Eur Radiol DOI 10.1007/s00330-016-4636-4
BREAST
Digital mammography screening: sensitivity of the programme dependent on breast density Stefanie Weigel 1
&
W. Heindel 1 & J. Heidrich 2 & H.-W. Hense 2,3 & O. Heidinger 2
Received: 5 September 2016 / Revised: 23 September 2016 / Accepted: 3 October 2016 # European Society of Radiology 2016
Abstract Objectives To analyse the impact of breast density on the sensitivity of a population-based digital mammography screening programme (SP) as key evaluation parameter. Methods 25,576 examinations were prospectively stratified from ACR category 1 to 4 for increments of 25 % density during independent double reading. SP was calculated as number of screen-detected cancers divided by the sum of screen-detected plus interval cancers (24-months period) per ACR category, related to the first reading (a), second reading (b) and highest stratification if discrepant (c). Chi-square tests were used for comparison. Results Overall sensitivity of the programme was 79.9 %. SP in ACR 4 (a: 50 %, b: 50 %, c: 50 %) was significantly lower than in ACR 3 (a: 72.9 %, b: 79.4 %, c: 80.7 %, p < 0.001), ACR 2 (a: 83.9 %, b: 85.7 %, c: 83.2 %, p < 0.001) and ACR 1 (a: 100 %, b: 88.8 %, c: 100 %; p < 0.001). Frequencies of ACR 4 were a: 5.0 %, b: 4.3 %, c: 6.9 %. Conclusion Digital mammography screening with independent double reading leads to a high overall SP. In the small group of women with breast density classified as ACR 4 SP is significantly reduced compared to all other ACR categories.
Key Points • Overall sensitivity of a population-based digital mammography screening programme (SP) was 79.9 %. • In women with ACR 1, 2, or 3, SP ranged between 72.9 %100 %. • ACR 4 was rare in participants (<7 %) and SP was only 50 %. • SP in ACR 4 differed significantly from ACR 3 (p < 0.001). Keywords Breast cancer screening . Digital mammography . Breast density . Interval cancer . Population surveillance
Abbreviations ACR American College of Radiology BI-RADS Breast Imaging Reporting and Data System REC Research Ethics Committee DCIS Ductal carcinoma in situ SP Sensitivity of the screening programme
Introduction Joint first (Weigel, Heindel) and joint last authorship (Hense, Heidinger) * Stefanie Weigel
[email protected]
1
Department of Clinical Radiology and Reference Center for Mammography, University of Muenster, Albert-Schweitzer-Campus 1, Building A1, 48149 Muenster, Germany
2
Cancer Registry of North Rhine-Westphalia, Muenster, Germany
3
Institute of Epidemiology and Social Medicine, University of Muenster, Muenster, Germany
Breast density is defined by mammographic appearance. The American College of Radiology (ACR) Breast Imaging Reporting and Data System (BI-RADS) classifies breasts as almost entirely fatty, scattered areas of fibroglandular density, heterogeneously dense, or extremely dense [1–3]. Heterogeneously dense and extremely dense are usually combined as dense breasts; per definition of the 4th edition of BIRADS, they were indicated by quartile ranges of percentage dense tissue over 50 %, and, per definition of the 5th edition, as possibly obscuring small masses and lowering the
Eur Radiol
sensitivity of mammography [1, 2]. Most of the dense breasts are classified as heterogeneously dense [2, 3]. By the lexicon, the frequency of breast densities is reported as 10 % fatty, 40 % scattered, 40 % heterogeneously, and 10 % extremely dense [2]. Higher breast density is associated with decreased mammographic sensitivity and specificity and also with increased breast cancer risk [3]. Breast density may also be of relevance as a factor that contributes differentially to screen-detected breast cancers and interval breast cancers: the latter are, by definition, diagnosed in screen-negative breasts outside of the screening programme within the interval between two screening rounds. Interval cancers are larger at diagnosis, of higher nuclear malignancy grades, more aggressive subtypes displaying more often a lymph node involvement, and they are associated with a poorer prognosis compared to screendetected cancers. Recent reports suggested that mammographic density may have an adverse effect by obscuring tumours during screening [4]. The purpose of this study was to analyse the sensitivity of a population-based digital mammography screening programme taking specifically into account the impact of mammographic breast densities.
Materials and methods This is a retrospective study carried out in a single digital screening unit of the National Mammography Screening Programme. The study was approved by the Research Ethics Committee (REC). No explicit waiver of REC opinion or informed consent was required under national law. Screening process The Mammography Screening Programme adheres to the European Guidelines for Quality Assurance in Breast Cancer Screening and Diagnosis, 4th edition [5]. Accordingly, the target population includes women aged 50–69 years who are invited at two-year intervals. Mammograms were obtained in two views (cranio-caudal, medio-lateral-oblique), using two full field systems (MicroDose Mammography, L30; Sectra Medical Systems, Linköping, Sweden, now Philips; Mammomat 3000 Nova, Siemens AG, Erlangen, Germany). All mammograms were double read by two of five readers; each had to read at least 5,000 screening mammograms per year during the complete study period and each had more than 5 years of experience in breast radiology. Abnormal findings detected by one or both readers resulted in a mandatory consensus meeting of the two readers with a third; during consensus it was decided which women were recalled for further assessment procedures. The results of both screening readings as well as all further procedures were documented using a
dedicated screening software. Breast density was classified, as nationally accepted, according to the American College of Radiology 4th edition by both readers prospectively as part of the independent double reading process. That edition indicates quartile ranges of percentage dense tissue (increments of 25 % density) for each of the four density categories: ACR 1 entirely fatty, ACR 2 scattered areas of fibroglandular density, ACR 3 heterogeneously dense, or ACR 4 extremely dense [1]. Categorization was based on the assessment of all views and was given per examined women. In cases of bilateral differing breast densities, the overall categorization was based on the breast side of higher density. Data of screen-detected cancers, including invasive cancers and ductal carcinoma in situ (DCIS), as well as the ACR classifications were extracted from the screening documentation system for the years 2009 and 2010. Cancers detected after further assessment at an early recall were regarded as screen-detected cancers [5] and classified as screen-detected cancers with a delayed diagnosis. The Cancer Registry receives notifications of all cancer cases, detected within and outside the mammography screening programme, electronically and stores all personal identifiers exclusively as doubly encrypted data. The completeness of breast cancer case notification is >95 %. For the purpose of this study, data from women participating in the included screening unit were linked with data from the Cancer Registry in order to determine interval cancers. Interval cancers were defined as all invasive cancers and DCIS diagnosed in screening participants after a negative screening and before the next regular screening or within a time period equal to the screening interval of 24 months [5]. ACR classifications and reading details of confirmed interval cancers were extracted from the screening documentation system. Study group During the study period (January 1 2009 to December 31 2010), 25,579 women were screened; 32.4 % of the total examinations participated in an initial screening (recall rate: 9.7 %) and 67.6 % in a subsequent screening (recall rate: 4.5 %). The data sets of 25,576 participants contained the visual classification of breast density as described above and were included in the study. There was no pre-defined, ordered assignment of mammographies to the five readers with a fixed association to the first or second reading. For 25,378 out of 25,576 examinations (99.2 %) ACR classifications were available resulting from the first reading and for 25,352 out of 25,576 examinations (99.1 %) resulting from the second reading. In 422 examinations, no double ACR assessment was available. For study purpose, we determined the distribution of ACR classifications of screening examinations separately
Eur Radiol
for the first reading (a), the second reading (b) and the highest stratification in cases of discrepancies between ACR classification of the first and second reading (c). Statistical analysis Screening evaluation parameters were evaluated separately for the four ACR categories based on the first reading (a), the second reading (b) and the highest ACR density reading (c). We calculated the detection rates of screen-detected cancers, interval-detected cancers and their sum (that is, the 2-year breast cancer incidence rates) per 1,000 women screened. The sensitivity of the screening programme (SP) was calculated as the ratio of the number of screen-detected cancers divided by the sum of screen-detected plus interval cancers occurring over the 2-year period [5]. Chi-square tests were used for pairwise comparisons of two proportions (MedCalc Statistical Software version 15.4; MedCalc Software bvba, Ostend, Belgium; http://www.medcalc.org). Statistical significance was assumed at p < 0.05.
Results During the study period 192 screen-detected cancers and 48 interval cancers were recorded out of a total of 25,576 examinations with a least one available ACR classification. The overall sensitivity of the screening programme was 79.9 %. The frequencies of ACR 1, ACR 2, ACR 3, and ACR 4 were 9.3 %, 43.4 %, 42.3 %, and 5.0 % for the first reading (n = 25,378), 12.1 %, 43.9 %, 39.6 %, and 4.3 %, for the second reading (n = 25,352) and 6.1 %, 38.7 %, 48.3 %, and 6.9 %, respectively, for the highest classification (n = 25,576) (Table 1).
The sensitivity of the programme in ACR 4 (SP a: 50 %, b: 50 %, c: 50 %) was significantly lower (each p < 0.001) than in ACR 3 (SP a: 72.9 %, b: 79.4 %, c: 80.7 %), ACR 2 (SP a: 83.9 %, b: 85.7 %, c: 83.2 %) and ACR 1 (SP a: 100 %, b: 88.8 %, c: 100 %) (Table 1). Numbers of screen-detected cancers and interval cancers are presented in Tables 2 and 3. Screen-detected breast cancer rates per 1,000 women screened were lowest within ACR 1 classifications (a: 2.14 ‰, b: 2.61 ‰, c: 1.28 ‰) and highest within ACR 3 classifications (a: 8.46 ‰, b: 8.93 ‰, c: 8.83 ‰). Differences reached no statistical clearly significance (Table 2). The rates of interval-detected breast cancers per 1,000 women screened increased from ACR 1 to ACR 4 based on the first reading, the second reading and the highest classification. This rate was significantly higher in ACR 4 compared to ACR 1 (p = 0.001) (Table 3). The 2-year breast cancer incidence rate in screening participants (screen-detected plus interval cancers) was lowest for women in ACR 1 and rose consistently from ACR 2 to ACR 3. Significance was present for ACR 4 versus ACR 1 (a: 6.30 ‰ vs. 2.14 ‰, p = 0.09; b: 10.96 ‰ vs. 2.94 ‰, p = 0.003; c: 7.90 ‰ vs. 1.28 ‰, p = 0.01) (Table 4). We reviewed prospectively documented mammographic abnormalities of screen-detected and interval cancers of cases with a classification ACR 4. Out of seven women with screen-detected cancers, five were recalled due to microcalcifications (n = 3 pTis, n = 1 pT1a, n = 1 pT1c), one due to an architectural distortion (pT1c) and one due to a superficial density outside the extreme dense tissue (pT1a). Table 5 summarizes tumour details of all cases of interval cancers including the results of the prospective independent double reading, the abnormalities at time of diagnosis and the categorisations of interval cancers.
Table 1 Numbers of ACR classifications, sensitivities of the screening programme of women screened, separated by mammographic breast density categories resulting from first and second independent double reading Breast First Second Highest case First reading density * reading reading reading Sensitivity of the Numbers Numbers Numbers programme ** (%)
Second reading Sensitivity of the programme ** (%)
Highest case reading Sensitivity of the programme ** (%)
First reading P-value (vs. ACR 4)***
Second reading P-value (vs. ACR 4)***
Highest case reading P-value (vs. ACR 4)***
ACR 1 ACR 2 ACR 3 ACR 4 Total
88.8 85.7 79.4 50.0 79.7
100.0 83.2 80.7 50.0 79.9
<0.001 <0.001 <0.001 Reference -
<0.001 <0.001 <0.001 Reference -
<0.001 <0.001 <0.001 Reference -
2,352 11,018 10,733 1,275 25,378
3,078 11,133 10,042 1,099 25,352
1,559 9,905 12,341 1,771 25,576
100.0 83.9 72.9 50.0 79.7
*Visual determination of breast densities according to the 4th edition of the BI-RADS lexicon at time of screening. **The sensitivity of the screening programme was defined as the ratio of screen-detected cancers divided by the sum of screen-detected plus interval cancers, evaluated for an interval of two years after negative screening. Numbers of screen- and interval-detected breast malignancies are given by Tables 2 and 3. ***p-values refer to a pairwise comparison. Reference refers to the reference group ACR 4
5 89 90 4 188
8 85 89 6 188
Numbers screen-detected breast cancers with available ACR classification related to the second reading
2 ***74 109 7 192
Numbers screen-detected breast cancers with available ACR classification related to the highest case reading
2.14 8.15 8.46 3.15 7.41
Screendetected breast cancer rate per 1,000 women screened First reading
2.61 7.70 8.93 5.48 7.42
Screendetected breast cancer rate per 1,000 women screened Second reading
1.28 7.47 8.83 3.95 7.51
Screendetected breast cancer rate per 1,000 women screened Highest case reading 0.82 0.08 0.06 Reference -
First reading P-value (vs. ACR 4)**
0.27 0.53 0.32 Reference -
Second reading P-value (vs. ACR 4)**
0.25 0.14 0.04 Reference -
Highest case reading P-value (vs. ACR 4)**
18 23 6 48
1
Numbers intervaldetected breast cancers with available ACR classification related to the second reading
15 26 7 48
0
Numbers interval-detected breast cancers with available ACR classification related to the highest case reading
**p-values refer to a pairwise comparison. Reference refers to the reference group ACR 4
*Visual determination of breast densities according to the 4th edition of the BI-RADS lexicon at time of screening
9,905 12,341 1,771 25,576
17 27 4 48
0
ACR 2 ACR 3 ACR 4 Total
1,559
3,061
11,035 9,964 1,094 25,352
ACR 1 2,336
10,914 10,636 1,268 25,378
Numbers intervaldetected breast cancers with available ACR classification related to the first reading
Second Highest Breast First density reading reading case Numbers Numbers reading * Numbers
1.56 2.54 3.15 1.89
0
Intervaldetected breast cancer rate per 1,000 women screened First reading
1.63 2.31 5.48 1.89
0.33
Intervaldetected breast cancer rate per 1,000 women screened Second reading
1.51 2.11 3.95 1.88
0
Intervaldetected breast cancer rate per 1,000 women screened Highest case reading
0.35 0.91 Reference -
0.001
First reading P-value (vs. ACR 4)**
0.02 0.10 Reference -
0.001
Second reading P-value (vs. ACR 4)**
0.06 0.22 Reference -
0.001
Highest case reading P-value (vs. ACR 4)**
Table 3 Numbers of ACR classifications, numbers of interval-detected breast cancers, interval-detected cancer rates separated by mammographic breast density categories resulting from first and second independent double reading
*** One out of 192 screen-detected cancers (0.5 %) was diagnosed with a delay due to an early recall more than 12 months after the initial screening examination date, histology: invasive ductal carcinoma of 15 mm in size
**p-values refer to a pairwise comparison. Reference refers to the reference group ACR 4.
*Visual determination of breast densities according to the 4th edition of the BI-RADS lexicon at time of screening.
1,559 9,905 12,341 1,771 25,576
ACR 1 2,336 ACR 2 10,914 ACR 3 10,636 ACR 4 1,268 Total 25,378
3,061 11,035 9,964 1,094 25,352
Numbers screen-detected breast cancers with available ACR classification related to the first reading
Breast First Second Highest density reading reading case * Numbers Numbers reading Numbers
Table 2 Numbers of ACR classifications, numbers of screen-detected breast cancers, screen-detected cancer rates separated by mammographic breast density categories resulting from first and second independent double reading
Eur Radiol
Eur Radiol Table 4 Numbers of ACR classifications and 2-year breast cancer incidences, separated by mammographic breast density categories resulting from first and second independent double reading Breast First Second Highest density reading reading case * Numbers Numbers reading Numbers
2-year breast cancer incidence rate per 1,000 women screened ** with available ACR classification related to the first reading
2-year breast cancer incidence rate per 1,000 women screened ** with available ACR classification related to the second reading
2-year breast cancer incidence rate per 1,000 women screened ** with available ACR classification related to the highest case reading
First reading P-value (vs. ACR 4)***
Second reading P-value (vs. ACR 4)***
Highest case reading P-value (vs. ACR 4)***
ACR 1 2,336
3,061
1,559
2.14
2.94
1.28
0.09
0.003
0.01
ACR 2 10,914
11,035
9,905
9.71
9.33
8.98
0.30
0.71
0.76
ACR 3 10,636 ACR 4 1,268
9,964 1,094
12,341 1,771
11.61 6.30
11.24 10.96
10.94 7.90
25,352
25,576
9.30
9.31
9.39
Total
25,378
0.12 0.95 0.30 Reference Reference Reference -
-
-
*Visual determination of breast densities according to the 4th edition of the BI-RADS lexicon at time of screening. **Includes screen- and interval-detected breast malignancies, numbers and details are given by Tables 2 and 3. ***p-values refer to a pairwise comparison. Reference refers to the reference group ACR 4.
Discussion Our study reveals relevant differences in both prevalence and breast cancer detection at screening between screening participants with mammographic breast densities ACR 3 and ACR 4 - as defined by the 4th edition of BI-RADS. ACR 3 was present in about 40 % of all participating women while ACR 4 was a rather rare condition with a frequency of below 7 %. The rate of breast cancers detected at screening was lower for breasts classified as ACR 4 than ACR 3 while the rate of interval cancers was higher, but these findings were not clearly significant. Worthy of note, however, the average sensitivity of the programme in women with ACR 3 ranged from 73 % to 81 % as contrasted by a sensitivity in ACR 4 of only 50 %; this difference was statistically significant irrespective of the reading-related classification selected. In other words, only one out two breast cancer cases occurring among screening participants with ACR 4 over a period of 24 months was detected in the screening examination. Most likely, a varying extent of masking and obscuring of tumour signs - if not associated with microcalcifications or architectural distortion - in breasts consisting of up to 75 % dense tissue (ACR 3) versus over 75 % dense tissue (ACR 4) might lead to such a discrepant screening performance. An inadequate proportion of missed cancers within the subgroup of interval cancers with ACR 4 classification was not found. The European guidelines recommend using the parameter sensitivity of the screening programme (SP) for evaluation purposes. As this parameter includes in the denominator also ‘genuine’ interval cancers whose preclinical detectable phase had not yet started at the time of the screening test, the SP
encompasses impacts of the sensitivity of the screening test, of the lead time and of the length of the screening interval [5]. Previous studies reported SP results of 78 % for initial screenings and a range from 67 % to 84 % for a pooled evaluation including subsequent screening [6, 7]. A recent Swedish study, including cancer diagnoses made from 2001 to 2008 in women aged 40–71 years, reported a SP of 70 % [4]. Despite methodological limitations and the absence of other reports that investigated the impact of ACR categories on SP in comparable detail, the SP for ACR 4 observed in this study was strikingly low in contrast to ACR 3. Of note, the order of magnitude of the SP from international reports was confirmed for classifications of ACR 1, ACR 2 and ACR 3 in the present study. Our results are in line with a congress presentation from the Dutch biennial screening programme (50–75 years) investigating the effect of volumetric mammographic density on performance of a breast cancer screening programme using full-field digital mammography. Within four density categories, comparable to the ACR breast density categories, they found that interval cancer rates were rising with increasing density categories, leading to consistently decreasing sensitivity of the screening programme from 85.0 % to 77.6 %, 69.0 % and 58.6 %, respectively [8]. As visual ACR assessment is a subjective classification, inter-observer variability affects the frequency of ACR categories and, therefore, also the evaluation of screening parameters per ACR category. A recent study showed an inter-exam agreement for readers using four density categories with kappa values of 0.76-0.82 while automated density measurement gave more consistent results [9]. To account for these variabilities, we report here the classifications separately for the first and the second reading, and for the highest density of the two readings. As there was no fixed order of assignments of the
N+
N-
N-
N-
N-
N+
pT3
pT1c
pT2
pT2
pT1b
pT2
Invasive ductal carcinoma (no special type) Invasive ductal carcinoma (no special type) Invasive ductal carcinoma (no special type)
Invasive ductal carcinoma (no special type) Mucinous carcinoma
19
13
10
9
13
None No abnormality on breast side with subsequent cancer, presented in consensus conference, No assessment Both BI-RADS 1 or 2 None
13
One BI-RADS 2 One abnormality presented in consensus conference, No assessment
Both BI-RADS 1 or 2
None
Only MRI available, mass
Mass
Mass
Mass in the retromamillary space
Asymmetry
Retrospective evaluation Abnormalities on diagnostic mammography
Amorphous microcalcifications in Increasing central and lateral parts of the microcalcifications in breast lateral parts with subtile increasing density
None
None No abnormality on breast side with subsequent cancer, presented in consensus conference, No assessment Both BI-RADS 1 or 2 None
None
Both BI-RADS 1 or 2
17
Retrospective evaluation Abnormalities on screening mammography
Results of independent double reading
Diagnosis following negative screening (month)
Minimal sign
Occult interval cancer
No categorisation performed
True interval cancer
True interval cancer
True interval cancer
True interval cancer
Internal review
**Pathological axillary nodal status: N+ (metastasis positive), N- (metastasis negative)
*Pathological tumour categories: pTis: carcinoma in situ, pTmic: microinvasion to 0.1 cm, pT1a: invasion > 0.1 cm to ≤ 0.5 cm, pT1b: invasion > 0.5 cm to 1 cm, pT1c: invasion > 1 cm to 2 cm, pT2: invasion > 2 cm to 5 cm, pT3: invasion > 5 cm
2
1
3
1
2
2
Invasive ductal carcinoma (no special type) Invasive ductal carcinoma (no special type)
N+
pT3
2
Axillary Grading Histological metastasis** tumour subtype
Histological and reading details of interval cancers with screening mammograms classified at least by one reader ACR 4
pT category*
Table 5
Eur Radiol
Eur Radiol
five radiologists to the first or second reading, and thus no systematic difference between them, the three classification methods reflect both the variability of readings and the consistency of their association with cancer detection. A lower sensitivity of the screening programme implies a shift of breast cancer diagnoses from the screening examination to the interval. In dense breasts, interval cancers were found to be phenotypically more similar to screen-detected breast cancers than interval cancers diagnosed in non-dense breasts [4]. Therefore, interval cancers in dense breasts might be detected earlier with extended screening protocols. This may be especially useful with regard to the lobular subtype, as previous analyses in our digital mammography screening programme revealed that SP was particularly low among young participants with lobular breast cancer [10]. Our results support recent studies from digital mammography screening programmes indicating that extremely dense tissue, that is ACR 4, is a rare constellation involving under 10 % of screening participants in the age groups 40 to over 70 years (1.9 % out of 10,728 participants) [11] as well as among first time participants aged 50 to 69 years (7.4 % out of 64,949 participants) [10]. Commonly, however, ACR 3 and ACR 4 are conflated to the joint category ‘dense breast’ which then comprises about 50 % of all screening participants as having this condition. In this context it is important to bear in mind that breast density legislation has rapidly been implemented around the United States and that women with ‘dense breasts’ according to this joint definition are informed that they might benefit from supplementary screening tests like - for example ultrasound of the breast [3]. Alternatively, the development of digital mammography to digital breast tomosynthesis has, when employed additively, the potential to increase screendetected invasive cancer rates and to reduce interval cancers [11]. However, the evidence supporting supplemental screening tests in women with dense breasts, especially in ACR 4 remains markedly limited [4, 12–15] and needs further evaluation [3, 16]. The strengths of our study are the prospective documentation of screening data gained in a clearly defined setting with validated protocols for all steps in the screening process. This was complemented by information from notifications submitted with high completeness to the cancer registry by statewide clinical and pathological sources for screen-detected and interval-detected breast cancer cases. As a limitation, the present data originate from only one screening unit and the generalizability of the results may thus be limited. Due to the constraint in numbers, we did also not calculate the SP separately for initial and subsequent screening examinations for different breast densities or age groups as recommended by the European guidelines for epidemiological programme evaluation [5]. Finally, automated software, which may possibly increase the consistency of ACR classifications, was also not available.
Conclusion At an age of 50 to 69 years, the ACR categories 1, 2 and 3 comprised more than 93 % of all participants in a digital mammography screening programme. The sensitivities of the screening programme, as previously reported for initial and subsequent screenings, were confirmed in these ACR categories, while there was a significant decrease in SP for category ACR 4. Therefore, we suppose that the screening performance should not be assessed in women with ‘dense breast’, commonly derived from conflating women with ACR 3 and ACR 4, but rather that it seems advisable to separate these two categories due to their different performance. Further research on criteria for objective standardization of ACR categorization and on the validation of these results in a multicentre setting is needed.
Acknowledgements The scientific guarantor of this publication are Stefanie Weigel and Hans-Werner Hense. The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article. The authors state that this work has not received any funding. One of the authors (Prof. Hense) has significant statistical expertise. Institutional Review Board approval was not required under national law. Written informed consent was waived by the Institutional Review Board. Some of the study subjects have been previously reported in: Weigel S, Biesheuvel C, Berkemeyer S, Kugel H, Heindel W. Digital mammography screening: how many breast cancers are additionally detected by bilateral ultrasound examination during assessment? Eur Radiol. 2013 Mar; 23(3):684–91. Weigel S, Berkemeyer S, Girnus R, Sommer A, Lenzen H, Heindel W. Digital mammography screening with photon-counting technique: can a high diagnostic performance be realized at low mean glandular dose? Radiology. 2014 May;271(2):345–55. Weigel S, Heindel W, Heidrich J, Heidinger O, Hense HW. Reduction of Advanced Breast Cancer Stages at Subsequent Participation in Mammography Screening. Rofo. 2016 Jan;188(1):33–7. Methodology: retrospective, observational, performed at one institution. We especially acknowledge the continuous and excellent support of our screening team.
References 1.
2.
3.
4.
American College of Radiology (2003) Breast imaging reporting and data system: BI-RADS atlas, 4th edn. American College of Radiology, Reston, VA Sickles EA, D’Orsi CJ, Bassett LW, et al (2013) ACR BI-RADS® Mammography. In: ACR BI-RADS® Atlas, Breast Imaging Reporting and Data System. American College of Radiology, Reston Melnikow J, Fenton JJ, Whitlock EP et al (2016) Supplemental screening for breast cancer in women with dense breasts: a systematic review for the U.S. preventive services task force. Ann Intern Med 164:268–278 Holm J, Humphreys K, Li J et al (2015) Risk factors and tumor characteristics of interval cancers by mammographic density. J Clin Oncol 33:1030–1037
Eur Radiol 5.
6.
7.
8.
9.
10.
Perry NM, Broeders M, de Wolf C et al (2006) European guidelines for quality assurance in breast cancer screening and diagnosis, 4th edn. Office for Official Publications of the European Communities, Luxembourg Heidinger O, Batzler WU, Krieg V et al (2012) The incidence of interval cancers in the German mammography screening programme: results from the population-based cancer registry in North Rhine-Westphalia. Dtsch Arztebl Int 109:781–787 Törnberg S, Kemetli L, Ascunce N et al (2010) A pooled analysis of interval cancer rates in six European countries. Eur J Cancer Prev 19:87–93 Wanders J, Holland K, Veldhuis W, et al (2015) Effect of volumetric mammographic density on performance of a breast cancer screening program using full-field digital mammography. In: European Congress of Radiology Holland K, van Zelst J, den Heeten GJ et al (2016) Consistency of breast density categories in serial screening mammograms: a comparison between automated and human assessment. Breast 29:49– 54 Heidinger O, Heidrich J, Batzler WU et al (2015) Digital mammography screening in Germany: impact of age and histological subtype on program sensitivity. Breast 24:191–196
11.
12.
13.
14.
15.
16.
McDonald ES, Oustimov A, Weinstein SP et al (2016) Effectiveness of digital breast tomosynthesis compared with digital mammography: outcomes analysis from 3 years of breast cancer screening. JAMA Oncol 2:737–743 Freer PE (2015) Mammographic breast density: impact on breast cancer risk and implications for screening. Radiographics 35:302– 315 Berg WA, Blume JD, Cormack JB et al (2008) Combined screening with ultrasound and mammography vs mammography alone in women at elevated risk of breast cancer. JAMA 299:2151–2163 Parris T, Wakefield D, Frimmer H (2013) Real world performance of screening breast ultrasound following enactment of Connecticut Bill 458. Breast J 19:64–70 Kerlikowske K, Zhu W, Tosteson AN et al (2015) Identifying women with dense breasts at high risk for interval cancer: a cohort study. Ann Intern Med 162:673–681 Tagliafico AS, Calabrese M, Mariscotti G, et al (2016) Adjunct screening with tomosynthesis or ultrasound in women with mammography-negative dense breasts: interim report of a prospective comparative trial. J Clin Oncol