Tomosynthesis and breast imaging update Michael J Michell... · Tomosynthesis and breast imaging update Dr Michael J Michell Consultant Radiologist King's College Hospital NHS Foundation

Tomosynthesis and breast imaging update Dr Michael J Michell Consultant Radiologist King's College Hospital NHS Foundation Trust

Breast imaging – new technology

BREAST CANCER – FLT PET shows different grades of tumour with the primary site Images courtesy of Wai Lup Wong, Paul Strickland Cancer Centre

Mount Vernon

Breast imaging:application of new technology

screening diagnosis/

assessment

local

staging

axilla response

to RX

systemic

disease

XRM ++ ++ ++

TOMO ? ++ ++ ?

CE-XRM ? ?

US ? ++ ++ ++ ++

CE-MRI ++ ++ ++

F MRI

MOL

+

CT/PET ++

Digital Breast Tomosynthesis

• Technology

• Accuracy

• Diagnosis/assessment

• Screening

Limitations of 2D mammography

• 3D information projected onto 2D image plane

superimposition of tissues

Clinical effect

- false positive cases screening recall

- false negative cases interval cancers

- mammographic features obscured

diagnostic uncertainty

understaging

DBT - technology

Hologic dimensions - 15 deg - 15 proj

Siemens innovation - 50 deg - 25 proj

GE - 40 deg - 15 proj

Sectra - 11 - continuous

IMS - 40 deg - non uniform sampling

DBT radiation dose 2V = up to 5 mGy . FFDM 2V = 4 mGy

reconstructed images are

displayed as 1mm slices

for viewing singly or in cine mode

number of slices varies

according to breast thickness

Does DBT improve diagnostic accuracy?

Accuracy of DBT recall from screening

bilateral 2D + tomo (combo)MLO/CC

Michell MJ et al, Clin Rad 2012; 67: 976-81

soft tissue lesions

microcalcification

Michell MJ et al Clin Rad 2012

DBT - technical developments

• tomo guided biopsy

• synthetic 2D

• contrast tomo

• combined imaging

• CAD

Tomosynthesis procedure - Case 2 Target

Synthetic 2D

the information from the

reconstructed tomo slices is

used to produce a 2D image

DBT - synthetic 2D

standard 2D synthetic 2D DBT

2D synthetic 2D

Synthetic 2D combined with DBT has

equivalent accuracy compared to FFDM

+ DBT

Skaane P et al Radiology. 2014, Jan

Zuley M et al Radiology.2014, Jan

• images are courtesy of the following sites:

– Rose Medical Center, Denver, CO, USA

– Hôpital Privé d’Antony, Antony, FR

– Kaohsiung Veterans General Hospital, Taiwan

Contrast Mammography

2D CE2D Sub 3D Slice

Assessment of screen detected lesions

Questions

1. is there a lesion?

2. where is it?

3. what is the risk of malignancy?

4. tumour size

5. unifocal/multifocal?

6. what further tests are needed?

1. is there a lesion?

2. Where is the lesion?

3. what’s the risk of malignancy?

Lesions

changing appearances

Mammographic features viewed on DBT

Circums-

cribed

Mass

Spiculated

mass Distortion

Asymmetric

Density None (no

abnormality)

Ma

mm

og

rap

hic

featu

res v

iew

ed

o

n F

FD

M

Circumscribed

mass 0 11 0 6 11 28

Spiculated

mass 3 0 3 0 5 11

Distortion 1 5 0 4 13 23

Asymmetric

Density 19 10 0 0 57 86

None

(no abnormality) 13 11 12 2 0 38

Total 36 37 15 12 86 186

Iqbal A, Michell MJ et al RSNA 2011

fibroadenoma

adenolipoma

mucinous ca

Comparison of DBT with spot compression

magnification views

spot compression

Mag MLO

DBT MLO

Invasive lobular cancer

difficult diagnostic work up

Mag CC DBT CC

US = U1

MRI +ve

DBT guided NLB

Gd 2 IDC

Non specific asymmetric density

Final diagnosis – high grade DCIS

Irregular mass on US

CB = Gd 3 IDC Palpable mass LT breast

- recalled from screening

Are all tumours

visible on DBT

Microcalcification

King’s study:DBT vs spot compression mag views

• 355 lesions in 342 patients

• 104 malignant

82 benign

169 normal

7 radiologists

standard + DBT vs

standard + Mag

Clin Rad, accepted for publication 2014

Zuley M, et al Radiology:266 Jan 2013

DBT vs supplemental views for evaluation of non calcified lesions

217 lesions, 72 cancers

AUC DBT = 0.87 suppl views = 0.83 p<.001

4. Measurement of tumour size

Iqbal A, Michell MJ et al, RSNA 2013

Tumour size - Invasive Ductal Carcinoma

Invasive Ductal Carcinoma, n = 108

DBT 0.655

(0.551 to 0.759)

FFDM 0.591

(0.471to 0.711)

US 0.535

( 0.424 0.647)

Values are presented as CCC [rho_c, 95%

CI]; CCC = Concordance correlation

coefficient; Figures in parentheses are

95% confidence intervals.

Accuracy of tumour measurement

correln coeff

(a)

correln coeff

(b)

MRI 0.92 -

DBT 0.89 0.86

DM 0.83 0.71

US 0.77 0.85

(a)Luparia A, Mariscotti G et al Radiol Med. 2013

(b)Fornvik D, Zackrisson S et al Acta Radiol. 2010 51(3)

5. How many lesions?

Tomosynthesis in diagnostic work up

- advantages vs spot compression

• whole breast imaged

• easier positioning

• some cancers less visible on spot compression

• 3D information on lesion position

• DBT in two projections recommended

• USS still necessary for soft tissue lesions

Diagnostic work up/ assessment

- current practice

• DBT for mammographic work

up of soft tissue lesions

• fine focus mag views for

microcalcification

DBT in screening

• specificity

• sensitivity

• combination of views

• case selection

• cost

DBT in screening trials

- OSLO - ( P Skaane)

- STORM - (S Ciatto )

- TOMMY - ( F Gilbert )

- MALMO - ( I Anderson )

OSLO Trial

• 36,000 women 50 - 69 years

• Recruitment 11/10 - 12/12

• Reading arms

FFDM, FFDM + CAD, FFDM + DBT, synth 2D + DBT

reading time - FFDM 45 secs

- Combo 91 secs

OSLO Trial : Comparison of digital mammography alone and

DM + tomosynthesis in a population based screening program P Skaane et al, Radiology, 2013 Jan

12,631 participants

DM 77 cancers 6.1/1000

DM+DBT 112 cancers 8/1000

27% increase in ca detection p =0.001 (95% CI 1.06-1.53)

40% increase in invasive cancers p < 0.001

Recall rate (prior to arbitration)

DM 6.1%

DM +DBT 5.3%

15% decrease recall rate p <0.001

7292 participants mean age – 58 yrs

Reading - 2D, 2D + 3D

Cancer detection

2D 39 cancers 5.3/1000 (95% CI 3.8-7.3)

2D + 3D 59 cancers 8.1/1000 (95% CI 6.2-10.4)

p<0.0001

Recall

Overall 395 (5.5%)

Conditional recall 254 (3.5%)

thelancet.com/oncology april 25 2015

Grade DM DM+DBT Diff

In situ low/int 4 4 0

high 17 16 -1

Invasive 1 17 32 15

2 29 35 6

3 9 13 4

Oslo Trial – screen detected cancers

MALMO 2 Trial

1V DBT vs 2V 2D Siemens

Preliminary results - 7500 women

2V DM 1V DBT

Cancer detection

/1000 screened

6.3 8.5

Recall rate

% screened

2.2 3.3

Zacrisson et al, ECR 2014

DBT in screening - expectations

• more screen detected cancers

• lower false positive recall rate

• ? one reader

BUT

What about ‘overdiagnosis’

Prospective RCT of tomosynthesis in screening

Control group - standard 2V FFDM

Study group - tomo + synthetic 2D

Primary end point - interval cancer rates

Sample size

30% improvement in cancer detection = 2-3/1000

UK interval cancer rate = 2.85 / 1000 36 mths

?intervention might reduce interval ca rate to 1.85/1000

For 90% power need 50,000 per arm

Prospective RCT of tomosynthesis in screening

Secondary endpoints

• cancer detection

• prognostic features

• recall rate

• one vs two readers

• economic analysis

Acknowledgements

NHSBSP

King’s breast radiology team

Conclusion

• New imaging technology will

make screening, diagnosis and

treatment of breast disease more

effective and more efficient