Histological subtype and receptor characteristics of cancers detected or missed by digital mammography and tomosynthesis: Results from MAITA randomized trials.

BACKGROUND: Screening with digital breast tomosynthesis (DBT) detects more cancers than with digital mammography (DM), but has small effect, if any, on interval cancer incidence.

METHODS: Data were pooled from two randomized trials within the MAITA Consortium, including 100743 women (664 cancers) screened at baseline and 82938 women (550 cancers) followed for cumulative incidence at subsequent round. At baseline, women were randomized to DBT+DM or DM alone; subsequent round used DM only. We compare the histological characteristics of cancers detected in women screened with DBT+DM versus DM alone, both at recruitment and after a negative screening result. No statistical tests are applied, except for interval cancers and cumulative incidence.

FINDINGS: At the first round, the detection rate was 50% higher in the DBT arm compared to the DM arm. The overall incidence of interval cancers was similar in the two arms (IRR 0.95 (0.69-1.32)). At the subsequent round, the detection was 14% lower in the DBT arm, but an overall residual 16% excess of cancers remained (IRR 1.16 (1.04-1.30)). The initial increase in detection with DBT was consistent across most tumor types, except for large tumors (≥20 mm), grade 3, HER2-positive and triple-negative cancers. The reduction in detection after the second round was appreciable for large tumors and node-positive disease, but not for aggressive subtypes (hormone receptors (estrogen- and progesterone)-negative, HER2-positive, Ki-67 positive, triple-negative).

INTERPRETATION: DBT likely detects tumours that would have presented as larger tumours in later screening rounds, particularly those with intrinsically favorable prognostic features.