Tumour growth rate and invasive interval cancer characteristics in a UK breast cancer screening population.

Abstract

Objectives: To estimate tumour volume doubling time (TVDT) of interval cancers (ICs).

Methods: Two radiologists retrospectively reviewed prior screening and diagnostic mammograms and measured mean diameter on "visible" ICs. Univariate analyses of clinicopathological variables (ER, HER2, grade, age at diagnosis, and breast density) were undertaken, and those with p < 0.1 were included in a generalised linear model to estimate TVDT, cancer size at screening, and time of cancer visibility for "non-visible" tumours.

Results: From 2011 to 2018, 476 ICs were diagnosed, almost half in the third year after screening with 86% grade 2 or 3. A visible abnormality at screening was identified in 281/476 (59%) cases. Significant differences in TVDT were found with age (p < 0.02), ER status (p < 0.0001). Median TVDTs of grade 1, 2 and 3 cancers were 317, 288, and 195 days, respectively (p < 0.001). For non-visible cancers, the median estimated size at screening was 1.7 mm (IQR 1.0-2.5) for grade 1, 2.5 mm (IQR 1.5-5.9) for grade 2, and 0.9 mm (IQR 0.4-2.0) for grade 3 cancers, p < 0.001. The estimated time for cancer visibility was 489 days (IQR 229-682) after screening and 645 days (IQR 527-798) for cancers diagnosed in the third year after screening.

Conclusion: Using TVDT of retrospectively visible interval cancers, non-visible interval cancer sizes can be estimated at the time of screening. Increasing the frequency of screening from three-yearly to two-yearly invitations would reduce the number of interval cancers significantly.

Key points: Question Growth modelling of visible interval cancers (ICs) at screening helps to track the likely progression of non-visible ICs over the screening interval. Findings Tumour doubling time of visible ICs at screening is positively associated with age and ER status and inversely associated with cancer grade. Clinical relevance Interval cancer characterisation and growth modelling can be helpful to better predict the benefits of supplemental screening and the frequency of screening, given a minimum detectable size.