Characteristics and prognosis of interval cancers after biennial screen-film or full-field digital screening mammography

Abstract

We determined the characteristics and prognosis of interval breast cancers (IC) at screen-film (SFM) and full-field digital (FFDM) screening mammography. The study population consisted of 417,746 consecutive screening mammograms (302,699 SFM screens and 115,047 FFDM screens), obtained between 2000 and 2011. During 2-year follow-up, we collected breast imaging reports, surgical reports, and pathology results. A total of 800 ICs had been diagnosed in the screened population, of which 288 detected in the first year (early ICs) and 512 in the second year (late ICs) after a negative screen. 31.3 % of early IC's and 19.1 % of late IC's, respectively, were visible in retrospect on the latest previous screens, but had been missed during screening (P < 0.001). Missed invasive ICs were larger (28.5 mm vs. 23.9 mm, P = 0.003) and showed a higher fraction of T3+cancers (16.9 vs. 8.5 %, P = 0.02) than true ICs (i.e., not visible at the latest screen). A higher portion of missed than true ICs underwent mastectomy (44.7 vs. 30.8 %, P = 0.002). We found no differences in mammographic and tumor characteristics for early ICs, detected either after SFM or FFDM. Late ICs following FFDM were more often true ICs than missed ICs (69.0 vs. 57.6 %, P = 0.03) and more often receptor triple negative (P = 0.02), compared to late ICs at SFM. Interval cancer subgroups showed comparable overall survival. Interval cancer subgroups show distinctive mammographic and tumor characteristics but a comparable overall survival.

The epidemiology, radiology and biological characteristics of interval breast cancers in population mammography screening

“…4, 13, 15–29 The table highlights the broad variability in both interval BC rates and cancer detection rates at screening, both of which are partly driven by underlying cancer rates in the populations reported in these studies. There is wide variability in the overall interval BC rates, ranging between 7.0 and 49.3 per 10,000 screens, partly explained by data shown for screening rounds (initial and repeat screens) and the duration and year of the inter-screening intervals; where reported, data for the inter-screening interval are presented by yearly rates for biennial or triennial screening.…”

“…An earlier study by Seigneurin et al 27 reported similar evidence from the French population breast screening program, based on women aged 50–69, in a comparison of two time frames reflecting transition from one-view to two-view mammography: a reduction of 8.6 interval BCs per 10,000 screens was associated with two-view mammography, with an estimated 36% difference in relative risk of interval BC at 24 months for two vs. one view mammography (Table 1). Weber et al 13 reported a study of the Southern screening region of the Dutch program, in which the use of digital (compared to film-screen) mammography was shown to be associated with a modest but significant reduction in interval BC rates of 3/10,000 screens ( p  = 0.02; Table 1). …”

“…1 Table 2 summarises findings from the literature search on radiological surveillance including the methods used to conduct mammographic review; 10, 13–15, 18, 26, 31–40 the latter substantially influences the distribution of radiological categories and can bias estimated proportions. 1, 33, 41 For example, a pilot study examining radiological review methods showed that informed vs. blinded (uninformed) review of interval BC leads to bias in classification whereby informed reviewers (aware they were reviewing mammograms containing interval cases) more frequently classified some interval BCs as positive, compared to reviewers who were unaware they were reading mammograms of interval BCs that had been added into routine screen-reading practice (‘uniformed’ review).…”

“…Radiological surveillance entails review of the mammograms taken at the time the interval BC is identified (usually at clinical presentation, hence the diagnostic mammograms) and the pre-diagnosis mammographic screen (the ‘negative’ index screen) and an interpretative judgement to classify each case into pre-defined categories. 1, 13–15 These categories may vary in definition, however, most include a ‘true interval’ category (where the cancer is not visible at the index screen but becomes visible at the diagnostic mammogram) and a ‘missed’ interval BC being the equivalent of a false-negative (where the cancer is visible on the index mammogram but is not recalled or is misinterpreted) and is at times referred to as screening error. 1, 13–15 Various methods have been used to perform radiological surveillance, as described in a review by Houssami et al 1 with potential biases inherent in the review strategy and the extent that readers are informed that they are evaluating interval BC cases.…”

“…1, 13–15 These categories may vary in definition, however, most include a ‘true interval’ category (where the cancer is not visible at the index screen but becomes visible at the diagnostic mammogram) and a ‘missed’ interval BC being the equivalent of a false-negative (where the cancer is visible on the index mammogram but is not recalled or is misinterpreted) and is at times referred to as screening error. 1, 13–15 Various methods have been used to perform radiological surveillance, as described in a review by Houssami et al 1 with potential biases inherent in the review strategy and the extent that readers are informed that they are evaluating interval BC cases. 1 Notwithstanding the methodological limitations of radiological surveillance, it provides insights into screening quality and on how screening could be improved.…”