A drop in study power
The new recommendations for TMIST come about 1 year after Medscape Medical News reported that the study was lagging in enrollment of both patients and participating sites/physicians.
Last year, two TMIST study investigators said it had been difficult enlisting sites, in part because many radiologists and facilities – informed by their experience and previous research results – already believe that the 3-D technology is superior.
Currently, most 3-D systems are used in conjunction with 2-D. First, two static images of the breast are taken (2-D), and then the unit moves in an arc, taking multiple images of the breast (3-D). Thus, 3-D is widely described as allowing clinicians to flip through the images like “pages in a book” and as offering a superior read of the breast.
The NCI working group concedes that “there is evidence that screening utilizing tomosynthesis may reduce recall rates and improve cancer detection,” but it says the trial is needed to address “questions that still remain regarding the overall benefit to patients.”
Furthermore, tomosynthesis “may carry higher out-of-pocket costs for women and is more labor intensive and costly for health care systems in that it requires about twice as much reader time for interpretation,” the working group said.
The “main hypothesis of TMIST” is that “tomosynthesis will decrease the cumulative incidence of advanced breast cancers, a surrogate for mortality, compared to standard digital mammography,” posits the group.
Advanced breast cancer is defined in TMIST as invasive breast cancers that meet any of the following criteria:
- Distant metastases
- At least one lymph node macrometastasis
- Tumor size >1 cm and triple-negative or positive for human epidermal growth factor receptor
- Tumor size ≥ 20 mm unless of pure mucinous or other favorable histologies.
In the original study design, the sample size was estimated to be sufficient to provide 90% power to detect a 20% relative reduction in the proportion of advanced cancers in the intervention arm (tomosynthesis, or 3-D) compared to the control arm (digital mammography, or 2-D) 4.5 years from randomization.
Now, with fewer patients and a revised analytic approach, the study’s statistical power will be decreased to 80% from the original 90%.
Also, an advanced cancer is counted “if it occurs at any time while the participant is on study.”
Dr. Kopans says that is a problem.
“That is a huge mistake, since digital breast tomosynthesis cannot impact prevalent cancers. They are already there. This means that their ‘power calculation’ is wrong, and they won’t have the power that they are claiming,” he said.
Dr. Kopans explained that the first screen in TMIST will have “no effect on the number of advanced cancers.” That’s because the cancers will have already grown enough to become advanced, he said.
On the other hand, an initial screening might detect and thus lead to the removal of nonadvanced, smaller cancers, which, had they not been detected and removed, would have grown to become advanced cancers by the next year. Thus, the screenings done after the first year are the ones that potentially prove the effect of the intervention.
However, the working group report says that changes to the study will not affect anything other than a 10% reduction in the study’s power.
The working group is concerned about TMIST going on for years and years. For that reason, they recommended that the NCI establish “strict criteria for termination of the study” if accrual goals are not met. However, those parameters have not been developed, and it was not part of the study group’s mission to establish them.
The working group was sponsored by the NCI. Dr. Kopans reports consulting with DART Imaging in China.
A version of this article first appeared on Medscape.com.
