BACKGROUND: Despite high invasive breast cancer detection rates, the effectiveness of digital breast tomosynthesis screening is still uncertain. We aimed to compare the invasive interval cancer rates in women screened with digital breast tomosynthesis plus synthesised two-dimensional (s2D) mammography with those screened with digital mammography.
METHODS: TOSYMA is a multicentre, open-label, parallel-group, randomised, controlled, superiority trial done at 17 screening sites across two federal states in Germany. Women aged 50-70 years were randomly allocated (1:1) to digital breast tomosynthesis plus s2D mammography or digital mammography. The first primary endpoint of TOSYMA (TOSYMA-1) was the detection rate of invasive breast cancer and has been previously reported. The second primary endpoint (TOSYMA-2) was the invasive interval cancer rate, defined as the proportion of women who developed an invasive breast cancer in the 24-month interval after being negative at screening (ie, no ductal carcinoma in situ or invasive breast cancer detection) in TOSYMA-1. The primary endpoint was analysed in all participants with a negative screening result in TOSYMA-1. The cumulative incidence of invasive interval cancers at 12 months in screen-negative women was the prespecified secondary outcome. Suspected interval cancers were identified by data linkage to cancer registries. The study is registered on ClinicalTrials.gov (NCT03377036) and is closed to accrual but follow-up is ongoing.
FINDINGS: Between July 5, 2018, and Dec 30, 2020, 99 689 women were enrolled in TOSYMA-1 and 99 558 participants included in the modified full analysis set of TOSYMA-1. The median age was 57 years (IQR 53-62) in the TOSYMA-2 modified full analysis set and all participants were female. 49 299 (99·1%) of 49 762 women assigned digital breast tomosynthesis plus s2D mammography screened negative in TOSYMA-1, as were 49 456 (99·3%) of 49 796 women assigned digital mammography who had evaluable data and were included in the TOSYMA-2 modified full analysis set. Invasive interval cancers were detected in 70 of 49 299 women in the digital breast tomosynthesis plus s2D mammography group (invasive interval rate 14·2 cases per 10 000 women screened with a negative result) and in 92 of 49 456 women in the digital mammography group (18·6 cases per 10 000 women screened with a negative result; adjusted odds ratio [OR] 0·76 [95% CI 0·55-1·06]; p=0·11). Invasive interval cancers were detected within 12 months in 22 women in the digital breast tomosynthesis plus s2D mammography group (4·5 cases per 10 000 women screened with a negative result) and in 31 women in the digital mammography group (6·3 cases per 10 000 women screened with a negative result; OR 0·71 [95% CI 0·52-0·97]).
INTERPRETATION: TOSYMA-2 indicates that about a quarter of invasive interval cancers occurring with digital mammography over 2 years might be avoided with digital breast tomosynthesis plus s2D mammography screening. Although the primary outcome of TOSYMA-2 did not reach statistical significance, this large, randomised trial adds substantial evidence to the use of digital breast tomosynthesis plus s2D mammography in breast cancer screening.
FUNDING: Deutsche Forschungsgemeinschaft (German Research Foundation).
| Discipline Area | Score |
|---|---|
| Public Health |  |
| Family Medicine (FM)/General Practice (GP) |  |
| General Internal Medicine-Primary Care(US) | |
This reinforces the current standard of care in the US.
The TOSYMA randomised superiority trial — the largest to date comparing digital breast tomosynthesis plus synthesised mammography (DBT+s2D) versus digital mammography (DM) alone — reports a clinically meaningful 24% relative reduction in 24-month invasive interval breast cancer rates, with a significant 29% reduction in the first year. The benefit was concentrated in women attending subsequent (incident) screening rounds rather than first-round (prevalence) screening, and was accompanied by an ~11% gain in programme sensitivity for invasive cancers without a loss in specificity. Although the primary endpoint did not reach statistical significance, these findings substantially strengthen the evidence favouring DBT+s2D screening.