Digital Tomosynthesis Mammography and Digital Mammography in Screening Patients for Breast Cancer

PRIMARY OBJECTIVES:

I. To compare the proportions of participants in the tomosynthesis mammography (TM) and digital mammography (DM) study arms experiencing the occurrence of an ?advanced? breast cancer at any time during a period of 4.5 years from randomization, including the period of active screening and a period of clinical follow-up after the last screen (T4).

SECONDARY OBJECTIVES:

I. To assess the potential effect of age, menopausal and hormonal status, breast density, and family cancer history on the primary endpoint difference between the two arms.

II. To compare the diagnostic performance of TM and DM, as measured by the area under the receiver operating characteristic (ROC) curve (AUC), sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV).

III. To compare the recall rates and biopsy rates for TM versus DM, with subset analyses by the same variables as listed in aim II.

IV. To compare the rate of interval cancers for TM and DM and to assess the mechanism of diagnosis for these interval cancers with categorization by symptomatic versus (vs) asymptomatic, and how detected: diagnosed via physical examination, mammography, ultrasound (US), magnetic resonance imaging (MRI) or other technologies.

V. To examine the correlation between Breast Imaging Reporting and Data System (BIRADS) imaging features and histologic and genetic features, such as invasive ductal and invasive lobular histology, high grade, high stage at diagnosis, and aggressive genetic subtypes.

VI. To assess different combinations of TM and synthesized 2 dimensional (2D) or DM in reader studies to assist in determining the optimum balance between diagnostic performance, radiation exposure and technique.

VII. To estimate and compare breast-cancer-specific mortality between the two study arms.

VIII. To estimate and compare the prevalence of breast cancer subtypes (luminal A, luminal B, HER2+, basal-like) low, medium or high proliferation via PAM50 proliferation signatures, and p53 mutant-like or wild-type-like according to a validated p53 dependent signature in the two arms, overall and stratified on whether cancers were detected through screening or as interval cancers, and whether cancers were invasive or in situ.

IX. To classify histologically malignant (true positive cases) and benign lesions (false positive cases) as normal-like or tumor-like using the PAM50 gene expression assay subtype (luminal A, luminal B, HER2, basal-like,), and low, medium, or high proliferation according to PAM50 proliferation signatures, and p53 mutant-like or wild-type-like according to a validated p53-dependent signature.

X. To assess the agreement between local and expert study pathologists for all breast lesions (benign and malignant) biopsied during the 4.5 years of screening with TM or DM.

XI. To create a blood and buccal cell biobank for future biomarker and genetic testing.

XII. To compare health care utilization (including cancer care received) and cost of an episode of breast cancer screening by TM versus DM, overall and within subsets.

XIII. To implement a centralized quality control (QC) monitoring program for both 2D digital mammography (DM) and tomosynthesis (TM), which provides rapid feedback on image quality, using quantitative tools, taking advantage of the automated analysis of digital images.

XIV. To assess temporal and site-to site variations in image quality, breast radiation dose, and other quality control parameters in TM vs. DM.

XV. To refine and implement task-based measures of image quality to assess the effects of technical parameters, including machine type, and detector spatial and contrast resolution on measures of diagnostic accuracy for TM.

XVI. To evaluate which QC tests are useful for determination of image quality and those that are predictive of device failure, in order to recommend an optimal QC program for TM.

OUTLINE: Patients are randomized to 1 of 2 arms.

ARM A: Patients undergo bilateral screening DM with standard craniocaudal (CC) and mediolateral oblique (MLO) views at baseline, 12, 24, 36, and 48 months if pre-menopausal or at baseline, 24, and 48 months if post-menopausal.

ARM B: Patients undergo manufacturer-defined screening TM at baseline, 12, 24, 36, and 48 months if pre-menopausal or at baseline, 24, and 48 months if post-menopausal.

After completion of study, patients are followed up for at least 3- 8 years after study entry.

• Digital Mammography (Procedure), also known as Full Field Digital Mammography, FFDM: Undergo DM
• Digital Tomosynthesis Mammography (Procedure), also known as DBT, Digital Breast Tomosynthesis, Digital Tomosynthesis of the Breast: Undergo TM
• Laboratory Biomarker Analysis (Other), also known as Biomarker analysis, genetic analysis, PAM50: Correlative studies

Primary outcomes

• Proportion of women diagnosed with an advanced breast cancer at any time during a period of 4.5 years from randomization, including the period of active screening and a period of follow up after the last screen [4.5 years after last registration]: The cumulative proportions of participants experiencing the primary endpoint in the two study arms will be compared. The primary comparison of the two study arms will be approached from an Intent-to-Treat perspective and will be based on a two-sided test for comparing binomial proportions.

Secondary outcomes

• Agreement between local and expert study pathologists for all breast lesions (benign and malignant) biopsied during the five years of screening [Up to 8 years]: Measures of agreement such as kappa statistics and concordance coefficients to assess the agreement of local and central pathology readings. In addition, the variability among local pathologists will be examined with respect to the degree of agreement with the central study interpretation. This analysis will utilize mixed models with random effects for local pathologists. There will be up to two central study independent pathologist interpretations for each representative diagnostic slide set.
• Breast Imaging-Reporting and Data System (BIRADS) imaging features [Up to 8 years]: The correlation between BIRADS imaging features and histologic and genetic features, such as invasive ductal and invasive lobular histology, high grade, high stage at diagnosis, and aggressive genetic subtypes will be examined. Using data on patients with cancer, estimates of the correlation between the two sets of features (BIRADS imaging features and histologic/genetic features) will be derived. Cluster analysis will be used to identify clusters of patients based on imaging features and will examine the association of these clusters with histology and genetic features. Using data from the fu
• Breast-cancer-specific mortality [Up to 8 years]: Breast-cancer-specific mortality between the two study arms will be estimated and compared. Information on cancer recurrence and mortality will be obtained for a period of at least 4.5-8 years on all study participants. Mortality rates will be estimated as the ratio of the number of breast cancer deaths during a time period to the number of person-years at risk. Person-years will be measured as time from randomization to breast cancer death or censoring. Cumulative mortality rates from breast cancer at the end of the study period in each arm will be compared via the relative risk (rate ratio).
• Centralized quality control (QC) monitoring program implementation [Up to 8 years]: Centralized QC monitoring program for both DM and TM, which provides rapid feedback on image quality, using quantitative tools, taking advantage of the automated analysis of digital images. The QC program will provide an auditable trail of QC activities and image quality parameters, while at the same time reducing QC effort required by the technologist at the site. Constant monitoring of data from all sites will occur, and Root Cause Analysis will be performed for non-compliant items. The remote monitoring system will be evaluated in terms of its percent ?up-time?, technologist compliance (vi
• Diagnostic and predictive performance of tomosynthesis mammography (TM) and digital mammography (DM) [AUC] [Up to 8 years]: ROC analysis will be performed to compare the performance characteristics of DM vs TM at each screening visit
• Assess the predictive performance of tomosynthesis mammography (TM) and digital mammography (DM) [Up to 8 years]: Compare the predictive characteristics (PPV,NPV,Sens, and Spec) of DM vs TM at each screening visit
• Health care costs (including diagnostic procedures and cancer care received) as the result of an episode of breast cancer screening by tomosynthesis mammography (TM) versus digital mammography (DM) [Up to 8 years]: Rates of utilization of key diagnostic procedures (e.g. extra TM or DM views, Ultrasound, Short-term interval follow-up, surgical consultation, percutaneous biopsy with, needle-localized open surgical biopsy, breast MRI) will be estimated; Medicare reimbursement costs will be used to derive a standardized measure of cost per participant; and these costs will be compared across the two study arms. These measures of cost will be compared across study arms using non-parametric methods.
• Health care utilization (including cancer care received) of an episode of breast cancer screening by tomosynthesis mammography (TM) versus digital mammography (DM) [Up to 8 years]: Rates of utilization of key diagnostic procedures (e.g. extra TM or DM views, Ultrasound, Short-term interval follow-up, surgical consultation, percutaneous biopsy with, needle-localized open surgical biopsy, breast MRI) will be estimated and compared across the two study arms. The comparisons will be made using regression modeling
• Histologically malignant (true positive cases) and benign lesions (false positive cases) [Up to 8 years]: Classification of histologically benign-appearing lesions (false positive cases) will be explored as normal-like or tumor-like using the PAM50 gene expression assay subtype and low, medium, or high proliferation according to a PAM50 proliferation signature, and p53 mutant-like or wild-type-like according to a validated p53-dependent signature, and according to histological features. The benign-appearing (false positive) biopsies will be characterized using digital histologic analysis tools that capture percent of area represented by stroma, epithelium, and fat as well as the density of nuclei
• Prevalence of breast cancer subtypes (luminal A, luminal B, HER2+, basal-like) and p53 signature in the two arms [Up to 8 years]: Prevalence of breast cancer subtypes (luminal A, luminal B, HER2+, basal-like) and p53 signature in the two arms will be estimated and compared, overall and stratified on whether cancers were detected through screening or as interval cancers. subtypes in each arm and to compare them across arms. The analysis will be performed overall, and stratified by screen detected or interval detected. Estimates of the prevalence of subtypes and confidence intervals will be developed for each screening round and for the full period of screening. The comparison of rates across arms will be based on multinom
• Proportion of women diagnosed with an ?advanced? breast cancer in the two arms [Up to 8 years]: The potential effect of age, menopausal and hormonal status, breast density, and family cancer history will be assessed on the primary endpoint difference between the two arms. Regression modeling will be used to assess the effect of age, menopausal and hormonal status, breast density, and family cancer history. Multiple imputation will be used to handle missing data in the response and the covariates, and a sensitivity analysis to assumptions about the missing data will be performed. An exploratory analysis using alternative definitions of the primary endpoint will also be conducted in order
• Quality control (QC) tests useful for determination of image quality and those that are predictive of device failure [Up to 8 years]: QC tests that are useful for determination of image quality and those that are predictive of device failure will be evaluated, in order to recommend an optimal QC program for TM. Tests that are most sensitive to changes in system performance will be established and tests that are inferior and/or redundant and can be eliminated. Changes will be tracked against site records of alterations or repairs to the system, recalibration and changes in imaging parameters. Changes in test results will be observed and if they are suggestive that remedial action is required, we will determine after such acti
• Rate of interval cancers [Up to 8 years]: The rate of interval cancers for TM and DM will be compared and the mechanism of diagnosis for these interval cancers will be assessed with categorization by symptomatic vs asymptomatic, and how detected: diagnosed via physical examination, mammography, ultrasound (US), magnetic resonance imaging (MRI) or other technologies. Interval cancers are those that occur between screening examinations. Interval cancer rates for each screening occasion and over the full set of screens will be estimated using Wilson intervals and compared across arms using two-sided tests for binomial proportions. The di
• Recall rates [Up to 8 years]: The recall rates for TM versus (vs) DM will be compared. Recall rates are defined as the number of screening examinations that are interpreted as BIRADS 0, 3, 4 and 5 divided by the total number of screening examinations. Recall rates for each screening occasion and over the full set of screens will be estimated using Wilson intervals and compared across arms using two-sided tests for binomial proportions. Logistic regression will be used to analyze potential differences across patient subsets.
• Biopsy rates [Up to 8 years]: The biopsy rates for TM versus (vs) DM will be compared. biopsy rates are defined as the number of biopsies divided by the total number of screening examinations. rates for each screening occasion and over the full set of screens will be estimated using Wilson intervals and compared across arms using two-sided tests for binomial proportions. Logistic regression will be used to analyze potential differences across patient subsets
• Task-based measure of image quality [Up to 8 years]: Task-based measures of image quality will be refined and implemented to assess the effects of technical parameters, including machine type, and detector spatial and contrast resolution on diagnostic accuracy for TM. the diagnostic accuracy of the resulting Task-based analysis, using mathematical observers, will be assessed from image information using techniques based on signal and noise transfer.
• Variability of quality control parameters [Up to 8 years]: The variability of standard quality control parameters will be assessed and compared temporally, within, and across sites for both DM and TM.

Sex: Female, ages: 45 Years to 74 Years, accepts healthy volunteers

Inclusion Criteria:

* Women of childbearing potential must not be known to be pregnant or lactating
* Patients must be scheduled for, or have intent to schedule, a screening mammogram
* Patients must be able to tolerate digital breast tomosynthesis and full-field digital mammographic imaging required by protocol.
* Patients must be willing and able to provide a written informed consent
* Patients must not have symptoms or signs of benign or malignant breast disease (eg, nipple discharge, breast lump) warranting a diagnostic rather than a screening mammogram, and/or other imaging studies (eg, sonogram); patients with breast pain are eligible as long as other criteria are met
* Patients must not have had a screening mammogram within the last 11 months prior to date of randomization
* Patients must not have previous personal history of breast cancer including ductal carcinoma in situ
* Patients must not have breast enhancements (e.g., implants or injections)
* ANNUAL SCREENING REGIMEN ELIGIBILITY CHECK
* To be eligible for inclusion in the annual screening regimen one of the following three conditions must be met in addition to the eligibility criteria above:

  * Patients are pre-menopausal; OR
  * Post-menopausal aged 45-69 with any of the following three risks factors:

    * Dense breasts (BIRADS density categories c-heterogeneously dense or d-extremely dense), or
    * Family history of breast cancer (first degree relative with breast cancer), or, positive genetic testing for any deleterious genes that indicate an increased risk for breast cancer, or
    * Currently on hormone therapy; OR
  * Post-menopausal ages 70-74 with either of the following two risk factors:

    * Dense breasts (BIRADS density categories c-heterogeneously dense or d-extremely dense), or
    * Currently on hormone therapy
* Postmenopausal women are defined as those with their last menstrual period more than 12 months prior to study entry; for the purpose of defining menopausal status for women who have had surgical cessation of their periods, women who no longer have menses due to hysterectomy and oophorectomy will be considered postmenopausal; women who no longer have menses due to hysterectomy without oophorectomy will be considered premenopausal until age 52 and postmenopausal thereafter
* All other postmenopausal women are eligible for inclusion in the biennial screening regimen
* For those women who cannot be assigned to annual or biennial screening at the time of study entry and randomization because they are postmenopausal, have no family history or known deleterious breast cancer mutation, are not on hormone therapy AND have not had a prior mammogram, breast density will be determined by the radiologist?s recording of it at the time of interpretation of the first study screening examination, either DM or TM; for those who are randomized to TM, radiologists will assign BI-RADS density through review of the DM or synthetic DM portion of the TM examination; such women cannot be part of the planned stratification by screening frequency and are expected to represent far less than 1% of the Tomosynthesis Mammographic Imaging Screening Trial (TMIST) population
* Breast density will be determined by prior mammography reports, when available; all other risk factors used to determine patient eligibility for annual or biennial screening will be determined by subject self-report

49 more sites not listed.

• Results publication: Lee JM, Ichikawa L, Su YR, Noguchi N, Hubbard RA, Kerlikowske K, Specht JM, O'Meara ES, Henderson LM, Houssami N. Comparative Accuracy of Digital Breast Tomosynthesis and Digital Mammography for Surveillance After Breast Cancer Treatment. J Natl Compr Canc Netw. 2025 Oct 16;23(11):445-452. doi: 10.6004/jnccn.2025.7057. PMID 41101341
• Results publication: Maki AK, Mawdsley GE, Mainprize JG, Pisano ED, Shen SZ, Alonzo-Proulx O, Yaffe MJ. Quality control for digital tomosynthesis in the ECOG-ACRIN EA1151 TMIST trial. Med Phys. 2023 Dec;50(12):7441-7461. doi: 10.1002/mp.16786. Epub 2023 Oct 13. PMID 37830895

• Etta Pisano, Principal investigator, ECOG-ACRIN Cancer Research Group

NIH grants: U10CA180820, U10CA180794