3D mammography is more precise, but does not detect more cancers

Patients and professionals have eagerly awaited a new mammography technology that allows for three-dimensional breast imaging in screening, but a recently published Norwegian trial finds that this new technology is not dramatically better than traditional mammography.

Published: May 9th 2019

New and advanced technology for breast cancer screening has been highly anticipated by professionals and patient organisations. Their hope has been that tomosynthesis will detect more of the aggressive and fatal breast cancers.

Radiologists Cecilia Sandvig and Hildegunn Aase, Haukeland University Hospital, interpreting tomosynthesis mammograms in the To-Be-trial. Photo by Haukeland University Hospital.

Research so far has shown that although tomosynthesis detects more tumours, they tend to be small and grow slowly.    

The principal investigator, Solveig Hofvind from Cancer Registry of Norway, and head of BreastScreen Norway, expects the trial’s results to raise some eyebrows.  

“We found that tomosynthesis and digital mammography detected about the same amount of breast cancers. At the same time, we had to recall fewer women screened with tomosynthesis for further examinations because of suspicious findings,” says Hofvind.

The Norwegian trial had a lower cancer detection than previous tomosynthesis studies. The study researchers believe the similar number of cancers detected of the new (3D) and the standard mammograms is due to the use of a stronger study design than previous studies. The lower than anticipated cancer detection may also be explained by the use of first-generation tomosynthesis equipment in the trial, or a lack of tomosynthesis experience in a screening setting among the radiologists at the start-up of the trial.

Hofvind says that although the results are surprising, it is safe to screen with tomosynthesis. She points out that we need further research on tomosynthesis in screening, as there are a number of areas where additional knowledge is needed.

“We still need to find out whether there are fewer cancers detected between screening rounds with tomosynthesis, and whether fewer cancers will be detected in the next screening round. Both of these outcomes can indicate whether tomosynthesis is better. We also need to compare the difference in costs between the two different techniques. We know reading tomosynthesis images is more time consuming, at least in the early implementation phase,” says Hofvind.

A group of health economists lead by professor Tron Moger from the University of Oslo are about to conduct extensive cost analyses as part of the study.


Towards more precise screening

Hofvind highlights several promising results in the trial. Among other things, the study does not support the notion that tomosynthesis detects the “wrong kind of cancers” – the small, slowly growing cancers. Fewer recalls also means less anxiety and inconvenience among women attending the screening program, as well as lower costs for the health system.

Co-author Dr. Lars A. Akslen agrees with Hofvind. Akslen is an experienced breast pathologist at the University in Bergen and Haukeland University Hospital. He says that the results from the To-Be trial indicate that the radiologists have found the “right” cancers through the use of tomosynthesis.

“The aim of mammographic screening is primarily to find the ‘dangerous’ breast cancers that affect the woman’s prognosis. When examining the tumour characteristics of the cancers in this trial, it would seem that the radiologists have been able to do just that”, Akslen says.

Image slices of a breast taken by thomosynthesis. The idea is to better expose any cancerous tumours "hiding" within glandular breast tissue. Photo by Haukeland University Hospital and Cancer Registry of Norway.

A better technique for some breasts?

Tomosynthesis is a form of 3D imaging that has been in clinical use for some years, both in Norway and internationally. The primary use of tomosynthesis in breast imaging has been as a diagnostic tool, for instance if a woman feels a suspicious lump or presents with other breast cancer symptoms.

Both patient organisations and professionals have long been eager to start screening with digital breast tomosynthesis in an organised population-based breast cancer-screening program for symptom-free women. However, there has not been enough evidence to support a major shift from digital mammography to digital breast tomosynthesis in screening.

A concern among health professionals is that tomosynthesis finds too many cancers. This can be harmful if the wrong type of (indolent) cancers are found – cancers that would never have caused any symptoms or disease, often referred to as overdiagnosis.

Professor Dr. Nehmat Houssami of the Faculty of Medicine and Health at the University of Sydney is a breast cancer screening expert, and co-author of the article in Lancet Oncology. She says that the To-Be trial’s results do not raise concerns that tomosynthesis will add to overdiagnosis.

“Still, it is crucial to pay close attention to results of the next screening round when the radiologists have more experience reading tomosynthesis images, because we may find that cancer detection will increase” she says, and adds that “tomosynthesis decreased recall in a program that already has low recall – this is good news for women”.

Houssami thinks there is definitely potential for tomosynthesis in screening and that well-designed studies (as done in BreastScreen Norway) comparing tomoysnthesis with mammography should be embedded in population screening programs because the technology performs differently in various screening settings.

“Standard digital mammography is not very accurate when screening women with ‘mammographically dense’ breasts, meaning that there is a lot of glandular tissue on the images. These women have a higher risk of breast cancer, and it is harder to find breast cancers on dense images because the tumour can ‘hide’ within the breast tissue”, she explains.

Mammograms showing varying breast densities, where A shows a large amount of fatty tissue (low mammographic breast density) and D shows a large amount of fibroglandular tissue (high mammographic breast density). Photo: Radiologyassistant

The research group is excited to investigate how tomosynthesis performs among women with dense breasts. An algorithm is currently analysing the images to estimate breast density. This is something the research group will look at next.

Principal investigator Hofvind explains that the randomised controlled trial design used in the To-Be trial is considered a gold standard in clinical research. In this trial, women participating in the screening program were randomly allocated to either screening with the standard method or the new method. This made it possible to isolate and analyse the specific effects of the new method.

A randomised controlled trial has never previously been used to compare digital breast tomosynthesis alone with digital mammography, the standard method. Hofvind explains that this, along with the high quality and complete data from the Cancer Registry of Norway, means that the results from the To-Be trial will have to be considered carefully when health authorities evaluate whether tomosynthesis should become the standard method for screening.

Tomosynthesis is time consuming

In Norway, one million screening examinations are performed every year. The time it takes a radiologist to read a screening mammogram is therefore a very important factor in an organised screening program.

“In the trial, we found that it took somewhat longer time for the radiologists to read the tomosynthesis images than the standard mammography images,” Dr. Hildegunn Aase says.

Aase is a chief physician in radiology at the breast centre at Haukeland University Hospital, as well as a PhD student in the To-Be trial.

“The time it took to read a standard mammography image was about 40 seconds, while it took about 70 seconds to read a tomosynthesis image,” she explains.

Aase is one of the radiologists at the breast centre in Bergen who has read the images in the trial. She explains that tomosynthesis shows more features and changes in the breast than standard mammography.

She underlines that the radiologists’ experience and ability to understand which of the sometimes miniscule changes in the breast indicate cancer will still be important in the future, even if using tomosynthesis.

Principal investigator Solveig Hofvind summarises by pointing out that there is still a need for more research before we can conclude whether tomosynthesis should be the standard method in breast cancer screening.

“We need to follow the women in Bergen over time to see whether this trial has caused fewer of the women to develop life-threatening cancers later in life. If that is the case, the new technology we tested may save lives, but the complete evaluation on whether or not to implement tomosynthesis as the new standard will be based on a balance between its benefits and harms”.

About the To-Be trial

Since 2015, the Cancer Registry of Norway, Haukeland University Hospital, the University of Bergen and the University of Oslo, have collaborated on a research project called the Tomosynthesis studies in Bergen (To-Be studies).

Results from this trial show that fewer women screened with tomosynthesis were recalled for further examination compared to those screened with digital mammography, but the amount of breast cancer found in these groups was similar.

Tomosynthesis is a new technique in mammographic imaging where the x-ray tube moves over the breast and captures dozens of images using low levels of radiation. These images are used to reconstruct a 2D image and a series of slices that make up a 3D image. The idea is to better expose any cancerous tumours “hiding” within glandular breast tissue.

Tomosynthesis. Illustrated by Åsne Holen, Cancer Registry of Norway.


To-Be 1: Randomised, controlled trial. Data collected during 2016 and 2017. Funding from the Research Council of Norway, the Cancer Registry of Norway and the Department of Radiology at the Haukeland University Hospital. 

To-Be 2: Prospective cohort study. Data collection during 2018 and 2019. Funding from the Norwegian Cancer Society, Cancer Registry of Norway and the Department of Radiology at the Haukeland University Hospital.