Final Research Plan

Breast Cancer: Screening

July 17, 2014

Recommendations made by the USPSTF are independent of the U.S. government. They should not be construed as an official position of the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services.

The final Research Plan is used to guide a systematic review of the evidence by researchers at an Evidence-based Practice Center. The resulting Evidence Review forms the basis of the USPSTF Recommendation Statement on this topic.

The draft Research Plan was available for comment from November 14 until December 11, 2013 at 5:00 p.m., ET.

* Excludes women with pre-existing breast cancer; clinically significant BRCA gene mutations, Li-Fraumeni syndrome, Cowden syndrome, hereditary diffuse gastric cancer, or other familial breast cancer syndromes; high-risk lesions (ductal or lobular carcinoma in situ, atypical ductal or lobular hyperplasia); or previous large doses of chest radiation therapy (≥20 Gy) before age 30 years.
† Addresses Contextual Question 1.

In the target population of women age 40 years and oldera:

  1. What is the effectiveness of routine mammography screening in reducing breast cancer–specific and all-cause mortality, and how does it differ by age, risk factorb, and screening interval?
  2. What is the effectiveness of routine mammography screening in reducing the incidence of advanced breast cancer and treatment-related morbidityc, and how does it differ by age, risk factorb, and screening interval?
  3. How does the effectiveness of routine breast cancer screening in reducing breast cancer–specific and all-cause mortality differ by screening strategyd?
  4. How does the effectiveness of routine breast cancer screening in reducing the incidence of advanced breast cancer and treatment-related morbidityc differ by screening strategyd?
  5. What are the harmse of routine mammography screening, and how do they differ by age, risk factorb, and screening interval?
  6. How do the harmse of routine breast cancer screening differ by screening strategyd?
  7. What are the test performance characteristics of tomosynthesis (3-D mammography) as a primary screening modality for breast cancer, performed either alone or simultaneously with 2-D digital mammography? How do these performance characteristics differ by age and risk factorb?
  8. What are the test performance characteristics of newer technologiesf for breast cancer screening when used as adjunctive screening after a negative screening mammogram in women found to have dense breasts? How do these performance characteristics differ by age and risk factorb?
  9. What is the accuracy and interrater reliability of the BI-RADS (Breast Imaging Reporting and Data System) determination of breast density?
  10. What is the effectiveness of adjunctive screening for breast cancer with newer technologiesf on intermediate breast cancer outcomesg (e.g., ductal carcinoma in situ [DCIS] detection rates, stage at diagnosis, and interval cancer rates) when performed after a negative screening mammogram in women found to have dense breasts?

a Excludes women with pre-existing breast cancer; clinically significant BRCA gene mutations, Li-Fraumeni syndrome, Cowden syndrome, hereditary diffuse gastric cancer, or other familial breast cancer syndromes; high-risk lesions (ductal or lobular carcinoma in situ, atypical ductal or lobular hyperplasia); or previous large doses of chest radiation therapy (≥20 Gy) before age 30 years.
b Risk factors include: age, family history, breast density, parity, race/ethnicity, age at first birth, menopausal status, current use of menopausal hormone therapy or oral contraception, prior benign breast biopsy, and, for women age >50 years, body mass index.
c Morbidity includes: physical adverse effects of treatment, quality of life measures, and other measures of impairment.
d Screening strategies include: mammography (digital, tomosynthesis), magnetic resonance imaging (MRI), ultrasonography, and clinical breast examination, either alone or in combination. One clinical situation in which a combined screening strategy may be employed is when a screening mammogram identifies a woman to have dense breasts, but is otherwise negative.
e Harms include: false-positive findings, anxiety, false-positive biopsies, false-negative findings, false reassurance, overdiagnosis and resulting overtreatment, and radiation exposure.
f Newer technologies include: MRI, hand-held ultrasonography, and whole breast ultrasonography.
g Intermediate breast cancer outcomes include: interval breast cancer incidence, DCIS diagnosis rate, stage at breast cancer diagnosis, and invasive breast cancer recurrence rate.

Contextual questions will not be systematically reviewed, but will be evaluated using a “best evidence” approach.

  1. What are the rates of specific adverse effects of current treatment regimens for invasive breast cancer and DCIS in the United States?
  2. What are the absolute incidence rates of DCIS and localized and advanced invasive breast cancer in screened and nonscreened populations in the United States?
  3. How do women weigh the harms and benefits of screening mammography and how do they use this information in their decisions to undergo screening?

The Research Approach identifies the study characteristics and criteria that the Evidence-based Practice Center will use to search for publications and to determine whether identified studies should be included or excluded from the Evidence Review. Criteria are overarching as well specific to each of the key questions (KQs).

Category Include Exclude
Population KQs 1–7: Women age ≥40 years.

KQs 8–10: Women age ≥40 years found to have dense breasts via mammography, but negative for abnormalities suspicious for cancer.

Men, women age <40 years, women with pre-existing breast cancer; women with clinically significant BRCA gene mutations, Li-Fraumeni syndrome, Cowden syndrome, hereditary diffuse gastric cancer, or other familial breast cancer syndromes; women with high-risk breast lesions (ductal or lobular carcinoma in situ, atypical ductal or lobular hyperplasia); or women with previous large doses of chest radiation therapy (≥20 Gy) before age 30 years.
Intervention KQs 1, 2, 5: Screening mammography, all methods (i.e., film, digital, tomosynthesis).

KQs 3, 4, 6: Screening mammography (all methods) combined with other modality; other screening modality (i.e., MRI, ultrasonography).

KQs 7, 8, 10: Breast tomosynthesis, breast MRI, hand-held ultrasonography, whole breast ultrasonography.

KQ 9: BI-RADS.

KQs 1, 2, 5: Mammography for diagnosis or surveillance.

KQs 3, 4, 6: Breast imaging or examinations for diagnosis or surveillance.

KQs 7, 8, 10: Mammography and use of other technologies for diagnosis or surveillance.

KQ 9: Other systems for measuring breast density.

Comparisons KQs 1, 2, 5: Mammography in women ages 40–49 vs. 50–59 vs. 60–69 vs. 70–79 years (or other age comparisons); annual mammography vs. biennial vs. triennial vs. alternate screening intervals vs. none; presence of risk factor vs. none (e.g., family history, extremely dense breasts).

KQs 3, 4, 6: Mammography, all types (i.e., film, digital, tomosynthesis) vs. other screening modality vs. mammography (all types) combined with other modality, including MRI and ultrasonography; screening interval and age differences by screening modality.

KQs 1, 2, 5: Data not provided by age, screening interval, or risk factor.

KQs 3, 4, 6: Data not provided by screening modality, age, or screening interval.

KQs 7, 8: Studies lacking comprehensively applied reference standard.

Outcomes Benefits

KQs 1, 3: Reduced breast cancer mortality and all-cause mortality.

KQs 2, 4: Reduced incidence of advanced breast cancer and treatment-related morbidity (i.e., physical adverse effects of treatment, quality of life measures, other measures of impairment).

KQ 10: Interval breast cancer incidence, DCIS diagnosis rate, stage at breast cancer diagnosis, invasive breast cancer recurrence rate.

Harms

KQs 5, 6: False-positive findings; anxiety; adverse effects on quality of life; false-positive biopsies; false-negative findings; false reassurance; overdiagnosis; overtreatment; radiation exposure.

Test Characteristics

KQs 7, 8: Sensitivity, specificity, positive predictive value, and negative predictive value for: invasive breast cancer, breast lesions (DCIS), total breast cancer, and breast cancer by stage.

KQ 9: BI-RADS determination reliability and consistency.

Outcomes not listed as included.
Timing Immediate, short-term, and long-term outcomes; duration of followup. No followup.
Setting Settings and populations of women applicable to U.S. primary care practice. Settings not applicable to U.S. primary care practice.
Study Design Effectiveness

KQs 1–4, 10: RCTs, prospective and retrospective cohort studies, meta-analyses.

Harms

KQs 5, 6: RCTs, prospective and retrospective cohort studies, case-control studies, cross-sectional studies, systematic reviews, meta-analyses, and modeling studies; others considered.

Test Characteristics

KQs 7, 8: Diagnostic accuracy studies with reference standard, meta-analyses.

KQ 9: Well-designed studies of intrarater and interrater reliability and accuracy of BI-RADS designations of breast density, as well as reproducibility of measures within women and stability of measures over time (e.g., training sets, diagnostic accuracy studies with reference standard).

Case reports, case series; studies outside of search dates unless updates of previous trials.
Language English-language abstracts (includes English-language abstracts of non–English-language papers) and papers. Non–English-language papers.
Contextual Question 1 U.S. rates of specific adverse effects of current treatment regimens for invasive breast cancer and DCIS from published sources and databases, obtained using a best evidence approach. Non-U.S. rates, older regimens (see search dates).
Contextual Question 2 Absolute incidence rates of DCIS and localized and advanced invasive breast cancer in screened and nonscreened populations in the United States from published sources and databases, obtained using a best evidence approach. Non-U.S. rates, older estimates (see search dates).
Contextual Question 3 Descriptive papers of how women's perceptions of the benefits and harms of breast cancer screening affect their clinical decisionmaking regarding breast cancer screening in the United States. Studies of women in other countries; older studies (see search dates).
Data Sources Ovid MEDLINE, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, Breast Cancer Surveillance Consortium database.

KQs 7–10 will additionally include grey literature and other supplemental sources.

Sources not listed as included.
Search Dates* Effectiveness KQs included in the previous report (KQs 1, 3): RCTs published between 2008 and February 2014 and updates of earlier trials; cohort studies published between 1996 and February 2014.

Effectiveness KQs not included in the previous report (KQs 2, 4): RCTs and cohort studies published between 1996 and February 2014.

Harms (KQs 5, 6): Studies published between 2008 and February 2014 and updates of earlier studies.

KQs 7–10: Studies published between January 2000 and July 2014.

Contextual questions: Studies published between 2010 and February 2014.

Studies published outside of the specified search dates that were not included in previous USPSTF systematic reviews.

* Search dates vary because some KQs (KQs 1, 3, 5, 6) were included in the previous systematic review and require only an update of studies published since the previous search in 2008 (the USPSTF will continue to consider and use evidence that was included in prior systematic reviews). Other KQs were not addressed by the previous review and require a search that covers a longer time period (KQs 2 and 4 and cohort studies for KQs 1 and 3). These searches extend to 1996 because this is the last time the USPSTF evaluated similar data and represents a period when practice was shifting to digital mammography in the United States. The contextual questions have a shorter time period for searches because they require current data.
Abbreviations: BI-RADS=Breast Imaging Reporting and Data System; DCIS=ductal carcinoma in situ; MRI=magnetic resonance imaging; RCT=randomized, controlled trial; USPSTF=U.S. Preventive Services Task Force.

The draft Research Plan was posted for public comment on the USPSTF Web site from November 14 to December 11, 2013. In response, the Task Force clarified the populations of women that are included in this review, as well as definitions related to risk. The Task Force also clarified that although the current review's search dates for trials of breast cancer screening effectiveness cover the years 2008 through 2014, it will consider evidence from older trials included in prior systematic reviews. That is, the current evidence review will primarily search for new studies or updates of previous trials, but the USPSTF, in making its recommendations, will consider the totality of evidence available, not just studies published since 2008. Contextual Question 3 (examining the woman's perspective when considering breast cancer screening) was refined to provide both descriptive data as well as related information about decision outcomes. The Research Approach was expanded to include information on the test characteristics of newer screening modalities, either in a primary screening capacity (such as breast tomosynthesis) or as a screening strategy in women who have a negative mammogram but are found to have dense breasts. To further explore evidence on the issue of clinical management of women with dense breasts, the Task Force added a question about the reliability and reproducibility of BI-RADS. It will also evaluate the available evidence on the effectiveness of screening in women with negative mammograms but dense breasts on additional intermediate outcomes, such as interval breast cancer incidence rate and rate of invasive breast cancer recurrence.

The USPSTF has commissioned decision models to supplement the systematic review of the evidence on breast cancer screening, as it did for the previous topic update in 2009. Decision models are mathematical simulations that project the population-level health outcomes that result from alternative interventions for prevention, diagnosis, and treatment. For example, a decision model can describe the expected population health outcomes of a clinical preventive service over a lifetime horizon rather than the more limited range of years generally captured by clinical trials. Inputs into the decision models will include data from the systematic evidence review, which will provide the best available information on the linkages between screening and outcomes of interest. In conjunction with the systematic review of the evidence, the decision model results will help the Task Force examine how the benefits and harms of breast cancer screening strategies may vary by different starting and stopping ages and the time interval between screening tests. The model will also investigate how a woman's personal factors may affect the benefits and harms of different breast cancer screening approaches.

For more information about how the USPSTF uses decision models in formulating its recommendations, please refer to the USPSTF Procedure Manual, available at http://www.uspreventiveservicestaskforce.org/Page/Name/procedure-manual.