Research Presentation Session

RPS 502 - Breast cancer screening scenarios with and without tomosynthesis

Lectures

1
RPS 502 - The early effects of a mammography screening program on advanced breast cancer incidence in the Friuli Venezia Giulia Italian region

RPS 502 - The early effects of a mammography screening program on advanced breast cancer incidence in the Friuli Venezia Giulia Italian region

07:21F. Giudici, Trieste / IT

Purpose:

Breast cancer (BC) screening aims to reduce BC mortality. Effects on mortality require more time to become evident making an evaluation of its effectiveness difficult. This study assessed incidence rates of advanced breast cancer (ABC) and a surrogate measure of mortality reduction based on the attendance to Friuli-Venezia Giulia (FVG) region mammography screening programme (MSP). It has been the first to evaluate, on a regional basis, the effectiveness of MPS implemented in Italy in 2000.

Methods and materials:

Administrative databases were record-linked with the FVG cancer registry. The cohort included women aged 50-69 years invited to the first screening round. Attenders were women who responded to at least one of the first two screening rounds (2006-2007 and 2008-2009) and non-attenders who responded to neither of them. ABC was defined as tumour ≥ 2 cm (T2+), lymph-node-positive (N+), or TNM Stage greater than II (stage II+). Age-standardised incidence rate ratios (IRR) of attenders versus non-attenders, with 95% confidence intervals (95%CI), were calculated. IRRs were adjusted for age and socio-economic status by fitting a Poisson regression.

Results:

The cohort included 104,488 attenders and 49,839 non-attenders (154,327 women). During the follow-up period (median 7.02 years), 3,866 invasive BC were diagnosed (2,717 and 1,149 among attender and non-attenders respectively). Attenders had lower rates of T2+ BC (IRR=0.64; 95% CI: 0.56-0.72), N+ BC (IRR=0.87; 95% CI: 0.78-0.98), and Stage II+ BC (IRR=0.78; 95% CI: 0.70-0.87). The self-selection bias was limited.

Conclusion:

The attendance to the FVG MPS was associated to a significant decrease in the risk of advanced BC that was independent of self-selection biases. This early effect is suggestive of a future impact of the screening programme on BC mortality.

Limitations:

Short follow-up.

Ethics committee approval

n/a

Funding:

No funding was received for this work.

2
RPS 502 - Delayed breast cancer diagnosis after repeated recall at biennial screening mammography: an observational follow-up study from the Netherlands

RPS 502 - Delayed breast cancer diagnosis after repeated recall at biennial screening mammography: an observational follow-up study from the Netherlands

06:11J. Lameijer, Eindhoven / NL

Purpose:

To investigate delayed breast cancer diagnosis after repeated recall for the same screening mammographic abnormality.

Methods and materials:

This was a retroprospective observational study performed in a cohort of women enrolled in a mammography screening programme in the Netherlands. All women aged 50-75 who underwent biennial screening mammography in the south of the Netherlands between January 1, 2007, and December 31, 2016, were included. Follow-up of recalled women was at least 2 years.

Results:

A delayed diagnosis of breast cancer was seen in 2.8% of all recalled women with breast cancer. Most delays were caused by incorrect BI-RADS classifications after recall (74.2%). Tumour characteristics of breast cancers with a diagnostic delay were comparable to those of cancers without delay, except for proportions of DCIS and tumour histology. The delayed confirmation of breast cancer significantly increased the mean tumour size (P<0.001). The proportion of women who experienced a diagnostic delay in the confirmation of their breast cancer significantly varied among hospitals (P=0.027).

Conclusion:

We found that the proportion of women with a long delay of their breast cancer confirmation following recall did not decrease during 20 years of screening. These delays worsen the tumour size and may negatively influence prognosis of survival. We suggest that quality assurance not only covers the screening programmes, but also the hospitals handling recall.

Limitations:

Screening mammography programmes are constantly subject to changes. These parameters influence accuracy and may limit extrapolation of our findings to other programmes.

Ethics committee approval

Ethical approval was not required for this observational follow-up study. The screening programme requires women to ‘opt-out’ of mammography screening. Three recalled women refused and were excluded.

Funding:

No funding was received for this work.

3
RPS 502 - Predicting the long-term impact of breast tomosynthesis on the cancer detection rate in a screening programme

RPS 502 - Predicting the long-term impact of breast tomosynthesis on the cancer detection rate in a screening programme

05:16F. Caumo, Padua / IT

Purpose:

To evaluate the long-term impact on cancer detection rate (CDR) of breast tomosynthesis (DBT) used as screening test.

Methods and materials:

From April 2015 to March 2017, 34,688 women were enrolled in a screening programme where digital mammography (FFDM) was replaced by DBT combined with synthetic mammography (SM). The rescreening round was split in two: about half the population was rescreened by DBT+SM and the second half by FFDM. CDR was calculated for the first screening round dividing between prevalent and incident cancers and compared with CDRs obtained by FFDM before using DBT. CDRprevalent found in the second round by DBT+SM and by FFDM respectively were compared with values found in the first round and with those associated with former FFDM screening. CDR differences were tested by the Chi-square test.

Results:

In the first DBT round, cancer detection of prevalent and incident cancers was respectively 11.8/1000 and 8.6/1000. Both values were significantly higher than CDRs found in the former FFDM screening (CDRprevalent = 8.4/1000; CDRincident = 5.7/1000; p<0.0001). At the rescreening with DBT+SM, CDRincident decreased at 8.1/1000 but the difference with the first round was not significant; a significant decrease was found at the rescreening by FFDM, with a CDRincident=3.5. The difference between incident CDRs by FFDM before and after the introduction of DBT was also significant (5.7/1000 vs. 3.5/1000, p<0.0001).

Conclusion:

Rescreening with DBT detected more cancers than FFDM, anticipating cancers which might have been detected in the following rounds. The decrease in CDR of rescreening by FFDM after DBT compared to FFDM CDR before DBT shows that DBT also found cancers which might have shown up within two years.

Limitations:

Case-control study.

Ethics committee approval

Approved by the IRB. Patient informed consents collected.

Funding:

No funding was received for this work.

4
RPS 502 - Delayed breast cancer diagnoses in women recalled at screening mammography: trends in the proportions and lengths of delay over two decades of screening

RPS 502 - Delayed breast cancer diagnoses in women recalled at screening mammography: trends in the proportions and lengths of delay over two decades of screening

08:03L. Duijm, Nijmegen / NL

Purpose:

To determine trends in the frequency and length of delay in breast cancer diagnosis in women recalled at screening mammography.

Methods and materials:

We included a consecutive series of 817,656 screens of women who received biennial screening mammography between January 1, 1997, and December 31, 2016. During at least 2.5 year follow-up, radiological reports and biopsy reports were collected of all recalled women. The 20 year inclusion period was divided into 4 cohorts of 5 years each (1997-2001, 2002-2006, 2007-2011, 2012-2016). For each cohort, we determined the number and tumour characteristics of screen detected cancers and assessed the proportion of recalled women who experienced a diagnostic delay of at least four months in breast cancer confirmation.

Results:

The proportion of recalled women who experienced a diagnostic delay of at least four months decreased from 7.5% in 1997-2001 (47/623) to 3.0% in 2012-2016 (67/2223, P<0.001). Among all delays, the proportion of women with a delay of at least two years increased from 27.7% (13/47) in 1997-2001 to 74.6% (50/67) in 2012-2016 (P<0.001), that is of all recalled women 2.1% (13/623) and 2.5% (50/2223), respectively. Most of these were only detected and re-recalled at the following screening round. Cancers with a diagnostic delay >2 years were more frequently invasive (P=0.009) and showed a less favourable B&R grading (P<0.001) than cancers with a diagnostic delay of 4-24 months.

Conclusion:

The proportion of recalled women with a diagnostic delay in confirmation of breast cancer has more than halved during two decades of screening mammography. Unfortunately, long delays now constitute the majority of women who face a delay, which may worsen breast cancer survival.

Limitations:

n/a

Ethics committee approval

n/a

Funding:

No funding was received for this work.

5
RPS 502 - First Australian pilot trial of digital breast tomosynthesis (3D-mammography) population-based screening in BreastScreen Victoria

RPS 502 - First Australian pilot trial of digital breast tomosynthesis (3D-mammography) population-based screening in BreastScreen Victoria

06:04D. Lockie, Southbank / AU

Purpose:

To assess cancer detection rates, recall rates, and reading time, as well as the feasibility of population-based tomosynthesis screening, in BreastScreen Victoria, Australia.

Methods and materials:

Women ≥40 years presenting for routine screening (August 2017–November 2018) to Maroondah BreastScreen were prospectively recruited and received tomosynthesis (with synthesised 2D-images) or standard 2D mammography for routine screening.

Reading outcome (clear or recall), final screening outcome for 3D and 2D cohorts, and screen reading times were obtained from the client information system (CIS).

Results:

Amongst 5,018 tomosynthesis screens, 49 cancers (40 invasive, 9 in situ) were detected representing a CDR of 9.77/1,000 (95%CI 7.23–12.89/1,000). Amongst 5,166 2D-mammography screens, 34 cancers were detected (30 invasive, 4 in situ), a CDR of 6.58/1,000 (4.56–9.18/1,000). The difference in CDR was 3.18/1,000 (-0.32–6.84); the difference in CDR between screening modalities was shown for repeat screening rounds (3.47/1,000; -0.26–7.52) and was evident in women ≥ 60 years (6.96/1,000; 1.05–13.91) in stratified analyses.

Tomosynthesis screening had higher recall (4.18%; 3.65–4.78%) than mammography (3.00%; 2.55–3.50%); the difference in recall was 1.18% (0.46–1.92%).

Screen-reading time for tomosynthesis was approximately 3 times that of mammography.

Approximately 5% of women opted out of tomosynthesis screening.

Conclusion:

Tomosynthesis screening seems feasible (given adequate infrastructure and service preparation). Tomosynthesis increased breast cancer detection, recall to assessment, and screen-reading time compared to mammography.

Our findings can underpin large-scale evaluation of tomosynthesis in Australia’s BreastScreen program that examines longer-term end-points.

Limitations:

Study design was intentionally not a RCT.

Ethics committee approval

Trial Registration: ACTRN-12617000947303.

Funding:

This study was funded by a National Breast Cancer Foundation (NBCF) Australia pilot study grant.

6
RPS 502 - A survey of technical repeats and recalls in the UK Breast Screening Service

RPS 502 - A survey of technical repeats and recalls in the UK Breast Screening Service

06:01M. Hill, Issy-Les-Moulineaux / FR

Purpose:

This study surveyed reasons for repeat examinations (TR) due to clinical image quality (IQ) deficiencies within the United Kingdom National Health Service Breast Screening Programme (NHSBSP). Understanding TR is a first step towards improving IQ, the patient experience, and reducing inefficiencies.

Methods and materials:

TR exams include technical repeats (TP), where additional views are acquired during screening, and technical recalls (TC), where women are invited to a second appointment to repeat views. Study cases coded with a repeat reason were obtained from the OPTIMAM image database among women with normal four-view screening exams acquired between 2011 and 2018. Distributions of repeats according to reason and breast laterality were analysed with repeat codes grouped into inadequate positioning (IP), image blurring (IB), and other factors (OF) categories.

Results:

In total, 3,821 cases were included, with 3,846 TR split between 3,228 TP and 618 TC. Of the TR, 2,053 and 1,438 were unilateral for right or left breast repeated views, respectively. The remaining 355 TR were for the same repeat reason bilaterally. Comparing repeat classifications by proportions of TP, TC, and total TR, IP accounted for 90%, 31%, and 80%, respectively, IB represented 8%, 66%, and 17%, respectively, and OF was consistently around 3%.

Conclusion:

The majority of TR were related to IP, but among TC, IB was the predominant repeat reason. A focused effort to reduce both IP and IB at the time of the screening exam has the potential to improve IQ and substantially reduce the number of TC in the NHSBSP.

Limitations:

Potential for population and practice bias due to case collection from two screening centres, predominantly using Hologic equipment.

Ethics committee approval

Source database REC approval: 19/SC/0284.

Funding:

Volpara Solutions Ltd. and CRUK support for OPTIMAM image database.

7
RPS 502 - A multicentre, retrospective analysis of interval cancers to determine clinicopathological factors that correlate with increased growth rates

RPS 502 - A multicentre, retrospective analysis of interval cancers to determine clinicopathological factors that correlate with increased growth rates

06:03N. Sharma, Leeds / UK

Purpose:

This multicentre, retrospective review of intervals cancers aimed to identify clinicopathological factors that correlate with increased tumour volume doubling times. The potential impact of delayed diagnosis was also explored.

Methods and materials:

Data from interval cancer diagnoses with retrospective screening mammographic changes classified as either uncertain or suspicious were collated from five English breast screening units. Tumour volume doubling time was calculated based on the time from screening to diagnosis and the observed size difference. Demographic data and pathology and radiological details were noted, including mammographic density, grade, and nodal and hormone receptor status.

Results:

Data was available for 306 patients with interval cancers. 81% were diagnosed at least 12 months after screening with 19% presenting within the first year. Average time to diagnosis was 644 days (SD 276 days). Average tumour volume doubling time, based on the difference in radiological sizes and assuming volume and size proportionality, was 167 days (95% CI 151-186). Age (p=0.01), grade (p<0.001), and ER status (p<0.001) correlated with TVDTs, with women under 60, grade 3 cancers, and ER negative cancers having significantly shorter doubling times. HER2 positive tumours had shorter doubling times than the negative, but this difference was not statistically significant. There was no correlation between breast density and tumour growth rate.

Conclusion:

Cancers diagnosed in women under the age of 60, Grade 3 and hormone receptor negative cancers were found to have shorter tumour volume doubling times. BIRADS breast density did not correlate with tumour growth rates. Delay in diagnosis in this group of screening patients is likely to result in a modest increase in breast cancer mortality.

Limitations:

Small numbers.

Ethics committee approval

n/a

Funding:

No funding was received for this work.

8
RPS 502 - An analysis of screen-detected invasive cancers by grade in the English breast cancer screening programme: are we failing to detect enough small grade 3 cancers?

RPS 502 - An analysis of screen-detected invasive cancers by grade in the English breast cancer screening programme: are we failing to detect enough small grade 3 cancers?

05:57M. Wallis, Cambridge / UK

Purpose:

Detection of small high grade invasive cancers contributes disproportionately to mortality reduction from mammographic screening. We have studied the relative sensitivity of detection of invasive cancers by grade in the English breast screening programme (NHSBSP).

Methods and materials:

This study examined data from national returns from the NHSBSP for 7 years, 2009/2010 to 2015/2016. Information on size and grade of invasive cancers was collected for first/prevalent (age 45-52) and routine repeat/incident (age 53-70) screens.

Results:

Data was analysed from 11.3 million screens when 67,681 invasive cancers were diagnosed. 29% of prevalent cancers were grade 1, 52% grade 2, and 18% grade 3. At all screens, the rate of detection of small (<15mm) grade 3 cancers and large (15mm) grade 1 cancers was disproportionately low. We estimate a relative sensitivity of mammography for grade 3 detection of 52% and grade 2 detection of 62% relative to the sensitivity for grade 1 cancers.

Conclusion:

Published data shows the majority of the mortality reduction from screening is attributable to the detection of grade 3 invasive cancers, and these are over-represented in interval cancers. In our study, the sensitivity of the NHSBSP for small high grade cancers appears similar to that reported from the Swedish Two Counties study. It has not been improved by the introduction of digital mammography. Technological advances should focus on improving high-grade cancer detection.

Limitations:

Retrospective data from a 3-year programme but prevalent/first screen is independent of screening interval.

Ethics committee approval

The study has no patient contact, intervention, or use of identifiable patient data, and is therefore exempted from ethical review in the UK.

Funding:

Roger Blanks and Rupert Alison received funding from Public Health England to perform this work.

9
RPS 502 - Trends in recall and outcome of screen-detected microcalcifications during two decades of screening mammography in the Netherlands

RPS 502 - Trends in recall and outcome of screen-detected microcalcifications during two decades of screening mammography in the Netherlands

06:29L. Duijm, Nijmegen / NL

Purpose:

To determine trends in recall and outcome of screen-detected microcalcifications during 20 years of screening mammography.

Methods and materials:

We included a consecutive series of 817,656 screens obtained in a Dutch breast cancer screening region between January 1997 and January 2017. In 2009/2010 (transition period) screen-film mammography (SFM) was gradually replaced by full-field digital mammography (FFDM). We focused on the recall of suspicious microcalcifications and obtained all radiology reports and pathology outcomes of recalled women, with a 2-year follow-up. Screening outcomes during the period of SFM (1997-2010) and FFDM (2009-2016) were compared.

Results:

A total of 18,565 women were recalled, of whom 3,556 had suspicious microcalcifications. The recall rate for microcalcifications per 1,000 screens increased from 1.0 in 1997-1998 to 7.2 during the transition (p<0.001) and afterwards gradually decreased to 5.2 at the latest digital screening period (p<0.001). The recalls yielding DCIS increased from 0.3 per 1,000 screens to 1.1 per 1,000 screens, respectively. More than half of the DCIS lesions were high grade (393/747, 52.6%). The distribution in DCIS grading was stable during the 20-year screening period (p= 0.362).

Conclusion:

The recall rate for suspicious microcalcifications at mammographic screening significantly increased over the past two decades, while the DCIS detection rate increased less rapidly, resulting in a lower positive predictive value of recall. Since it is not yet clear to which degree histological features of DCIS can be estimated by the patterns of microcalcifications alone, a stereotactic core-needle biopsy is currently still mandatory in the workup of these lesions.

Limitations:

n/a

Ethics committee approval

Ethical approval was not required for the current study.

Funding:

No funding was received for this work.

PEP Subscription Required

This course is only accessible for ESR Premium Education Package subscribers.