MRI screening for breast cancer: Earlier detection but more false positives
Annual breast cancer screening with magnetic resonance imaging (MRI) led to detection of fewer late-stage and node-positive cancers than annual mammography, according to the FaMRIsc* study from the Netherlands. However, specificity was lower and number of false-positives higher with MRI vs mammography.
“MRI screening can result in an important and favourable shift in tumour stage at time of breast cancer detection compared with mammography screening, reducing the incidence of late-stage cancers and thus reducing the need for adjuvant chemotherapy and the risk of mortality,” said the researchers.
“[However, with increasing breast density], MRI screening would come at the cost of lower specificity,” they cautioned.
Study participants were 1,355 women aged 30–55 years (mean age 44.7 years) with a cumulative lifetime breast cancer risk of ≥20 percent due to familial risk, BRCA 1/2 and TP53 wildtype, and no history of invasive cancer. They were randomized to undergo MRI or mammography annually (n=675 and 680, respectively) in addition to clinical breast examination. Women in the MRI group also underwent biennial mammography.
The women underwent a mean 4.3 rounds of screening each and were followed up for a median 5.2 years, during which time 55 cancer cases were identified. A separate group of 231 women formed the registration (non-randomized) group, with 218 and 13, respectively, undergoing mammography or MRI based on joint decision with their doctors.
Breast cancer was more frequently detected among women who underwent MRI than mammography (n=40 vs 15; p=0.0017). However, this significance was lost in subsequent screening rounds. [Lancet Oncol 2019;doi;10.1016/S1470-2045(19)30275-X]
There were 24 and eight cases of invasive cancer detected by MRI and mammography, respectively. Invasive cancers detected by MRI were smaller than those detected by mammography (median 9 vs 17 mm; p=0.010) and were less frequently node-positive (17 percent vs 63 percent; p=0.023). Sixty-one percent (n=14) of MRI-detected invasive cancers were detected by MRI only.
MRI detected tumours at an earlier stage of disease than mammography (48 percent vs 7 percent of cancers detected were stage T1a and T1b cancers and 6 percent vs 25 percent were stage ≥T2; p=0.035). Node-positive tumours were less frequently detected in the MRI vs mammography group (11 percent vs 63 percent; p=0.014).
One cancer in the MRI group and two in the mammography group were interval cancers, ie, cancers that were diagnosed based on symptoms between two screening rounds with a negative result obtained in the previous screening. Sensitivity did not differ significantly between MRI and mammography (97.5 percent vs 86.7 percent; p=0.18), though specificity was greater with mammography (91.0 percent vs 83.8 percent; p<0.0001).
There were 449 false-positive results in the MRI group, 22 percent of which were from a positive mammogram despite a negative MRI result.
When analysis included women in both the randomization and registration groups, the number of breast cancer cases detected increased with breast density in the mammography but not the MRI group (p=0.018 and 0.92, respectively). All stage ≥T2 and three of the five interval cancers were only diagnosed at higher breast density. MRI and mammography specificity reduced with increasing breast density (p<0.0001 and p=0.00015, respectively), while the rate of false positives increased. Breast density had no significant effect on MRI or mammography sensitivity (p=0.080 and p=0.18, respectively).
Age may play a role in false positivity incidence, given the reduced occurrence in women aged ≥50 years, said the researchers. However, the increase in false positivity with increasing breast density suggests that breast density should be accounted for, more so than age, when selecting screening method, they suggested.
Regarding the view of MRI leading to overdiagnosis of breast cancer, the emphasis should be on the danger of underdiagnosis, said Professor Christiane Kuhl from University Hospital Aachen, Aachen, Germany, who was not affiliated with the study. [Lancet Oncol 2019;doi:10.1016/S1470-2045(19)30314-6]
“Rather than assuming that most additional cancers detected in the MRI group reflect overdiagnosis until proven otherwise, we should assume that not detecting these cancers by mammographic screening leads to harm,” she said.