publication date: May. 24, 2019
Trials & Tribulations
Why I choose to play a dual role in TMIST
By Worta McCaskill-Stevens
Division of Cancer Prevention, National Cancer Institute
As a physician-scientist, I have spent my career advancing research in women’s health. I was a part of the planning effort to design and launch TMIST—the Tomosynthesis Mammographic Imaging Screening Trial.
Today, as a woman and a person of color, I have the opportunity to contribute on a much more personal level.
In 2012, the National Cancer Institute and experts in breast cancer research, public health, clinical and advocacy communities and related disciplines began to consider designs for the first randomized breast cancer screening trial in decades.
The rationale for such a screening trial encompassed:
the introduction of new screening technology;
the lack of evidence-based information to guide women and providers about benefits and harms in the current screening environment; and,
evolving evidence with respect to screening intervals.
Moreover, maintaining equipoise and winning over trial champions among the radiology community are critical to the success of the trial.
The NCI recognized that the mantra for decades, that “increased sensitivity for smaller and smaller tumors” didn’t necessarily translate into increased benefits from breast cancer screening. Over recent decades, emerging data have deepened our knowledge about the balance of risks and harms for women who are at “average risk.”
NCI realized that a trial design which included no screening as a control arm would not be feasible, given that in the U.S., over 67% of women aged 40 and over report use of mammography within the past two years. Furthermore, as the clinical and scientific community moves toward tailoring care for individuals, breast cancer screening to fit all seems inconsistent with emerging science.
An important goal of screening is to reduce the incidence of late-stage cancers, rather than to simply increase the incidence of early-stage disease. Therefore, TMIST is designed to determine whether fewer advanced cancers are found over time through routine screening of the “average woman” with 3-dimensional mammography compared to 2-dimensional mammography.
To evaluate this research question, only a randomized trial would provide the highest level of evidence to inform practice.
The primary hypothesis is that 3-D will decrease the cumulative incidence of advanced cancers, compared to standard digital mammography. If there is no difference in the incidence of advanced cancers, there would be limited data to support the implementation of the new and more costly screening procedure.
Some studies, but not all, have shown increases in cancer detection rates and decreases in recall rates, but do not provide direct proof of benefit. Any increase in sensitivity with a new technology could be of benefit, harm, or a mixture, which only the randomized trial can sort out.
Breast cancer mortality, as the primary endpoint, would have required a sample size and study duration of impracticable magnitude, and the cost of such a trial would be immense.
Nevertheless, TMIST is designed to provide important evidence short of breast cancer mortality—the reduction in incidence of life threatening, advanced cancers. It will also provide evidence of important secondary or ancillary endpoints: false positives, recall rates, biopsy rates, and overdiagnosis.
The trial design incorporates both the assumption that there is a benefit to screening mammography, and the assumption of associated harms with screening, including false-positives and overdiagnosis. The trial fills in an important research gap with respect to the risks associated with the two technologies.
Overdiagnosis is a challenging and counterintuitive concept for clinicians and the general population. An important additional research gap is the biology of the breast cancers that are detected in the context of screening.
Moreover, the TMIST tissue biorepository can be used to identify non-life-threatening cancers (overdiagnosis at the individual level). Increased sensitivity may include an unknown amount of overdiagnosis.
Currently, this information is not available to integrate into the decision-making and guidance for screening intervals and management for the average-risk woman.
A discussion of overdiagnosis is published in the context of trial design for breast cancer screening in Trials & Tribulations (The Clinical Cancer Letter, May 10).
Studies using annual screening do not show clear differences in benefit compared to longer intervals of screening. That is why triennial screening is used in several European countries. Simulation models assessing screening intervals suggest little reduction in breast cancer mortality, but an increase in false positives, benign biopsies, and overdiagnosis.
One recent model suggests that all but 1% of women aged 50 to 65 years would benefit from biennial screening, compared to annual. A recently published Guidance Statement from the American College of Physicians provides advice on screening in average-risk women based on review of existing guidelines and evidence.
The guidance recommends the following timeline:
Ages 40-49, begin discussions about screening with benefits and harms of screening prior to 50;
Ages 50-74, offer biennial screening; and
Ages >75, discontinue with less than a 10-year life expectancy.
To assure generalizability of its fundings, TMIST is being conducted as an academic and community partnership, funded through the NCI Community Oncology Research Program (NCORP) and conducted by the ECOG/ACRIN Research Base.
The diversity of the participating oncology practices reflects the real-world setting in which mammography is conducted. Lessons learned include the the need for involving primary care physicians and advocates, particularly in providing referrals and prescriptions for mammograms.
In this regard, we conducted a detailed survey of potential mammography centers in NCORP to gauge their level of comfort in participant recruitment.
Public health messaging is critical at the local and national levels to recruit to a randomized screening trial and to appreciate the outstanding research questions in breast cancer screening. Women who are scheduled to receive routine mammograms are the intended participants.
However, since many women are not adherent to recommended screening intervals, with disparities based on socioeconomic status, language, and access to screening, the recruitment strategies are dynamic and customized locally.
NCI is working with advocacy groups, developing media communications and education materials that are also customized to the participating institutions. We are providing funding for uninsured women to participate and conducting the trial in diverse and underserved populations. For example, African American women have a higher adherence to annual screening compared to other groups, but Hispanic women have lower screening rates compared to other groups.
An important rationale for the trial is to study the new and more expensive technology before its widespread acceptance. From a public health perspective, estimated cost savings if there was no trial, compared to TMIST being better than 2-D would be $1.4 billion per year and $1.71 billion if there is no difference, based upon assumptions from the Centers for Medicare and Medicaid Services enrollment and CY2016 Medicare payments.
Early detection, screening, diagnosis, treatment, surveillance, and survivorship represent bread-and-butter issues in the cancer management spectrum.
The general population of average-risk women is diverse and in need of personalized guidance to inform practice within that spectrum. Clinical annotation of specimens with subsequent molecular characterization and the accompanying demographic information collected will move us closer to personalizing screening recommendations.
Importantly, most studies that evaluate 3-D in the literature conclude that further investigation is needed to determine the clinical outcomes in comparison to standard 2-D mammography in terms of higher cancer detection rates, characterization of the benign, premalignant and invasive cancers, and incidence of late stage disease.
By participating in TMIST, women will now have the opportunity to help fill in the gaps in knowledge about the benefits and harms from the two screening modalities.
I am one of these women.
The author is a medical oncologist and chief of the Community Oncology and Prevention Trials Research Group, which houses the NCI Community Oncology Research Program, a community-based clinical trials network launched in 2014. As NCORP director, McCaskill-Stevens oversees the program supporting community hospitals, physicians and others to participate in NCI-approved cancer treatment, prevention, screening, and control clinical trials, as well as cancer care delivery studies.