ from the tumours in BRCA2 gene mutation carriers and other high-risk groups; the tumours in BRCA1 carriers have poorer prognostic features and higher rates of growth.
Level 1: A2 Rijnsburger 2010, Tilanus-Linthorst 2007
Sensitivity / specificity
Lord's systematic review (2007) is qualitatively the best implemented and therefore provides the most reliable estimate of the diagnostic accuracy of MRI for this indication. It shows that, based on 5 studies, the sensitivity increases when MRI is added to mammography, namely 58% (95% CI 47-70) incremental sensitivity. In absolute terms, the sensitivity of MRI in combination with mammography is 94% (95% CI 86-98).
The specificity was too heterogeneous to be able to pool in a meta-analysis; for MRI combined with conventional tests it varied from 77-96%. The authors estimate that for every 1,000 screens, 10-24 additional cancers are detected by adding MRI.
Two studies published after this systematic review confirm these results for sensitivity and specificity [Bigenwald, 2008; Kuhl 2010]. But Hoogerbrugge (2008) reports a much lower sensitivity of 71% for MRI combined with mammography and 60% for just MRI. The reason for this lower sensitivity is not clear. Weinstein (2009) found a sensitivity of 71% for MRI and a specificity of 79%.
In Warner's systematic review (2008) the sensitivity of MRI for BI-RADS 3 is not significantly different from that for BI-RADS 4; the specificity is, however, significantly lower for BI-RADS 3 than for BI-RADS 4 (86% vs. 96%). Bigenwald (2008) also reports the sensitivity according to the BI-RADS score, with an apparent trend of decreasing sensitivity as the BI-RADS score increases, but with greater inaccuracy due to the small sample size. This study does not give specificity statistics.
Bigenwald (2008) reports the sensitivity of MRI vs. mammography based on the density of the breast tissue. Unfortunately, this study is small, so the estimates are imprecise. Their results show a possible trend of sensitivity decreasing as breast tissue density increases, but the confidence intervals are very wide, so the trend is not significant.
Some studies included a few women with a BRCA1/2 mutation [Hoogerbrugge, 2008], others included women with an elevated risk based on a mutation or family history [Bigenwald, 2008; Kriege, 2006; Kriege, 2007], or breast cancer in their own previous medical history [Kuhl, 2010; Weinstein, 2009]. There is no clear difference in sensitivity and specificity between these different groups. The heterogeneity within each group is at least as significant as the heterogeneity between the groups: for MRIs, Hoogerbrugge reports a sensitivity of 60% and Weinstein 71%, whereas the other studies report sensitivity ranging from 57-100%.
Effect of the screening round
All the studies included women with an elevated risk, who in many cases had already had a mammogram before the study began, but had never had an MRI. This distorts the comparison between MRI and mammography, because in the case of MRI scans, prevalent tumours are found in the first round, while in the case of mammograms it is no longer the first round.
Two studies evaluated a possible effect based on screening round [Hoogerbrugge, 2008; Kriege, 2006]. The first study found a decrease in the number of positive MRI scans, namely 18% after the first round and 9% in subsequent rounds. Kriege (2006) found a significant trend (p<0.001) for the number of positive MRI scans over the various rounds: 12.9% in the first round, 11.3% in the second round, 12.7% in the third round, 9.3% in the fourth round and 6.9% in the fifth and subsequent rounds. This study also evaluated the effect of undergoing a mammogram for inclusion in the study, and found 14.9% positive MRI scans in women who had had a mammogram previously, and 8.3% positive MRI scans in women without previous mammograms (p<0.001). Notably, for MRI scanning in women who had not had any previous mammograms, a sensitivity of 57% was found for invasive breast cancer. Even for mammography a significant difference was found: there were more abnormal mammograms (7.6%) in women who were receiving mammography for the first time than in women who had already had a mammogram (5.6%, p=0.09).
The predictive values are directly dependent on the prevalence of the outcome in the study population. The prevalence varied from 2.6 to 9.5%. This prevalence is of course dependent on the follow-up time, since most of the studies used follow-up to verify negative tests. The study with the lowest prevalence [Kriege, 2007] had a follow-up of 2.9 years; the study with the highest prevalence [Bigenwald, 2008] reported no follow-up duration, only the study duration, which was 9 years.
Two systematic reviews [Granader; 2008; Warner, 2008] also reported wide ranges in positive predictive values: 3-79% for MRIs, 6-100% for mammograms. The combination MRI and mammogram has a PPV between 3 and 79%.
Hoogerbrugge reports a positive predictive value of 17% for MRI; a different study [Kuhl, 2010] found a PPV of 48% for MRI, compared to 39% for mammography and 40% for the combination of MRI and mammography.
Kriege (2006) reports a PPV of 3.7-10.8% for MRI, depending on the round but without a significant trend. For invasive carcinomas, the same study did find a significantly higher PPV for mammography when it was the woman's first mammogram (22%) than when she had had a previous mammogram (3.8%, p=0.003). No significant difference was found for MRI: PPV 16% for a previous mammogram compared to 6.6% without a previous mammogram (p=0.18).
The negative predictive values are not reported in the two systematic reviews [Granader, 2008; Warner, 2008], due to the difficulty in verifying negative tests.
Only one study reports the negative predictive value [Kuhl, 2010]: MRI 99.9%, mammography 98.9%, MRI + mammography 100%.
Effect on quality of life
We found a study that evaluated the effect of screening on quality of life in this high risk group [Rijnsburger, 2004]. The authors reported that the screening modality had no effect on quality of life, but they did find a significant effect from additional diagnostic testing, which changed over time.
Also, an apparently greater proportion of women reported pain and inconvenience with mammography than with MRI, and an apparently greater proportion of women reported distress/anxiety with MRI than with mammography (no statistical hypothesis testing).
Effect on morbidity, including treatments for breast cancer
One study found no significant difference between cancers detected by MRI and cancers detected by a different method, in terms of the size, grade of tumour differentiation, estrogen/progesterone receptor and tumour type [Kriege, 2007]. The tumours that were found using MRI were significantly less node positive (6% vs. 44%, p=0.02). Compared with mammography, MRI is significantly more sensitive for T1 tumours, N0, non-ductal tumours and estrogen receptor positive tumours. Another study did find a significant difference in size of the invasive tumour when screening by MRI in comparison with screening without MRI: 6 mm vs. 22 mm, p<0.0001 [Chereau, 2010].
A systematic review found that the risk of having to be re-tested because of false-positive results increases by a factor 3 (RR 3.43-4.86), which is equal to 71-74 additional call-backs for false-positive results per 1,000 screens. This involves 7-46 additional benign percutaneous biopsies (RR 1.22-9.50), and 2 additional benign surgical biopsies (RR 2.0; 95% CI 0.5-8.0) [Lord 2007]. A later study also found an increase in the number of biopsies, but without statistical hypothesis testing [Weinstein, 2009].
As far as treatments are concerned, there were fewer axillary node dissections (43% vs. 68%, p=0.03), less adjuvant chemotherapy (43% vs. 86%, p=0.0001), and less radiotherapy (62% vs. 81%, p=0.05) [Chereau, 2010] using MRI screening compared to screening without MRI.
A false-positive result from MRI plus mammography has no effect on the woman's preference for having a prophylactic mastectomy or surveillance [Hoogerbrugge, 2008].
Effects on mortality
There are no randomised studies on the effect of adding MRI to the screening program. It is therefore unknown whether the higher sensitivity of MRI for diagnosing breast cancer also translates into a lower mortality: either breast cancer-related mortality or general mortality.
MRI was already included in various screening programs, which complicates conducting an RCT, hence an RCT may never be conducted. Because of this, it may never be possible to quantify the risk of overdiagnosis and overtreatment in this high-risk group.
Chereau (2010) found no significant difference in three-year disease-free survival, metastasis-free survival and total survival using MRI screening compared to screening without MRI. It should be noted that with screening, survival (as opposed to mortality) is a poor measure of outcome, because it is distorted by lead-time bias. MRI scans can accelerate the time of diagnosis but do not change the ultimate mortality rate.
MRISC follow-up study
After the clinical question appeared, the long-term results of the MRISC (MRI Screening in women with familial or genetic predisposition for breast cancer) study were published [Rijnsburger, 2010]. This study found that MRI has a sensitivity of 71%, a specificity of 90% and a PPV of 7.7%. Mammography has a sensitivity of 41%, a specificity of 95% and a PPV of 8.5%. The percentage of tumours of 1 cm or less is 40.5%, the percentage of node-negative tumours is 70%. The overall five-year survival of 93% is higher than the 74% survival of historic cohorts who were not screened by MRI.
The detection rate in the gene mutation carriers was 62/1,000 screens, in very high-risk women 24/1,000 screens, and in moderately high-risk women 31/1,000 screens.
The study shows that breast cancers in the BRCA1 gene mutation carriers form a separate group. In almost all cases these were invasive (93.5%), high-grade (grade 3, 78%) cancers, occurring at a young age (58% under age 40). The sensitivity of mammography was very low, at 25%, and the sensitivity of MRI was higher, at 67%. The percentage of interval cancers was 32%. This indicates a higher rate of growth, as described previously by Tilanus-Linthorst (2007).
MRI screening requires radiological expertise, especially because of its low specificity. This expertise is best guaranteed in hospitals with a clinical geneticist, because surveillance of mutation carriers is concentrated there.
It is gradually becoming clear that cancers that occur with BRCA1 have characteristics associated with a poorer prognosis than cancers in other women with elevated risk due to family history. There is discussion of changing the screening schedule, with the idea of alternating screening by mammography or MRI respectively with an interval of 6 months. Another concept is increasing the frequency of MRI: every 6 months until age 40. However, insight inthe consequences for the women (long-term effects of additional use of Gadolineum and false-positive findings) is lacking. Detection of cancers in very high risk groups and moderately high risk groups lags behind detection in mutation carriers. An RCT was started in November 2010 at Erasmus Medical Centre to obtain more insight into the relationships between breast tissue density, cancer risk and diagnostic accuracy of MRI in these women. This is the FaMRISC study, to be conducted at 9 centres. The intent is to include 2,000 women with a lifetime risk (LTR) of more than 20%, with the goal of detecting 50 cancers in 4 years. In one arm women undergo annual clinical breast examination and mammography. In the other arm women undergo annual clinical breast examination and MRI. Every two years, an additional mammogram will be done, because of the lower sensitivity of MRI for DCIS.
Authorization date and validity
Last review : 13-02-2012
Last authorization : 13-02-2012
The national Breast Cancer guideline 2012 is a living guideline, in other words there is no standard term of revision. NABON continually watches at new developments and clinical problems in the areas of screening, diagnostics, treatment and aftercare, and whether this requires an update.
Initiative and authorization
Initiative : Nationaal Borstkanker Overleg NederlandAuthorized by:
- Nederlandse Internisten Vereniging
- Nederlandse Vereniging voor Heelkunde
- Nederlandse Vereniging voor Psychiatrie
- Nederlandse Vereniging voor Radiologie
- Nederlandse Vereniging voor Radiotherapie en Oncologie
Approximately 14,000 women (and 100 men) are diagnosed with invasive breast cancer each year in the Netherlands, and about 1,900 have an in situ carcinoma. A woman's risk of having breast cancer over the course of her life is 12-13%. This means that breast cancer is the most common form of cancer in women in the Netherlands. Early detection, particularly via national breast cancer screening, combined with adjuvant therapy followed by locoregional treatment, improves the prognosis in women with breast cancer
The guideline on Breast Cancer Screening and Diagnostics, published in 2000, was updated in 2007. In 2002, the first multidisciplinary National Breast Cancer Guideline was published, it was revised in 2004, 2005 and 2006. In 2008 both guidelines were combined to Breast Cancer Guideline, which 2012 revision is now effected.
Scope and target group
This guideline is written for all the members of the professional groups that have contributed to its development.
This guideline is a document with recommendations and instructions to support daily practice. The guideline is based on the results of scientific research and expert opinion, with the aim of establishing good medical practice. It specifies the best general care for women with (suspected) breast cancer and for those who are eligible for screening. The guideline aims to serve as a guide for the daily practice of breast cancer screening, diagnostics, treatment and aftercare. This guideline is also used in the creation of informational materials for patients, in cooperation with the KWF (Dutch Cancer Society).
A core group consisting of a radiologist, surgeon, pathologist, medical oncologist and radiation therapist began preparing for the revision of the breast cancer practice guidelines in 2009. A multidisciplinary guideline development group was formed in early 2010 to implement the revision. This group consisted of mandated representatives from all of the relevant specialisations concerned with breast cancer, plus two delegates from the BVN (Dutch Breast Cancer Society) (see list of guideline development group members). The benefits of such a multidisciplinary approach are obvious: not only does it best reflect the care, but it offers the greatest possible expertise for the guideline. In composing the development group, geographic distribution of the members, balanced representation of the various organisations and agencies concerned, and a fair distribution in academic background were taken into account as much as possible.
The guideline development group received procedural and administrative support from IKNL (Comprehensive Cancer Centre for the Netherlands) and support on methodology from Bureau ME-TA. Partial funding was obtained from SKMS (Quality Funds Foundation of Dutch Medical Specialists). This subsidy would not have been possible without the extensive assistance provided by the NVvR (Radiological Society of the Netherlands).
Declaration of interest
Partial funding for the guideline revision was obtained from the Society of Dutch Medical Specialists in the framework of the SKMS. IKNL sponsored some of the cost. On two occasions, as well as at the beginning and end of the process, all of the members of the guideline development group were asked to fill out a statement of potential conflicts of interest, in which they stated their relationship with the pharmaceutical industry. A list of these statements of interest can be found in the appendices.
In developing this guideline, four clinical questions were formulated. These questions emerge from an inventory of clinical problems collected in the field from professionals, patients and patient representatives.
Also, A multidisciplinary guideline development group was formed in early 2010 to create and implement the revision. This group consisted of mandated representatives from all of the relevant specialisations concerned with breast cancer, plus two delegates from the BVN (Dutch Breast Cancer Society).
Method of development
Feasibility has been taken into account in developing the guideline. This included attention to factors that could promote or hinder putting the advice into practice. Examples include the implementation of an analysis of problems, the multidisciplinary composition of the guideline development group, and making active use of support from the guideline development group members. Presenting the draft guideline to the field and communicating what, if anything, is being done with the responses, also promotes implementation. In this manner, a guideline has been developed that answers current questions in the field.
The guideline is distributed widely and is available in digital form on the Dutch Guideline Database. The guideline may also be brought to the attention of a wider audience in other periodicals or continuing education sessions, for example. To promote use of the guideline, we recommend that the regional tumour working groups and group practices, as well as scientific and professional organisations, repeatedly bring the guideline to the attention of their members. Any problems that may arise in using the guidelines can then be discussed and, when appropriate, submitted to the national guideline development group, as it is a "living" guideline. If desirable, parts of the guideline can be made more explicit by formulating regional additions or translation to the local situation in departmental and/or hospital protocols.
In principle, indicators are determined during development of the guideline that can be used to monitor implementation of the recommendations. Via a documentation project, these indicators can then be used to determine the extent of compliance with the guideline. The information from the documentation project becomes input for the revision of the guideline.
Methods and proces
This module has been evidence-based revised in 2008 and consensus based updated in 2012.
A revision of an existing guideline consists of revised and updated text. Revised text is new text based on an evidence-based review of the medical literature; updated text is the old guideline text which has been edited by the experts without performing a review of medical literature. Each section of the guideline states what type of revision has taken place. Each chapter of the guideline is structured according to a set format, given below. The purpose of this is to make the guideline transparent, so that each user can see on what literature and considerations the recommendations are based on.
Description of the literature
To the greatest extent possible, the answers to the fundamental questions (and therefore the recommendations in this guideline) were based on published scientific research. The articles selected were evaluated by an expert in methodology for their research quality, and graded in proportion to evidence using the following classification system:
Classification of research results based on level of evidence
Research on the effects of diagnostics on clinical outcomes in a prospectively monitored, well-defined patient group, with a predefined policy based on the test outcomes to be investigated, or decision analysis research into the effects of diagnostics on clinical outcomes based on results of a study of A2-level and sufficient consideration is given to the interdependency of diagnostic tests.
Research relative to a reference test, where criteria for the test to be investigated and for a reference test are predefined, with a good description of the test and the clinical population to be investigated; this must involve a large enough series of consecutive patients; predefined upper limits must be used, and the results of the test and the "gold standard" must be assessed independently. Interdependence is normally a feature of situations involving multiple diagnostic tests, and their analysis must be adjusted accordingly, for example using logistic regression.
Comparison with a reference test, description of the test and population researched, but without the other features mentioned in level A.
Opinions of experts, such as guideline development group members
Based on the medical literature, one or more relevant conclusions are made for each section. The most important literature is listed according to the level of evidential strength, allowing conclusions to be drawn based on the level of
evidence. All the medical literature included in the conclusion is described in the bibliography.
Classification of conclusions based on literature analysis
Based on 1 systematic review (A1) or at least 2 independent A2 reviews.
Based on at least 2 independent B reviews
Based on 1 level A2 of B research, or any level C research
Opinions of experts, such as guideline development group members
Based on the conclusion(s), recommendations are made. However, there are other considerations that contribute to formulation of the recommendation besides literature evidence, such as safety, the patients' preferences, professional expertise, cost-effectiveness, organisational aspects and social consequences. The other considerations are mentioned separately. In this manner, it is clear how the guideline development group arrived at a particular recommendation.
The final wording of the recommendation is the result of the scientific conclusion, taking into account the other considerations. The purpose of following this procedure and drawing up the guidelines in this format is to increase transparency.
An alphabetical list of literature references can be found at the end of the guideline.
All draft texts have been discussed by the guideline development group.
Searches are available upon request. Please contact the Richtlijnendatabase.
In het programma Doen of Laten? Het terugdringen van onnodige zorg zijn aanbevelingen geïdentificeerd die adviseren iets niet te doen. Doel is aandacht te vragen voor terughoudendheid en te helpen bij het maken van weloverwogen keuzes. Meer over Doen of Laten?