6/7/2017 Disclosures Breast Density in the Tomosynthesis World - - PowerPoint PPT Presentation

▶
6 7 2017
SMART_READER_LITE
LIVE PREVIEW

6/7/2017 Disclosures Breast Density in the Tomosynthesis World - - PowerPoint PPT Presentation

Nov 19, 2022 330 likes •491 views

6/7/2017 Disclosures Breast Density in the Tomosynthesis World Travel support and speakers fees SOPHIA ZACKRISSON MD PH.D, ASSOCIATE PROFESSOR SENIOR LECTURER, DIAGNOSTIC RADIOLOGY Siemens Healthineers SKNE UNIVERSITY HOSPITAL

slide-1
SLIDE 1

6/7/2017 1

Breast Density in the Tomosynthesis World

SOPHIA ZACKRISSON MD PH.D, ASSOCIATE PROFESSOR SENIOR LECTURER, DIAGNOSTIC RADIOLOGY SKÅNE UNIVERSITY HOSPITAL MALMÖ SWEDEN

Disclosures

  • Travel support and speaker’s fees

Siemens Healthineers Astra Zeneca

Outline

① Prospective trials- results and differences ② Retrospective studies – added value? ③ DBT and density ④ What evidence is needed to switch to tomoscreening?

slide-2
SLIDE 2

6/7/2017 2

54-y-old woman 13 mm inv ductal ca grade 2; DCIS

DM DBT

slide-3
SLIDE 3

6/7/2017 3

The major limitation in 2D mammography is

  • verlapping tissue = anatomical noise = dense breast tissue

MAMMOGRAPHY CC MLO

ULTRASOUND

slide-4
SLIDE 4

6/7/2017 4

BT

MAMMOGRAPHY TOMOSYNTHESIS

3 cm invasive ductal carcinoma grade 3 with metastasis to axilla

Tomosynthesis is a better mammography

So why wait?

  • Evidence based medicine
  • European guidelines for breast cancer screening
  • -> screening of healthy women
  • -> majority do not have breast cancer
  • -> we must know what we are doing!

Outline

① Prospective trials- results and differences ② Retrospective studies – added value? ③ DBT and density ④ What evidence is needed to switch to tomoscreening?

slide-5
SLIDE 5

6/7/2017 5

1 Prospective trials - published

① STORM 1 (N=7,300) ② OSLO (interim N=12,500) ③ MBTST (interim N=7,500) ④ STORM 2 (N=9,600)

Ciatto et al, Lancet Oncol 2013 Houssami et al, Euro J Ca 2014 Skaane et al, Radiol 2013 Skaane et al, Euro Radiol 2013 Lång et al, Eur Radiol 2016 Lång et al, Eur Radiol 2016 Bernardi et al, Lancet Oncol 2016

Summary prospective trials

  • 30% increased cancer detection
  • DBT better in all density categories
  • Predominantly invasive cancers
  • Low base line recall rates
  • Acceptable recall rates
  • Prevalence effect

1 Prospective trials - ongoing

① OSLO (N=25000) ② STORM (N=7300) ③ MBTST (N=15000) ④ STORM 2 (N=9600) ⑤ TMIST, Canada/US? ⑥ TOBE, Bergen, Norway ⑦ Piemonte, Italy ⑧ Reggio Emilia, Italy ⑨ UK, Netherlands??…

Trial and publications Study design (n screens) Screen-reading practice & context

Ciatto et al 2013 [STORM trial] Lancet Oncology Houssami et al 2014 [STORM 12-month follow-up] Euro J Ca

  • Prospective trial (7,292) in Italian

screening program

  • 2D compared to integrated 2D/3D
  • Sequential read of 2D

then 2D/3D

  • Independent double-

reading

  • No arbitration

Bernardi et al 2016 [STORM 2 trial]

  • Prospective trial (9,672) in Italian

screening program

  • 2D compared to 3D + synthetic 2D and

integrated 2D/3D

  • Sequential read of 2D

then 2D/3D or 3D+ synthetic

  • Independent double-

reading

  • No arbitration

Skaane et al, 2013 [Interim report Oslo trial] Radiol Skaane et al 2013 Euro Radiol

  • Prospective trial (12,631 in interim,

25000 full study) in Norwegian screening program

  • 2D compared to several strategies:

2D+3D/3D+CAD/3D+ synthetic 2D

  • Four reading arms (2

readers per arm)

  • Independent double-

reading

  • Arbitration

Lång et al 2016 [Interim report MBTST]** Eur Radiol Lång et al 2016 [Interim report MBTST FP recall] Eur Radiol

  • Prospective trial (7,500 in interim

analysis, 15000 in full study) random sample invited in Swedish screening program

  • 2D versus stand-alone 1-view 3D
  • 1-view 3D
  • Independent double-

reading

  • Arbitration
slide-6
SLIDE 6

6/7/2017 6

Detection rate/1000

Double reading

Increased detection DM DM+DBT STORM

(Ciatto; Lancet Oncol 2013)

5.3 8.1 27% OTST

(Skaane; Eur Radiol 2013)

7.1 9.4 30% MBTST

(Lang; Eur Radiol 2015))

6.3 8.9

(1 view DBT)

43% STORM 2

(Bernardi; Lancet Oncol 2016)

DM DBT + synth DM 40% 6.3 8.8

Detection rates – prospective trials

Recall rate (%)

Double reading

Change DM DM+DBT STORM (estimated, not actual)

(Ciatto; Lancet Oncol 2013)

4.4 3.5 ↓ OTST

(Skaane; Eur Radiol 2013)

2.9 3.7 ↑ MBTST

(Lang; Eur Radiol 2015))

2.6 1 view DBT ↑ 3.8 STORM 2

(Bernardi; Lancet Oncol 2016)

DM DBT + synth DM ↑ 3.4* 4.5*

Recall rates – prospective trials

* Reported as false positive recall rate

  • Not comparable study design
  • Not representing “true screening situation”
  • Similar results
  • Prevalence effect

MAMMOGRAPHY TOMOSYNTHESIS FALSE POSITIVE 1-yr follow-up

MBTST

  • Women aged 40-74 ys
  • 1-view (MLO)
  • Wide-angle system
  • Reduced compression (30-50%)
  • Full study results 2017
slide-7
SLIDE 7

6/7/2017 7

Less compression

  • Vigorous (standard) compression vs half compression force
  • No significant difference in image quality
  • Pressure measurements

Standard 50% Distribution

Förnvik D et al. Radiat Prot Dosim 2010 Dustler M et al. Acta Radiol 2012

Outline

① Prospective trials- results and differences ② Retrospective studies – added value? ③ DBT and density ④ What evidence is needed to switch to tomoscreening?

2 Retrospective studies

  • 2D compared to 2D/3D
  • Compare different cohorts with 2D and

2D/3D

  • Different time periods
  • Annual screening
  • No specified inclusion criteria

Rose AJR 2013 Haas Radiology 2013 McCarthy JNCI 2014 Greenberg AJR 2014 Lourenco Radiol 2015 McDonald JAMA Onc 2016 Conant Br Res Treat 2016 Zuckerman Radiol 2016 (Synthetic 2D+DBT)

N DM vs DM+DBT Cancer detection/1000 women Recall rate % Change recall rate

Rose AJR 2013 13.000 vs 9.500 4.0 vs 5.4 8.7 vs 5.5 *

  • 37%

Friedewald JAMA 2014 270.000 vs 174.000 4.2 vs 5.4 * 11 vs 9.1 *

  • 17%

Haas Radiology 2013 13.100 vs 6.100 5.2 vs 5.7 12 vs 8.4 *

  • 30%

McCarthy JNCI 2014 10.700 vs 15.500 4.6 vs 5.5 10.4 vs 8.8 *

  • 15%

Greenberg AJR 2014 54.600 vs 23.100 4.9 vs 6.3 * 16.2 vs 13.6

  • 16%

Lourenco Radiol 2015 12.921 vs 12.577 5.4 vs 4.6 9.3 vs 6.4 *

  • 45%

McDonald JAMA Onc 2016 44.468 (repeated DBT screens) year 0-3 4.6 vs 5.5, 5.8, and 6.1 10.4 vs 8-9 *

  • 23 to -34%

Conant Br Res Treat 2016 142.883 vs. 55.998 5.9 vs 4.4* 10.4 vs 8.7*

  • 20%

* Statistically significant

Retrospective studies – DM vs DM+DBT

slide-8
SLIDE 8

6/7/2017 8

Retrospective studies – DM vs DM+DBT

  • + Large studies
  • - Selection bias
  • Not organized screening
  • None or small effect on detection rates
  • Higher base-line recall rates (8-16%) -> large reductions with DBT

Rose AJR 2013 Haas Radiology 2013 McCarthy JNCI 2014 Greenberg AJR 2014 Lourenco Radiol 2015 McDonald JAMA Onc 2016 Conant Br Res Treat 2016 Zuckerman S 2016 (Synthetic 2D + DBT)

Outline

① Prospective trials- results and differences ② Retrospective studies – added value? ③ DBT and density ④ What evidence is needed to switch to tomoscreening?

Cancer detection by density in MBTST (n=7,500)

Breast density BIRADS 4th Detected DBT total Detected DBT alone Detected DM total Detected DM alone Fatty (<25%) 7 (10) 3 (14) 4 (9) Scattered (25-50%) 17 (25) 5 (24) 12 (26) Heterogeneous (51-75%) 33 (49) 10 (48) 24 (51) 1 (100) Dense (>75%) 10 (15) 3 (14) 7 (15)

Lång 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 Malmö Breast Tomosynthesis Screening Trial, a population-based

  • study. Eur Radiol. 2016 Jan;26(1):184-90.

38% in non-dense breasts

68-y-old asymptomatic woman (screening))

CC MLO

slide-9
SLIDE 9

6/7/2017 9

Pathology: 1.0 cm ductal carcinoma gr 1

2D 3D Interim report of ASTOUND (Tagliafico et al JCO 2016)

Italian population-based: recruitment from 5 Italian centres with dedicated breast imaging facilities Eligible: Asymptomatic women self-referring to screening * Breast density 3-4 Age ≥ 38 years * Live density reporting to action tomosynthesis acquisition at same breast compression

Prof Nehmat Houssami

Detected 24 additional BCs 12 detected on tomosynthesis & ultrasound 1 dx only on tomosynthesis 11 dx only on ultrasound Incremental CDR tomosynthesis 4.0/1000 screens (1.8–6.2) versus ultrasound 7.1/1000 screens (4.2–10.0), P=0.006 Incremental FPR 3.33%

Interim report of ASTOUND (N=3,231)

Prof Nehmat Houssami

Tagliafico et al JCO 2016

slide-10
SLIDE 10

6/7/2017 10

ASTOUND, illustrative case comparative cancer detection

3

Case: MLO of left breast, 67-year-old

  • Mammogram not recalled (‘negative’).
  • Tomosynthesis (positive/recalled):

architectural distortion .

  • Ultrasound (positive): small irregular

hypoechoic mass/indistinct margins, posterior acoustic shadowing

  • Histology: 5mm invasive ductal

carcinoma, negative node

MAMMO TOMO US

Prof Nehmat Houssami

Interim report ASTOUND (Tagliafico et al JCO 2016)

Study limitations:

  • Study of screen-detection: Impact of adjunct screening not assessed

(reduce interval cancers ? More benefit vs more overdiagnosis?)

  • Radiologists were very experienced in ultrasound screening (FPR

data may not transfer to other settings)

  • First screening round with adjunct tomosynthesis versus mix of first

& repeat ultrasound screening

  • Detailed characteristics of women (other than age & density) not

known (not high-risk but may have included above population risk)

Prof Nehmat Houssami

Interim report ASTOUND (Tagliafico et al JCO 2016) Summary

  • Each of tomosynthesis & ultrasound detected BC in women with negative 2D-mammogram

& dense breasts

  • Ultrasound detected more BCs than tomosynthesis however >50% of additional BCs in 2D-

negative cohort detected by integrating tomosynthesis (3D) in same screening procedure

  • Tomosynthesis more feasible / more cost-effective (research needed)
  • What are the data on tomosynthesis screening in dense breasts?

3 Prof Nehmat Houssami

Other data on tomosynthesis versus US screening in dense breasts

Retrospective study with only subset of women with dense breasts receiving adjunct imaging – therefore prone to selection problems:

4

From: Starikov et al Clin Imaging 2015

Screens with dense breasts CDR/1000 screens Recall All screens N= 10915 4.7 18.6% 7117 who had 2D-mammograhy alone 3.8 19.9% 1875 who had 2D + tomosynthesis 5.3 10.4% 1397 who had 2D + Ultrasound 7.2 20.8% 526 who had 2D + tomosynthesis + Ultrasound 7.6 23.4%

Prof Nehmat Houssami

slide-11
SLIDE 11

6/7/2017 11

Tomosynthesis screening in dense breasts

  • Rapid review: streamlined evidence synthesis (emerging evidence)
  • 8 relevant population screening studies identified (July 2016)
  • 2 groups of studies reporting slightly different estimates (for CDR & recall)

4 Prof Nehmat Houssami

Tomosynthesis screening in dense breasts: prospective studies

(Houssami & Turner; Breast 2016) Study (author, year published) Recruitment timeframe Study population median (or mean) age; range, years Number of screens: heterogeneously dense or extremely dense breasts Effect on BC detection in dense breasts attributed to tomosynthesis per 1000 screens Effect on recall in dense breasts, per 1000 screens Prospective trials comparing tomosynthesis (2D+3D) screening with 2D-mammography alone in the same screening participants at same screening episode: Estimates represent incremental rates in the same women Ciatto 2013 (STORM)

2011-2012

58; 48-71 1215

+2.5/1000

  • 26.0/1000 (false recall

conditional analysis) Lång 2016 (Malmö)

2010-2012

56; 40-76 3150

+3.8/1000

NA* Bernardi 2016 (STORM 2)

2013-2015

58; 49-71 2592

+5.4/1000

+10.5/1000 (false recall) Tagliafico 2016 (ASTOUND)

2012-2015

51; 38-88 3231

+4.0/1000

NA*

Prof Nehmat Houssami

Tomosynthesis screening in dense breasts: prospective studies Incremental BC detection (Houssami & Turner, 2016)

4

Meta-analysis: Incremental CDR attributed to tomosynthesis in 10,188 screening participants with heterogeneously dense or extremely dense breasts (P<0.001 versus 2D alone)

Prof Nehmat Houssami

Tomosynthesis screening in dense breasts: Retrospective studies (Houssami & Turner; Breast 2016)

4

Retrospective studies comparing tomosynthesis (2D+3D) screening with 2D-mammography alone in different groups

  • f screening participants: Estimates of effect represent differences in rates between different groups of women

Number of screens with dense breasts Difference in cancer detection (for T-M) Difference in recall rates (for T-M) Rose 2013 M: 2010 T: 2011-2012 M (mean) 53.8 T (mean) 54.5 M: 7009 T: 4006

+1.4/1000

  • 36.8/1000

McCarthy 2014 M: 2010-2011 T: 2011-2013 M (mean) 56.9 T (mean) 56.7 M: 3489 T: 5056

+1.8/1000

  • 19.4/1000

Conant 2016 2011-2014 Range: 40-74 M: 35320 [44303]‡ T: 9265 [21133]‡

+2.1/1000

  • 22.1/1000

Rafferty 2016 2011-2012 Not reported M: 131996 T: 84243

+1.4/1000

  • 18.4/1000

Prof Nehmat Houssami

slide-12
SLIDE 12

6/7/2017 12

Tomosynthesis screening in dense breasts: Retrospective studies

(Houssami & Turner; Breast 2016)

4

Meta-analysis: difference in CDR between 103,230 tomosynthesis-screened versus 177,814 mammography- screened women with heterogeneously dense or extremely dense breasts (P<0.001)

Prof Nehmat Houssami

Tomosynthesis screening in dense breasts: Retrospective studies

(Houssami & Turner; Breast 2016)

  • Pooled difference of – 23.3 recalls/1000 screens, a significant (P<0.001) reduction in

recall from using tomosynthesis compared with 2D alone screening in dense breasts

Meta-analysis of recall data

4 Prof Nehmat Houssami

Measuring density on tomosynthesis?

  • Differences in breast density between modalities

(MRI, FFDM, DBT)

– Machida et al Springer Plus 2016, Pertuz et al Radiology 2015, Tagliafico et al Breast Cancer Res Treat 2013

  • Differences in breast density between software and radiologists

– Timberg et al Proceedings IWDM 2016, Ekpo et al The Breast 2016

Volumetric density in mammography vs tomosynthesis (central projection):

  • high correlation (r = 0.94) substantial agreement (κ = 0.76)

Substantial agreement between radiologists (BI-RADS 5th Ed) and software

  • mammography images (κ = 0.73) tomosynthesis images (κ =0.62)
slide-13
SLIDE 13

6/7/2017 13

Outline

① Prospective trials- results and differences ② Retrospective studies – added value? ③ DBT and density ④ What evidence is needed to switch to tomoscreening?

  • 3. What evidence is needed?

RCT with mortality endpoints?

  • 3. What evidence is needed?
  • Await full reports OSLO and MBTST
  • Interval cancer rates
  • Analysis of DBT-detected cancer characteristics
  • Incidence screening with DBT (OSLO cont, coming RCTs )
  • 3. What evidence is needed?

Practicalities…

  • Reading time doubled?
  • Double or single reading?
  • CAD?
  • Data storage
  • Cost-benefit analysis
slide-14
SLIDE 14

6/7/2017 14

DBT screening in my view

DBT only - NOT combo 2D/3D 1 or 2-view DBT (+ synthetic 2D for comparison with prior DM) Consistent evidence that DBT increases cancer detection in all breasts and especially in dense breasts DBT not as sensitive as MRI or US although more cost- effective on population level 11 mm invasive lobular cancer, grade 2, N0

Lund University Cancer Imaging Group, LUCI

Medical Physics

Anders Tingberg MSc, PhD Daniel Förnvik MSc, PhD Pontus Timberg MSc PhD Magnus Dustler MSc, PhD Hannie Petersson, MSc, PhD candidate

  • Hannie Petersson MSc, PhD

Diagnostic Radiology

Sophia Zackrisson MD PhD, PI Ingvar Andersson, MD PhD Kristina Lång, MD, PhD Hanna Sartor MD, PhD Kristin Johnsson, MD, PhD candidate Aldana Rosso, MSc, PhD, statistician

  • Svenska läkaresällskapet
  • Allmänna Sjukhusets i Malmö stiftelse för

bekämpande av cancer

  • Stiftelsen för forskning, Onk klin Malmö

Tack så mycket!

slide-15
SLIDE 15

6/7/2017 15