A comparison of two digital mammography systems: are there any differences?

To compare the performance of two newly introduced digital mammography technologies (Sectra and Hologic) to the Welsh breast-screening service; specifically, whether there are differences in the number, type, size, and grade of tumour identified.

This was a retrospective study of a prospectively collected database of 50,000 consecutive screening episodes from 2012; clients were aged 49-88 years (mean 61.9 years). All studies were double-blind read by two readers. All tumours identified in the two arms of the study were detailed and compared specifically with regards to type (ductal or lobular) size, grade, and whether invasive or non-invasive. Performance was analysed for any statistically significant differences.

Twenty-five thousand consecutive women were screened with Hologic (recall rate 5.9% of which 18% were cancer) and 25,000 were screened with Sectra digital mammography (recall rate 4.3% and 22% were cancer). Five hundred tumours were found with no significant difference in invasive cancer detection or between ductal or lobular subtypes. The Hologic system detected 267 tumours; of which 81 (30.33%) were non-invasive (3.24 per 1000), compared to the Sectra system with 233 cancers overall including 36 non-invasive (15.45%, 1.44 per 1000). The difference in non-invasive lesions (mainly ductal carcinoma in situ [DCIS]) detection was significant (p<0.001); 38% of which were high nuclear grade (HNG) using Hologic and 50% HNG lesions using Sectra. There was no significant difference in non-invasive size between the two technologies. The mean glandular dose received using the Sectra system was significantly less (0.7 mGy) compared to the Hologic system (1.6 mGy) for a 50-60 mm breast thickness.

Population breast screening is frequently criticised for identifying lesions irrelevant to long-term outcomes or life expectancy and although the two systems seem comparable in terms of invasive cancer detection, a statistically significant difference in the detection of non-invasive lesions was seen, not reported in previous studies. This is a contentious issue, as identifying more DCIS has the potential to over-diagnose screened women leading to increased morbidity, higher "cancer detection rates", longer cancer waiting times, and reduced patient psychological wellbeing. The Sectra system is able to deliver a similar invasive detection cancer delivering a much lower dose mammogram, which is important in limiting overall population radiation dose. Further study as to whether the differences in tumour detection rates are clinically significant long term are now required.