Comparison of quadrant-specific breast cancer incidence trends in the United States and England between 1975 and 2013.

BACKGROUND

UK breast cancer incidence rates suggest that upper outer quadrant (UOQ) cancers have risen disproportionately compared with other areas over time. We aimed to provide a comparison of the trend in quadrant-specific breast cancer incidence between the United States (US) and England, and determine whether a disproportionate UOQ increase is present.

METHODS

Surveillance Epidemiology and End Results (SEER) cancer registry data were obtained on 630,007 female breast cancers from 1975 to 2013. English cancer registry data were obtained on 1,121,134 female breast cancers from 1979 to 2013. Temporal incidence changes were analysed using negative binomial regression. Interaction terms determined whether incidence changes were similar between sites.

RESULTS

English breast cancer incidence in the UOQ rose significantly from 13% to 28% from 1979 to 2013 whereas no significant increase was observed among SEER data. The significant interaction between quadrant and year of diagnosis (p<0.001) in both SEER and English data indicates that breast cancer incidence in each quadrant changed at a different rate. Incidence in the UOQ rose disproportionately compared to the nipple (SEER IRR=0.81, p<0.001; England IRR=0.78, p<0.001) and axillary tail (SEER IRR=0.87, p=0.018; England IRR=0.69, p<0.001) in both SEER and England. In addition, incidence rose disproportionately in the UOQ compared to non-site-specific tumours in England (Overlapping lesions IRR=0.81, p=0.002; NOS IRR=0.78, p<0.001). The proportion of non-site-specific tumours was substantially higher in England than SEER throughout the study period (62% in England; 39% in SEER).

CONCLUSIONS

Breast cancer incidence in the UOQ increased disproportionately compared to non-site-specific tumours in England but not in SEER, likely due to the decrease in non-site-specific tumours observed in England over time. There may be real differences in incidence between the two countries, possibly due to differences in aetiology, but is much more likely to be an artefact of changing data collection methods and improvements in site coding in either country.