University of Groningen Breast cancer screening in women at elevated risk Phí, Xuân Anh

(1)

University of Groningen

Breast cancer screening in women at elevated risk

Phí, Xuân Anh

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2018

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Phí, X. A. (2018). Breast cancer screening in women at elevated risk: Comparative evaluation of screening modalities to inform practice. University of Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

Digital breast tomosynthesis for

breast cancer screening and diagnosis

in women withdense breasts – a systematic

review and meta-analysis

Alberto Taglia ico Marcel J.W. Greuter Nehmat Houssami Geertruida H. De Bock BMC Cancer. 2018 Apr 3;18(1):380

(3)

ABSTRACT

Background: This study aimed to systema cally review and to meta-analyse the accuracy of

digital breast tomosynthesis (DBT) versus digital mammography (DM) in women with mammographically dense breasts in screening and diagnosis.

1

Methods: Two independent reviewers iden ﬁed screening or diagnos c studies repor ng at

least one of four outcomes (cancer detec on rate-CDR, recall rate, sensi vity and specificity) for DBT and DM in women with mammographically dense breasts. Study quality was assessed using QUADAS-2. Meta-analysis of CDR and recall rate used a random effects model. Summary ROC curve summarized sensi vity and specificity.

1

Results: Sixteen studies were included (ﬁve diagnos c; eleven screening). In diagnosis, DBT

increased sensi vity (84%-90%) versus DM alone (69%-86%) but not speciﬁcity. DBT improved CDR versus DM alone (RR: 1.16, 95%CI 1.02-1.31). In screening, DBT+DM increased CDR versus DM alone (RR: 1.33, 95%CI 1.20-1.47 for retrospec ve studies; RR: 1.52, 95%CI 1.08-2.11 for prospec ve studies). Recall rate was signiﬁcantly reduced by DBT+DM in retrospec ve studies (RR: 0.72, 95%CI 0.64-0.80) but not in two prospec ve studies (RR: 1.12, 95%CI 0.76-1.63).

1

Conclusion: In women with mammographically dense breasts, DBT+/-DM increased CDR

significantly (versus DM) in screening and diagnosis. In diagnosis, DBT+/-DM increased sensi vity but not specificity. The effect of DBT+DM on recall rate in screening dense breasts varied between studies.

1

Key words: breast neoplasm, digital mammography, digital breast tomosynthesis, review,

meta-analysis, breast density

(4)

Background

1

Breast cancer (BC) is the most common cancer in women and the leading cause of cancer death among women in Europe (1). Many countries have adopted popula on-wide BC screening, ini ally with ﬁlm-screen and subsequently with digital mammography (DM), aiming to lower mortality from BC by earlier detec on of the disease (2,3). However, DM has moderate sensi vity, for which es mates vary from 67.3% to 93.3% (4). High breast ssue density, deﬁned as having more than 50% density on mammography, categories 3 and 4 or categories c and d in the BI-RADS 4 or 5 edi on respec vely (5,6) reduces the sensi vity of th th

mammography due to its masking eﬀect, and may increase false-posi ves due to superimposi on of dense parenchyma. It is es mated that about half of all women taking part in screening have dense breasts although the propor on diﬀers in age-groups (7,8). Breast density is also considered an independent risk factor for BC (9).

1

Digital Breast Tomosynthesis (DBT) enables pseudo-3D imaging of the breast, resul ng in be er discrimina on of ssue structures and poten ally improved visualisa on of cancer (10,11). Hence, DBT has the poten al to improve both sensi vity and speciﬁcity of imaging in BC screening, leading to more detected cancers with fewer false-posi ves (10). However, using both DM and DBT increases radia on dose to the breast, if both acquisi ons are obtained. Improved screening accuracy using DBT has been shown in several prospec ve and retrospec ve studies, in screening popula ons and in studies using BC-enriched mammogram series (12,13). Very few reviews have examined the role of DBT in women with dense breasts, and these were either concise reports or did not use systema c methodology (14,15). Therefore, in this work we aimed to systema cally review the literature on the accuracy of DBT compared to DM in women with dense breasts. A secondary objec ve was to perform a meta-analysis on four outcomes (cancer detec on rate - CDR, recall rates, sensi vity and speciﬁcity) of DBT compared to DM in women with dense breasts.

Methods 1

A systema c review and meta-analysis were performed by two independent reviewers (XAP and GHdB or AT), following a predetermined review protocol based on the PRISMA guidelines (h p://www.prisma-statement.org) (supplementary 3, online only). Discordance throughout the process was discussed between the two reviewers and if consensus was not reached then a third reviewer (GHdB or NH) was consulted.

(5)

We searched for studies that included women older than 18 years, who underwent breast imaging using DBT and DM and were classiﬁed as having dense breasts on mammography. Studies comparing DBT to DM and repor ng at least one accuracy measure were considered. Prospec ve as well as retrospec ve compara ve studies could be included.

Data Sources and searches

1

PubMed and the Web of Science were searched for relevant English-language ar cles published between January 2007 up to and including May 2017. Addi onally, references of iden ﬁed eligible ar cles and reviews were manually screened for addi onal relevant sources. The search strategy included three main key words: Tomosynthesis or 3D mammography in the tle or abstract and BC in all ﬁelds (see Figure 1).

Eligibility criteria 1

Eligible studies were: studies that compared the accuracy of DBT and DM in a screening se ng or diagnos c se ng; reported data on at least one of 4 outcomes (CDR, recall rate, sensi vity and speciﬁcity) for both DBT and DM (where data reported or could be calculated); included at least 100 women with dense breasts who were asymptoma c (screening se ng) or recalled a er screening (diagnos c se ng); and where 'dense breast' was deﬁned as more than 50% density [BI-RADS 3 and 4 (4 edi on) or BI-RADS c and d (5 edi on)]. Only English th th

publica ons were considered. Studies which did not contain original data, or simula on studies, were excluded. If mul ple publica ons were based on the same study popula on, the most extensive study in terms of data reported was chosen.

Study selec on 1

Ar cles iden fied from the search were loaded into RefWorks (2016, ProQuest LLC) and duplicates were removed. Titles/ abstracts, followed by full text, were reviewed based on predefined criteria and a final set of eligible studies were selected.

Data collec on process 1

A predeﬁned form was developed, and used to extract informa on from included studies: type of study (prospec ve or retrospec ve), study se ng (screening or diagnos c), number of women with dense breasts, inclusion and exclusion criteria, age of women with dense breast (age of whole study popula on if not speciﬁed for dense breasts), number of screening

Chapter 7 Cost-eﬀectiveness of MRI screening in BRCA carriers aged 60+

(6)

rounds if applicable, length of follow-up, method of repor ng breast density, number of BCs, reading protocol (single or double reading), deﬁni ons for recall and for posi ve test, DBT manufacturer, number of DBT views (one or two), u lisa on of addi onal modali es (DM or none), and reported outcomes for DBT and for DM in women with dense breasts (CDR, recall rate sensi vity and/or speciﬁcity).

Risk of bias and quality appraisal 1

The quality of included studies was assessed using the QUADAS-2 tool which was modified to ensure assessment was appropriate for the breast screening or diagnos c context. The domains considered were: pa ent selec on, index test, reference standard, flow and ming and applicability. This was performed by two reviewers independently and final quality assessment was based on consensus.

Data analysis 1

Meta-analysis was performed to es mate the rela ve risk of cancer detec on and of recall for DBT and DM using a random effects model in RevMan 5.3. This analysis was performed separately for screening and diagnos c studies, and also separately for studies comparing two groups of women (unpaired data) and those comparing detec on within one group of women (paired data). Subgroup analyses were carried out to examine the effect of covariates, modality (whether stand-alone DBT, or DBT with DM), outcome defini ons, and reading protocol (single or double-reading). A summary ROC was produced for DBT and DM for sensi vity and specificity where studies reported these outcomes. For computa on, it was assumed that all screens were independent, even if there were mul ple screens for some pa ents in some studies. Heterogeneity across studies was quan fied with I2 measure for CDR and recall rate.

(7)

Figure 1: Flow-chart of study inclusion

Search strategies:

Pubmed: (((tomosynthesis[Title/Abstract] OR 3D

mammography[Title/Abstract])) AND (breast cancer[MeSH Major Topic] OR breast cancer[ tle/abstract])) AND ("2007/01/01"[Date - Publica on] : "3000” [Date - Publica on])

Web of knowledge: (tomosynthesis or 3d mammography) and breast cancer, me span: 1997 -2016

Publica ons iden ﬁed through search in Web of

Knowledge

Publica ons iden ﬁed through search

in PubMed

Title and Abstracts screened a er removal of duplicates (n = 608) Excluded = 545 studies : Not including DBT (n= 13) Not including DM (n=2 ) No breast study (n= 8) No prospec ve/retrospec ve study on breast cancer screening or diagnos c work-up: (n = 477)

Other outcomes than sensi vity, speciﬁcity, CDR or RR (n= 31)

High risk popula on (n= 1)

Full text assessed for eligibility (n = 63) Exclude 47 studies No report on dense breast popula on (n=43) Diﬀerent deﬁni on of dense breast (n=1) Same study popula on (n=1) No full text available (n=1, Chinese) Included in the qualita ve

synthesis (n = 16)

Included in the meta-analysis (n = 16)

(8)

Results 1

Study inclusion 1

A total of 608 unique studies were eligible for tle and abstract screening, and 63 studies were checked at full-text reading (details in Figure 1). Sixteen studies (12,16-30) met our predeﬁned inclusion criteria and were included in the evidence synthesis. The meta-analysis was performed separately for 5 diagnos c studies, and for 11 screening studies (these were examined separately for 8 screening studies that used two independent study groups, and the 3 screening studies that used one study group). Details about study inclusion with reasons for exclusion are described in the ﬂow-chart (Figure 1).

Overview of included studies 1

Characteris cs of 16 included studies are presented in supplementary 1 (online only). Studies differed in terms of study se ng, threshold defini ons, breast density categoriza on, reading protocol and whether DBT was used alone or with DM. Among the five diagnos c studies, four studies using DBT and DM reported sensi vity and specificity (12,17,21,30) and one study reported recall rate (18). It was possible to calculate CDR from three studies which reported sensi vity and specificity (12,17,21). All but two of 11 screening studies performed one screening round. Nine screening studies reported CDR (16,19,20,23-27,29) and nine studies reported recall rate (16,19,20,22,24,26-29).

Quality assessment 1

Six of the 16 included studies were at high risk of bias in terms of pa ent selec on. In five studies, DBT was performed more o en in women with dense breast, with family history of BC or it was performed based on availability and women's preferences (20,22,27-29). Three studies did not specify density classifica on (22,26,28). Three of 16 studies were at high risk of bias in terms of index test due to unspecified outcome defini on (26,28,29). One study did not specify the reference standard for nega ve images (21). One third of the studies were at high risk of bias regarding flow and ming domain because DBT and DM were performed in different periods of me (24-27,29). Overview of risk of bias and applicability is shown in table 1.

(9)

(10)

Comparing tomosynthesis and digital mammography in women with dense breasts in diagnos c se ngs (N=5)

1

In the diagnos c se ng, sensi vity of DBT ranged from 84% (95%CI 71-93) (30) to 89% (95%CI 81-95) (12), being higher than the sensi vity of DM which ranged from 69% (95C%I 58-79) (12) to 86% (95%CI 81-89) (21) (Figure 2). The specificity of DBT ranged from 72% (95%CI 68-72) (21) to 93% (95%CI 89-96) (17) and was not different from the specificity of DM which ranged from 57% (95%CI 55-59) (21) to 94% (95%CI 91-97) (17) (Figure 2). Using DBT with or without DM improved CDR with a ra o of 1.12 (95%CI 1.01-1.24) compared to DM alone (Figure 3a). Heterogeneity across studies was small (I2 = 8%). Only one study reported a significant reduc on in recall rate using DBT plus DM compared to DM alone (RR=0.56, 95%CI 0.47-0.66) (18). Due to the small number of studies (n=5), subgroup analysis was not performed.

Comparing tomosynthesis and digital mammography in women with dense breasts in screening se ng

1

Studies comparing two groups of par cipants (N=8) 1

All but two studies used single-reading. Six studies were included in the analysis of CDR and were homogeneous in repor ng CDR (I2=0%). CDR was es mated to be significantly higher when using DBT with or without DM compared to DM alone (RR= 1.33, 95%CI 1.20-1.47) (Figure 3b). Seven studies reported recall rates with high heterogeneity (I2=93%). Pooled es mate for these studies showed a significant reduc on in recall rate when using DBT with or without DM compared to DM alone (RR= 0.72, 95%CI 0.64-0.80) (Figure 4a). Subgroup analysis based on test defini on showed consistently reduced recall rates (RR= 0.59, 95%CI 0.52-0.67 for BI-RADS 0 as recalled and RR= 0.84, 95%CI 0.81-0.87 for other defini ons). Subgroup analysis also reduced heterogeneity in subgroup es mates (I2: 62% and 37%, respec vely) (Supplementary 2, online only).

Studies comparing within the same group of par cipants (paired data) (N=3) 1

These studies used DBT combined with DM (16-19) or DBT alone (23), double-reading protocol and similar deﬁni ons for recall. Pooled es mates from three studies showed improved CDR when using DBT with DM compared to DM alone, RR = 1.52, 95%CI 1.08-2.12 (Figure 3c) with homogeneity across studies (I2=0%). Using DBT with DM did not reduce recall rate based on two studies (I2=76%). The pooled RR was 1.12, 95%CI 0.76 -1.63 (Figure 4b). Subgroup analysis was not performed due to a small number of studies.

(11)

Figur

e 2. F

or

es

t plot and Summary r

eceiv er oper ang char act eris c plot of DB T and DM in diagnos c se ng Chapter 7 Cost-eﬀectiveness of MRI screening in BRCA carriers aged 60+

(12)

Risk Ratio Favours [DM] Favours [DBT] Risk Ratio M-H, Random, 95% Cl M-H, Random, 95% Cl DM DBT Study or Subgroup Events Total Events Total Weight

Gilbert 2015 Chae 2016 Shin 2015 299 214 76 2126 472 139 288 188 59 2126 472 139 41.3% 42.2% 16.5% 1.05[0.90, 1.22] 1.14[0.98, 1.32] 1.29[1.01, 1.65] Total (95% Cl) 2737 2737 100% 1.12[1.01, 1.24] Total events 589 533

Heterogeneity: Tau = 0.00; Chi = 2.17, df = 2(P=0.34); l = 8%

z

Test for over all effect: Z = 2.21 (P=0.03)

0.1 0.2 0.5 1 2 5 10 Figur e 3. Cancer de tecon r at e of D B T and DM in diagnos c and scr eening s tudies Risk Ratio Favours [DM] Favours [DBT] Risk Ratio M-H, Random, 95% Cl M-H, Random, 95% Cl DM DBT Study or Subgroup Events Total Events Total Weight

Bernardi 2016 Ciatto 2013 Lang 2016 34 8 43 2592 1215 3150 20 5 31 2592 1215 3150 37.4% 9.1% 53.5% 1.70[0.98, 2.95] 1.60[0.52, 4.88] 1.39[0.88, 2.20] Total (95% Cl) 6957 6957 100% 1.12[1.01, 1.24] Total events 85 56

z

0.01 0.1 1 10 100 Risk Ratio Risk Ratio M-H, Random, 95% Cl M-H, Random, 95% Cl DM DBT Study or Subgroup Events Total Events Total Weight

Conant 2016 McCarthy 2014 McDonald 2016 Rafferty 2016 Rose 2013 Starikov 2016 63 35 81 ₄₉₅ 25 10 9265 5056 10733 84243 4666 1875 166 18 18 597 28 27 35319 3489 3489 ₁₃₁₉₉₆ 7009 7117 12.6% 3.3% 4.1% 74.5% 3.6% 2.0% 1.45[1.08, 1.93] 1.34[0.76, 2.37] 1.46[0.88, 2.43] 1.30[1.15, 1.46] 1.34[0.78, 2.30] 1.41[0.68, 2.90] Total (95% Cl) 115838 188419 100.0% 1.33[1.20, 1.47] Total events 709 854

z

Test for over all effect: Z = 5.43 (P<0.00001)

Favours [DM] Favours [DBT] 0.1 0.2 0.5 1 2 5 10

(13)

Figur e 4. R ec all r at e of DB T and DM in scr eening s tudies Chapter 7 Cost-eﬀectiveness of MRI screening in BRCA carriers aged 60+ Risk Ratio Risk Ratio M-H, Random, 95% Cl M-H, Random, 95% Cl DM DBT Study or Subgroup Events Total Events Total Weight

Conant 2016 Haas 2013 McCarthy 2014 Rafferty 2016 Rose 2013 Sharpe 2016 Starikov 2016 2186 257 547 9030 321 169 195 21133 2639 5056 84243 4666 2603 1875 5561 358 445 ₁₆₅₈₂ 740 2706 1416 44135 2158 3489 ₁₃₁₉₉₆ 7009 31063 7117 16.5% 12.8% 14.1% 16.9% 13.8% 12.7% 13.1% 0.82[0.78, 0.86] 0.59[0.51, 0.68] 0.85[0.75, 0.95] 0.85[0.83, 0.87] 0.54[0.57, 0.74] 0.75[0.64, 0.87] 0.52[0.45, 0.60] Total (95% Cl) 12705 226967 100.0% 0.72[0.64, 0.80] Total events 12705 27808

Heterogeneity: Tau = 0.02; Chi = 83.30, df = 6(P<0.00001); l = 93%

z

Test for over all effect: Z = 6.10 (P<0.00001)

Favours [DM] Favours [DBT] 0.1 0.2 0.5 1 2 5 10 Risk Ratio Favours [DM] Favours [DBT] Risk Ratio M-H, Random, 95% Cl M-H, Random, 95% Cl DM DBT Study or Subgroup Events Total Events Total Weight Bernardi 2016 Ciatto 2013 162 80 2592 1215 121 88 2592 1215 52.9% 47.1% 1.34[1.06, 1.68] 0.91[0.68, 1.22] Total (95% Cl) 3807 3807 100% 1.12[1.01, 1.24] Total events 242 209

z

0.1 0.2 0.5 1 2 5 10

(14)

Discussion 1

This systema c review iden fied 16 studies (5 diagnos c and 11 screening studies) comparing accuracy measures, such as CDR, recall rate, sensi vity and specificity, of DBT and DM in women with dense breasts at mammography. We found that in diagnos c studies, DBT with or without DM improved CDR (RR= 1.12, 95%CI 1.01-1.24) and sensi vity compared to DM alone (84%-89% vs 69%-86%) in women with dense breasts, whereas specificity did not increase when DBT was used (72-93% vs 57-94%). In the screening se ng, CDR was improved when using DBT with or without DM, in studies comparing within one study group (RR= 1.52, 95%CI 1.08-2.12) or comparing two study groups of par cipants (RR= 1.33, 95%CI 1.20-1.47). Recall rate was reduced when using DBT compared to DM alone in screening studies using two study groups (RR= 0.72, 95%CI 0.64-0.80), though heterogeneity across studies was very high (I2=93%) and par ally explained by the two different defini ons of outcome.

1

Almost all of the reviews in the literature comparing DBT with DM in BC screening do not dis nctly report on women with dense breasts. One review, not restricted to women with dense breasts, reported that DBT with DM increased CDR with a RR of 1.29 (95%CI 1.16-1.43) (31) which is comparable to our es mate. We iden fied only two reviews repor ng on women with dense breasts, one was a quan ta ve rapid review and one was a narra ve review without analyses (14,15). The rapid review iden fied eight studies comparing CDR and recall rate of DBT plus DM to DM alone in women with dense breasts but was restricted to screening studies. The rapid review reported a significantly increased CDR when pooling studies comparing within same group of par cipants (incremental cancer detec on per 1000 screens: 3.9, 95%CI 2.7-5.1) as well as when pooling studies comparing two groups of par cipants (incremental cancer detec on per 1000 screens: 1.4, 95%CI 0.9-2.0) (14). Although our results are generally in line with these previous reviews, by performing a systema c search and by considering both screening and diagnos c studies, we were able to iden fy more data sources for the comparison of DBT with DM, and were also able to present data on a broader range of outcomes (sensi vity and specificity as well as CDR and recall rate) hence we extend on exis ng reviews. In addi on, we conducted quality assessment of the included evidence which was not done in the other reviews on this issue.

1

Studies included in our review were heterogeneous in several aspects. Firstly, some studies included asymptoma c or symptoma c popula on. Although aiming to inves gate the accuracy of DBT and DM in BC screening, some studies included women who were recalled a er screening (12,18,21,30). By doing so, they obtained popula ons recalled to assessment which have higher cancer rates than unselected asymptoma c popula ons.

(15)

assessment which have higher cancer rates than unselected asymptoma c popula ons. However, the results from the screening and diagnos c se ngs had generally comparable findings. Secondly, retrospec ve studies tended to perform single-reading whereas the prospec ve studies performed double-reading (reflec ng screening prac ce in various se ngs) which may increase CDRs in the la er. In the two STORM trials, screen-reading results were based on recall by either reader, making the recall rate of integrated DM and DBT higher than DM in STORM-2 (16) and non-significantly lower in STORM-1 (19). Addi onally, all but one (23) screening study used DBT together with DM while among five diagnos c studies two studies used DBT as stand-alone modality (18,30).

1

Another difference among studies was the outcome defini ons which may be contribu ng to some of the observed heterogeneity. Studies performed in the United States used the BI-RADS system for repor ng recall whereas the European studies used a simplified 'recall or no recall' repor ng for screen-readings. Amongst studies using BI-RADS, different thresholds were used to define recall or posi ve test. When analysing data for different thresholds, the result of recall rate in screening studies using two study groups remained significantly lower for DBT compared to DM but the heterogeneity decreased. Studies defining BI-RADS 0 as recall (22,27,29) showed a larger decrease in recall rate than studies using a different defini on (BI-RADS 0,3,4,5 (20) or BI-RADS 0,4,5 (25) or where unspecified (26,28). Among four diagnos c studies repor ng sensi vity and specificity, one study from the UK (21) used a lower threshold (BI-RADS 3 instead of BI-RADS 4) and reported lower specificity than other diagnos c studies which may account for more false posi ves.

1

The main limita on of the data used in our analysis is that all but two screening studies (20,23) used only a single screening (likely to be first round) for DBT. When only first DBT screening rounds are used (29,24,26-29), more prevalent cases are usually detected, increasing CDR and poten ally exaggera ng the contribu on of DBT to screen-detec on measures. Another limita on is the short temporal perspec ve in all these studies: because of the recent introduc on of DBT, the lack of long follow-up makes it impossible to assess whether the improved CDR and sensi vity of DBT screening further reduces BC mortality through screening compared to screening with DM alone. The retrospec ve studies had one major limita on in that they used two study groups (DM group versus DBT group) which were not randomly assigned to screening modali es. The study groups were from different me periods (or services) and in the DBT implementa on phase, or due to the limit of DBT availability, there may have been selec on to DBT screening and hence poten al bias. In those studies, DBT groups were more likely to include women with dense breasts and family

(16)

history indica ng high cancer rate. The incremental value of DBT in those studies may be par ally due to possible selec on bias. Finally, in order to be able to compute the results, we made an assump on of independent screens, which might not be the case in the few studies that included more than one screening round (20,25) or where study popula ons might overlap (16,19). However, es mates from those studies were similar to the other studies, thus we do not foresee that this assump on aﬀected our reported ﬁndings.

1

We performed a systema c review to summarize current evidence on the use of DBT in BC screening and diagnosis specifically in women with dense breasts on mammography. We iden fied and systema cally examined data for women with dense breasts from 16 eligible studies to report the most extensive review so far on the accuracy of DBT in women with dense breasts. Moreover, this is the first review assessing the quality of evidence and bias in the studies on DBT in women with dense breasts.

Conclusion 1

We found that in both the screening and diagnos c se ngs, DBT improved CDR (versus DM) in women with dense breasts. In the diagnos c se ng, using DBT with or without DM increased sensi vity but did not change specificity. There was a significant reduc on in recall rate when using DBT with DM (versus DM) in retrospec ve screening studies comparing between two study groups, although heterogeneity across studies was rela vely high. A small number of prospec ve studies conducted in organized screening programs did not show reduced recall from using DBT. Improved CDR and reduced recall rate from DBT may imply a more effec ve screening test or diagnos c work-up for women with dense breasts. However, the cri cal issue is that more studies with longer follow-up and more screening rounds are necessary to draw definite conclusions on whether this improvement in cancer detec on has an impact on interval cancer rates and poten ally on BC mortality.

List of abbrevia ons

1

DBT: Digital breast tomosynthesis; DM: Digital mammography; CDR: Cancer detec on rate; BC: Breast cancer; RR: Rela ve risk; BI-RADS: Breast Imaging Repor ng and Data System.

Declara ons

1

Ethics approval and consent to par cipate: Not applicable

1

(17)

Availability of data and materials: The authors declare that all the data suppor ng the ﬁndings of this

study are available within the ar cle and its supplementary informa on ﬁles.

1

Compe ng interests: All authors have no conﬂict of interest to report.

1

Funding source: The work had no speciﬁc funding. N. Houssami receives support through a Na onal

Breast Cancer Founda on (NBCF Australia) Breast Cancer Research Leadership Fellowship.

1

Authors' contribu ons

1

X.A.Phi: methodology, valida on, formal analysis, inves ga on, data cura on, wri ng-original dra , wri ng-reviewing and edi ng, project administra on. A. Tagliaﬁco: inves ga on, wri ng-reviewing and edi ng. N. Houssami: methodology, valida on, inves ga on, wri ng-reviewing and edi ng. M.J.W Greuter: conceptualiza on, methodology, valida on, inves ga on, wri ng-reviewing and edi ng. G.H.de Bock: conceptualiza on, methodology, valida on, formal analysis, inves ga on, data cura on, wri ng-original dra , wri ng-reviewing and edi ng, project administra on, supervision.

1

Acknowledgment: We acknowledge A.M. Daszczuk for her contribu on to ar cle reviewing and data

extrac on for this work.

(18)

(19)

(20)

(21)

(22)

* R eport ed da ta ar e r ela ted t o all w omen parcipa ng in this s tudy as the da ta w er e not pr esen ted f or w

omen with dense br

eas

ts separ

& This inf

orma on w as not e xtr act ed dir ectly fr om the included s tudy but fr om pr evious public

aon on the same popula

on. N S: Not speciﬁed; BC: br eas t c ancer; DM: digit al mammogr aph y; DB T: digit al br eas t t omos yn thesis; vs: v er sus

(23)

Supplemen tary 2. R ec all r at e of DB T and DM in scr eening s tudies using tw o s tudy gr oup s b y out come de ﬁnion

(24)

Supplemen

(25)

(26)

Fr

om: Moher D

, Liber

a A, T

etzlaﬀ J

, Altman DG, The PRISMA Gr

oup (2009). Pr ef err ed R eporng It ems f or S ys tema c R eview s and Met Analy

ses: The PRISMA St

at

emen

(27)

Reference

1

(1) Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, Rosso S, Coebergh JW, Comber H, et al. Cancer incidence and mortality pa erns in Europe: es mates for 40 countries in 2012. Eur J Cancer 2013 Apr;49(6):1374-1403.

1

(2) Lauby-Secretan B, Scoccian C, Loomis D, Benbrahim-Tallaa L, Bouvard V, Bianchini F, et al. Breast-cancer screening--viewpoint of the IARC Working Group. N Engl J Med 2015 Jun 11;372(24):2353-2358.

1

(3) Chen L, Linden HM, Anderson BO, Li CI. Trends in 5-year survival rates among breast cancer pa ents by hormone receptor status and stage. Breast Cancer Res Treat 2014 Oct;147(3):609-616.

1

(4) Souza FH, Wendland EM, Rosa MI, Polanczyk CA. Is full-ﬁeld digital mammography more accurate than screen-ﬁlm mammography in overall popula on screening? A systema c review and meta-analysis. Breast 2013 Jun;22(3):217-224.

1

(5) D'Orsi CJ, Mendelson EB, Ikeda DM, et al. Breast Imaging Repor ng and Data System: ACR BI-RADS - breast imaging atlas . 2003.

1

(6) D'Orsi CJ, Sickles EA, Mendelson EB. ACR BI-RADS Atlas, Breast Imaging Repor ng and Data system. Reston, VA: American College of Radiology; 2013.

1

(7) Kerlikowske K, Zhu W, Hubbard RA, Geller B, Di us K, Braithwaite D, et al. Outcomes of screening mammography by frequency, breast density, and postmenopausal hormone therapy. JAMA Intern Med 2013 May 13;173(9):807-816.

1

(8) Sprague BL, Gangnon RE, Burt V, Trentham-Dietz A, Hampton JM, Wellman RD, et al. Prevalence of mammographically dense breasts in the United States. J Natl Cancer Inst 2014 Sep 12;106(10):10.1093/jnci/dju255. Print 2014 Oct.

1

(9) Boyd NF, Guo H, Mar n LJ, Sun L, Stone J, Fishell E, et al. Mammographic density and the risk and detec on of breast cancer. N Engl J Med 2007 Jan 18;356(3):227-236.

1

(10) Gilbert FJ, Tucker L, Young KC. Digital breast tomosynthesis (DBT): a review of the evidence for use as a screening tool. Clin Radiol 2016 Feb;71(2):141-150.

1

(11) Houssami N, Skaane P. Overview of the evidence on digital breast tomosynthesis in breast cancer detec on. Breast 2013 Apr;22(2):101-108.

1

(12) Shin SU, Chang JM, Bae MS, Lee SH, Cho N, Seo M, et al. Compara ve evalua on of average glandular dose and breast cancer detec on between single-view digital breast tomosynthesis (DBT) plus single-view digital mammography (DM) and two-view DM: correla on with breast thickness and density. Eur Radiol 2015 Jan;25(1):1-8.

1

(28)

mammography plus tomosynthesis in a popula on-based screening program. Radiology 2013 Apr;267(1):47-56.

1

(14) Houssami N, Turner RM. Rapid review: Es mates of incremental breast cancer detec on from tomosynthesis (3D-mammography) screening in women with dense breasts. Breast 2016 Oct 6;30:141-145.

1

(15) Durand MA, Philpo s LE. Can Tomosynthesis Replace 2D Mammography as a Future Breast Screening Tool?. Curr Breast Cancer Rep 2016;8:213-220.

1

(16) Bernardi D, Macaskill P, Pellegrini M, Valen ni M, Fanto C, Os llio L, et al. Breast cancer screening with tomosynthesis (3D mammography) with acquired or synthe c 2D mammography compared with 2D mammography alone (STORM-2): a popula on-based prospec ve study. Lancet Oncol 2016 Aug;17(8):1105-1113.

1

(17) Chae EY, Kim HH, Cha JH, Shin HJ, Choi WJ. Detec on and characteriza on of breast lesions in a selec ve diagnos c popula on: diagnos c accuracy study for comparison between one-view digital breast tomosynthesis and two-view full-ﬁeld digital mammography. Br J Radiol 2016 Jun;89(1062):20150743.

1

(18) Carbonaro LA, Di Leo G, Clauser P, Trimboli RM, Verardi N, Fedeli MP, et al. Impact on the recall rate of digital breast tomosynthesis as an adjunct to digital mammography in the screening se ng. A double reading experience and review of the literature. Eur J Radiol 2016 Apr;85(4):808-814.

1

(19) Cia o S, Houssami N, Bernardi D, Caumo F, Pellegrini M, Brunelli S, et al. Integra on of 3D digital mammography with tomosynthesis for popula on breast-cancer screening (STORM): a prospec ve comparison study. Lancet Oncol 2013 Jun;14(7):583-589.

1

(20) Conant EF, Beaber EF, Sprague BL, Herschorn SD, Weaver DL, Onega T, et al. Breast cancer screening using tomosynthesis in combina on with digital mammography compared to digital mammography alone: a cohort study within the PROSPR consor um. Breast Cancer Res Treat 2016 Feb;156(1):109-116.

1

(21) Gilbert FJ, Tucker L, Gillan MG, Willsher P, Cooke J, Duncan KA, et al. The TOMMY trial: a comparison of TOMosynthesis with digital MammographY in the UK NHS Breast Screening Programme--a mul centre retrospec ve reading study comparing the diagnos c performance of digital breast tomosynthesis and digital mammography with digital mammography alone. Health Technol Assess 2015 Jan;19(4):i-xxv, 1-136.

1

(22) Haas BM, Kalra V, Geisel J, Raghu M, Durand M, Philpo s LE. Comparison of tomosynthesis plus digital mammography and digital mammography alone for breast cancer screening. Radiology 2013 Dec;269(3):694-700.

1

(23) Lang K, Andersson I, Rosso A, Tingberg A, Timberg P, Zackrisson S. Performance of one-view breast tomosynthesis as a stand-alone breast cancer screening modality: results from the Malmo Breast Tomosynthesis Screening Trial, a popula on-based study. Eur Radiol 2016 Jan;26(1):184-190.

(29)

(24) McCarthy AM, Kontos D, Synnestvedt M, Tan KS, Heitjan DF, Schnall M, et al. Screening outcomes following implementa on of digital breast tomosynthesis in a general-popula on screening program. J Natl Cancer Inst 2014 Oct 13;106(11):10.1093/jnci/dju316. Print 2014 Nov.

1

(25) McDonald ES, Ous mov A, Weinstein SP, Synnestvedt MB, Schnall M, Conant EF. Eﬀec veness of Digital Breast Tomosynthesis Compared With Digital Mammography: Outcomes Analysis From 3 Years of Breast Cancer Screening. JAMA Oncol 2016 Jun 1;2(6):737-743.

1

(26) Raﬀerty EA, Durand MA, Conant EF, Copit DS, Friedewald SM, Plecha DM, et al. Breast Cancer Screening Using Tomosynthesis and Digital Mammography in Dense and Nondense Breasts. JAMA 2016 Apr 26;315(16):1784-1786.

1

(27) Rose SL, Tidwell AL, Bujnoch LJ, Kushwaha AC, Nordmann AS, Sexton R,Jr. Implementa on of breast tomosynthesis in a rou ne screening prac ce: an observa onal study. AJR Am J Roentgenol 2013 Jun;200(6):1401-1408.

1

(28) Sharpe RE,Jr., Venkataraman S, Phillips J, Dialani V, Fein-Zachary VJ, Prakash S, et al. Increased Cancer Detec on Rate and Varia ons in the Recall Rate Resul ng from Implementa on of 3D Digital Breast Tomosynthesis into a Popula on-based Screening Program. Radiology 2016 MAR 2016;278(3):698-706.

1

(29) Starikov A, Drotman M, Hentel K, Katzen J, Min RJ, Arleo EK. 2D mammography, digital breast tomosynthesis, and ultrasound: which should be used for the diﬀerent breast densi es in breast cancer screening? Clin Imaging 2016 Jan-Feb;40(1):68-71.

1

(30) Waldherr C, Cerny P, Alterma HJ, Berclaz G, Ciriolo M, Buser K, et al. Value of one-view breast tomosynthesis versus two-view mammography in diagnos c workup of women with clinical signs and symptoms and in women recalled from screening. AJR Am J Roentgenol 2013 Jan;200(1):226-231.

1

(31) Yun SJ, Ryu CW, Rhee SJ, Ryu JK, Oh JY. Beneﬁt of adding digital breast tomosynthesis to digital mammography for breast cancer screening focused on cancer characteris cs: a meta-analysis. Breast Cancer Res Treat 2017 May 18

(30)