Elodia Cole, MS
New advances in breast cancer research are increasingly emerging at the intersection of imaging, biology and clinical outcomes. Multimodal datasets, such as the Tomosynthesis Mammographic Imaging Screening Trial (TMIST), will enable investigators to integrate mammographic imaging with pathology, tumor characteristics, biospecimens and long-term follow-up. This endeavor will present an unprecedented resource for developing and validating next-generation AI models, identifying novel biomarkers and uncovering new insights into breast cancer risk and disease progression.
TMIST is a prospective clinical trial led by ECOG-ACRIN Cancer Research Group open to sites participating in the National Cancer Institute National Clinical Trials Network. This trial compares the impact of digital breast tomosynthesis and digital mammography on the number of advanced breast cancers detected in the population of women who undergo annual or biennial routine screening over several years.
The 10-year study, which began in 2017, reached its enrollment goal of 108,508 women in December 2024, with endpoint events continuing to accrue through December 2027.
Asymptomatic women aged 45–74 were randomized to undergo either digital breast tomosynthesis or digital mammography at an assigned annual or biennial screening frequency based on risk factors for their first five years of the study with up to an additional three years of long-term follow-up (maximum of eight years).
A total of 134 participating sites in the following countries participated: The United States, Canada, Argentina, Chile, Peru, Spain, Italy, Thailand, Taiwan and South Korea.
TMIST includes a diverse population of participants: 21% of U.S. enrollments are African American and 40% of the global enrollment is Hispanic.
Beyond its primary objective of evaluating screening impact on advanced breast cancer incidence, TMIST has generated an extensive repository of imaging, clinical, pathology and biospecimen data that will support breast cancer research long after the trial's primary analyses are complete.
At the heart of TMIST is one of the largest prospective mammography imaging datasets ever assembled. The image archive includes digital mammograms and digital breast tomosynthesis examinations in Digital Imaging and Communications in Medicine (DICOM) format.
In the tomosynthesis arm, researchers collected projection data from the craniocaudal and mediolateral oblique views, along 2D mammograms acquired in the “Combo“ or “2D-3D” mode.
The dataset can support a wide range of investigations, including AI development and validation, imaging biomarker discovery, reader performance studies, risk prediction and screening effectiveness research.
By combining large-scale imaging data with robust clinical annotation, TMIST offers a powerful resource for advancing breast cancer detection and improving screening strategies.
TMIST uniquely combines breast imaging data with detailed clinical information and long-term clinical outcomes, allowing researchers to examine not only what was detected on a screening mammogram, but also what happened next.
Available data include participant demographics, breast cancer risk factors, screening and diagnostic imaging follow-up, cancer diagnoses, cancer treatment and long-term follow up.
The TMIST dataset allows researchers to connect screening findings with detailed pathology and tumor biology data. For women who underwent a breast biopsy, researchers collected pathology materials, including digitized diagnostic slides and formalin-fixed, paraffin-embedded tissue blocks. All cancerous lesions, intermediate benign lesions and a subset of other benign lesions underwent genetic analysis.
By linking imaging features to underlying tumor biology, investigators can explore why certain cancers are more readily detected than others, identify imaging patterns associated with aggressive disease and evaluate differences in screening performance across tumor subtypes.
The collection of blood and buccal biospecimens create opportunities for translational research that can deepen our understanding of breast cancer biology. Nearly two-thirds of women consented to provide blood samples (65%) and buccal samples (67%). These specimens enabled investigators to explore genetic, genomic, molecular and biomarker-based factors that may influence cancer development, detection and outcomes.
Due to the breadth and depth of the TMIST trial, the healthcare field is well-positioned to pursue new research opportunities across multiple disciplines:
By bringing together imaging, clinical outcomes, pathology data and biospecimens within a single research platform, TMIST can help investigators address questions that would be difficult or impossible to study using isolated datasets. In fact, many of the trial’s most impactful applications and discoveries may be those that extend beyond the trial's original objectives.
The true value of TMIST will be measured not only by the answers it provides to its planned primary and secondary study objectives, but also by the discoveries it enables for years to come. Built through the participation of tens of thousands of women and the efforts of hundreds of investigators, research staff and institutions across the world, TMIST represents a substantial investment capable of supporting future breast cancer research.
By making its imaging, clinical, pathology, tumor biology and biospecimen resources available to qualified investigators, the study will create opportunities for new collaborations and support innovative secondary studies across disciplines. Whether the goal is to develop AI tools, identify novel biomarkers, improve risk prediction or refine screening strategies, TMIST offers a powerful platform for scientific exploration.
RADICAL: Advancing Imaging Research Without Moving the Data
RADICAL demonstrates how federated research enables multisite imaging studies, AI analysis and data collaboration without moving clinical data.
Read moreTurn Your Ideas Into Practice-Changing Research
The ACR Center for Research and Innovation™️ helps researchers move from concept to publication with expert support and strategic partnerships.
Read moreMed Student Guide to Virtual Research
Medical students without a home radiology program can find virtual research mentors, build projects and grow careers through ACR resources.
Read more