Articles

Integration of 3D digital mammography with

tomosynthesis for population breast-cancer screening
(STORM): a prospective comparison study
Stefano Ciatto†, Nehmat Houssami, Daniela Bernardi, Francesca Caumo, Marco Pellegrini, Silvia Brunelli, Paola Tuttobene, Paola Bricolo,
Carmine Fantò, Marvi Valentini, Stefania Montemezzi, Petra Macaskill

Background Digital breast tomosynthesis with 3D images might overcome some of the limitations of conventional 2D Lancet Oncol 2013; 14: 583–89
mammography for detection of breast cancer. We investigated the effect of integrated 2D and 3D mammography in Published Online
population breast-cancer screening. April 25, 2013
http://dx.doi.org/10.1016/
S1470-2045(13)70134-7
Methods Screening with Tomosynthesis OR standard Mammography (STORM) was a prospective comparative study.
UO Senologia Clinica e
We recruited asymptomatic women aged 48 years or older who attended population-based breast-cancer screening Screening Mammografico,
through the Trento and Verona screening services (Italy) from August, 2011, to June, 2012. We did screen-reading in Department of Diagnostics,
two sequential phases—2D only and integrated 2D and 3D mammography—yielding paired data for each screen. Azienda Provinciale Servizi
Sanitari, Trento, Italy
Standard double-reading by breast radiologists determined whether to recall the participant based on positive
(S Ciatto PhD, D Bernardi MD,
mammography at either screen read. Outcomes were measured from final assessment or excision histology. Primary M Pellegrini MD,
outcome measures were the number of detected cancers, the number of detected cancers per 1000 screens, the P Tuttobene MD, C Fantò MD,
number and proportion of false positive recalls, and incremental cancer detection attributable to integrated 2D and M Valentini MD); Centro di
Prevenzione Senologica,
3D mammography. We compared paired binary data with McNemar’s test.
Marzana, Verona, Italy (S Ciatto,
F Caumo MD, S Brunelli MD,
Findings 7292 women were screened (median age 58 years [IQR 54–63]). We detected 59 breast cancers (including P Bricolo MD,
52 invasive cancers) in 57 women. Both 2D and integrated 2D and 3D screening detected 39 cancers. We detected S Montemezzi MD); and
Screening and Test Evaluation
20 cancers with integrated 2D and 3D only versus none with 2D screening only (p<0·0001). Cancer detection rates were
Program, School of Public
5·3 cancers per 1000 screens (95% CI 3·8–7·3) for 2D only, and 8·1 cancers per 1000 screens (6·2–10·4) for integrated Health, Sydney Medical School,
2D and 3D screening. The incremental cancer detection rate attributable to integrated 2D and 3D mammography was University of Sydney, Sydney,
2·7 cancers per 1000 screens (1·7–4·2). 395 screens (5·5%; 95% CI 5·0–6·0) resulted in false positive recalls: 181 at NSW, Australia
(Prof N Houssami MBBS/PhD,
both screen reads, and 141 with 2D only versus 73 with integrated 2D and 3D screening (p<0·0001). We estimated that Prof P Macaskill PhD)
conditional recall (positive integrated 2D and 3D mammography as a condition to recall) could have reduced false
†Died May, 2012
positive recalls by 17·2% (95% CI 13·6–21·3) without missing any of the cancers detected in the study population.
Correspondence to:
Prof Nehmat Houssami,
Interpretation Integrated 2D and 3D mammography improves breast-cancer detection and has the potential to reduce School of Public Health, Sydney
false positive recalls. Randomised controlled trials are needed to compare integrated 2D and 3D mammography with Medical School, University of
Sydney, Sydney 2006, NSW,
2D mammography for breast cancer screening.
Australia
[email protected]
Funding National Breast Cancer Foundation, Australia; National Health and Medical Research Council, Australia;
Hologic, USA; Technologic, Italy.

Introduction low-dose radiographs are used to reconstruct a pseudo-

Although controversial, mammography screening is the 3D image of the breast.4–6
only population-level early detection strategy that has Initial clinical studies of 3D mammography, 6–10 though
been shown to reduce breast-cancer mortality in based on small or selected series, suggest that addition of
randomised trials.1,2 Irrespective of which side of the 3D to 2D mammography could improve cancer detection
mammography screening debate one supports,1–3 efforts and reduce the number of false positives. However,
should be made to investigate methods that enhance the previous assessments of breast tomosynthesis might have
quality of (and hence potential benefit from) mam- been constrained by selection biases that distorted the
mography screening. A limitation of standard 2D potential effect of 3D mammography; thus, screening
mammography is the superimposition of breast tissue or trials of integrated 2D and 3D mammography are needed.6
parenchymal density, which can obscure cancers or make We report the results of a large prospective study
normal structures appear suspicious. This shortcoming (Screening with Tomosynthesis OR standard Mammog-
reduces the sensitivity of mammography and increases raphy [STORM]) of 3D digital mammography. We investi-
false-positive screening. Digital breast tomosynthesis gated the effect of screen-reading using both standard 2D
with 3D images might help to overcome these limitations. and 3D imaging with tomosynthesis compared with
Several reviews4,5 have described the development of screening with standard 2D digital mammography only for
breast tomosynthesis technology, in which several population breast-cancer screening.

www.thelancet.com/oncology Vol 14 June 2013 583

 Articles

Methods analytical combinations. Radiologists had to record

Study design and participants whether or not to recall the participant at each screen-
STORM is a prospective population-screening study reading phase before progressing to the next phase of the
that compares mammography screen-reading in two sequence. For each screen, data were also collected for
sequential phases (figure)—2D only versus integrated 2D breast density (at the 2D screen-read), and the side and
and 3D mammography with tomosynthesis—yielding quadrant for any recalled abnormality (at each screen-
paired results for each screening examination. Women read). All eight radiologists were breast radiologists with
aged 48 years or older who attended population-based a mean of 8 years (range 3–13 years) experience in
screening through the Trento and Verona screening mammography screening, and had received basic
services, Italy, from August, 2011, to June, 2012, were training in integrated 2D and 3D mammography. Several
invited to be screened with integrated 2D and 3D of the radiologists had also used 2D and 3D
mammography. Participants in routine screening mammography for patients recalled after positive
mammography (once every 2 years) were asymptomatic conventional mammography screening as part of
women at standard (population) risk for breast cancer. previous studies of tomosynthesis.8,13
The study was granted institutional ethics approval at each Mammograms were interpreted in two independent
centre, and participants gave written informed consent. screen-reads done in parallel, as practised in most
Women who opted not to participate in the study received population breast-screening programmes in Europe. A
standard 2D mammography. Digital mammography has screen was considered positive and the woman recalled
been used in the Trento breast-screening programme for further investigations if either screen-reader recorded
since 2005, and in the Verona programme since 2007; a positive result at either 2D or integrated 2D and 3D
each service monitors outcomes and quality indicators as screening (figure). When previous screening
dictated by European standards, and both have published mammograms were available, these were shown to the
data for screening performance.11,12 radiologist at the time of screen-reading, as is standard
practice. For assessment of breast density, we used Breast
Procedures Imaging Reporting and Data System (BI-RADS)14
All participants had digital mammography using a classification, with participants allocated to one of two
Selenia Dimensions Unit with integrated 2D and groups (1–2 [low density] or 3–4 [high density]).
3D mammography done in the COMBO mode (Hologic, Disagreement between readers about breast density was
Bedford, MA, USA): this setting takes 2D and 3D images resolved by assessment by a third reader.
at the same screening examination with a single breast Our primary outcomes were the number of cancers
position and compression. Each 2D and 3D image detected, the number of cancers detected per
consisted of a bilateral two-view (mediolateral oblique 1000 screens, the number and percentage of false posi-
and craniocaudal) mammogram. Screening mammo- tive recalls, and the incremental cancer detection rate
grams were interpreted sequentially by radiologists, first attributable to integrated 2D and 3D mammography
on the basis of standard 2D mammography alone, and screening. We compared the number of cancers that
then by the same radiologist (on the same day) on the were detected only at 2D mammography screen-reading
basis of integrated 2D and 3D mammography (figure). and those that were detected only at 2D and 3D
Thus, integrated 2D and 3D mammography screening mammography screen-reading; we also did this analysis
refers to non-independent screen reading based on joint for false positive recalls. To explore the potential effect of
interpretation of 2D and 3D images, and does not refer to integrated 2D and 3D screening on false-positive recalls,
we also estimated how many false-positive recalls would
have resulted from using a hypothetical conditional false-
Enrolment and mammography
positive recall approach; ie—positive integrated 2D and
3D mammography as a condition of recall (screening
recalled at 2D mammography only would not be recalled).
Sequential screen-reading phase 1: Sequential screen-reading phase 1:
radiologist A reports whether to recall or not radiologist B reports whether to recall or not Pre-planned secondary analyses were comparison of
based on standard 2D mammograms only based on standard 2D mammograms only outcome measures by age group and breast density.
Outcomes were assessed by excision histology for
participants who had surgery, or the complete assessment
Sequential screen-reading phase 2: Sequential screen-reading phase 2:
radiologist A reports whether to recall or not radiologist B reports whether to recall or not outcome (including investigative imaging with or
based on 2D and 3D mammograms based on 2D and 3D mammograms without histology from core needle biopsy) for all recalled
participants. Because our study focuses on the difference
in detection by the two screening methods, some cancers
Recall (positive screen) based on decision to recall might have been missed by both 2D and integrated 2D
by either reader at either screen-reading phase and 3D mammography; this possibility could be assessed
at future follow-up to identify interval cancers. However,
Figure: Study design this outcome is not assessed in the present study and

584 www.thelancet.com/oncology Vol 14 June 2013

 Articles

does not affect estimates of our primary outcomes—ie,

Cancers Cancers
comparative true or false positive detection for 2D-only detected at detected
versus integrated 2D and 3D mammography. both 2D only only at
and integrated integrated
2D and 3D 2D and 3D
Statistical analysis screening screening
The sample size was chosen to provide 80% power to (n=39) (n=20)
detect a difference of 20% in cancer detection, assuming pT category
a detection probability of 80% for integrated 2D and pTis (in-situ cancer) 4 (10%) 3 (15%)
3D screening mammography and 60% for 2D only pT1a (≤5 mm) 3 (8%) 0 (0%)
screening, with a two-sided significance threshold of 5%.
pT1b (>5 mm but ≤10 mm) 10 (26%) 8 (40%)
Based on the method of Lachenbruch15 for estimating
pT1c (>10 mm but ≤20 mm) 20 (51%) 8 (40%)
sample size for studies that use McNemar’s test for
pT2 (>20 mm but ≤50 mm) 2 (5%) 1 (5%)
paired binary data, a minimum of 40 cancers were
Node status
needed. Because most screens in the participating
Negative for metastases 24 (62%) 11 (55%)
centres were incident (repeat) screening (75%–80%),
Positive for metastases 9 (23%) 3 (15%)
we used an underlying breast-cancer prevalence of 0·5%
Micrometastases or isolated tumour cells 2 (5%) 1 (5%)
to estimate that roughly 7500–8000 screens would be
No node surgery or data not available 4 (10%) 5 (25%)
needed to identify 40 cancers in the study population.
We calculated the Wilson CI for the false-positive recall Data are n (%). pT=pathological tumour size.
ratio for integrated 2D and 3D screening with conditional
Table 1: Breast-cancer characteristics
recall compared with 2D only screening.16 All of the other
analyses were done with SAS/STAT (version 9.2), using
exact methods to compute 95 CIs and p-values.
Breast cancer No breast cancer (including false recalls)
Integrated 2D Integrated 2D Total Integrated 2D Integrated 2D Total
Role of the funding source and 3D positive and 3D negative and 3D positive and 3D negative
The sponsors of the study had no role in study design,
2D positive 39 0 39 181 141 322
data collection, data analysis, data interpretation, or
2D negative 20 0* 20 73 6840 6913
writing of the report. The corresponding author (NH)
had full access to all the data in the study and had final Total 59 0 59 254 6981 7235

responsibility for the decision to submit for publication. p value <0·0001 <0·0001

p values are exact for McNemar’s test for paired binary data. *Does not include follow-up data on interval cancers.
Results
Table 2: Outcomes of screening
7292 participants with a median age of 58 years (IQR
54–63, range 48–71) were screened between Aug 12, 2011,
and June 29, 2012. Roughly 5% of invited women
Number Cancer detection p value Incremental cancer
declined integrated 2D and 3D screening and received of cancers rate (cancers per detection rate
standard 2D mammography. We present data for 1000 screens; attributed to
7294 screens because two participants had bilateral 95% CI) integrated 2D and 3D
screening (95% CI)
cancer (detected with different screen-reading
techniques for one participant). We detected 59 breast Overall (7294 screens)
cancers in 57 participants (52 invasive cancers and seven 2D mammography 39 5·3 (3·8–7·3) ·· ··
ductal carcinoma in-situ). Of the invasive cancers, most Integrated 2D and 3D mammography 59 8·1 (6·2–10·4) <0·0001 2·7 (1·7–4·2)
were invasive ductal (n=37); others were invasive special Age <60 years (4044 screens)
types (n=7), invasive lobular (n=4), and mixed invasive 2D mammography 20 4·9 (3·0–7·6) ·· ··
types (n=4). Table 1 shows the characteristics of the Integrated 2D and 3D mammography 27 6·7 (4·4–9·7) 0·016 1·7 (0·7–3·6)
cancers. Mean tumour size (for the invasive cancers with Age ≥60 years (3250 screens)
known exact size) was 13·7 mm (SD 5·8) for cancers 2D mammography 19 5·8 (3·5–9·1) ·· ··
detected with both 2D alone and integrated 2D and 3D Integrated 2D and 3D mammography 32 9·8 (6·7–13·9) <0·0001 4·0 (2·1–6·8)
screening (n=29), and 13·5 mm (SD 6·7) for cancers Breast density 1–2 (6079 screens)
detected only with integrated 2D and 3D screening 2D mammography 34 5·6 (3·9–7·8) ·· ··
(n=13). Integrated 2D and 3D mammography 51 8·4 (6·3–11·0) <0·0001 2·8 (1·6–4·5)
Of the 59 cancers, 39 were detected at both 2D and Breast density 3–4 (1215 screens)
integrated 2D and 3D screening (table 2). 20 cancers were 2D mammography 5 4·1 (3·1–9·6) ·· ··
detected with only integrated 2D and 3D screening Integrated 2D and 3D mammography 8 6·6 (4·1–18·6) 0·25 2·5 (0·5–7·2)
compared with none detected with only 2D screening
Table 3: Breast-cancer detection rates, and incremental detection from integrated 2D and 3D screening
(p<0·0001; table 2). 395 screens were false positive (5·5%, mammography
95% CI 5·0–6·0); 181 occurred at both screen-readings,

www.thelancet.com/oncology Vol 14 June 2013 585

 Articles

(95% CI 1·4–3·8). The stratified results show that

Age <60 years Age ≥60 years
integrated 2D and 3D mammography was associated
Integrated Integrated Total Integrated Integrated Total with an incrementally increased cancer detection rate in
2D and 3D 2D and 3D 2D and 3D 2D and 3D
positive negative positive negative both age-groups and density categories (tables 3–5). A
minority (16·7%) of breasts were of high density (category
Breast cancer
3–4) reducing the power of statistical comparisons in this
2D positive 20 0 20 19 0 19
subgroup (table 5). The incremental cancer detection rate
2D negative 7 0* 7 13 0* 13
was much the same in low density versus high density
Total 27 0 27 32 0 32
groups (2·8 per 1000 vs 2·5 per 1000; p=0·84; table 3).
p value 0·016 <0·0001
Overall recall—any recall resulting in true or false
No breast cancer (including false recalls)
positive screens—was 6·2% (95% CI 5·7–6·8), and the
2D positive 129 89 218 52 52 104
false-positive rate for the 7235 screens of participants
2D negative 41 3758 3799 32 3082 3114
who did not have breast cancer was 5·5% (5·0–6·0).
Total 170 3847 4017 84 3134 3218 Table 6 shows the contribution to false-positive recalls
p value <0·0001 0·038 from 2D mammography only, integrated 2D and 3D
p values are exact for McNemar’s test for paired binary data. *Does not include follow-up data on interval cancers, this mammography only, and both, and the estimated
does not affect the comparative detection data. number of false positives if positive integrated 2D and
3D mammography was a condition for recall (positive 2D
Table 4: Outcomes of screening, stratified by age
only not recalled). Overall, more of the false-positive rate
was driven by 2D mammography only than by integrated
Breast density 1–2 (low density) Breast density 3–4 (high density) 2D and 3D, although almost half of the false-positive rate
was a result of false positives recalled at both screen-
Integrated Integrated Total Integrated Integrated Total
2D and 3D 2D and 3D 2D and 3D 2D and 3D reading phases (table 6). The findings were much the
positive negative positive negative same when stratified by age and breast density (table 6).
Breast cancer Had a conditional recall rule been applied, we estimate
2D positive 34 0 34 5 0 5 that the false-positive rate would have been 3·5% (95% CI
2D negative 17 0* 17 3 0* 3
3·1–4·0%; table 6) and could have potentially prevented
Total 51 0 51 8 0 8
68 of the 395 false positives (a reduction of 17·2%;
p value <0·0001 0·25
95% CI 13·6–21·3). The ratio between the number of
No breast cancer (including false recalls)
false positives with integrated 2D and 3D screening with
conditional recall (n=254) versus 2D only screening
2D positive 130 109 239 51 32 83
(n=322) was 0·79 (95% CI 0·71–0·87).
2D negative 52 5737 5789 21 1103 1124
Total 182 5846 6028 72 1135 1207
p value <0·0001 0·17
Discussion
Our study showed that integrated 2D and 3D mam-
p values are exact for McNemar’s test for paired binary data. *Does not include follow-up data on interval cancers, this mography screening significantly increases detection of
does not affect the comparative detection data. breast cancer compared with conventional mammog-
Table 5: Outcomes of screening, stratified by breast density raphy screening. There was consistent evidence of an
incremental improvement in detection from integrated
2D and 3D mammography across age-group and breast
and 141 occurred at 2D screening only compared with density strata, although the analysis by breast density was
73 at integrated 2D and 3D screening (p<0·0001; table 2). limited by low number of women with breasts of high
These differences were still significant in sensitivity density.
analyses that excluded the two participants with bilateral One should note that we investigated comparative
cancer (data not shown). cancer detection, and not absolute screening sensitivity.
5·3 cancers per 1000 screens (95% CI 3·8–7·3; table 3) By integrating 2D and 3D mammography using the
were detected with 2D mammography only versus study screen-reading protocol, 1% of false-positive recalls
8·1 cancers per 1000 screens (95% CI 6·2–10·4) with resulted from 2D and 3D screen-reading only (table 6).
integrated 2D and 3D mammography (p<0·0001). The However, significantly more false positives resulted from
incremental cancer detection rate attributable to inte- 2D only mammography compared with integrated 2D
grated 2D and 3D screening was 2·7 cancers per and 3D mammography, both overall and in the stratified
1000 screens (95% CI 1·7–4·2), which is 33·9% (95% CI analyses. Application of a conditional recall rule would
22·1–47·4) of the cancers detected in the study popu- have resulted in a false-positive rate of 3·5% instead of
lation. In a sensitivity analysis that excluded the two the actual false-positive rate of 5·5%. The estimated false
participants with bilateral cancer the estimated incre- positive recall ratio of 0·79 for integrated 2D and 3D
mental cancer detection rate attributable to integrated 2D screening with conditional recall compared with 2D only
and 3D screening was 2·6 cancers per 1000 screens screening suggests that integrated 2D and 3D screening

586 www.thelancet.com/oncology Vol 14 June 2013

 Articles

could reduce false recalls by roughly a fifth. Had such a

n % (95% CI)
condition been adopted, none of the cancers detected in
the study would have been missed because no cancers Overall (72 35 screens*)

were detected by 2D mammography only, although this Recalled at either 2D or integrated 2D and 3D mammography 395 5·5% (5·0–6·0)
result might be because our design allowed an Recalled at both 2D and integrated 2D and 3D mammography 181 2·5% (2·2–2·9)
independent read for 2D only mammography whereas Recalled at 2D mammography only 141 2·0% (1·6–2·3)
the integrated 2D and 3D read was an interpretation of a Recalled at integrated 2D and 3D mammography only 73 1·0% (0·8–1·3)
combination of 2D and 3D imaging. We do not Conditional false positive recalls† 254 3·5% (3·1–4·0)
recommend that such a conditional recall rule be used in Age <60 years (4017 screens)
breast-cancer screening until our findings are replicated Recalled at either 2D or integrated 2D and 3D mammography 259 6·5% (5·7–7·3)
in other mammography screening studies—STORM Recalled at both 2D and integrated 2D and 3D mammography 129 3·2% (2·7–3·8)
involved double-reading by experienced breast Recalled at 2D mammography only 89 2·2% (1·8–2·7)
radiologists, and our results might not apply to other Recalled at integrated 2D and 3D mammography only 41 1·0% (0·7–1·4)
screening settings. Using a test set of 130 mammograms, Conditional false positive recalls† 170 4·2% (3·6–4·9)
Wallis and colleagues7 report that adding tomosynthesis Age ≥60 years (3218 screens)
to 2D mammography increased the accuracy of Recalled at either 2D or integrated 2D and 3D mammography 136 4·2% (3·6–5·0)
inexperienced readers (but not of experienced readers), Recalled at both 2D and integrated 2D and 3D mammography 52 1·6% (1·2–2·1)
therefore having experienced radiologists in STORM Recalled at 2D mammography only 52 1·6% (1·2–2·1)
could have underestimated the effect of integrated 2D Recalled at integrated 2D and 3D mammography only 32 1·0% (0·7–1·4)
and 3D screen-reading. Conditional false positive recalls† 84 2·6% (2·1–3·2)
No other population screening trials of integrated Breast density 1–2 (6028 screens)
2D and 3D mammography have reported final results Recalled at either 2D or integrated 2D and 3D mammography 291 4·8% (4·3–5·4)
(panel); however, an interim analysis of the Oslo trial17—a Recalled at both 2D and integrated 2D and 3D mammography 130 2·2% (1·8–2·6)
large population screening study—has shown that Recalled at 2D mammography only 109 1·8% (1·5–2·2)
integrated 2D and 3D mammography substantially Recalled at integrated 2D and 3D mammography only 52 0·9% (0·6–1·1)
increases detection of breast cancer. The Oslo study Conditional false positive recalls† 182 3·0% (2·6–3·5)
investigators screened women with both 2D and Breast density 3–4 (1207 screens)
3D mammography, but randomised reading strategies Recalled at either 2D or integrated 2D and 3D mammography 104 8·6% (7·1–10·3)
(with vs without 3D mammograms) and adjusted for the Recalled at both 2D and integrated 2D and 3D mammography 51 4·2% (3·2–5·5)
different screen-readers,17 whereas we used sequential
Recalled at 2D mammography only 32 2·7% (1·8–3·7)
screen-reading to keep the same reader for each exam-
Recalled at integrated 2D and 3D mammography only 21 1·7% (1·1–2·7)
ination. Our estimates for comparative cancer detection
Conditional false positive recalls† 72 6·0% (4·7–7·5)
and for cancer detection rates are consistent with those
of the interim analysis of the Oslo study.17 The applied *Did not have breast cancer. †False-positive recalls using positive integrated 2D and 3D mammography as a condition
recall methods differed between the Oslo study (which to recall (ie—positive 2D mammography only would not be recalled).

used an arbitration meeting to decide recall) and the Table 6: False-positive recalls for mammography screening
STORM study (we recalled based on a decision by either
screen-reader), yet both studies show that 3D mammog-
raphy reduces false-positive recalls when added to 2D images. Second, although most screening in STORM
standard mammography. was incident screening, the substantial increase in cancer
An editorial in The Lancet18 might indeed signal the detection rate with integrated 2D and 3D mammography
closing of a chapter of debate about the benefits and results from the enhanced sensitivity of integrated 2D and
harms of screening. We hope that our work might be the 3D screening and is probably also a result of a prevalence
beginning of a new chapter for mammography screening: effect (ie, the effect of a first screening round with
our findings should encourage new assessments of integrated 2D and 3D mammography). We did not assess
screening using 2D and 3D mammography and should the effect of repeat (incident) screening with integrated
factor several issues related to our study. First, we 2D and 3D mammography on cancer detection—it might
compared standard 2D mammography with integrated provide a smaller effect on cancer detection rates than
2D and 3D mammography—the 3D mammograms were what we report. Third, STORM was not designed to
not interpreted independently of the 2D mammograms— measure biological differences between the cancers
therefore 3D mammography only (without the 2D images) detected at integrated 2D and 3D screening compared
might not provide the same results. Our experience with with those detected at both screen-reading phases.
breast tomosynthesis—and a review6 of 3D Descriptive analyses suggest that, generally, breast cancers
mammography—underscore the importance of detected only at integrated 2D and 3D screening had
2D images in integrated 2D and 3D screen-reading. similar features (eg, histology, pathological tumour size,
The 2D images form the basis of the radiologist’s ability to node status) as those detected at both screen-reading
integrate the information from 3D images with that from phases. Thus, some of the cancers detected only at 2D and

www.thelancet.com/oncology Vol 14 June 2013 587

 Articles

cancer rates from our study will only apply to breast

Panel: Research in context cancers that were not identified using 2D only or
Systematic review integrated 2D and 3D screening. We know of two patients
NH searched the English language published work up to October, 2012, to assess the from our study who have developed interval cancers
evidence of the accuracy of digital breast tomosynthesis (3D mammography), using the (follow-up range 8–16 months). We did not get this
search strategy and eligibility criteria detailed in a systematic review.6 The search information from cancer registries and follow-up was
consisted of a Medline search (exploded ”breast neoplasm”, combined with “tomosyn$” very short, so these data should be interpreted very
in title) and contact with experts. We did not identify any randomised controlled trials of cautiously, especially because interval cancers would be
digital breast tomosynthesis, or any population screening trials of digital breast expected to occur in the second year of the standard 2 year
tomosynthesis that had reported final results.6 Studies of digital breast tomosynthesis interval between screening rounds. Studies of interval
were generally small test-set (observer) studies with a high proportion of patients with cancer rates after integrated 2D and 3D mammography
cancer or were based on selected clinical series with various limitations, however these would need to be randomised controlled trials and have a
studies provided evidence that addition of digital breast tomosynthesis to standard very large sample size. Additionally, the development of
mammography increases the accuracy of interpretation.6 An interim analysis of the Oslo reconstructed 2D images from a 3D mammogram23
trial—a large screening trial in which women had both standard mammography and provides a timely solution to concerns about radiation by
digital breast tomosynthesis with randomly assigned screen-reading strategies—reported providing both the 2D and 3D images from tomosynthesis,
that the addition of digital breast tomosynthesis to digital mammography significantly eliminating the need for two acquisitions.
increased detection of breast cancer and reduced false positive recalls.17 We have shown that integrated 2D and 3D mammog-
raphy in population breast-cancer screening increases
Interpretation detection of breast cancer and can reduce false-positive
Our work is, to the best of our knowledge, the first population breast-cancer screening trial recalls depending on the recall strategy. Our results do
of integrated 2D and 3D mammography screening to report its final results. Our findings— not warrant an immediate change to breast-screening
that integrated 2D and 3D screening significantly increased breast cancer detection (relative practice, instead, they show the urgent need for random-
to standard 2D mammography) and has the potential to reduce false-positive recalls— ised controlled trials of integrated 2D and 3D versus 2D
accord with the interim analysis of the Oslo trial.17 These two screening trials provide mammography, and for further translational research in
evidence that integrated 2D and 3D mammography screening could be used as part of breast tomosynthesis. We envisage that future screening
population screening, provided that screening outcome measures are carefully monitored. trials investigating this issue will include measures of
To assess whether the measured increase in breast cancer detection with integrated 2D and breast cancer detection, and will be designed to assess
3D mammography is likely to translate into improved screening efficacy, a randomised interval cancer rates as a surrogate endpoint for screening
controlled trial is needed to compare integrated 2D and 3D mammography with standard efficacy.
2D mammography with interval cancer rates as the endpoint.
Contributors
SC had the idea for and designed the study, and collected and interpreted
3D screening might represent early detection (and would data. NH advised on study concepts and methods, analysed and
interpreted data, searched the published work, and wrote and revised the
be expected to receive screening benefit) whereas some report. DB and FC were lead radiologists, recruited participants,
might represent over-detection and a harm from collected data, and commented on the draft report. MP, SB, PT, PB, PT,
screening, as for conventional screening mammography.1,19 CF, and MV did the screen-reading, collected data, and reviewed the
The absence of consensus about over-diagnosis in breast- draft report. SM collected data and reviewed the draft report. PM
planned the statistical analysis, analysed and interpreted data, and wrote
cancer screening should not detract from the importance and revised the report.
of our study findings to applied screening research and to
Conflicts of interest
screening practice; however, our trial was not done to SC, DB, FC, MP, SB, PT, PB, CF, MV, and SM received assistance from
assess the extent to which integrated 2D and 3D mam- Hologic (Hologic USA; Technologic Italy) in the form of tomosynthesis
mography might contribute to over-diagnosis. technology and technical support for the duration of the study, and travel
The average dose of glandular radiation from the many support to attend collaborators’ meetings. NH receives research support
from a National Breast Cancer Foundation (NBCF Australia) Practitioner
low-dose projections taken during a single acquisition of Fellowship, and has received travel support from Hologic to attend a
3D mammography is roughly the same as that from collaborators’ meeting. PM receives research support through Australia’s
2D mammography.6,20–22 Using integrated 2D and 3D en- National Health and Medical Research Council programme grant 633003
tails both a 2D and 3D acquisition in one breast com- to the Screening & Test Evaluation Program.

pression, which roughly doubles the radiation dose to the References

1 Independent UK Panel on Breast Cancer Screening. The benefits
breast. Therefore, integrated 2D and 3D mammography and harms of breast cancer screening: an independent review.
for population screening might only be justifiable if Lancet 2012; 380: 1778–86.
improved outcomes were not defined solely in terms of 2 Glasziou P, Houssami N. The evidence base for breast cancer
screening. Prev Med 2011; 53: 100–102.
improved detection. For example, it would be valuable to
3 Autier P, Esserman LJ, Flowers CI, Houssami N. Breast cancer
show that the increased detection with integrated 2D and screening: the questions answered. Nat Rev Clin Oncol 2012;
3D screening leads to reduced interval cancer rates at 9: 599–605.
follow-up. A limitation of our study might be that data for 4 Baker JA, Lo JY. Breast tomosynthesis: state-of-the-art and review of
the literature. Acad Radiol 2011; 18: 1298–310.
interval cancers were not available; however, because of 5 Helvie MA. Digital mammography imaging: breast tomosynthesis
the paired design we used, future evaluation of interval and advanced applications. Radiol Clin North Am 2010; 48: 917–29.

588 www.thelancet.com/oncology Vol 14 June 2013

 Articles

6 Houssami N, Skaane P. Overview of the evidence on digital breast 15 Lachenbruch PA. On the sample size for studies based on
tomosynthesis in breast cancer detection. Breast 2013; 22: 101–08. McNemar’s test. Stat Med 1992; 11: 1521–25.
7 Wallis MG, Moa E, Zanca F, Leifland K, Danielsson M. Two-view 16 Bonett DG, Price RM. Confidence intervals for a ratio of binomial
and single-view tomosynthesis versus full-field digital proportions based on paired data. Stat Med 2006; 25: 3039–47.
mammography: high-resolution X-ray imaging observer study. 17 Skaane P, Bandos AI, Gullien R, et al. Comparison of digital
Radiology 2012; 262: 788–96. mammography alone and digital mammography plus
8 Bernardi D, Ciatto S, Pellegrini M, et al. Prospective study of breast tomosynthesis in a population-based screening program. Radiology
tomosynthesis as a triage to assessment in screening. 2013; published online Jan 3. http://dx.doi.org/10.1148/
Breast Cancer Res Treat 2012; 133: 267–71. radiol.12121373.
9 Michell MJ, Iqbal A, Wasan RK, et al. A comparison of the accuracy 18 The Lancet. The breast cancer screening debate: closing a chapter?
of film-screen mammography, full-field digital mammography, and Lancet 2012; 380: 1714.
digital breast tomosynthesis. Clin Radiol 2012; 67: 976–81. 19 Biesheuvel C, Barratt A, Howard K, Houssami N, Irwig L. Effects of
10 Skaane P, Gullien R, Bjorndal H, et al. Digital breast tomosynthesis study methods and biases on estimates of invasive breast cancer
(DBT): initial experience in a clinical setting. Acta Radiol 2012; overdetection with mammography screening: a systematic review.
53: 524–29. Lancet Oncol 2007; 8: 1129–38.
11 Pellegrini M, Bernardi D, Di MS, et al. Analysis of proportional 20 Tagliafico A, Astengo D, Cavagnetto F, et al. One-to-one comparison
incidence and review of interval cancer cases observed within the between digital spot compression view and digital breast
mammography screening programme in Trento province, Italy. tomosynthesis. Eur Radiol 2012; 22: 539–44.
Radiol Med 2011; 116: 1217–25. 21 Tingberg A, Fornvik D, Mattsson S, Svahn T, Timberg P,
12 Caumo F, Vecchiato F, Pellegrini M, Vettorazzi M, Ciatto S, Zackrisson S. Breast cancer screening with tomosynthesis—initial
Montemezzi S. Analysis of interval cancers observed in an Italian experiences. Radiat Prot Dosimetry 2011; 147: 180–83.
mammography screening programme (2000–2006). Radiol Med 22 Feng SS, Sechopoulos I. Clinical digital breast tomosynthesis
2009; 114: 907–14. system: dosimetric characterization. Radiology 2012; 263: 35–42.
13 Bernardi D, Ciatto S, Pellegrini M, et al. Application of breast 23 Gur D, Zuley ML, Anello MI, et al. Dose reduction in digital breast
tomosynthesis in screening: incremental effect on mammography tomosynthesis (DBT) screening using synthetically reconstructed
acquisition and reading time. Br J Radiol 2012; 85: e1174–78. projection images: an observer performance study. Acad Radiol
14 American College of Radiology. ACR BI-RADS: breast imaging 2012; 19: 166–71.
reporting and data system, Breast Imaging Atlas. Reston: American
College of Radiology, 2003.

www.thelancet.com/oncology Vol 14 June 2013 589