Mammography in combination with breast ultrasonography versus mammography for breast cancer screening in women at average risk
Abstract
Background
Breast cancer is the most common malignant disease diagnosed in women worldwide. Screening with mammography has the ability to detect breast cancer at an early stage. The diagnostic accuracy of mammography screening largely depends on the radiographic density of the imaged breasts. In radiographically dense breasts, non‐calcified breast cancers are more likely to be missed than in fatty breasts. As a consequence, some cancers are not detected by mammography screening. Supporters of adjunct ultrasonography to the screening regimen for breast cancer argue that it might be a safe and inexpensive approach to reduce the false negative rates of the screening process. Critics, however, are concerned that performing supplemental ultrasonography on women at average risk will also increase the rate of false positive findings and can lead to unnecessary biopsies and treatments.
Objectives
To assess the comparative effectiveness and safety of mammography in combination with breast ultrasonography versus mammography for breast cancer screening for women at average risk of breast cancer.
Search methods
We searched the Cochrane Breast Cancer Group's Specialised Register, MEDLINE (via OvidSP) and EMBASE up until February 2012.
To detect ongoing or unpublished studies, we searched the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP), ClinicalTrials.gov and the National Cancer Institute's clinical trial database until June 2012. In addition, we conducted grey literature searches using the following sources: OpenGrey; National Institute of Health RePORTER; Health Services Research Projects in Progress (HSRPROJ); Hayes, Inc. Health Technology Assessment; The New York Academy of Medicine’s Grey Literature Index and Conference Papers Index.
Selection criteria
For efficacy, we considered randomised controlled trials (RCTs), with either individual or cluster randomisation, and prospective, controlled non‐randomised studies with a low risk of bias and a sample size of at least 500 participants.
In addition to studies eligible for efficacy, we considered any controlled, non‐randomised study with a low risk of bias and a study size of at least 500 participants for the assessment of harms.
Our population of interest were women between the ages of 40 and 75 years who were at average risk for breast cancer.
Data collection and analysis
Two review authors screened abstracts and full‐text publications against the inclusion criteria. None of the studies met our inclusion criteria.
Main results
Our review did not detect any controlled studies on the use of adjunct ultrasonography for screening in women at average risk for breast cancer. One ongoing randomised controlled trial was identified (J‐START, Japan).
Authors' conclusions
Presently, there is no methodologically sound evidence available justifying the routine use of ultrasonography as an adjunct screening tool in women at average risk for breast cancer.
PICOs
Plain language summary
Mammography followed by ultrasonography compared to mammography alone for breast cancer screening in women at average risk of breast cancer
Worldwide, breast cancer is the most common malignancy in women. Evidence shows that mammography in healthy women 50 to 70 years of age can detect breast cancer early and reduce the risk of dying from breast cancer. Mammography, however, is not a perfect tool to detect breast cancer and misses some tumours in some women, particularly in women who have dense breasts. In women with dense breasts, the normal breast tissue and the tumour are difficult to distinguish from each other on the mammogram. Because of this, some supporters feel that the addition of ultrasonography screening of these women in addition to the mammography screening will detect those tumours that are missed by mammography alone. Others feel that this will increase the rate of false positive tumours and increase the number of biopsies and unnecessary treatment.
The benefit of ultrasound as an additional examination for women who do not have especially dense breasts and who have normal mammographies is uncertain. This review sought to examine the evidence for and against adding ultrasonography screening to mammograms for women at average risk for breast cancer. It is important to weigh positive and negative sides of screening because the detection of more tumours by screening does not necessarily mean that more women will have their lives saved. We need to assess whether the few additional cancers that may be detected by ultrasonography lead to a real decrease in mortality from breast cancer and then balance any benefit against the harm caused by many women being incorrectly alarmed or diagnosed.
We did not find any trials that addressed our review question. One randomised controlled trial is currently underway in Japan (called J‐START). Because it is unclear whether ultrasonography in women with normal mammographies can reduce the risk of dying from breast cancer, they should not be used on a routine basis. If screening with ultrasonography is performed it should be as part of a clinical trial designed to test the effect of additional screening on mortality and the harms experienced by women who have a positive ultrasonography screening test.
Authors' conclusions
Background
Description of the condition
Breast cancer is the most common malignant disease diagnosed in women worldwide, comprising 16% of all female cancers (World Health Organization 2011). The risk of developing breast cancer increases with age and certain risk factors such as dense breasts, family history of breast or ovarian cancer, or familial breast cancer gene mutations of BRCA1 (BReast CAncer 1, early onset) and BRCA2 (BReast CAncer 2, susceptibility protein).
Screening with mammography has the ability to detect breast cancer at an early stage. Subsequent effective diagnostic pathways and treatment regimens can reduce the burden of disease of breast cancer, most importantly mortality in women aged 50 to 70 years (USPSTF). A Cochrane review estimated a relative reduction of mortality from breast cancer of 15%, corresponding to an absolute risk reduction of 0.05 per cent in women aged 50 and older (Gøtzsche 2011). In addition to a reduction in mortality, studies evaluating the efficacy of mammographic screening have repeatedly reported a reduction in breast cancer morbidity (Griffin 2010). The sensitivity of mammography ranges between 77% and 95% and the specificity ranges between 94% and 97% (Nelson 2009). The diagnostic accuracy of mammography screening largely depends on the radiographic density of the imaged breasts (Carney 2003). In radiographically dense breasts, non‐calcified breast cancers are more likely to be missed than in fatty breasts. As a consequence, some cancers are not detected by mammography screening.
Ultrasonography of the breast is currently not recommended in screening of women at average risk for breast cancer (Elmore 2005; Griffin 2010). Most clinical practice guidelines specify ultrasonography of the breast as a supplementary examination for further clarification of ambiguous findings (Albert 2009). The European Guidelines on Quality Assurance in Breast Cancer Screening and Diagnosis state that ultrasonography should be carried out in the presence of a discrete clinical mass even if negative on mammography (Perry 2008).
Supporters of supplemental ultrasonography to the screening regimen for breast cancer argue that it might be a safe and inexpensive approach to reduce the false negative rates of the screening process. Critics, however, are concerned that performing supplemental ultrasonography on women at average risk will also increase the rate of false positive findings and can lead to unnecessary biopsies and treatments. Authors of a 2009 systematic review of six observational studies in ultrasonography noted the increased biopsy rate in women at intermediate risk, finding a mean positive predictive value of 15% (range 2% to 28%) from four of the six studies, that is, the percentage of positively classified findings for which no carcinoma was subsequently found ranged from 72% to 98% (Nothacker 2009).
Description of the intervention
The intervention entails any form of mammography screening (for example one view, two views, digital, etc) that meets the technical standards of the European Guidelines for Quality Assurance in Breast Cancer Screening and Diagnosis (Perry 2008) with adjunct breast ultrasonography used as a sequential screening test (that is breast ultrasonography that is conducted in women with negative screening mammograms). Breast ultrasonography as a diagnostic test following a positive mammogram is not of interest for this systematic review. We draw this distinction because only ultrasonographies conducted in women with negative mammograms are true screening tests because their goal is to increase the sensitivity of the screening procedure.
To be eligible for this report, ultrasonography needs to be performed with a high‐frequency transducer of 7.5 MHz or higher.
How the intervention might work
To increase either sensitivity or specificity, two or more screening tests may be applied in the same individuals. These tests can be used sequentially or simultaneously. Sequential screening tests are applied in a proportion of the population with a specific result of the first screening test. In sequential screening the post‐test probability of the first screening test becomes the pre‐test probability of the second screening test. The goal of sequential screening is usually to increase sensitivity. By contrast, simultaneous screening applies two (or more) tests to the screened individuals without knowledge of the results of each individual test. Therefore, the pre‐test probability remains the same for all tests.
Breast ultrasonography is used routinely as a diagnostic measure to distinguish benign from malignant lesions because it can differentiate between cysts and solid tumours and thus lowers the number of indeterminate mammographical findings. A 2008 study found an increase in diagnostic accuracy when using breast ultrasonography in addition to mammography (accuracy of 0.78 (95% confidence interval (CI) 0.67 to 0.87) for mammography alone compared with 0.91 (95% CI 0.84 to 0.96) when mammography is combined with ultrasonography) (Berg 2008). Thus, breast ultrasonography as an adjunct screening tool to mammography might also be able to detect cancer lesions that mammography screening misses. We consider adjunct breast ultrasonography a sequential screening test because it is administered as an add‐on test in women with a negative mammogram. Women with a positive mammogram will also receive breast ultrasonography, but for this population ultrasonography is a diagnostic test.
In women at increased risk for breast cancer, defined by high breast density or other risk factors, several studies have demonstrated that supplemental screening with ultrasonography can increase the detection rates of cancer, particularly in women with dense breasts (Berg 2008; Nothacker 2009). Mammographically dense breast tissue is an independent risk factor for breast cancer and is associated with a high risk of interval cancers, that is cancers that become clinically apparent between screening tests (Boyd 2007). Ultrasonography, therefore, has the potential to detect mammographically occult cancers at an earlier stage and to improve surrogate outcomes such as tumour size and lymph node status, which have been linked to a poor prognosis of breast cancer (Michaelson 2002; Michaelson 2003).
Why it is important to do this review
In women at increased risk for breast cancer, adjunct ultrasonography can improve the diagnostic yield of breast cancer screening (Berg 2008). Based on these findings, ultrasonography is sometimes used routinely as an adjunct screening tool in women at average risk. It is unclear whether the use of ultrasonography as an adjunct screening tool in women at average risk corresponds to a reduction in mortality and morbidity (the ultimate goal of any screening programme) or to an increase in screening‐related harms.
Objectives
To assess the comparative effectiveness and safety of mammography in combination with breast ultrasonography versus mammography for breast cancer screening for women at average risk of breast cancer. Figure 1 depicts the analytic pathway of the research question.
Analytic pathway of the comparative efficacy and risk of harms of mammography screening with and without supplemental ultrasonography
Methods
Criteria for considering studies for this review
Types of studies
For efficacy we considered RCTs with either individual or cluster randomisation and prospective, controlled non‐randomised studies with a low risk of bias and a sample size of at least 500 participants.
In addition to studies eligible for efficacy, we considered any controlled, non‐randomised study with a low risk of bias and a study size of at least 500 participants for the assessment of harms.
Studies needed to have a follow‐up period of at least one year and had to include at least one relevant outcome.
Types of participants
Women between the age of 40 and 75 years who are at average risk for breast cancer, have not previously had breast cancer, and who participate in a breast cancer screening program or undergo mammography screening.
We define women at average risk as those who have a lifetime risk of less than 15% or who have dense breasts without any additional risk factors for breast cancer.
Types of interventions
Any form of mammography screening (for example one view, two views, digital, etc) that meets the technical standards of the European Guidelines for Quality Assurance in Breast Cancer Screening and Diagnosis (Perry 2008) with additional breast ultrasonography compared with mammography screening without breast ultrasonography.
Types of outcome measures
Primary outcomes
-
Breast cancer mortality
Secondary outcomes
-
All‐cause mortality
-
Incremental cancer detection rate
-
Incremental detection rate of invasive cancers
-
Rate of interval cancers
-
Lymph node status
-
Size of detected cancers
-
Health‐related quality of life
-
False positive rate
-
False negative rate
-
Rate of biopsies
-
Screening associated harm (psychological distress, adverse effects caused by subsequent diagnostic or therapeutic interventions, others)
Search methods for identification of studies
Electronic searches
We searched the following databases:
-
The Cochrane Breast Cancer Group (CBCG) searched their Specialised Register. Details of the search strategies used by the Group for the identification of studies and the procedure used to code references are outlined in the Group's module: http://www.mrw.interscience.wiley.com/cochrane/clabout/articles/BREASTCA/frame.html.
-
MEDLINE (via OvidSP) (from July 2008 to February 2012). See Appendix 1 for the full search strategy.
-
EMBASE (via Embase.com) (2008 to February 2012). See Appendix 2 for the full search strategy.
-
The WHO International Clinical Trials Registry Platform (ICTRP) search portal (http://apps.who.int/trialsearch/AdvSearch.aspx) for all prospectively registered and ongoing trials to January 2012. See Appendix 3 for the search strategy.
Searching other resources
We manually searched reference lists of pertinent reviews and relevant background articles on this topic to look for any relevant citations that our searches might have missed.
We searched for grey literature (through June 2012) in the following databases:
-
OpenGrey;
-
ClinicalTrials.gov;
-
National Cancer Institute's clinical trial database;
-
National Institute of Health RePORTER;
-
Health Services Research Projects in Progress (HSRPROJ);
-
Hayes Inc. Health Technology Assessment;
-
The New York Academy of Medicine’s Grey Literature Index; and
-
Conference Papers Index.
Data collection and analysis
Selection of studies
We developed and pilot‐tested literature review forms for abstract and full‐text reviews. Two authors (AC, AK, DB, GG, KT, THH, MvN) independently reviewed abstracts. We retrieved full‐text copies of all studies that potentially met the inclusion criteria based on the abstract review. Studies marked for possible inclusion by either review author underwent a full‐text review. For studies that lacked adequate information to determine inclusion or exclusion, we retrieved the full text and then made the determination. If the necessary information in the full‐text articles was unclear or missing, we contacted authors of the publications. Two trained members of the research team (GG, KT) independently reviewed each full‐text article for inclusion or exclusion based on the eligibility criteria described above. If both review authors agreed that a study did not meet the eligibility criteria, we excluded it. If the review authors disagreed, they resolved conflicts by discussion and consensus or by consulting a third member of the review team. All results were tracked in an EndNote X5 database.
Data extraction and management
We designed, pilot‐tested and used structured data extraction forms to gather pertinent information from relevant articles; this included characteristics of study populations, settings, interventions, comparators, study designs, methods and results.
Assessment of risk of bias in included studies
We planned to assess the risk of bias of included randomised trials using the Cochrane 'Risk of bias' tool as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). The tool includes assessment of: sequence generation; allocation sequence concealment; blinding of participants, personnel and outcome assessors; incomplete outcome data; selective outcome reporting and other potential threats to validity. In addition, we planned to assess whether all relevant outcomes for the trial were reported in the published articles. We intended to rate each domain as high risk of bias, low risk of bias or unclear risk of bias.
For non‐randomised studies, we planned to use criteria involving selection of cases or cohorts and controls, adjustment for confounders, methods of outcomes assessment, length of follow‐up and statistical analysis (Higgins 2011).
Measures of treatment effect
We planned to use extracted data from the original studies to construct 2 x 2 tables. Where multiple studies would have allowed for quantitative analysis, we planned to calculate the risk ratio or odds ratio with 95% confidence intervals for each outcome. In addition, we planned to pool continuous data using the mean difference or standardised mean difference. For time‐to‐event data, we planned to calculate a pooled hazard ratio where this was available or to dichotomise data at multiple time points into response/no response (e.g. at one week, two weeks, four weeks, etc).
Unit of analysis issues
The unit of our analyses was intended to include women (not cancer lesions).
Dealing with missing data
We intended to use intention‐to‐treat analysis where data were missing from participants who dropped out of trials before completion. Where data regarding an outcome of interest were not reported, we planned to contact authors of publications to obtain missing results.
Assessment of heterogeneity
We planned to use the Cochran Chi2 test (Q‐test) to assess heterogeneity. We intended to use the I2 statistic to estimate the degree of heterogeneity. This measure describes the percentage of total variation across studies that results from heterogeneity rather than chance. We would have interpreted the importance of any heterogeneity in terms of its magnitude and the direction of effects. We would not have used thresholds; instead we would have adopted the overlapping bands suggested in the Cochrane Handbook. For example, we planned to consider an I2 of between 0% and 40% as probably not important, between 30% and 60% as representing moderate heterogeneity, between 50% and 90% as substantial heterogeneity, and between 75% and 100% as considerable heterogeneity (Higgins 2011).
Assessment of reporting biases
We checked trial registries (for example WHO ICTRP and www.clinicaltrials.gov) to detect completed but unpublished trials.
Data synthesis
We planned to analyse data using Review Manager 5.1 (RevMan 5.1). We would have pooled data for meta‐analysis where the participant groups were similar and the studies assessed the same treatments with the same comparator and had similar definitions of outcome measures over a similar duration of treatment. We planned to use a fixed‐effect model where heterogeneity was low and a random‐effects model where the presence of heterogeneity resulted in a higher I2 unless too few studies were included in the analysis. We planned to rate the strength of the evidence based on the system developed by the GRADE Working Group.
Subgroup analysis and investigation of heterogeneity
We planned to perform subgroup analyses based on breast density.
Sensitivity analysis
We had planned to conduct sensitivity analyses excluding small studies, studies with a high risk of bias and studies published in abstract form.
Results
Description of studies
Results of the search
We identified 2731 citations from searches and reviews of reference lists. Overall, we did not find any completed studies that met our eligibility criteria. Figure 2 depicts the numbers of search results and the flow of the literature for this report.
Study flow diagram.
Included studies
We did not find any controlled studies assessing the incremental benefits and harms of adjunct screening ultrasonography in women at average risk for breast cancer. Our searches in clinical trial registries detected one ongoing RCT in Japan that is potentially relevant. This trial, termed J‐START (Japan‐Strategic Anti‐cancer Randomized Trial), is a large‐scale study that will randomly assign 100,000 Japanese women aged 40 to 49 years to either mammography or mammography with adjunct ultrasonography (Ohuchi 2011). The primary endpoints of this trial are sensitivity and specificity; the secondary endpoint is the accumulated incidence rate of advanced cancers during the four‐year follow‐up period.
Excluded studies
Overall, we excluded 45 studies. The main reasons for exclusion were study populations that did not meet the eligibility criteria (e.g. women with high risk of breast cancer) or study designs that ascertained only the diagnostic yield of adjunct ultrasonography without taking screening‐relevant health outcomes into consideration. Reasons for excluding studies after full‐text review are summarised under Characteristics of excluded studies.
Risk of bias in included studies
We did not include any studies.
Allocation
We did not include any studies.
Blinding
We did not include any studies.
Incomplete outcome data
We did not include any studies.
Selective reporting
We did not include any studies.
Other potential sources of bias
We did not include any studies.
Effects of interventions
We did not include any studies.
Discussion
Summary of main results
Overall, our review did not detect any controlled studies that provided evidence for (or against) the use of adjunct ultrasonography for screening in women at average risk for breast cancer. The only available evidence regarding adjunct breast ultrasonography in women at average risk is limited to one uncontrolled observational study of women with normal screening mammograms who received sequential ultrasonography screening (Buchberger 2000). This study, conducted in Tyrol, Austria, reported an incremental diagnostic yield of 2.6 cancers per 1000 women without a personal history of breast cancer. Although this study provides some information regarding the performance of adjunct ultrasonography as a sequential screening test, i.e. how well it diagnoses illness, it cannot provide information on the overall usefulness as a screening test, i.e. whether adjunct ultrasonography results in a reduction of morbidity and mortality. Simply diagnosing more cases of illness does not necessarily result in lower mortality or less morbidity. Screening for neuroblastoma in children in Japan in the late 1990s provides a dramatic historical example of how inferences on the usefulness of cancer screening, when based solely on incremental cancer detection rates, can result in a screening programme that causes more harm than benefits (Soderstrom 2005).
Overall completeness and applicability of evidence
Despite extensive searches of the grey literature, we did not find any eligible studies. A separate publication will extrapolate findings of results from women at elevated risk for breast cancer to estimate the false positive rates in women at average risk who were recalled because of positive ultrasonographies (Gartlehner 2013).
Quality of the evidence
We did not include any studies.
Potential biases in the review process
Publication bias is a threat for any systematic review. Although we have conducted extensive searches of grey literature, we cannot be sure that we have detected each study conducted in this field. We have identified one study that has been registered but has not yet been published (Ohuchi 2011).
Agreements and disagreements with other studies or reviews
We did not find any other studies that addressed the research question. A systematic review of adjunct ultrasonography in women at high risk for breast cancer, defined by increased breast density or other risk factors, demonstrated that supplemental screening with ultrasonography can increase the detection rates of cancer at the cost of a high false positive rate (Nothacker 2009).
Analytic pathway of the comparative efficacy and risk of harms of mammography screening with and without supplemental ultrasonography
